linux/net/bluetooth/ecc.c
<<
>>
Prefs
   1/*
   2 * Copyright (c) 2013, Kenneth MacKay
   3 * All rights reserved.
   4 *
   5 * Redistribution and use in source and binary forms, with or without
   6 * modification, are permitted provided that the following conditions are
   7 * met:
   8 *  * Redistributions of source code must retain the above copyright
   9 *   notice, this list of conditions and the following disclaimer.
  10 *  * Redistributions in binary form must reproduce the above copyright
  11 *    notice, this list of conditions and the following disclaimer in the
  12 *    documentation and/or other materials provided with the distribution.
  13 *
  14 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  15 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  16 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
  17 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
  18 * HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
  19 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
  20 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
  21 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
  22 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
  23 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
  24 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  25 */
  26
  27#include <linux/random.h>
  28
  29#include "ecc.h"
  30
  31/* 256-bit curve */
  32#define ECC_BYTES 32
  33
  34#define MAX_TRIES 16
  35
  36/* Number of u64's needed */
  37#define NUM_ECC_DIGITS (ECC_BYTES / 8)
  38
  39struct ecc_point {
  40        u64 x[NUM_ECC_DIGITS];
  41        u64 y[NUM_ECC_DIGITS];
  42};
  43
  44typedef struct {
  45        u64 m_low;
  46        u64 m_high;
  47} uint128_t;
  48
  49#define CURVE_P_32 {    0xFFFFFFFFFFFFFFFFull, 0x00000000FFFFFFFFull, \
  50                        0x0000000000000000ull, 0xFFFFFFFF00000001ull }
  51
  52#define CURVE_G_32 { \
  53                {       0xF4A13945D898C296ull, 0x77037D812DEB33A0ull,   \
  54                        0xF8BCE6E563A440F2ull, 0x6B17D1F2E12C4247ull }, \
  55                {       0xCBB6406837BF51F5ull, 0x2BCE33576B315ECEull,   \
  56                        0x8EE7EB4A7C0F9E16ull, 0x4FE342E2FE1A7F9Bull }  \
  57}
  58
  59#define CURVE_N_32 {    0xF3B9CAC2FC632551ull, 0xBCE6FAADA7179E84ull,   \
  60                        0xFFFFFFFFFFFFFFFFull, 0xFFFFFFFF00000000ull }
  61
  62static u64 curve_p[NUM_ECC_DIGITS] = CURVE_P_32;
  63static struct ecc_point curve_g = CURVE_G_32;
  64static u64 curve_n[NUM_ECC_DIGITS] = CURVE_N_32;
  65
  66static void vli_clear(u64 *vli)
  67{
  68        int i;
  69
  70        for (i = 0; i < NUM_ECC_DIGITS; i++)
  71                vli[i] = 0;
  72}
  73
  74/* Returns true if vli == 0, false otherwise. */
  75static bool vli_is_zero(const u64 *vli)
  76{
  77        int i;
  78
  79        for (i = 0; i < NUM_ECC_DIGITS; i++) {
  80                if (vli[i])
  81                        return false;
  82        }
  83
  84        return true;
  85}
  86
  87/* Returns nonzero if bit bit of vli is set. */
  88static u64 vli_test_bit(const u64 *vli, unsigned int bit)
  89{
  90        return (vli[bit / 64] & ((u64) 1 << (bit % 64)));
  91}
  92
  93/* Counts the number of 64-bit "digits" in vli. */
  94static unsigned int vli_num_digits(const u64 *vli)
  95{
  96        int i;
  97
  98        /* Search from the end until we find a non-zero digit.
  99         * We do it in reverse because we expect that most digits will
 100         * be nonzero.
 101         */
 102        for (i = NUM_ECC_DIGITS - 1; i >= 0 && vli[i] == 0; i--);
 103
 104        return (i + 1);
 105}
 106
 107/* Counts the number of bits required for vli. */
 108static unsigned int vli_num_bits(const u64 *vli)
 109{
 110        unsigned int i, num_digits;
 111        u64 digit;
 112
 113        num_digits = vli_num_digits(vli);
 114        if (num_digits == 0)
 115                return 0;
 116
 117        digit = vli[num_digits - 1];
 118        for (i = 0; digit; i++)
 119                digit >>= 1;
 120
 121        return ((num_digits - 1) * 64 + i);
 122}
 123
 124/* Sets dest = src. */
 125static void vli_set(u64 *dest, const u64 *src)
 126{
 127        int i;
 128
 129        for (i = 0; i < NUM_ECC_DIGITS; i++)
 130                dest[i] = src[i];
 131}
 132
 133/* Returns sign of left - right. */
 134static int vli_cmp(const u64 *left, const u64 *right)
 135{
 136    int i;
 137
 138    for (i = NUM_ECC_DIGITS - 1; i >= 0; i--) {
 139            if (left[i] > right[i])
 140                    return 1;
 141            else if (left[i] < right[i])
 142                    return -1;
 143    }
 144
 145    return 0;
 146}
 147
 148/* Computes result = in << c, returning carry. Can modify in place
 149 * (if result == in). 0 < shift < 64.
 150 */
 151static u64 vli_lshift(u64 *result, const u64 *in,
 152                           unsigned int shift)
 153{
 154        u64 carry = 0;
 155        int i;
 156
 157        for (i = 0; i < NUM_ECC_DIGITS; i++) {
 158                u64 temp = in[i];
 159
 160                result[i] = (temp << shift) | carry;
 161                carry = temp >> (64 - shift);
 162        }
 163
 164        return carry;
 165}
 166
 167/* Computes vli = vli >> 1. */
 168static void vli_rshift1(u64 *vli)
 169{
 170        u64 *end = vli;
 171        u64 carry = 0;
 172
 173        vli += NUM_ECC_DIGITS;
 174
 175        while (vli-- > end) {
 176                u64 temp = *vli;
 177                *vli = (temp >> 1) | carry;
 178                carry = temp << 63;
 179        }
 180}
 181
 182/* Computes result = left + right, returning carry. Can modify in place. */
 183static u64 vli_add(u64 *result, const u64 *left,
 184                        const u64 *right)
 185{
 186        u64 carry = 0;
 187        int i;
 188
 189        for (i = 0; i < NUM_ECC_DIGITS; i++) {
 190                u64 sum;
 191
 192                sum = left[i] + right[i] + carry;
 193                if (sum != left[i])
 194                        carry = (sum < left[i]);
 195
 196                result[i] = sum;
 197        }
 198
 199        return carry;
 200}
 201
 202/* Computes result = left - right, returning borrow. Can modify in place. */
 203static u64 vli_sub(u64 *result, const u64 *left, const u64 *right)
 204{
 205        u64 borrow = 0;
 206        int i;
 207
 208        for (i = 0; i < NUM_ECC_DIGITS; i++) {
 209                u64 diff;
 210
 211                diff = left[i] - right[i] - borrow;
 212                if (diff != left[i])
 213                        borrow = (diff > left[i]);
 214
 215                result[i] = diff;
 216        }
 217
 218        return borrow;
 219}
 220
 221static uint128_t mul_64_64(u64 left, u64 right)
 222{
 223        u64 a0 = left & 0xffffffffull;
 224        u64 a1 = left >> 32;
 225        u64 b0 = right & 0xffffffffull;
 226        u64 b1 = right >> 32;
 227        u64 m0 = a0 * b0;
 228        u64 m1 = a0 * b1;
 229        u64 m2 = a1 * b0;
 230        u64 m3 = a1 * b1;
 231        uint128_t result;
 232
 233        m2 += (m0 >> 32);
 234        m2 += m1;
 235
 236        /* Overflow */
 237        if (m2 < m1)
 238                m3 += 0x100000000ull;
 239
 240        result.m_low = (m0 & 0xffffffffull) | (m2 << 32);
 241        result.m_high = m3 + (m2 >> 32);
 242
 243        return result;
 244}
 245
 246static uint128_t add_128_128(uint128_t a, uint128_t b)
 247{
 248        uint128_t result;
 249
 250        result.m_low = a.m_low + b.m_low;
 251        result.m_high = a.m_high + b.m_high + (result.m_low < a.m_low);
 252
 253        return result;
 254}
 255
 256static void vli_mult(u64 *result, const u64 *left, const u64 *right)
 257{
 258        uint128_t r01 = { 0, 0 };
 259        u64 r2 = 0;
 260        unsigned int i, k;
 261
 262        /* Compute each digit of result in sequence, maintaining the
 263         * carries.
 264         */
 265        for (k = 0; k < NUM_ECC_DIGITS * 2 - 1; k++) {
 266                unsigned int min;
 267
 268                if (k < NUM_ECC_DIGITS)
 269                        min = 0;
 270                else
 271                        min = (k + 1) - NUM_ECC_DIGITS;
 272
 273                for (i = min; i <= k && i < NUM_ECC_DIGITS; i++) {
 274                        uint128_t product;
 275
 276                        product = mul_64_64(left[i], right[k - i]);
 277
 278                        r01 = add_128_128(r01, product);
 279                        r2 += (r01.m_high < product.m_high);
 280                }
 281
 282                result[k] = r01.m_low;
 283                r01.m_low = r01.m_high;
 284                r01.m_high = r2;
 285                r2 = 0;
 286        }
 287
 288        result[NUM_ECC_DIGITS * 2 - 1] = r01.m_low;
 289}
 290
 291static void vli_square(u64 *result, const u64 *left)
 292{
 293        uint128_t r01 = { 0, 0 };
 294        u64 r2 = 0;
 295        int i, k;
 296
 297        for (k = 0; k < NUM_ECC_DIGITS * 2 - 1; k++) {
 298                unsigned int min;
 299
 300                if (k < NUM_ECC_DIGITS)
 301                        min = 0;
 302                else
 303                        min = (k + 1) - NUM_ECC_DIGITS;
 304
 305                for (i = min; i <= k && i <= k - i; i++) {
 306                        uint128_t product;
 307
 308                        product = mul_64_64(left[i], left[k - i]);
 309
 310                        if (i < k - i) {
 311                                r2 += product.m_high >> 63;
 312                                product.m_high = (product.m_high << 1) |
 313                                                 (product.m_low >> 63);
 314                                product.m_low <<= 1;
 315                        }
 316
 317                        r01 = add_128_128(r01, product);
 318                        r2 += (r01.m_high < product.m_high);
 319                }
 320
 321                result[k] = r01.m_low;
 322                r01.m_low = r01.m_high;
 323                r01.m_high = r2;
 324                r2 = 0;
 325        }
 326
 327        result[NUM_ECC_DIGITS * 2 - 1] = r01.m_low;
 328}
 329
 330/* Computes result = (left + right) % mod.
 331 * Assumes that left < mod and right < mod, result != mod.
 332 */
 333static void vli_mod_add(u64 *result, const u64 *left, const u64 *right,
 334                        const u64 *mod)
 335{
 336        u64 carry;
 337
 338        carry = vli_add(result, left, right);
 339
 340        /* result > mod (result = mod + remainder), so subtract mod to
 341         * get remainder.
 342         */
 343        if (carry || vli_cmp(result, mod) >= 0)
 344                vli_sub(result, result, mod);
 345}
 346
 347/* Computes result = (left - right) % mod.
 348 * Assumes that left < mod and right < mod, result != mod.
 349 */
 350static void vli_mod_sub(u64 *result, const u64 *left, const u64 *right,
 351                        const u64 *mod)
 352{
 353        u64 borrow = vli_sub(result, left, right);
 354
 355        /* In this case, p_result == -diff == (max int) - diff.
 356         * Since -x % d == d - x, we can get the correct result from
 357         * result + mod (with overflow).
 358         */
 359        if (borrow)
 360                vli_add(result, result, mod);
 361}
 362
 363/* Computes result = product % curve_p
 364   from http://www.nsa.gov/ia/_files/nist-routines.pdf */
 365static void vli_mmod_fast(u64 *result, const u64 *product)
 366{
 367        u64 tmp[NUM_ECC_DIGITS];
 368        int carry;
 369
 370        /* t */
 371        vli_set(result, product);
 372
 373        /* s1 */
 374        tmp[0] = 0;
 375        tmp[1] = product[5] & 0xffffffff00000000ull;
 376        tmp[2] = product[6];
 377        tmp[3] = product[7];
 378        carry = vli_lshift(tmp, tmp, 1);
 379        carry += vli_add(result, result, tmp);
 380
 381        /* s2 */
 382        tmp[1] = product[6] << 32;
 383        tmp[2] = (product[6] >> 32) | (product[7] << 32);
 384        tmp[3] = product[7] >> 32;
 385        carry += vli_lshift(tmp, tmp, 1);
 386        carry += vli_add(result, result, tmp);
 387
 388        /* s3 */
 389        tmp[0] = product[4];
 390        tmp[1] = product[5] & 0xffffffff;
 391        tmp[2] = 0;
 392        tmp[3] = product[7];
 393        carry += vli_add(result, result, tmp);
 394
 395        /* s4 */
 396        tmp[0] = (product[4] >> 32) | (product[5] << 32);
 397        tmp[1] = (product[5] >> 32) | (product[6] & 0xffffffff00000000ull);
 398        tmp[2] = product[7];
 399        tmp[3] = (product[6] >> 32) | (product[4] << 32);
 400        carry += vli_add(result, result, tmp);
 401
 402        /* d1 */
 403        tmp[0] = (product[5] >> 32) | (product[6] << 32);
 404        tmp[1] = (product[6] >> 32);
 405        tmp[2] = 0;
 406        tmp[3] = (product[4] & 0xffffffff) | (product[5] << 32);
 407        carry -= vli_sub(result, result, tmp);
 408
 409        /* d2 */
 410        tmp[0] = product[6];
 411        tmp[1] = product[7];
 412        tmp[2] = 0;
 413        tmp[3] = (product[4] >> 32) | (product[5] & 0xffffffff00000000ull);
 414        carry -= vli_sub(result, result, tmp);
 415
 416        /* d3 */
 417        tmp[0] = (product[6] >> 32) | (product[7] << 32);
 418        tmp[1] = (product[7] >> 32) | (product[4] << 32);
 419        tmp[2] = (product[4] >> 32) | (product[5] << 32);
 420        tmp[3] = (product[6] << 32);
 421        carry -= vli_sub(result, result, tmp);
 422
 423        /* d4 */
 424        tmp[0] = product[7];
 425        tmp[1] = product[4] & 0xffffffff00000000ull;
 426        tmp[2] = product[5];
 427        tmp[3] = product[6] & 0xffffffff00000000ull;
 428        carry -= vli_sub(result, result, tmp);
 429
 430        if (carry < 0) {
 431                do {
 432                        carry += vli_add(result, result, curve_p);
 433                } while (carry < 0);
 434        } else {
 435                while (carry || vli_cmp(curve_p, result) != 1)
 436                        carry -= vli_sub(result, result, curve_p);
 437        }
 438}
 439
 440/* Computes result = (left * right) % curve_p. */
 441static void vli_mod_mult_fast(u64 *result, const u64 *left, const u64 *right)
 442{
 443        u64 product[2 * NUM_ECC_DIGITS];
 444
 445        vli_mult(product, left, right);
 446        vli_mmod_fast(result, product);
 447}
 448
 449/* Computes result = left^2 % curve_p. */
 450static void vli_mod_square_fast(u64 *result, const u64 *left)
 451{
 452        u64 product[2 * NUM_ECC_DIGITS];
 453
 454        vli_square(product, left);
 455        vli_mmod_fast(result, product);
 456}
 457
 458#define EVEN(vli) (!(vli[0] & 1))
 459/* Computes result = (1 / p_input) % mod. All VLIs are the same size.
 460 * See "From Euclid's GCD to Montgomery Multiplication to the Great Divide"
 461 * https://labs.oracle.com/techrep/2001/smli_tr-2001-95.pdf
 462 */
 463static void vli_mod_inv(u64 *result, const u64 *input, const u64 *mod)
 464{
 465        u64 a[NUM_ECC_DIGITS], b[NUM_ECC_DIGITS];
 466        u64 u[NUM_ECC_DIGITS], v[NUM_ECC_DIGITS];
 467        u64 carry;
 468        int cmp_result;
 469
 470        if (vli_is_zero(input)) {
 471                vli_clear(result);
 472                return;
 473        }
 474
 475        vli_set(a, input);
 476        vli_set(b, mod);
 477        vli_clear(u);
 478        u[0] = 1;
 479        vli_clear(v);
 480
 481        while ((cmp_result = vli_cmp(a, b)) != 0) {
 482                carry = 0;
 483
 484                if (EVEN(a)) {
 485                        vli_rshift1(a);
 486
 487                        if (!EVEN(u))
 488                                carry = vli_add(u, u, mod);
 489
 490                        vli_rshift1(u);
 491                        if (carry)
 492                                u[NUM_ECC_DIGITS - 1] |= 0x8000000000000000ull;
 493                } else if (EVEN(b)) {
 494                        vli_rshift1(b);
 495
 496                        if (!EVEN(v))
 497                                carry = vli_add(v, v, mod);
 498
 499                        vli_rshift1(v);
 500                        if (carry)
 501                                v[NUM_ECC_DIGITS - 1] |= 0x8000000000000000ull;
 502                } else if (cmp_result > 0) {
 503                        vli_sub(a, a, b);
 504                        vli_rshift1(a);
 505
 506                        if (vli_cmp(u, v) < 0)
 507                                vli_add(u, u, mod);
 508
 509                        vli_sub(u, u, v);
 510                        if (!EVEN(u))
 511                                carry = vli_add(u, u, mod);
 512
 513                        vli_rshift1(u);
 514                        if (carry)
 515                                u[NUM_ECC_DIGITS - 1] |= 0x8000000000000000ull;
 516                } else {
 517                        vli_sub(b, b, a);
 518                        vli_rshift1(b);
 519
 520                        if (vli_cmp(v, u) < 0)
 521                                vli_add(v, v, mod);
 522
 523                        vli_sub(v, v, u);
 524                        if (!EVEN(v))
 525                                carry = vli_add(v, v, mod);
 526
 527                        vli_rshift1(v);
 528                        if (carry)
 529                                v[NUM_ECC_DIGITS - 1] |= 0x8000000000000000ull;
 530                }
 531        }
 532
 533        vli_set(result, u);
 534}
 535
 536/* ------ Point operations ------ */
 537
 538/* Returns true if p_point is the point at infinity, false otherwise. */
 539static bool ecc_point_is_zero(const struct ecc_point *point)
 540{
 541        return (vli_is_zero(point->x) && vli_is_zero(point->y));
 542}
 543
 544/* Point multiplication algorithm using Montgomery's ladder with co-Z
 545 * coordinates. From http://eprint.iacr.org/2011/338.pdf
 546 */
 547
 548/* Double in place */
 549static void ecc_point_double_jacobian(u64 *x1, u64 *y1, u64 *z1)
 550{
 551        /* t1 = x, t2 = y, t3 = z */
 552        u64 t4[NUM_ECC_DIGITS];
 553        u64 t5[NUM_ECC_DIGITS];
 554
 555        if (vli_is_zero(z1))
 556                return;
 557
 558        vli_mod_square_fast(t4, y1);   /* t4 = y1^2 */
 559        vli_mod_mult_fast(t5, x1, t4); /* t5 = x1*y1^2 = A */
 560        vli_mod_square_fast(t4, t4);   /* t4 = y1^4 */
 561        vli_mod_mult_fast(y1, y1, z1); /* t2 = y1*z1 = z3 */
 562        vli_mod_square_fast(z1, z1);   /* t3 = z1^2 */
 563
 564        vli_mod_add(x1, x1, z1, curve_p); /* t1 = x1 + z1^2 */
 565        vli_mod_add(z1, z1, z1, curve_p); /* t3 = 2*z1^2 */
 566        vli_mod_sub(z1, x1, z1, curve_p); /* t3 = x1 - z1^2 */
 567        vli_mod_mult_fast(x1, x1, z1);    /* t1 = x1^2 - z1^4 */
 568
 569        vli_mod_add(z1, x1, x1, curve_p); /* t3 = 2*(x1^2 - z1^4) */
 570        vli_mod_add(x1, x1, z1, curve_p); /* t1 = 3*(x1^2 - z1^4) */
 571        if (vli_test_bit(x1, 0)) {
 572                u64 carry = vli_add(x1, x1, curve_p);
 573                vli_rshift1(x1);
 574                x1[NUM_ECC_DIGITS - 1] |= carry << 63;
 575        } else {
 576                vli_rshift1(x1);
 577        }
 578        /* t1 = 3/2*(x1^2 - z1^4) = B */
 579
 580        vli_mod_square_fast(z1, x1);      /* t3 = B^2 */
 581        vli_mod_sub(z1, z1, t5, curve_p); /* t3 = B^2 - A */
 582        vli_mod_sub(z1, z1, t5, curve_p); /* t3 = B^2 - 2A = x3 */
 583        vli_mod_sub(t5, t5, z1, curve_p); /* t5 = A - x3 */
 584        vli_mod_mult_fast(x1, x1, t5);    /* t1 = B * (A - x3) */
 585        vli_mod_sub(t4, x1, t4, curve_p); /* t4 = B * (A - x3) - y1^4 = y3 */
 586
 587        vli_set(x1, z1);
 588        vli_set(z1, y1);
 589        vli_set(y1, t4);
 590}
 591
 592/* Modify (x1, y1) => (x1 * z^2, y1 * z^3) */
 593static void apply_z(u64 *x1, u64 *y1, u64 *z)
 594{
 595        u64 t1[NUM_ECC_DIGITS];
 596
 597        vli_mod_square_fast(t1, z);    /* z^2 */
 598        vli_mod_mult_fast(x1, x1, t1); /* x1 * z^2 */
 599        vli_mod_mult_fast(t1, t1, z);  /* z^3 */
 600        vli_mod_mult_fast(y1, y1, t1); /* y1 * z^3 */
 601}
 602
 603/* P = (x1, y1) => 2P, (x2, y2) => P' */
 604static void xycz_initial_double(u64 *x1, u64 *y1, u64 *x2, u64 *y2,
 605                                u64 *p_initial_z)
 606{
 607        u64 z[NUM_ECC_DIGITS];
 608
 609        vli_set(x2, x1);
 610        vli_set(y2, y1);
 611
 612        vli_clear(z);
 613        z[0] = 1;
 614
 615        if (p_initial_z)
 616                vli_set(z, p_initial_z);
 617
 618        apply_z(x1, y1, z);
 619
 620        ecc_point_double_jacobian(x1, y1, z);
 621
 622        apply_z(x2, y2, z);
 623}
 624
 625/* Input P = (x1, y1, Z), Q = (x2, y2, Z)
 626 * Output P' = (x1', y1', Z3), P + Q = (x3, y3, Z3)
 627 * or P => P', Q => P + Q
 628 */
 629static void xycz_add(u64 *x1, u64 *y1, u64 *x2, u64 *y2)
 630{
 631        /* t1 = X1, t2 = Y1, t3 = X2, t4 = Y2 */
 632        u64 t5[NUM_ECC_DIGITS];
 633
 634        vli_mod_sub(t5, x2, x1, curve_p); /* t5 = x2 - x1 */
 635        vli_mod_square_fast(t5, t5);      /* t5 = (x2 - x1)^2 = A */
 636        vli_mod_mult_fast(x1, x1, t5);    /* t1 = x1*A = B */
 637        vli_mod_mult_fast(x2, x2, t5);    /* t3 = x2*A = C */
 638        vli_mod_sub(y2, y2, y1, curve_p); /* t4 = y2 - y1 */
 639        vli_mod_square_fast(t5, y2);      /* t5 = (y2 - y1)^2 = D */
 640
 641        vli_mod_sub(t5, t5, x1, curve_p); /* t5 = D - B */
 642        vli_mod_sub(t5, t5, x2, curve_p); /* t5 = D - B - C = x3 */
 643        vli_mod_sub(x2, x2, x1, curve_p); /* t3 = C - B */
 644        vli_mod_mult_fast(y1, y1, x2);    /* t2 = y1*(C - B) */
 645        vli_mod_sub(x2, x1, t5, curve_p); /* t3 = B - x3 */
 646        vli_mod_mult_fast(y2, y2, x2);    /* t4 = (y2 - y1)*(B - x3) */
 647        vli_mod_sub(y2, y2, y1, curve_p); /* t4 = y3 */
 648
 649        vli_set(x2, t5);
 650}
 651
 652/* Input P = (x1, y1, Z), Q = (x2, y2, Z)
 653 * Output P + Q = (x3, y3, Z3), P - Q = (x3', y3', Z3)
 654 * or P => P - Q, Q => P + Q
 655 */
 656static void xycz_add_c(u64 *x1, u64 *y1, u64 *x2, u64 *y2)
 657{
 658        /* t1 = X1, t2 = Y1, t3 = X2, t4 = Y2 */
 659        u64 t5[NUM_ECC_DIGITS];
 660        u64 t6[NUM_ECC_DIGITS];
 661        u64 t7[NUM_ECC_DIGITS];
 662
 663        vli_mod_sub(t5, x2, x1, curve_p); /* t5 = x2 - x1 */
 664        vli_mod_square_fast(t5, t5);      /* t5 = (x2 - x1)^2 = A */
 665        vli_mod_mult_fast(x1, x1, t5);    /* t1 = x1*A = B */
 666        vli_mod_mult_fast(x2, x2, t5);    /* t3 = x2*A = C */
 667        vli_mod_add(t5, y2, y1, curve_p); /* t4 = y2 + y1 */
 668        vli_mod_sub(y2, y2, y1, curve_p); /* t4 = y2 - y1 */
 669
 670        vli_mod_sub(t6, x2, x1, curve_p); /* t6 = C - B */
 671        vli_mod_mult_fast(y1, y1, t6);    /* t2 = y1 * (C - B) */
 672        vli_mod_add(t6, x1, x2, curve_p); /* t6 = B + C */
 673        vli_mod_square_fast(x2, y2);      /* t3 = (y2 - y1)^2 */
 674        vli_mod_sub(x2, x2, t6, curve_p); /* t3 = x3 */
 675
 676        vli_mod_sub(t7, x1, x2, curve_p); /* t7 = B - x3 */
 677        vli_mod_mult_fast(y2, y2, t7);    /* t4 = (y2 - y1)*(B - x3) */
 678        vli_mod_sub(y2, y2, y1, curve_p); /* t4 = y3 */
 679
 680        vli_mod_square_fast(t7, t5);      /* t7 = (y2 + y1)^2 = F */
 681        vli_mod_sub(t7, t7, t6, curve_p); /* t7 = x3' */
 682        vli_mod_sub(t6, t7, x1, curve_p); /* t6 = x3' - B */
 683        vli_mod_mult_fast(t6, t6, t5);    /* t6 = (y2 + y1)*(x3' - B) */
 684        vli_mod_sub(y1, t6, y1, curve_p); /* t2 = y3' */
 685
 686        vli_set(x1, t7);
 687}
 688
 689static void ecc_point_mult(struct ecc_point *result,
 690                           const struct ecc_point *point, u64 *scalar,
 691                           u64 *initial_z, int num_bits)
 692{
 693        /* R0 and R1 */
 694        u64 rx[2][NUM_ECC_DIGITS];
 695        u64 ry[2][NUM_ECC_DIGITS];
 696        u64 z[NUM_ECC_DIGITS];
 697        int i, nb;
 698
 699        vli_set(rx[1], point->x);
 700        vli_set(ry[1], point->y);
 701
 702        xycz_initial_double(rx[1], ry[1], rx[0], ry[0], initial_z);
 703
 704        for (i = num_bits - 2; i > 0; i--) {
 705                nb = !vli_test_bit(scalar, i);
 706                xycz_add_c(rx[1 - nb], ry[1 - nb], rx[nb], ry[nb]);
 707                xycz_add(rx[nb], ry[nb], rx[1 - nb], ry[1 - nb]);
 708        }
 709
 710        nb = !vli_test_bit(scalar, 0);
 711        xycz_add_c(rx[1 - nb], ry[1 - nb], rx[nb], ry[nb]);
 712
 713        /* Find final 1/Z value. */
 714        vli_mod_sub(z, rx[1], rx[0], curve_p); /* X1 - X0 */
 715        vli_mod_mult_fast(z, z, ry[1 - nb]); /* Yb * (X1 - X0) */
 716        vli_mod_mult_fast(z, z, point->x);   /* xP * Yb * (X1 - X0) */
 717        vli_mod_inv(z, z, curve_p);          /* 1 / (xP * Yb * (X1 - X0)) */
 718        vli_mod_mult_fast(z, z, point->y);   /* yP / (xP * Yb * (X1 - X0)) */
 719        vli_mod_mult_fast(z, z, rx[1 - nb]); /* Xb * yP / (xP * Yb * (X1 - X0)) */
 720        /* End 1/Z calculation */
 721
 722        xycz_add(rx[nb], ry[nb], rx[1 - nb], ry[1 - nb]);
 723
 724        apply_z(rx[0], ry[0], z);
 725
 726        vli_set(result->x, rx[0]);
 727        vli_set(result->y, ry[0]);
 728}
 729
 730static void ecc_bytes2native(const u8 bytes[ECC_BYTES],
 731                             u64 native[NUM_ECC_DIGITS])
 732{
 733        int i;
 734
 735        for (i = 0; i < NUM_ECC_DIGITS; i++) {
 736                const u8 *digit = bytes + 8 * (NUM_ECC_DIGITS - 1 - i);
 737
 738                native[NUM_ECC_DIGITS - 1 - i] =
 739                                ((u64) digit[0] << 0) |
 740                                ((u64) digit[1] << 8) |
 741                                ((u64) digit[2] << 16) |
 742                                ((u64) digit[3] << 24) |
 743                                ((u64) digit[4] << 32) |
 744                                ((u64) digit[5] << 40) |
 745                                ((u64) digit[6] << 48) |
 746                                ((u64) digit[7] << 56);
 747        }
 748}
 749
 750static void ecc_native2bytes(const u64 native[NUM_ECC_DIGITS],
 751                             u8 bytes[ECC_BYTES])
 752{
 753        int i;
 754
 755        for (i = 0; i < NUM_ECC_DIGITS; i++) {
 756                u8 *digit = bytes + 8 * (NUM_ECC_DIGITS - 1 - i);
 757
 758                digit[0] = native[NUM_ECC_DIGITS - 1 - i] >> 0;
 759                digit[1] = native[NUM_ECC_DIGITS - 1 - i] >> 8;
 760                digit[2] = native[NUM_ECC_DIGITS - 1 - i] >> 16;
 761                digit[3] = native[NUM_ECC_DIGITS - 1 - i] >> 24;
 762                digit[4] = native[NUM_ECC_DIGITS - 1 - i] >> 32;
 763                digit[5] = native[NUM_ECC_DIGITS - 1 - i] >> 40;
 764                digit[6] = native[NUM_ECC_DIGITS - 1 - i] >> 48;
 765                digit[7] = native[NUM_ECC_DIGITS - 1 - i] >> 56;
 766        }
 767}
 768
 769bool ecc_make_key(u8 public_key[64], u8 private_key[32])
 770{
 771        struct ecc_point pk;
 772        u64 priv[NUM_ECC_DIGITS];
 773        unsigned int tries = 0;
 774
 775        do {
 776                if (tries++ >= MAX_TRIES)
 777                        return false;
 778
 779                get_random_bytes(priv, ECC_BYTES);
 780
 781                if (vli_is_zero(priv))
 782                        continue;
 783
 784                /* Make sure the private key is in the range [1, n-1]. */
 785                if (vli_cmp(curve_n, priv) != 1)
 786                        continue;
 787
 788                ecc_point_mult(&pk, &curve_g, priv, NULL, vli_num_bits(priv));
 789        } while (ecc_point_is_zero(&pk));
 790
 791        ecc_native2bytes(priv, private_key);
 792        ecc_native2bytes(pk.x, public_key);
 793        ecc_native2bytes(pk.y, &public_key[32]);
 794
 795        return true;
 796}
 797
 798bool ecdh_shared_secret(const u8 public_key[64], const u8 private_key[32],
 799                        u8 secret[32])
 800{
 801        u64 priv[NUM_ECC_DIGITS];
 802        u64 rand[NUM_ECC_DIGITS];
 803        struct ecc_point product, pk;
 804
 805        get_random_bytes(rand, ECC_BYTES);
 806
 807        ecc_bytes2native(public_key, pk.x);
 808        ecc_bytes2native(&public_key[32], pk.y);
 809        ecc_bytes2native(private_key, priv);
 810
 811        ecc_point_mult(&product, &pk, priv, rand, vli_num_bits(priv));
 812
 813        ecc_native2bytes(product.x, secret);
 814
 815        return !ecc_point_is_zero(&product);
 816}
 817