]>
Commit | Line | Data |
---|---|---|
3c4b2390 SB |
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 | ||
4a2289da | 27 | #include <linux/module.h> |
3c4b2390 SB |
28 | #include <linux/random.h> |
29 | #include <linux/slab.h> | |
30 | #include <linux/swab.h> | |
31 | #include <linux/fips.h> | |
32 | #include <crypto/ecdh.h> | |
6755fd26 | 33 | #include <crypto/rng.h> |
3c4b2390 SB |
34 | |
35 | #include "ecc.h" | |
36 | #include "ecc_curve_defs.h" | |
37 | ||
38 | typedef struct { | |
39 | u64 m_low; | |
40 | u64 m_high; | |
41 | } uint128_t; | |
42 | ||
43 | static inline const struct ecc_curve *ecc_get_curve(unsigned int curve_id) | |
44 | { | |
45 | switch (curve_id) { | |
46 | /* In FIPS mode only allow P256 and higher */ | |
47 | case ECC_CURVE_NIST_P192: | |
48 | return fips_enabled ? NULL : &nist_p192; | |
49 | case ECC_CURVE_NIST_P256: | |
50 | return &nist_p256; | |
51 | default: | |
52 | return NULL; | |
53 | } | |
54 | } | |
55 | ||
56 | static u64 *ecc_alloc_digits_space(unsigned int ndigits) | |
57 | { | |
58 | size_t len = ndigits * sizeof(u64); | |
59 | ||
60 | if (!len) | |
61 | return NULL; | |
62 | ||
63 | return kmalloc(len, GFP_KERNEL); | |
64 | } | |
65 | ||
66 | static void ecc_free_digits_space(u64 *space) | |
67 | { | |
68 | kzfree(space); | |
69 | } | |
70 | ||
71 | static struct ecc_point *ecc_alloc_point(unsigned int ndigits) | |
72 | { | |
73 | struct ecc_point *p = kmalloc(sizeof(*p), GFP_KERNEL); | |
74 | ||
75 | if (!p) | |
76 | return NULL; | |
77 | ||
78 | p->x = ecc_alloc_digits_space(ndigits); | |
79 | if (!p->x) | |
80 | goto err_alloc_x; | |
81 | ||
82 | p->y = ecc_alloc_digits_space(ndigits); | |
83 | if (!p->y) | |
84 | goto err_alloc_y; | |
85 | ||
86 | p->ndigits = ndigits; | |
87 | ||
88 | return p; | |
89 | ||
90 | err_alloc_y: | |
91 | ecc_free_digits_space(p->x); | |
92 | err_alloc_x: | |
93 | kfree(p); | |
94 | return NULL; | |
95 | } | |
96 | ||
97 | static void ecc_free_point(struct ecc_point *p) | |
98 | { | |
99 | if (!p) | |
100 | return; | |
101 | ||
102 | kzfree(p->x); | |
103 | kzfree(p->y); | |
104 | kzfree(p); | |
105 | } | |
106 | ||
107 | static void vli_clear(u64 *vli, unsigned int ndigits) | |
108 | { | |
109 | int i; | |
110 | ||
111 | for (i = 0; i < ndigits; i++) | |
112 | vli[i] = 0; | |
113 | } | |
114 | ||
115 | /* Returns true if vli == 0, false otherwise. */ | |
4a2289da | 116 | bool vli_is_zero(const u64 *vli, unsigned int ndigits) |
3c4b2390 SB |
117 | { |
118 | int i; | |
119 | ||
120 | for (i = 0; i < ndigits; i++) { | |
121 | if (vli[i]) | |
122 | return false; | |
123 | } | |
124 | ||
125 | return true; | |
126 | } | |
4a2289da | 127 | EXPORT_SYMBOL(vli_is_zero); |
3c4b2390 SB |
128 | |
129 | /* Returns nonzero if bit bit of vli is set. */ | |
130 | static u64 vli_test_bit(const u64 *vli, unsigned int bit) | |
131 | { | |
132 | return (vli[bit / 64] & ((u64)1 << (bit % 64))); | |
133 | } | |
134 | ||
135 | /* Counts the number of 64-bit "digits" in vli. */ | |
136 | static unsigned int vli_num_digits(const u64 *vli, unsigned int ndigits) | |
137 | { | |
138 | int i; | |
139 | ||
140 | /* Search from the end until we find a non-zero digit. | |
141 | * We do it in reverse because we expect that most digits will | |
142 | * be nonzero. | |
143 | */ | |
144 | for (i = ndigits - 1; i >= 0 && vli[i] == 0; i--); | |
145 | ||
146 | return (i + 1); | |
147 | } | |
148 | ||
149 | /* Counts the number of bits required for vli. */ | |
150 | static unsigned int vli_num_bits(const u64 *vli, unsigned int ndigits) | |
151 | { | |
152 | unsigned int i, num_digits; | |
153 | u64 digit; | |
154 | ||
155 | num_digits = vli_num_digits(vli, ndigits); | |
156 | if (num_digits == 0) | |
157 | return 0; | |
158 | ||
159 | digit = vli[num_digits - 1]; | |
160 | for (i = 0; digit; i++) | |
161 | digit >>= 1; | |
162 | ||
163 | return ((num_digits - 1) * 64 + i); | |
164 | } | |
165 | ||
166 | /* Sets dest = src. */ | |
167 | static void vli_set(u64 *dest, const u64 *src, unsigned int ndigits) | |
168 | { | |
169 | int i; | |
170 | ||
171 | for (i = 0; i < ndigits; i++) | |
172 | dest[i] = src[i]; | |
173 | } | |
174 | ||
175 | /* Returns sign of left - right. */ | |
4a2289da | 176 | int vli_cmp(const u64 *left, const u64 *right, unsigned int ndigits) |
3c4b2390 SB |
177 | { |
178 | int i; | |
179 | ||
180 | for (i = ndigits - 1; i >= 0; i--) { | |
181 | if (left[i] > right[i]) | |
182 | return 1; | |
183 | else if (left[i] < right[i]) | |
184 | return -1; | |
185 | } | |
186 | ||
187 | return 0; | |
188 | } | |
4a2289da | 189 | EXPORT_SYMBOL(vli_cmp); |
3c4b2390 SB |
190 | |
191 | /* Computes result = in << c, returning carry. Can modify in place | |
192 | * (if result == in). 0 < shift < 64. | |
193 | */ | |
194 | static u64 vli_lshift(u64 *result, const u64 *in, unsigned int shift, | |
195 | unsigned int ndigits) | |
196 | { | |
197 | u64 carry = 0; | |
198 | int i; | |
199 | ||
200 | for (i = 0; i < ndigits; i++) { | |
201 | u64 temp = in[i]; | |
202 | ||
203 | result[i] = (temp << shift) | carry; | |
204 | carry = temp >> (64 - shift); | |
205 | } | |
206 | ||
207 | return carry; | |
208 | } | |
209 | ||
210 | /* Computes vli = vli >> 1. */ | |
211 | static void vli_rshift1(u64 *vli, unsigned int ndigits) | |
212 | { | |
213 | u64 *end = vli; | |
214 | u64 carry = 0; | |
215 | ||
216 | vli += ndigits; | |
217 | ||
218 | while (vli-- > end) { | |
219 | u64 temp = *vli; | |
220 | *vli = (temp >> 1) | carry; | |
221 | carry = temp << 63; | |
222 | } | |
223 | } | |
224 | ||
225 | /* Computes result = left + right, returning carry. Can modify in place. */ | |
226 | static u64 vli_add(u64 *result, const u64 *left, const u64 *right, | |
227 | unsigned int ndigits) | |
228 | { | |
229 | u64 carry = 0; | |
230 | int i; | |
231 | ||
232 | for (i = 0; i < ndigits; i++) { | |
233 | u64 sum; | |
234 | ||
235 | sum = left[i] + right[i] + carry; | |
236 | if (sum != left[i]) | |
237 | carry = (sum < left[i]); | |
238 | ||
239 | result[i] = sum; | |
240 | } | |
241 | ||
242 | return carry; | |
243 | } | |
244 | ||
245 | /* Computes result = left - right, returning borrow. Can modify in place. */ | |
4a2289da | 246 | u64 vli_sub(u64 *result, const u64 *left, const u64 *right, |
3c4b2390 SB |
247 | unsigned int ndigits) |
248 | { | |
249 | u64 borrow = 0; | |
250 | int i; | |
251 | ||
252 | for (i = 0; i < ndigits; i++) { | |
253 | u64 diff; | |
254 | ||
255 | diff = left[i] - right[i] - borrow; | |
256 | if (diff != left[i]) | |
257 | borrow = (diff > left[i]); | |
258 | ||
259 | result[i] = diff; | |
260 | } | |
261 | ||
262 | return borrow; | |
263 | } | |
4a2289da | 264 | EXPORT_SYMBOL(vli_sub); |
3c4b2390 SB |
265 | |
266 | static uint128_t mul_64_64(u64 left, u64 right) | |
267 | { | |
268 | u64 a0 = left & 0xffffffffull; | |
269 | u64 a1 = left >> 32; | |
270 | u64 b0 = right & 0xffffffffull; | |
271 | u64 b1 = right >> 32; | |
272 | u64 m0 = a0 * b0; | |
273 | u64 m1 = a0 * b1; | |
274 | u64 m2 = a1 * b0; | |
275 | u64 m3 = a1 * b1; | |
276 | uint128_t result; | |
277 | ||
278 | m2 += (m0 >> 32); | |
279 | m2 += m1; | |
280 | ||
281 | /* Overflow */ | |
282 | if (m2 < m1) | |
283 | m3 += 0x100000000ull; | |
284 | ||
285 | result.m_low = (m0 & 0xffffffffull) | (m2 << 32); | |
286 | result.m_high = m3 + (m2 >> 32); | |
287 | ||
288 | return result; | |
289 | } | |
290 | ||
291 | static uint128_t add_128_128(uint128_t a, uint128_t b) | |
292 | { | |
293 | uint128_t result; | |
294 | ||
295 | result.m_low = a.m_low + b.m_low; | |
296 | result.m_high = a.m_high + b.m_high + (result.m_low < a.m_low); | |
297 | ||
298 | return result; | |
299 | } | |
300 | ||
301 | static void vli_mult(u64 *result, const u64 *left, const u64 *right, | |
302 | unsigned int ndigits) | |
303 | { | |
304 | uint128_t r01 = { 0, 0 }; | |
305 | u64 r2 = 0; | |
306 | unsigned int i, k; | |
307 | ||
308 | /* Compute each digit of result in sequence, maintaining the | |
309 | * carries. | |
310 | */ | |
311 | for (k = 0; k < ndigits * 2 - 1; k++) { | |
312 | unsigned int min; | |
313 | ||
314 | if (k < ndigits) | |
315 | min = 0; | |
316 | else | |
317 | min = (k + 1) - ndigits; | |
318 | ||
319 | for (i = min; i <= k && i < ndigits; i++) { | |
320 | uint128_t product; | |
321 | ||
322 | product = mul_64_64(left[i], right[k - i]); | |
323 | ||
324 | r01 = add_128_128(r01, product); | |
325 | r2 += (r01.m_high < product.m_high); | |
326 | } | |
327 | ||
328 | result[k] = r01.m_low; | |
329 | r01.m_low = r01.m_high; | |
330 | r01.m_high = r2; | |
331 | r2 = 0; | |
332 | } | |
333 | ||
334 | result[ndigits * 2 - 1] = r01.m_low; | |
335 | } | |
336 | ||
337 | static void vli_square(u64 *result, const u64 *left, unsigned int ndigits) | |
338 | { | |
339 | uint128_t r01 = { 0, 0 }; | |
340 | u64 r2 = 0; | |
341 | int i, k; | |
342 | ||
343 | for (k = 0; k < ndigits * 2 - 1; k++) { | |
344 | unsigned int min; | |
345 | ||
346 | if (k < ndigits) | |
347 | min = 0; | |
348 | else | |
349 | min = (k + 1) - ndigits; | |
350 | ||
351 | for (i = min; i <= k && i <= k - i; i++) { | |
352 | uint128_t product; | |
353 | ||
354 | product = mul_64_64(left[i], left[k - i]); | |
355 | ||
356 | if (i < k - i) { | |
357 | r2 += product.m_high >> 63; | |
358 | product.m_high = (product.m_high << 1) | | |
359 | (product.m_low >> 63); | |
360 | product.m_low <<= 1; | |
361 | } | |
362 | ||
363 | r01 = add_128_128(r01, product); | |
364 | r2 += (r01.m_high < product.m_high); | |
365 | } | |
366 | ||
367 | result[k] = r01.m_low; | |
368 | r01.m_low = r01.m_high; | |
369 | r01.m_high = r2; | |
370 | r2 = 0; | |
371 | } | |
372 | ||
373 | result[ndigits * 2 - 1] = r01.m_low; | |
374 | } | |
375 | ||
376 | /* Computes result = (left + right) % mod. | |
377 | * Assumes that left < mod and right < mod, result != mod. | |
378 | */ | |
379 | static void vli_mod_add(u64 *result, const u64 *left, const u64 *right, | |
380 | const u64 *mod, unsigned int ndigits) | |
381 | { | |
382 | u64 carry; | |
383 | ||
384 | carry = vli_add(result, left, right, ndigits); | |
385 | ||
386 | /* result > mod (result = mod + remainder), so subtract mod to | |
387 | * get remainder. | |
388 | */ | |
389 | if (carry || vli_cmp(result, mod, ndigits) >= 0) | |
390 | vli_sub(result, result, mod, ndigits); | |
391 | } | |
392 | ||
393 | /* Computes result = (left - right) % mod. | |
394 | * Assumes that left < mod and right < mod, result != mod. | |
395 | */ | |
396 | static void vli_mod_sub(u64 *result, const u64 *left, const u64 *right, | |
397 | const u64 *mod, unsigned int ndigits) | |
398 | { | |
399 | u64 borrow = vli_sub(result, left, right, ndigits); | |
400 | ||
401 | /* In this case, p_result == -diff == (max int) - diff. | |
402 | * Since -x % d == d - x, we can get the correct result from | |
403 | * result + mod (with overflow). | |
404 | */ | |
405 | if (borrow) | |
406 | vli_add(result, result, mod, ndigits); | |
407 | } | |
408 | ||
409 | /* Computes p_result = p_product % curve_p. | |
410 | * See algorithm 5 and 6 from | |
411 | * http://www.isys.uni-klu.ac.at/PDF/2001-0126-MT.pdf | |
412 | */ | |
413 | static void vli_mmod_fast_192(u64 *result, const u64 *product, | |
414 | const u64 *curve_prime, u64 *tmp) | |
415 | { | |
416 | const unsigned int ndigits = 3; | |
417 | int carry; | |
418 | ||
419 | vli_set(result, product, ndigits); | |
420 | ||
421 | vli_set(tmp, &product[3], ndigits); | |
422 | carry = vli_add(result, result, tmp, ndigits); | |
423 | ||
424 | tmp[0] = 0; | |
425 | tmp[1] = product[3]; | |
426 | tmp[2] = product[4]; | |
427 | carry += vli_add(result, result, tmp, ndigits); | |
428 | ||
429 | tmp[0] = tmp[1] = product[5]; | |
430 | tmp[2] = 0; | |
431 | carry += vli_add(result, result, tmp, ndigits); | |
432 | ||
433 | while (carry || vli_cmp(curve_prime, result, ndigits) != 1) | |
434 | carry -= vli_sub(result, result, curve_prime, ndigits); | |
435 | } | |
436 | ||
437 | /* Computes result = product % curve_prime | |
438 | * from http://www.nsa.gov/ia/_files/nist-routines.pdf | |
439 | */ | |
440 | static void vli_mmod_fast_256(u64 *result, const u64 *product, | |
441 | const u64 *curve_prime, u64 *tmp) | |
442 | { | |
443 | int carry; | |
444 | const unsigned int ndigits = 4; | |
445 | ||
446 | /* t */ | |
447 | vli_set(result, product, ndigits); | |
448 | ||
449 | /* s1 */ | |
450 | tmp[0] = 0; | |
451 | tmp[1] = product[5] & 0xffffffff00000000ull; | |
452 | tmp[2] = product[6]; | |
453 | tmp[3] = product[7]; | |
454 | carry = vli_lshift(tmp, tmp, 1, ndigits); | |
455 | carry += vli_add(result, result, tmp, ndigits); | |
456 | ||
457 | /* s2 */ | |
458 | tmp[1] = product[6] << 32; | |
459 | tmp[2] = (product[6] >> 32) | (product[7] << 32); | |
460 | tmp[3] = product[7] >> 32; | |
461 | carry += vli_lshift(tmp, tmp, 1, ndigits); | |
462 | carry += vli_add(result, result, tmp, ndigits); | |
463 | ||
464 | /* s3 */ | |
465 | tmp[0] = product[4]; | |
466 | tmp[1] = product[5] & 0xffffffff; | |
467 | tmp[2] = 0; | |
468 | tmp[3] = product[7]; | |
469 | carry += vli_add(result, result, tmp, ndigits); | |
470 | ||
471 | /* s4 */ | |
472 | tmp[0] = (product[4] >> 32) | (product[5] << 32); | |
473 | tmp[1] = (product[5] >> 32) | (product[6] & 0xffffffff00000000ull); | |
474 | tmp[2] = product[7]; | |
475 | tmp[3] = (product[6] >> 32) | (product[4] << 32); | |
476 | carry += vli_add(result, result, tmp, ndigits); | |
477 | ||
478 | /* d1 */ | |
479 | tmp[0] = (product[5] >> 32) | (product[6] << 32); | |
480 | tmp[1] = (product[6] >> 32); | |
481 | tmp[2] = 0; | |
482 | tmp[3] = (product[4] & 0xffffffff) | (product[5] << 32); | |
483 | carry -= vli_sub(result, result, tmp, ndigits); | |
484 | ||
485 | /* d2 */ | |
486 | tmp[0] = product[6]; | |
487 | tmp[1] = product[7]; | |
488 | tmp[2] = 0; | |
489 | tmp[3] = (product[4] >> 32) | (product[5] & 0xffffffff00000000ull); | |
490 | carry -= vli_sub(result, result, tmp, ndigits); | |
491 | ||
492 | /* d3 */ | |
493 | tmp[0] = (product[6] >> 32) | (product[7] << 32); | |
494 | tmp[1] = (product[7] >> 32) | (product[4] << 32); | |
495 | tmp[2] = (product[4] >> 32) | (product[5] << 32); | |
496 | tmp[3] = (product[6] << 32); | |
497 | carry -= vli_sub(result, result, tmp, ndigits); | |
498 | ||
499 | /* d4 */ | |
500 | tmp[0] = product[7]; | |
501 | tmp[1] = product[4] & 0xffffffff00000000ull; | |
502 | tmp[2] = product[5]; | |
503 | tmp[3] = product[6] & 0xffffffff00000000ull; | |
504 | carry -= vli_sub(result, result, tmp, ndigits); | |
505 | ||
506 | if (carry < 0) { | |
507 | do { | |
508 | carry += vli_add(result, result, curve_prime, ndigits); | |
509 | } while (carry < 0); | |
510 | } else { | |
511 | while (carry || vli_cmp(curve_prime, result, ndigits) != 1) | |
512 | carry -= vli_sub(result, result, curve_prime, ndigits); | |
513 | } | |
514 | } | |
515 | ||
516 | /* Computes result = product % curve_prime | |
517 | * from http://www.nsa.gov/ia/_files/nist-routines.pdf | |
518 | */ | |
519 | static bool vli_mmod_fast(u64 *result, u64 *product, | |
520 | const u64 *curve_prime, unsigned int ndigits) | |
521 | { | |
d5c3b178 | 522 | u64 tmp[2 * ECC_MAX_DIGITS]; |
3c4b2390 SB |
523 | |
524 | switch (ndigits) { | |
525 | case 3: | |
526 | vli_mmod_fast_192(result, product, curve_prime, tmp); | |
527 | break; | |
528 | case 4: | |
529 | vli_mmod_fast_256(result, product, curve_prime, tmp); | |
530 | break; | |
531 | default: | |
532 | pr_err("unsupports digits size!\n"); | |
533 | return false; | |
534 | } | |
535 | ||
536 | return true; | |
537 | } | |
538 | ||
539 | /* Computes result = (left * right) % curve_prime. */ | |
540 | static void vli_mod_mult_fast(u64 *result, const u64 *left, const u64 *right, | |
541 | const u64 *curve_prime, unsigned int ndigits) | |
542 | { | |
d5c3b178 | 543 | u64 product[2 * ECC_MAX_DIGITS]; |
3c4b2390 SB |
544 | |
545 | vli_mult(product, left, right, ndigits); | |
546 | vli_mmod_fast(result, product, curve_prime, ndigits); | |
547 | } | |
548 | ||
549 | /* Computes result = left^2 % curve_prime. */ | |
550 | static void vli_mod_square_fast(u64 *result, const u64 *left, | |
551 | const u64 *curve_prime, unsigned int ndigits) | |
552 | { | |
d5c3b178 | 553 | u64 product[2 * ECC_MAX_DIGITS]; |
3c4b2390 SB |
554 | |
555 | vli_square(product, left, ndigits); | |
556 | vli_mmod_fast(result, product, curve_prime, ndigits); | |
557 | } | |
558 | ||
559 | #define EVEN(vli) (!(vli[0] & 1)) | |
560 | /* Computes result = (1 / p_input) % mod. All VLIs are the same size. | |
561 | * See "From Euclid's GCD to Montgomery Multiplication to the Great Divide" | |
562 | * https://labs.oracle.com/techrep/2001/smli_tr-2001-95.pdf | |
563 | */ | |
4a2289da | 564 | void vli_mod_inv(u64 *result, const u64 *input, const u64 *mod, |
3c4b2390 SB |
565 | unsigned int ndigits) |
566 | { | |
d5c3b178 KC |
567 | u64 a[ECC_MAX_DIGITS], b[ECC_MAX_DIGITS]; |
568 | u64 u[ECC_MAX_DIGITS], v[ECC_MAX_DIGITS]; | |
3c4b2390 SB |
569 | u64 carry; |
570 | int cmp_result; | |
571 | ||
572 | if (vli_is_zero(input, ndigits)) { | |
573 | vli_clear(result, ndigits); | |
574 | return; | |
575 | } | |
576 | ||
577 | vli_set(a, input, ndigits); | |
578 | vli_set(b, mod, ndigits); | |
579 | vli_clear(u, ndigits); | |
580 | u[0] = 1; | |
581 | vli_clear(v, ndigits); | |
582 | ||
583 | while ((cmp_result = vli_cmp(a, b, ndigits)) != 0) { | |
584 | carry = 0; | |
585 | ||
586 | if (EVEN(a)) { | |
587 | vli_rshift1(a, ndigits); | |
588 | ||
589 | if (!EVEN(u)) | |
590 | carry = vli_add(u, u, mod, ndigits); | |
591 | ||
592 | vli_rshift1(u, ndigits); | |
593 | if (carry) | |
594 | u[ndigits - 1] |= 0x8000000000000000ull; | |
595 | } else if (EVEN(b)) { | |
596 | vli_rshift1(b, ndigits); | |
597 | ||
598 | if (!EVEN(v)) | |
599 | carry = vli_add(v, v, mod, ndigits); | |
600 | ||
601 | vli_rshift1(v, ndigits); | |
602 | if (carry) | |
603 | v[ndigits - 1] |= 0x8000000000000000ull; | |
604 | } else if (cmp_result > 0) { | |
605 | vli_sub(a, a, b, ndigits); | |
606 | vli_rshift1(a, ndigits); | |
607 | ||
608 | if (vli_cmp(u, v, ndigits) < 0) | |
609 | vli_add(u, u, mod, ndigits); | |
610 | ||
611 | vli_sub(u, u, v, ndigits); | |
612 | if (!EVEN(u)) | |
613 | carry = vli_add(u, u, mod, ndigits); | |
614 | ||
615 | vli_rshift1(u, ndigits); | |
616 | if (carry) | |
617 | u[ndigits - 1] |= 0x8000000000000000ull; | |
618 | } else { | |
619 | vli_sub(b, b, a, ndigits); | |
620 | vli_rshift1(b, ndigits); | |
621 | ||
622 | if (vli_cmp(v, u, ndigits) < 0) | |
623 | vli_add(v, v, mod, ndigits); | |
624 | ||
625 | vli_sub(v, v, u, ndigits); | |
626 | if (!EVEN(v)) | |
627 | carry = vli_add(v, v, mod, ndigits); | |
628 | ||
629 | vli_rshift1(v, ndigits); | |
630 | if (carry) | |
631 | v[ndigits - 1] |= 0x8000000000000000ull; | |
632 | } | |
633 | } | |
634 | ||
635 | vli_set(result, u, ndigits); | |
636 | } | |
4a2289da | 637 | EXPORT_SYMBOL(vli_mod_inv); |
3c4b2390 SB |
638 | |
639 | /* ------ Point operations ------ */ | |
640 | ||
641 | /* Returns true if p_point is the point at infinity, false otherwise. */ | |
642 | static bool ecc_point_is_zero(const struct ecc_point *point) | |
643 | { | |
644 | return (vli_is_zero(point->x, point->ndigits) && | |
645 | vli_is_zero(point->y, point->ndigits)); | |
646 | } | |
647 | ||
648 | /* Point multiplication algorithm using Montgomery's ladder with co-Z | |
649 | * coordinates. From http://eprint.iacr.org/2011/338.pdf | |
650 | */ | |
651 | ||
652 | /* Double in place */ | |
653 | static void ecc_point_double_jacobian(u64 *x1, u64 *y1, u64 *z1, | |
654 | u64 *curve_prime, unsigned int ndigits) | |
655 | { | |
656 | /* t1 = x, t2 = y, t3 = z */ | |
d5c3b178 KC |
657 | u64 t4[ECC_MAX_DIGITS]; |
658 | u64 t5[ECC_MAX_DIGITS]; | |
3c4b2390 SB |
659 | |
660 | if (vli_is_zero(z1, ndigits)) | |
661 | return; | |
662 | ||
663 | /* t4 = y1^2 */ | |
664 | vli_mod_square_fast(t4, y1, curve_prime, ndigits); | |
665 | /* t5 = x1*y1^2 = A */ | |
666 | vli_mod_mult_fast(t5, x1, t4, curve_prime, ndigits); | |
667 | /* t4 = y1^4 */ | |
668 | vli_mod_square_fast(t4, t4, curve_prime, ndigits); | |
669 | /* t2 = y1*z1 = z3 */ | |
670 | vli_mod_mult_fast(y1, y1, z1, curve_prime, ndigits); | |
671 | /* t3 = z1^2 */ | |
672 | vli_mod_square_fast(z1, z1, curve_prime, ndigits); | |
673 | ||
674 | /* t1 = x1 + z1^2 */ | |
675 | vli_mod_add(x1, x1, z1, curve_prime, ndigits); | |
676 | /* t3 = 2*z1^2 */ | |
677 | vli_mod_add(z1, z1, z1, curve_prime, ndigits); | |
678 | /* t3 = x1 - z1^2 */ | |
679 | vli_mod_sub(z1, x1, z1, curve_prime, ndigits); | |
680 | /* t1 = x1^2 - z1^4 */ | |
681 | vli_mod_mult_fast(x1, x1, z1, curve_prime, ndigits); | |
682 | ||
683 | /* t3 = 2*(x1^2 - z1^4) */ | |
684 | vli_mod_add(z1, x1, x1, curve_prime, ndigits); | |
685 | /* t1 = 3*(x1^2 - z1^4) */ | |
686 | vli_mod_add(x1, x1, z1, curve_prime, ndigits); | |
687 | if (vli_test_bit(x1, 0)) { | |
688 | u64 carry = vli_add(x1, x1, curve_prime, ndigits); | |
689 | ||
690 | vli_rshift1(x1, ndigits); | |
691 | x1[ndigits - 1] |= carry << 63; | |
692 | } else { | |
693 | vli_rshift1(x1, ndigits); | |
694 | } | |
695 | /* t1 = 3/2*(x1^2 - z1^4) = B */ | |
696 | ||
697 | /* t3 = B^2 */ | |
698 | vli_mod_square_fast(z1, x1, curve_prime, ndigits); | |
699 | /* t3 = B^2 - A */ | |
700 | vli_mod_sub(z1, z1, t5, curve_prime, ndigits); | |
701 | /* t3 = B^2 - 2A = x3 */ | |
702 | vli_mod_sub(z1, z1, t5, curve_prime, ndigits); | |
703 | /* t5 = A - x3 */ | |
704 | vli_mod_sub(t5, t5, z1, curve_prime, ndigits); | |
705 | /* t1 = B * (A - x3) */ | |
706 | vli_mod_mult_fast(x1, x1, t5, curve_prime, ndigits); | |
707 | /* t4 = B * (A - x3) - y1^4 = y3 */ | |
708 | vli_mod_sub(t4, x1, t4, curve_prime, ndigits); | |
709 | ||
710 | vli_set(x1, z1, ndigits); | |
711 | vli_set(z1, y1, ndigits); | |
712 | vli_set(y1, t4, ndigits); | |
713 | } | |
714 | ||
715 | /* Modify (x1, y1) => (x1 * z^2, y1 * z^3) */ | |
716 | static void apply_z(u64 *x1, u64 *y1, u64 *z, u64 *curve_prime, | |
717 | unsigned int ndigits) | |
718 | { | |
d5c3b178 | 719 | u64 t1[ECC_MAX_DIGITS]; |
3c4b2390 SB |
720 | |
721 | vli_mod_square_fast(t1, z, curve_prime, ndigits); /* z^2 */ | |
722 | vli_mod_mult_fast(x1, x1, t1, curve_prime, ndigits); /* x1 * z^2 */ | |
723 | vli_mod_mult_fast(t1, t1, z, curve_prime, ndigits); /* z^3 */ | |
724 | vli_mod_mult_fast(y1, y1, t1, curve_prime, ndigits); /* y1 * z^3 */ | |
725 | } | |
726 | ||
727 | /* P = (x1, y1) => 2P, (x2, y2) => P' */ | |
728 | static void xycz_initial_double(u64 *x1, u64 *y1, u64 *x2, u64 *y2, | |
729 | u64 *p_initial_z, u64 *curve_prime, | |
730 | unsigned int ndigits) | |
731 | { | |
d5c3b178 | 732 | u64 z[ECC_MAX_DIGITS]; |
3c4b2390 SB |
733 | |
734 | vli_set(x2, x1, ndigits); | |
735 | vli_set(y2, y1, ndigits); | |
736 | ||
737 | vli_clear(z, ndigits); | |
738 | z[0] = 1; | |
739 | ||
740 | if (p_initial_z) | |
741 | vli_set(z, p_initial_z, ndigits); | |
742 | ||
743 | apply_z(x1, y1, z, curve_prime, ndigits); | |
744 | ||
745 | ecc_point_double_jacobian(x1, y1, z, curve_prime, ndigits); | |
746 | ||
747 | apply_z(x2, y2, z, curve_prime, ndigits); | |
748 | } | |
749 | ||
750 | /* Input P = (x1, y1, Z), Q = (x2, y2, Z) | |
751 | * Output P' = (x1', y1', Z3), P + Q = (x3, y3, Z3) | |
752 | * or P => P', Q => P + Q | |
753 | */ | |
754 | static void xycz_add(u64 *x1, u64 *y1, u64 *x2, u64 *y2, u64 *curve_prime, | |
755 | unsigned int ndigits) | |
756 | { | |
757 | /* t1 = X1, t2 = Y1, t3 = X2, t4 = Y2 */ | |
d5c3b178 | 758 | u64 t5[ECC_MAX_DIGITS]; |
3c4b2390 SB |
759 | |
760 | /* t5 = x2 - x1 */ | |
761 | vli_mod_sub(t5, x2, x1, curve_prime, ndigits); | |
762 | /* t5 = (x2 - x1)^2 = A */ | |
763 | vli_mod_square_fast(t5, t5, curve_prime, ndigits); | |
764 | /* t1 = x1*A = B */ | |
765 | vli_mod_mult_fast(x1, x1, t5, curve_prime, ndigits); | |
766 | /* t3 = x2*A = C */ | |
767 | vli_mod_mult_fast(x2, x2, t5, curve_prime, ndigits); | |
768 | /* t4 = y2 - y1 */ | |
769 | vli_mod_sub(y2, y2, y1, curve_prime, ndigits); | |
770 | /* t5 = (y2 - y1)^2 = D */ | |
771 | vli_mod_square_fast(t5, y2, curve_prime, ndigits); | |
772 | ||
773 | /* t5 = D - B */ | |
774 | vli_mod_sub(t5, t5, x1, curve_prime, ndigits); | |
775 | /* t5 = D - B - C = x3 */ | |
776 | vli_mod_sub(t5, t5, x2, curve_prime, ndigits); | |
777 | /* t3 = C - B */ | |
778 | vli_mod_sub(x2, x2, x1, curve_prime, ndigits); | |
779 | /* t2 = y1*(C - B) */ | |
780 | vli_mod_mult_fast(y1, y1, x2, curve_prime, ndigits); | |
781 | /* t3 = B - x3 */ | |
782 | vli_mod_sub(x2, x1, t5, curve_prime, ndigits); | |
783 | /* t4 = (y2 - y1)*(B - x3) */ | |
784 | vli_mod_mult_fast(y2, y2, x2, curve_prime, ndigits); | |
785 | /* t4 = y3 */ | |
786 | vli_mod_sub(y2, y2, y1, curve_prime, ndigits); | |
787 | ||
788 | vli_set(x2, t5, ndigits); | |
789 | } | |
790 | ||
791 | /* Input P = (x1, y1, Z), Q = (x2, y2, Z) | |
792 | * Output P + Q = (x3, y3, Z3), P - Q = (x3', y3', Z3) | |
793 | * or P => P - Q, Q => P + Q | |
794 | */ | |
795 | static void xycz_add_c(u64 *x1, u64 *y1, u64 *x2, u64 *y2, u64 *curve_prime, | |
796 | unsigned int ndigits) | |
797 | { | |
798 | /* t1 = X1, t2 = Y1, t3 = X2, t4 = Y2 */ | |
d5c3b178 KC |
799 | u64 t5[ECC_MAX_DIGITS]; |
800 | u64 t6[ECC_MAX_DIGITS]; | |
801 | u64 t7[ECC_MAX_DIGITS]; | |
3c4b2390 SB |
802 | |
803 | /* t5 = x2 - x1 */ | |
804 | vli_mod_sub(t5, x2, x1, curve_prime, ndigits); | |
805 | /* t5 = (x2 - x1)^2 = A */ | |
806 | vli_mod_square_fast(t5, t5, curve_prime, ndigits); | |
807 | /* t1 = x1*A = B */ | |
808 | vli_mod_mult_fast(x1, x1, t5, curve_prime, ndigits); | |
809 | /* t3 = x2*A = C */ | |
810 | vli_mod_mult_fast(x2, x2, t5, curve_prime, ndigits); | |
811 | /* t4 = y2 + y1 */ | |
812 | vli_mod_add(t5, y2, y1, curve_prime, ndigits); | |
813 | /* t4 = y2 - y1 */ | |
814 | vli_mod_sub(y2, y2, y1, curve_prime, ndigits); | |
815 | ||
816 | /* t6 = C - B */ | |
817 | vli_mod_sub(t6, x2, x1, curve_prime, ndigits); | |
818 | /* t2 = y1 * (C - B) */ | |
819 | vli_mod_mult_fast(y1, y1, t6, curve_prime, ndigits); | |
820 | /* t6 = B + C */ | |
821 | vli_mod_add(t6, x1, x2, curve_prime, ndigits); | |
822 | /* t3 = (y2 - y1)^2 */ | |
823 | vli_mod_square_fast(x2, y2, curve_prime, ndigits); | |
824 | /* t3 = x3 */ | |
825 | vli_mod_sub(x2, x2, t6, curve_prime, ndigits); | |
826 | ||
827 | /* t7 = B - x3 */ | |
828 | vli_mod_sub(t7, x1, x2, curve_prime, ndigits); | |
829 | /* t4 = (y2 - y1)*(B - x3) */ | |
830 | vli_mod_mult_fast(y2, y2, t7, curve_prime, ndigits); | |
831 | /* t4 = y3 */ | |
832 | vli_mod_sub(y2, y2, y1, curve_prime, ndigits); | |
833 | ||
834 | /* t7 = (y2 + y1)^2 = F */ | |
835 | vli_mod_square_fast(t7, t5, curve_prime, ndigits); | |
836 | /* t7 = x3' */ | |
837 | vli_mod_sub(t7, t7, t6, curve_prime, ndigits); | |
838 | /* t6 = x3' - B */ | |
839 | vli_mod_sub(t6, t7, x1, curve_prime, ndigits); | |
840 | /* t6 = (y2 + y1)*(x3' - B) */ | |
841 | vli_mod_mult_fast(t6, t6, t5, curve_prime, ndigits); | |
842 | /* t2 = y3' */ | |
843 | vli_mod_sub(y1, t6, y1, curve_prime, ndigits); | |
844 | ||
845 | vli_set(x1, t7, ndigits); | |
846 | } | |
847 | ||
848 | static void ecc_point_mult(struct ecc_point *result, | |
849 | const struct ecc_point *point, const u64 *scalar, | |
3da2c1df | 850 | u64 *initial_z, const struct ecc_curve *curve, |
3c4b2390 SB |
851 | unsigned int ndigits) |
852 | { | |
853 | /* R0 and R1 */ | |
d5c3b178 KC |
854 | u64 rx[2][ECC_MAX_DIGITS]; |
855 | u64 ry[2][ECC_MAX_DIGITS]; | |
856 | u64 z[ECC_MAX_DIGITS]; | |
3da2c1df VC |
857 | u64 sk[2][ECC_MAX_DIGITS]; |
858 | u64 *curve_prime = curve->p; | |
3c4b2390 | 859 | int i, nb; |
3da2c1df VC |
860 | int num_bits; |
861 | int carry; | |
862 | ||
863 | carry = vli_add(sk[0], scalar, curve->n, ndigits); | |
864 | vli_add(sk[1], sk[0], curve->n, ndigits); | |
865 | scalar = sk[!carry]; | |
866 | num_bits = sizeof(u64) * ndigits * 8 + 1; | |
3c4b2390 SB |
867 | |
868 | vli_set(rx[1], point->x, ndigits); | |
869 | vli_set(ry[1], point->y, ndigits); | |
870 | ||
871 | xycz_initial_double(rx[1], ry[1], rx[0], ry[0], initial_z, curve_prime, | |
872 | ndigits); | |
873 | ||
874 | for (i = num_bits - 2; i > 0; i--) { | |
875 | nb = !vli_test_bit(scalar, i); | |
876 | xycz_add_c(rx[1 - nb], ry[1 - nb], rx[nb], ry[nb], curve_prime, | |
877 | ndigits); | |
878 | xycz_add(rx[nb], ry[nb], rx[1 - nb], ry[1 - nb], curve_prime, | |
879 | ndigits); | |
880 | } | |
881 | ||
882 | nb = !vli_test_bit(scalar, 0); | |
883 | xycz_add_c(rx[1 - nb], ry[1 - nb], rx[nb], ry[nb], curve_prime, | |
884 | ndigits); | |
885 | ||
886 | /* Find final 1/Z value. */ | |
887 | /* X1 - X0 */ | |
888 | vli_mod_sub(z, rx[1], rx[0], curve_prime, ndigits); | |
889 | /* Yb * (X1 - X0) */ | |
890 | vli_mod_mult_fast(z, z, ry[1 - nb], curve_prime, ndigits); | |
891 | /* xP * Yb * (X1 - X0) */ | |
892 | vli_mod_mult_fast(z, z, point->x, curve_prime, ndigits); | |
893 | ||
894 | /* 1 / (xP * Yb * (X1 - X0)) */ | |
895 | vli_mod_inv(z, z, curve_prime, point->ndigits); | |
896 | ||
897 | /* yP / (xP * Yb * (X1 - X0)) */ | |
898 | vli_mod_mult_fast(z, z, point->y, curve_prime, ndigits); | |
899 | /* Xb * yP / (xP * Yb * (X1 - X0)) */ | |
900 | vli_mod_mult_fast(z, z, rx[1 - nb], curve_prime, ndigits); | |
901 | /* End 1/Z calculation */ | |
902 | ||
903 | xycz_add(rx[nb], ry[nb], rx[1 - nb], ry[1 - nb], curve_prime, ndigits); | |
904 | ||
905 | apply_z(rx[0], ry[0], z, curve_prime, ndigits); | |
906 | ||
907 | vli_set(result->x, rx[0], ndigits); | |
908 | vli_set(result->y, ry[0], ndigits); | |
909 | } | |
910 | ||
911 | static inline void ecc_swap_digits(const u64 *in, u64 *out, | |
912 | unsigned int ndigits) | |
913 | { | |
914 | int i; | |
915 | ||
916 | for (i = 0; i < ndigits; i++) | |
917 | out[i] = __swab64(in[ndigits - 1 - i]); | |
918 | } | |
919 | ||
2eb4942b VC |
920 | static int __ecc_is_key_valid(const struct ecc_curve *curve, |
921 | const u64 *private_key, unsigned int ndigits) | |
3c4b2390 | 922 | { |
2eb4942b VC |
923 | u64 one[ECC_MAX_DIGITS] = { 1, }; |
924 | u64 res[ECC_MAX_DIGITS]; | |
3c4b2390 SB |
925 | |
926 | if (!private_key) | |
927 | return -EINVAL; | |
928 | ||
2eb4942b | 929 | if (curve->g.ndigits != ndigits) |
3c4b2390 SB |
930 | return -EINVAL; |
931 | ||
2eb4942b VC |
932 | /* Make sure the private key is in the range [2, n-3]. */ |
933 | if (vli_cmp(one, private_key, ndigits) != -1) | |
3c4b2390 | 934 | return -EINVAL; |
2eb4942b VC |
935 | vli_sub(res, curve->n, one, ndigits); |
936 | vli_sub(res, res, one, ndigits); | |
937 | if (vli_cmp(res, private_key, ndigits) != 1) | |
3c4b2390 SB |
938 | return -EINVAL; |
939 | ||
940 | return 0; | |
941 | } | |
942 | ||
2eb4942b VC |
943 | int ecc_is_key_valid(unsigned int curve_id, unsigned int ndigits, |
944 | const u64 *private_key, unsigned int private_key_len) | |
945 | { | |
946 | int nbytes; | |
947 | const struct ecc_curve *curve = ecc_get_curve(curve_id); | |
948 | ||
949 | nbytes = ndigits << ECC_DIGITS_TO_BYTES_SHIFT; | |
950 | ||
951 | if (private_key_len != nbytes) | |
952 | return -EINVAL; | |
953 | ||
954 | return __ecc_is_key_valid(curve, private_key, ndigits); | |
955 | } | |
4a2289da | 956 | EXPORT_SYMBOL(ecc_is_key_valid); |
2eb4942b | 957 | |
6755fd26 TDA |
958 | /* |
959 | * ECC private keys are generated using the method of extra random bits, | |
960 | * equivalent to that described in FIPS 186-4, Appendix B.4.1. | |
961 | * | |
962 | * d = (c mod(n–1)) + 1 where c is a string of random bits, 64 bits longer | |
963 | * than requested | |
964 | * 0 <= c mod(n-1) <= n-2 and implies that | |
965 | * 1 <= d <= n-1 | |
966 | * | |
967 | * This method generates a private key uniformly distributed in the range | |
968 | * [1, n-1]. | |
969 | */ | |
970 | int ecc_gen_privkey(unsigned int curve_id, unsigned int ndigits, u64 *privkey) | |
971 | { | |
972 | const struct ecc_curve *curve = ecc_get_curve(curve_id); | |
d5c3b178 | 973 | u64 priv[ECC_MAX_DIGITS]; |
6755fd26 TDA |
974 | unsigned int nbytes = ndigits << ECC_DIGITS_TO_BYTES_SHIFT; |
975 | unsigned int nbits = vli_num_bits(curve->n, ndigits); | |
976 | int err; | |
977 | ||
978 | /* Check that N is included in Table 1 of FIPS 186-4, section 6.1.1 */ | |
d5c3b178 | 979 | if (nbits < 160 || ndigits > ARRAY_SIZE(priv)) |
6755fd26 TDA |
980 | return -EINVAL; |
981 | ||
982 | /* | |
983 | * FIPS 186-4 recommends that the private key should be obtained from a | |
984 | * RBG with a security strength equal to or greater than the security | |
985 | * strength associated with N. | |
986 | * | |
987 | * The maximum security strength identified by NIST SP800-57pt1r4 for | |
988 | * ECC is 256 (N >= 512). | |
989 | * | |
990 | * This condition is met by the default RNG because it selects a favored | |
991 | * DRBG with a security strength of 256. | |
992 | */ | |
993 | if (crypto_get_default_rng()) | |
4c0e22c9 | 994 | return -EFAULT; |
6755fd26 TDA |
995 | |
996 | err = crypto_rng_get_bytes(crypto_default_rng, (u8 *)priv, nbytes); | |
997 | crypto_put_default_rng(); | |
998 | if (err) | |
999 | return err; | |
1000 | ||
2eb4942b VC |
1001 | /* Make sure the private key is in the valid range. */ |
1002 | if (__ecc_is_key_valid(curve, priv, ndigits)) | |
6755fd26 TDA |
1003 | return -EINVAL; |
1004 | ||
1005 | ecc_swap_digits(priv, privkey, ndigits); | |
1006 | ||
1007 | return 0; | |
1008 | } | |
4a2289da | 1009 | EXPORT_SYMBOL(ecc_gen_privkey); |
6755fd26 | 1010 | |
7380c56d TDA |
1011 | int ecc_make_pub_key(unsigned int curve_id, unsigned int ndigits, |
1012 | const u64 *private_key, u64 *public_key) | |
3c4b2390 SB |
1013 | { |
1014 | int ret = 0; | |
1015 | struct ecc_point *pk; | |
d5c3b178 | 1016 | u64 priv[ECC_MAX_DIGITS]; |
3c4b2390 SB |
1017 | const struct ecc_curve *curve = ecc_get_curve(curve_id); |
1018 | ||
d5c3b178 | 1019 | if (!private_key || !curve || ndigits > ARRAY_SIZE(priv)) { |
3c4b2390 SB |
1020 | ret = -EINVAL; |
1021 | goto out; | |
1022 | } | |
1023 | ||
ad269597 | 1024 | ecc_swap_digits(private_key, priv, ndigits); |
3c4b2390 SB |
1025 | |
1026 | pk = ecc_alloc_point(ndigits); | |
1027 | if (!pk) { | |
1028 | ret = -ENOMEM; | |
1029 | goto out; | |
1030 | } | |
1031 | ||
3da2c1df | 1032 | ecc_point_mult(pk, &curve->g, priv, NULL, curve, ndigits); |
3c4b2390 SB |
1033 | if (ecc_point_is_zero(pk)) { |
1034 | ret = -EAGAIN; | |
1035 | goto err_free_point; | |
1036 | } | |
1037 | ||
ad269597 TDA |
1038 | ecc_swap_digits(pk->x, public_key, ndigits); |
1039 | ecc_swap_digits(pk->y, &public_key[ndigits], ndigits); | |
3c4b2390 SB |
1040 | |
1041 | err_free_point: | |
1042 | ecc_free_point(pk); | |
1043 | out: | |
1044 | return ret; | |
1045 | } | |
4a2289da | 1046 | EXPORT_SYMBOL(ecc_make_pub_key); |
3c4b2390 | 1047 | |
ea169a30 | 1048 | /* SP800-56A section 5.6.2.3.4 partial verification: ephemeral keys only */ |
4a2289da VC |
1049 | int ecc_is_pubkey_valid_partial(const struct ecc_curve *curve, |
1050 | struct ecc_point *pk) | |
ea169a30 SM |
1051 | { |
1052 | u64 yy[ECC_MAX_DIGITS], xxx[ECC_MAX_DIGITS], w[ECC_MAX_DIGITS]; | |
1053 | ||
1054 | /* Check 1: Verify key is not the zero point. */ | |
1055 | if (ecc_point_is_zero(pk)) | |
1056 | return -EINVAL; | |
1057 | ||
1058 | /* Check 2: Verify key is in the range [1, p-1]. */ | |
1059 | if (vli_cmp(curve->p, pk->x, pk->ndigits) != 1) | |
1060 | return -EINVAL; | |
1061 | if (vli_cmp(curve->p, pk->y, pk->ndigits) != 1) | |
1062 | return -EINVAL; | |
1063 | ||
1064 | /* Check 3: Verify that y^2 == (x^3 + a·x + b) mod p */ | |
1065 | vli_mod_square_fast(yy, pk->y, curve->p, pk->ndigits); /* y^2 */ | |
1066 | vli_mod_square_fast(xxx, pk->x, curve->p, pk->ndigits); /* x^2 */ | |
1067 | vli_mod_mult_fast(xxx, xxx, pk->x, curve->p, pk->ndigits); /* x^3 */ | |
1068 | vli_mod_mult_fast(w, curve->a, pk->x, curve->p, pk->ndigits); /* a·x */ | |
1069 | vli_mod_add(w, w, curve->b, curve->p, pk->ndigits); /* a·x + b */ | |
1070 | vli_mod_add(w, w, xxx, curve->p, pk->ndigits); /* x^3 + a·x + b */ | |
1071 | if (vli_cmp(yy, w, pk->ndigits) != 0) /* Equation */ | |
1072 | return -EINVAL; | |
1073 | ||
1074 | return 0; | |
ea169a30 | 1075 | } |
4a2289da | 1076 | EXPORT_SYMBOL(ecc_is_pubkey_valid_partial); |
ea169a30 | 1077 | |
8f44df15 | 1078 | int crypto_ecdh_shared_secret(unsigned int curve_id, unsigned int ndigits, |
ad269597 TDA |
1079 | const u64 *private_key, const u64 *public_key, |
1080 | u64 *secret) | |
3c4b2390 SB |
1081 | { |
1082 | int ret = 0; | |
1083 | struct ecc_point *product, *pk; | |
d5c3b178 KC |
1084 | u64 priv[ECC_MAX_DIGITS]; |
1085 | u64 rand_z[ECC_MAX_DIGITS]; | |
1086 | unsigned int nbytes; | |
3c4b2390 SB |
1087 | const struct ecc_curve *curve = ecc_get_curve(curve_id); |
1088 | ||
d5c3b178 KC |
1089 | if (!private_key || !public_key || !curve || |
1090 | ndigits > ARRAY_SIZE(priv) || ndigits > ARRAY_SIZE(rand_z)) { | |
3c4b2390 SB |
1091 | ret = -EINVAL; |
1092 | goto out; | |
1093 | } | |
1094 | ||
d5c3b178 | 1095 | nbytes = ndigits << ECC_DIGITS_TO_BYTES_SHIFT; |
3c4b2390 | 1096 | |
d5c3b178 | 1097 | get_random_bytes(rand_z, nbytes); |
3c4b2390 SB |
1098 | |
1099 | pk = ecc_alloc_point(ndigits); | |
1100 | if (!pk) { | |
1101 | ret = -ENOMEM; | |
d5c3b178 | 1102 | goto out; |
3c4b2390 SB |
1103 | } |
1104 | ||
ea169a30 SM |
1105 | ecc_swap_digits(public_key, pk->x, ndigits); |
1106 | ecc_swap_digits(&public_key[ndigits], pk->y, ndigits); | |
1107 | ret = ecc_is_pubkey_valid_partial(curve, pk); | |
1108 | if (ret) | |
1109 | goto err_alloc_product; | |
1110 | ||
1111 | ecc_swap_digits(private_key, priv, ndigits); | |
1112 | ||
3c4b2390 SB |
1113 | product = ecc_alloc_point(ndigits); |
1114 | if (!product) { | |
1115 | ret = -ENOMEM; | |
1116 | goto err_alloc_product; | |
1117 | } | |
1118 | ||
3da2c1df | 1119 | ecc_point_mult(product, pk, priv, rand_z, curve, ndigits); |
3c4b2390 | 1120 | |
ad269597 | 1121 | ecc_swap_digits(product->x, secret, ndigits); |
3c4b2390 SB |
1122 | |
1123 | if (ecc_point_is_zero(product)) | |
1124 | ret = -EFAULT; | |
1125 | ||
1126 | ecc_free_point(product); | |
1127 | err_alloc_product: | |
1128 | ecc_free_point(pk); | |
1129 | out: | |
1130 | return ret; | |
1131 | } | |
4a2289da VC |
1132 | EXPORT_SYMBOL(crypto_ecdh_shared_secret); |
1133 | ||
1134 | MODULE_LICENSE("Dual BSD/GPL"); |