4a6c3cc964f754086c95a0e9a34a525106d6df0f
[openssl.git] / crypto / bn / bn_nist.c
1 /* crypto/bn/bn_nist.c */
2 /*
3  * Written by Nils Larsch for the OpenSSL project
4  */
5 /* ====================================================================
6  * Copyright (c) 1998-2005 The OpenSSL Project.  All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  *
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer. 
14  *
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in
17  *    the documentation and/or other materials provided with the
18  *    distribution.
19  *
20  * 3. All advertising materials mentioning features or use of this
21  *    software must display the following acknowledgment:
22  *    "This product includes software developed by the OpenSSL Project
23  *    for use in the OpenSSL Toolkit. (http://www.openssl.org/)"
24  *
25  * 4. The names "OpenSSL Toolkit" and "OpenSSL Project" must not be used to
26  *    endorse or promote products derived from this software without
27  *    prior written permission. For written permission, please contact
28  *    openssl-core@openssl.org.
29  *
30  * 5. Products derived from this software may not be called "OpenSSL"
31  *    nor may "OpenSSL" appear in their names without prior written
32  *    permission of the OpenSSL Project.
33  *
34  * 6. Redistributions of any form whatsoever must retain the following
35  *    acknowledgment:
36  *    "This product includes software developed by the OpenSSL Project
37  *    for use in the OpenSSL Toolkit (http://www.openssl.org/)"
38  *
39  * THIS SOFTWARE IS PROVIDED BY THE OpenSSL PROJECT ``AS IS'' AND ANY
40  * EXPRESSED OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
41  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
42  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE OpenSSL PROJECT OR
43  * ITS CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
44  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
45  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
46  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
47  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
48  * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
49  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
50  * OF THE POSSIBILITY OF SUCH DAMAGE.
51  * ====================================================================
52  *
53  * This product includes cryptographic software written by Eric Young
54  * (eay@cryptsoft.com).  This product includes software written by Tim
55  * Hudson (tjh@cryptsoft.com).
56  *
57  */
58
59 #include "bn_lcl.h"
60 #include "cryptlib.h"
61
62
63 #define BN_NIST_192_TOP (192+BN_BITS2-1)/BN_BITS2
64 #define BN_NIST_224_TOP (224+BN_BITS2-1)/BN_BITS2
65 #define BN_NIST_256_TOP (256+BN_BITS2-1)/BN_BITS2
66 #define BN_NIST_384_TOP (384+BN_BITS2-1)/BN_BITS2
67 #define BN_NIST_521_TOP (521+BN_BITS2-1)/BN_BITS2
68
69 /* pre-computed tables are "carry-less" values of modulus*(i+1) */
70 #if BN_BITS2 == 64
71 static const BN_ULONG _nist_p_192[][BN_NIST_192_TOP] = {
72         {0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFEULL,0xFFFFFFFFFFFFFFFFULL},
73         {0xFFFFFFFFFFFFFFFEULL,0xFFFFFFFFFFFFFFFDULL,0xFFFFFFFFFFFFFFFFULL},
74         {0xFFFFFFFFFFFFFFFDULL,0xFFFFFFFFFFFFFFFCULL,0xFFFFFFFFFFFFFFFFULL}
75         };
76 static const BN_ULONG _nist_p_192_sqr[] = {
77         0x0000000000000001ULL,0x0000000000000002ULL,0x0000000000000001ULL,
78         0xFFFFFFFFFFFFFFFEULL,0xFFFFFFFFFFFFFFFDULL,0xFFFFFFFFFFFFFFFFULL
79         };
80 static const BN_ULONG _nist_p_224[][BN_NIST_224_TOP] = {
81         {0x0000000000000001ULL,0xFFFFFFFF00000000ULL,
82          0xFFFFFFFFFFFFFFFFULL,0x00000000FFFFFFFFULL},
83         {0x0000000000000002ULL,0xFFFFFFFE00000000ULL,
84          0xFFFFFFFFFFFFFFFFULL,0x00000001FFFFFFFFULL} /* this one is "carry-full" */
85         };
86 static const BN_ULONG _nist_p_224_sqr[] = {
87         0x0000000000000001ULL,0xFFFFFFFE00000000ULL,
88         0xFFFFFFFFFFFFFFFFULL,0x0000000200000000ULL,
89         0x0000000000000000ULL,0xFFFFFFFFFFFFFFFEULL,
90         0xFFFFFFFFFFFFFFFFULL
91         };
92 static const BN_ULONG _nist_p_256[][BN_NIST_256_TOP] = {
93         {0xFFFFFFFFFFFFFFFFULL,0x00000000FFFFFFFFULL,
94          0x0000000000000000ULL,0xFFFFFFFF00000001ULL},
95         {0xFFFFFFFFFFFFFFFEULL,0x00000001FFFFFFFFULL,
96          0x0000000000000000ULL,0xFFFFFFFE00000002ULL},
97         {0xFFFFFFFFFFFFFFFDULL,0x00000002FFFFFFFFULL,
98          0x0000000000000000ULL,0xFFFFFFFD00000003ULL},
99         {0xFFFFFFFFFFFFFFFCULL,0x00000003FFFFFFFFULL,
100          0x0000000000000000ULL,0xFFFFFFFC00000004ULL},
101         {0xFFFFFFFFFFFFFFFBULL,0x00000004FFFFFFFFULL,
102          0x0000000000000000ULL,0xFFFFFFFB00000005ULL},
103         };
104 static const BN_ULONG _nist_p_256_sqr[] = {
105         0x0000000000000001ULL,0xFFFFFFFE00000000ULL,
106         0xFFFFFFFFFFFFFFFFULL,0x00000001FFFFFFFEULL,
107         0x00000001FFFFFFFEULL,0x00000001FFFFFFFEULL,
108         0xFFFFFFFE00000001ULL,0xFFFFFFFE00000002ULL
109         };
110 static const BN_ULONG _nist_p_384[][BN_NIST_384_TOP] = {
111         {0x00000000FFFFFFFFULL,0xFFFFFFFF00000000ULL,0xFFFFFFFFFFFFFFFEULL,
112          0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL},
113         {0x00000001FFFFFFFEULL,0xFFFFFFFE00000000ULL,0xFFFFFFFFFFFFFFFDULL,
114          0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL},
115         {0x00000002FFFFFFFDULL,0xFFFFFFFD00000000ULL,0xFFFFFFFFFFFFFFFCULL,
116          0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL},
117         {0x00000003FFFFFFFCULL,0xFFFFFFFC00000000ULL,0xFFFFFFFFFFFFFFFBULL,
118          0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL},
119         {0x00000004FFFFFFFBULL,0xFFFFFFFB00000000ULL,0xFFFFFFFFFFFFFFFAULL,
120          0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL},
121         };
122 static const BN_ULONG _nist_p_384_sqr[] = {
123         0xFFFFFFFE00000001ULL,0x0000000200000000ULL,0xFFFFFFFE00000000ULL,
124         0x0000000200000000ULL,0x0000000000000001ULL,0x0000000000000000ULL,
125         0x00000001FFFFFFFEULL,0xFFFFFFFE00000000ULL,0xFFFFFFFFFFFFFFFDULL,
126         0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL
127         };
128 static const BN_ULONG _nist_p_521[] =
129         {0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,
130         0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,
131         0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,
132         0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,
133         0x00000000000001FFULL};
134 static const BN_ULONG _nist_p_521_sqr[] = {
135         0x0000000000000001ULL,0x0000000000000000ULL,0x0000000000000000ULL,
136         0x0000000000000000ULL,0x0000000000000000ULL,0x0000000000000000ULL,
137         0x0000000000000000ULL,0x0000000000000000ULL,0xFFFFFFFFFFFFFC00ULL,
138         0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,
139         0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,
140         0xFFFFFFFFFFFFFFFFULL,0x000000000003FFFFULL
141         };
142 #elif BN_BITS2 == 32
143 static const BN_ULONG _nist_p_192[][BN_NIST_192_TOP] = {
144         {0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFE,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
145         {0xFFFFFFFE,0xFFFFFFFF,0xFFFFFFFD,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
146         {0xFFFFFFFD,0xFFFFFFFF,0xFFFFFFFC,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF}
147         };
148 static const BN_ULONG _nist_p_192_sqr[] = {
149         0x00000001,0x00000000,0x00000002,0x00000000,0x00000001,0x00000000,
150         0xFFFFFFFE,0xFFFFFFFF,0xFFFFFFFD,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF
151         };
152 static const BN_ULONG _nist_p_224[][BN_NIST_224_TOP] = {
153         {0x00000001,0x00000000,0x00000000,0xFFFFFFFF,
154          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
155         {0x00000002,0x00000000,0x00000000,0xFFFFFFFE,
156          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF}
157         };
158 static const BN_ULONG _nist_p_224_sqr[] = {
159         0x00000001,0x00000000,0x00000000,0xFFFFFFFE,
160         0xFFFFFFFF,0xFFFFFFFF,0x00000000,0x00000002,
161         0x00000000,0x00000000,0xFFFFFFFE,0xFFFFFFFF,
162         0xFFFFFFFF,0xFFFFFFFF
163         };
164 static const BN_ULONG _nist_p_256[][BN_NIST_256_TOP] = {
165         {0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0x00000000,
166          0x00000000,0x00000000,0x00000001,0xFFFFFFFF},
167         {0xFFFFFFFE,0xFFFFFFFF,0xFFFFFFFF,0x00000001,
168          0x00000000,0x00000000,0x00000002,0xFFFFFFFE},
169         {0xFFFFFFFD,0xFFFFFFFF,0xFFFFFFFF,0x00000002,
170          0x00000000,0x00000000,0x00000003,0xFFFFFFFD},
171         {0xFFFFFFFC,0xFFFFFFFF,0xFFFFFFFF,0x00000003,
172          0x00000000,0x00000000,0x00000004,0xFFFFFFFC},
173         {0xFFFFFFFB,0xFFFFFFFF,0xFFFFFFFF,0x00000004,
174          0x00000000,0x00000000,0x00000005,0xFFFFFFFB},
175         };
176 static const BN_ULONG _nist_p_256_sqr[] = {
177         0x00000001,0x00000000,0x00000000,0xFFFFFFFE,
178         0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFE,0x00000001,
179         0xFFFFFFFE,0x00000001,0xFFFFFFFE,0x00000001,
180         0x00000001,0xFFFFFFFE,0x00000002,0xFFFFFFFE
181         };
182 static const BN_ULONG _nist_p_384[][BN_NIST_384_TOP] = {
183         {0xFFFFFFFF,0x00000000,0x00000000,0xFFFFFFFF,0xFFFFFFFE,0xFFFFFFFF,
184          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
185         {0xFFFFFFFE,0x00000001,0x00000000,0xFFFFFFFE,0xFFFFFFFD,0xFFFFFFFF,
186          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
187         {0xFFFFFFFD,0x00000002,0x00000000,0xFFFFFFFD,0xFFFFFFFC,0xFFFFFFFF,
188          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
189         {0xFFFFFFFC,0x00000003,0x00000000,0xFFFFFFFC,0xFFFFFFFB,0xFFFFFFFF,
190          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
191         {0xFFFFFFFB,0x00000004,0x00000000,0xFFFFFFFB,0xFFFFFFFA,0xFFFFFFFF,
192          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
193         };
194 static const BN_ULONG _nist_p_384_sqr[] = {
195         0x00000001,0xFFFFFFFE,0x00000000,0x00000002,0x00000000,0xFFFFFFFE,
196         0x00000000,0x00000002,0x00000001,0x00000000,0x00000000,0x00000000,
197         0xFFFFFFFE,0x00000001,0x00000000,0xFFFFFFFE,0xFFFFFFFD,0xFFFFFFFF,
198         0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF
199         };
200 static const BN_ULONG _nist_p_521[] = {0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,
201         0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,
202         0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,
203         0xFFFFFFFF,0x000001FF};
204 static const BN_ULONG _nist_p_521_sqr[] = {
205         0x00000001,0x00000000,0x00000000,0x00000000,0x00000000,0x00000000,
206         0x00000000,0x00000000,0x00000000,0x00000000,0x00000000,0x00000000,
207         0x00000000,0x00000000,0x00000000,0x00000000,0xFFFFFC00,0xFFFFFFFF,
208         0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,
209         0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,
210         0xFFFFFFFF,0xFFFFFFFF,0x0003FFFF
211         };
212 #else
213 #error "unsupported BN_BITS2"
214 #endif
215
216
217 static const BIGNUM _bignum_nist_p_192 =
218         {
219         (BN_ULONG *)_nist_p_192[0],
220         BN_NIST_192_TOP,
221         BN_NIST_192_TOP,
222         0,
223         BN_FLG_STATIC_DATA
224         };
225
226 static const BIGNUM _bignum_nist_p_224 =
227         {
228         (BN_ULONG *)_nist_p_224[0],
229         BN_NIST_224_TOP,
230         BN_NIST_224_TOP,
231         0,
232         BN_FLG_STATIC_DATA
233         };
234
235 static const BIGNUM _bignum_nist_p_256 =
236         {
237         (BN_ULONG *)_nist_p_256[0],
238         BN_NIST_256_TOP,
239         BN_NIST_256_TOP,
240         0,
241         BN_FLG_STATIC_DATA
242         };
243
244 static const BIGNUM _bignum_nist_p_384 =
245         {
246         (BN_ULONG *)_nist_p_384[0],
247         BN_NIST_384_TOP,
248         BN_NIST_384_TOP,
249         0,
250         BN_FLG_STATIC_DATA
251         };
252
253 static const BIGNUM _bignum_nist_p_521 =
254         {
255         (BN_ULONG *)_nist_p_521,
256         BN_NIST_521_TOP,
257         BN_NIST_521_TOP,
258         0,
259         BN_FLG_STATIC_DATA
260         };
261
262
263 const BIGNUM *BN_get0_nist_prime_192(void)
264         {
265         return &_bignum_nist_p_192;
266         }
267
268 const BIGNUM *BN_get0_nist_prime_224(void)
269         {
270         return &_bignum_nist_p_224;
271         }
272
273 const BIGNUM *BN_get0_nist_prime_256(void)
274         {
275         return &_bignum_nist_p_256;
276         }
277
278 const BIGNUM *BN_get0_nist_prime_384(void)
279         {
280         return &_bignum_nist_p_384;
281         }
282
283 const BIGNUM *BN_get0_nist_prime_521(void)
284         {
285         return &_bignum_nist_p_521;
286         }
287
288
289 static void nist_cp_bn_0(BN_ULONG *buf, BN_ULONG *a, int top, int max)
290         {
291         int i;
292         BN_ULONG *_tmp1 = (buf), *_tmp2 = (a);
293
294 #ifdef BN_DEBUG
295         OPENSSL_assert(top <= max);
296 #endif
297         for (i = (top); i != 0; i--)
298                 *_tmp1++ = *_tmp2++;
299         for (i = (max) - (top); i != 0; i--)
300                 *_tmp1++ = (BN_ULONG) 0;
301         }
302
303 static void nist_cp_bn(BN_ULONG *buf, BN_ULONG *a, int top)
304         { 
305         int i;
306         BN_ULONG *_tmp1 = (buf), *_tmp2 = (a);
307         for (i = (top); i != 0; i--)
308                 *_tmp1++ = *_tmp2++;
309         }
310
311 #if BN_BITS2 == 64
312 #define bn_cp_64(to, n, from, m)        (to)[n] = (m>=0)?((from)[m]):0;
313 #define bn_64_set_0(to, n)              (to)[n] = (BN_ULONG)0;
314 /*
315  * two following macros are implemented under assumption that they
316  * are called in a sequence with *ascending* n, i.e. as they are...
317  */
318 #define bn_cp_32_naked(to, n, from, m)  (((n)&1)?(to[(n)/2]|=((m)&1)?(from[(m)/2]&BN_MASK2h):(from[(m)/2]<<32))\
319                                                 :(to[(n)/2] =((m)&1)?(from[(m)/2]>>32):(from[(m)/2]&BN_MASK2l)))
320 #define bn_32_set_0(to, n)              (((n)&1)?(to[(n)/2]&=BN_MASK2l):(to[(n)/2]=0));
321 #define bn_cp_32(to,n,from,m)           ((m)>=0)?bn_cp_32_naked(to,n,from,m):bn_32_set_0(to,n)
322 #else
323 #define bn_cp_64(to, n, from, m) \
324         { \
325         bn_cp_32(to, (n)*2, from, (m)*2); \
326         bn_cp_32(to, (n)*2+1, from, (m)*2+1); \
327         }
328 #define bn_64_set_0(to, n) \
329         { \
330         bn_32_set_0(to, (n)*2); \
331         bn_32_set_0(to, (n)*2+1); \
332         }
333 #if BN_BITS2 == 32
334 #define bn_cp_32(to, n, from, m)        (to)[n] = (m>=0)?((from)[m]):0;
335 #define bn_32_set_0(to, n)              (to)[n] = (BN_ULONG)0;
336 #endif
337 #endif /* BN_BITS2 != 64 */
338
339
340 #define nist_set_192(to, from, a1, a2, a3) \
341         { \
342         bn_cp_64(to, 0, from, (a3) - 3) \
343         bn_cp_64(to, 1, from, (a2) - 3) \
344         bn_cp_64(to, 2, from, (a1) - 3) \
345         }
346
347 int BN_nist_mod_192(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
348         BN_CTX *ctx)
349         {
350         int      top = a->top, i;
351         int      carry;
352         register BN_ULONG *r_d, *a_d = a->d;
353         BN_ULONG t_d[BN_NIST_192_TOP],
354                  buf[BN_NIST_192_TOP],
355                  c_d[BN_NIST_192_TOP],
356                 *res;
357         PTR_SIZE_INT mask;
358         static const BIGNUM _bignum_nist_p_192_sqr = {
359                 (BN_ULONG *)_nist_p_192_sqr,
360                 sizeof(_nist_p_192_sqr)/sizeof(_nist_p_192_sqr[0]),
361                 sizeof(_nist_p_192_sqr)/sizeof(_nist_p_192_sqr[0]),
362                 0,BN_FLG_STATIC_DATA };
363
364         field = &_bignum_nist_p_192; /* just to make sure */
365
366         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_192_sqr)>=0)
367                 return BN_nnmod(r, a, field, ctx);
368
369         i = BN_ucmp(field, a);
370         if (i == 0)
371                 {
372                 BN_zero(r);
373                 return 1;
374                 }
375         else if (i > 0)
376                 return (r == a) ? 1 : (BN_copy(r ,a) != NULL);
377
378         if (r != a)
379                 {
380                 if (!bn_wexpand(r, BN_NIST_192_TOP))
381                         return 0;
382                 r_d = r->d;
383                 nist_cp_bn(r_d, a_d, BN_NIST_192_TOP);
384                 }
385         else
386                 r_d = a_d;
387
388         nist_cp_bn_0(buf, a_d + BN_NIST_192_TOP, top - BN_NIST_192_TOP, BN_NIST_192_TOP);
389
390         nist_set_192(t_d, buf, 0, 3, 3);
391         carry = (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP);
392         nist_set_192(t_d, buf, 4, 4, 0);
393         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP);
394         nist_set_192(t_d, buf, 5, 5, 5)
395         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP);
396
397         if (carry > 0)
398                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_192[carry-1],BN_NIST_192_TOP);
399         else
400                 carry = 1;
401
402         /*
403          * we need 'if (carry==0 || result>=modulus) result-=modulus;'
404          * as comparison implies subtraction, we can write
405          * 'tmp=result-modulus; if (!carry || !borrow) result=tmp;'
406          * this is what happens below, but without explicit if:-) a.
407          */
408         mask  = 0-(PTR_SIZE_INT)bn_sub_words(c_d,r_d,_nist_p_192[0],BN_NIST_192_TOP);
409         mask &= 0-(PTR_SIZE_INT)carry;
410         res   = (BN_ULONG *)
411          (((PTR_SIZE_INT)c_d&~mask) | ((PTR_SIZE_INT)r_d&mask));
412         nist_cp_bn(r_d, res, BN_NIST_192_TOP);
413         r->top = BN_NIST_192_TOP;
414         bn_correct_top(r);
415
416         return 1;
417         }
418
419 typedef BN_ULONG (*bn_addsub_f)(BN_ULONG *,const BN_ULONG *,const BN_ULONG *,int);
420
421 #define nist_set_224(to, from, a1, a2, a3, a4, a5, a6, a7) \
422         { \
423         bn_cp_32(to, 0, from, (a7) - 7) \
424         bn_cp_32(to, 1, from, (a6) - 7) \
425         bn_cp_32(to, 2, from, (a5) - 7) \
426         bn_cp_32(to, 3, from, (a4) - 7) \
427         bn_cp_32(to, 4, from, (a3) - 7) \
428         bn_cp_32(to, 5, from, (a2) - 7) \
429         bn_cp_32(to, 6, from, (a1) - 7) \
430         }
431
432 int BN_nist_mod_224(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
433         BN_CTX *ctx)
434         {
435         int     top = a->top, i;
436         int     carry;
437         BN_ULONG *r_d, *a_d = a->d;
438         BN_ULONG t_d[BN_NIST_224_TOP],
439                  buf[BN_NIST_224_TOP],
440                  c_d[BN_NIST_224_TOP],
441                 *res;
442         PTR_SIZE_INT mask;
443         union { bn_addsub_f f; PTR_SIZE_INT p; } u;
444         static const BIGNUM _bignum_nist_p_224_sqr = {
445                 (BN_ULONG *)_nist_p_224_sqr,
446                 sizeof(_nist_p_224_sqr)/sizeof(_nist_p_224_sqr[0]),
447                 sizeof(_nist_p_224_sqr)/sizeof(_nist_p_224_sqr[0]),
448                 0,BN_FLG_STATIC_DATA };
449
450
451         field = &_bignum_nist_p_224; /* just to make sure */
452
453         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_224_sqr)>=0)
454                 return BN_nnmod(r, a, field, ctx);
455
456         i = BN_ucmp(field, a);
457         if (i == 0)
458                 {
459                 BN_zero(r);
460                 return 1;
461                 }
462         else if (i > 0)
463                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
464
465         if (r != a)
466                 {
467                 if (!bn_wexpand(r, BN_NIST_224_TOP))
468                         return 0;
469                 r_d = r->d;
470                 nist_cp_bn(r_d, a_d, BN_NIST_224_TOP);
471                 }
472         else
473                 r_d = a_d;
474
475 #if BN_BITS2==64
476         /* copy upper 256 bits of 448 bit number ... */
477         nist_cp_bn_0(t_d, a_d + (BN_NIST_224_TOP-1), top - (BN_NIST_224_TOP-1), BN_NIST_224_TOP);
478         /* ... and right shift by 32 to obtain upper 224 bits */
479         nist_set_224(buf, t_d, 14, 13, 12, 11, 10, 9, 8);
480         /* truncate lower part to 224 bits too */
481         r_d[BN_NIST_224_TOP-1] &= BN_MASK2l;
482 #else
483         nist_cp_bn_0(buf, a_d + BN_NIST_224_TOP, top - BN_NIST_224_TOP, BN_NIST_224_TOP);
484 #endif
485         nist_set_224(t_d, buf, 10, 9, 8, 7, 0, 0, 0);
486         carry = (int)bn_add_words(r_d, r_d, t_d, BN_NIST_224_TOP);
487         nist_set_224(t_d, buf, 0, 13, 12, 11, 0, 0, 0);
488         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_224_TOP);
489         nist_set_224(t_d, buf, 13, 12, 11, 10, 9, 8, 7);
490         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_224_TOP);
491         nist_set_224(t_d, buf, 0, 0, 0, 0, 13, 12, 11);
492         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_224_TOP);
493
494 #if BN_BITS2==64
495         carry = (int)(r_d[BN_NIST_224_TOP-1]>>32);
496 #endif
497         u.f = bn_sub_words;
498         if (carry > 0)
499                 {
500                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_224[carry-1],BN_NIST_224_TOP);
501 #if BN_BITS2==64
502                 carry=(int)(~(r_d[BN_NIST_224_TOP-1]>>32))&1;
503 #endif
504                 }
505         else if (carry < 0)
506                 {
507                 /* it's a bit more comlicated logic in this case.
508                  * if bn_add_words yields no carry, then result
509                  * has to be adjusted by unconditionally *adding*
510                  * the modulus. but if it does, then result has
511                  * to be compared to the modulus and conditionally
512                  * adjusted by *subtracting* the latter. */
513                 carry = (int)bn_add_words(r_d,r_d,_nist_p_224[-carry-1],BN_NIST_224_TOP);
514                 mask = 0-(PTR_SIZE_INT)carry;
515                 u.p = ((PTR_SIZE_INT)bn_sub_words&mask) |
516                  ((PTR_SIZE_INT)bn_add_words&~mask);
517                 }
518         else
519                 carry = 1;
520
521         /* otherwise it's effectively same as in BN_nist_mod_192... */
522         mask  = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_224[0],BN_NIST_224_TOP);
523         mask &= 0-(PTR_SIZE_INT)carry;
524         res   = (BN_ULONG *)(((PTR_SIZE_INT)c_d&~mask) |
525          ((PTR_SIZE_INT)r_d&mask));
526         nist_cp_bn(r_d, res, BN_NIST_224_TOP);
527         r->top = BN_NIST_224_TOP;
528         bn_correct_top(r);
529
530         return 1;
531         }
532
533 #define nist_set_256(to, from, a1, a2, a3, a4, a5, a6, a7, a8) \
534         { \
535         bn_cp_32(to, 0, from, (a8) - 8) \
536         bn_cp_32(to, 1, from, (a7) - 8) \
537         bn_cp_32(to, 2, from, (a6) - 8) \
538         bn_cp_32(to, 3, from, (a5) - 8) \
539         bn_cp_32(to, 4, from, (a4) - 8) \
540         bn_cp_32(to, 5, from, (a3) - 8) \
541         bn_cp_32(to, 6, from, (a2) - 8) \
542         bn_cp_32(to, 7, from, (a1) - 8) \
543         }
544
545 int BN_nist_mod_256(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
546         BN_CTX *ctx)
547         {
548         int     i, top = a->top;
549         int     carry = 0;
550         register BN_ULONG *a_d = a->d, *r_d;
551         BN_ULONG t_d[BN_NIST_256_TOP],
552                  buf[BN_NIST_256_TOP],
553                  c_d[BN_NIST_256_TOP],
554                 *res;
555         PTR_SIZE_INT mask;
556         union { bn_addsub_f f; PTR_SIZE_INT p; } u;
557         static const BIGNUM _bignum_nist_p_256_sqr = {
558                 (BN_ULONG *)_nist_p_256_sqr,
559                 sizeof(_nist_p_256_sqr)/sizeof(_nist_p_256_sqr[0]),
560                 sizeof(_nist_p_256_sqr)/sizeof(_nist_p_256_sqr[0]),
561                 0,BN_FLG_STATIC_DATA };
562
563         field = &_bignum_nist_p_256; /* just to make sure */
564
565         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_256_sqr)>=0)
566                 return BN_nnmod(r, a, field, ctx);
567
568         i = BN_ucmp(field, a);
569         if (i == 0)
570                 {
571                 BN_zero(r);
572                 return 1;
573                 }
574         else if (i > 0)
575                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
576
577         if (r != a)
578                 {
579                 if (!bn_wexpand(r, BN_NIST_256_TOP))
580                         return 0;
581                 r_d = r->d;
582                 nist_cp_bn(r_d, a_d, BN_NIST_256_TOP);
583                 }
584         else
585                 r_d = a_d;
586
587         nist_cp_bn_0(buf, a_d + BN_NIST_256_TOP, top - BN_NIST_256_TOP, BN_NIST_256_TOP);
588
589         /*S1*/
590         nist_set_256(t_d, buf, 15, 14, 13, 12, 11, 0, 0, 0);
591         /*S2*/
592         nist_set_256(c_d, buf, 0, 15, 14, 13, 12, 0, 0, 0);
593         carry = (int)bn_add_words(t_d, t_d, c_d, BN_NIST_256_TOP);
594         /* left shift */
595                 {
596                 register BN_ULONG *ap,t,c;
597                 ap = t_d;
598                 c=0;
599                 for (i = BN_NIST_256_TOP; i != 0; --i)
600                         {
601                         t= *ap;
602                         *(ap++)=((t<<1)|c)&BN_MASK2;
603                         c=(t & BN_TBIT)?1:0;
604                         }
605                 carry <<= 1;
606                 carry  |= c;
607                 }
608         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP);
609         /*S3*/
610         nist_set_256(t_d, buf, 15, 14, 0, 0, 0, 10, 9, 8);
611         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP);
612         /*S4*/
613         nist_set_256(t_d, buf, 8, 13, 15, 14, 13, 11, 10, 9);
614         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP);
615         /*D1*/
616         nist_set_256(t_d, buf, 10, 8, 0, 0, 0, 13, 12, 11);
617         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
618         /*D2*/
619         nist_set_256(t_d, buf, 11, 9, 0, 0, 15, 14, 13, 12);
620         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
621         /*D3*/
622         nist_set_256(t_d, buf, 12, 0, 10, 9, 8, 15, 14, 13);
623         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
624         /*D4*/
625         nist_set_256(t_d, buf, 13, 0, 11, 10, 9, 0, 15, 14);
626         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
627
628         /* see BN_nist_mod_224 for explanation */
629         u.f = bn_sub_words;
630         if (carry > 0)
631                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_256[carry-1],BN_NIST_256_TOP);
632         else if (carry < 0)
633                 {
634                 carry = (int)bn_add_words(r_d,r_d,_nist_p_256[-carry-1],BN_NIST_256_TOP);
635                 mask = 0-(PTR_SIZE_INT)carry;
636                 u.p = ((PTR_SIZE_INT)bn_sub_words&mask) |
637                  ((PTR_SIZE_INT)bn_add_words&~mask);
638                 }
639         else
640                 carry = 1;
641
642         mask  = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_256[0],BN_NIST_256_TOP);
643         mask &= 0-(PTR_SIZE_INT)carry;
644         res   = (BN_ULONG *)(((PTR_SIZE_INT)c_d&~mask) |
645          ((PTR_SIZE_INT)r_d&mask));
646         nist_cp_bn(r_d, res, BN_NIST_256_TOP);
647         r->top = BN_NIST_256_TOP;
648         bn_correct_top(r);
649
650         return 1;
651         }
652
653 #define nist_set_384(to,from,a1,a2,a3,a4,a5,a6,a7,a8,a9,a10,a11,a12) \
654         { \
655         bn_cp_32(to, 0, from,  (a12) - 12) \
656         bn_cp_32(to, 1, from,  (a11) - 12) \
657         bn_cp_32(to, 2, from,  (a10) - 12) \
658         bn_cp_32(to, 3, from,  (a9) - 12)  \
659         bn_cp_32(to, 4, from,  (a8) - 12)  \
660         bn_cp_32(to, 5, from,  (a7) - 12)  \
661         bn_cp_32(to, 6, from,  (a6) - 12)  \
662         bn_cp_32(to, 7, from,  (a5) - 12)  \
663         bn_cp_32(to, 8, from,  (a4) - 12)  \
664         bn_cp_32(to, 9, from,  (a3) - 12)  \
665         bn_cp_32(to, 10, from, (a2) - 12)  \
666         bn_cp_32(to, 11, from, (a1) - 12)  \
667         }
668
669 int BN_nist_mod_384(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
670         BN_CTX *ctx)
671         {
672         int     i, top = a->top;
673         int     carry = 0;
674         register BN_ULONG *r_d, *a_d = a->d;
675         BN_ULONG t_d[BN_NIST_384_TOP],
676                  buf[BN_NIST_384_TOP],
677                  c_d[BN_NIST_384_TOP],
678                 *res;
679         PTR_SIZE_INT mask;
680         union { bn_addsub_f f; PTR_SIZE_INT p; } u;
681         static const BIGNUM _bignum_nist_p_384_sqr = {
682                 (BN_ULONG *)_nist_p_384_sqr,
683                 sizeof(_nist_p_384_sqr)/sizeof(_nist_p_384_sqr[0]),
684                 sizeof(_nist_p_384_sqr)/sizeof(_nist_p_384_sqr[0]),
685                 0,BN_FLG_STATIC_DATA };
686
687
688         field = &_bignum_nist_p_384; /* just to make sure */
689
690         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_384_sqr)>=0)
691                 return BN_nnmod(r, a, field, ctx);
692
693         i = BN_ucmp(field, a);
694         if (i == 0)
695                 {
696                 BN_zero(r);
697                 return 1;
698                 }
699         else if (i > 0)
700                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
701
702         if (r != a)
703                 {
704                 if (!bn_wexpand(r, BN_NIST_384_TOP))
705                         return 0;
706                 r_d = r->d;
707                 nist_cp_bn(r_d, a_d, BN_NIST_384_TOP);
708                 }
709         else
710                 r_d = a_d;
711
712         nist_cp_bn_0(buf, a_d + BN_NIST_384_TOP, top - BN_NIST_384_TOP, BN_NIST_384_TOP);
713
714         /*S1*/
715         nist_set_256(t_d, buf, 0, 0, 0, 0, 0, 23-4, 22-4, 21-4);
716                 /* left shift */
717                 {
718                 register BN_ULONG *ap,t,c;
719                 ap = t_d;
720                 c=0;
721                 for (i = 3; i != 0; --i)
722                         {
723                         t= *ap;
724                         *(ap++)=((t<<1)|c)&BN_MASK2;
725                         c=(t & BN_TBIT)?1:0;
726                         }
727                 *ap=c;
728                 }
729         carry = (int)bn_add_words(r_d+(128/BN_BITS2), r_d+(128/BN_BITS2), 
730                 t_d, BN_NIST_256_TOP);
731         /*S2 */
732         carry += (int)bn_add_words(r_d, r_d, buf, BN_NIST_384_TOP);
733         /*S3*/
734         nist_set_384(t_d,buf,20,19,18,17,16,15,14,13,12,23,22,21);
735         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
736         /*S4*/
737         nist_set_384(t_d,buf,19,18,17,16,15,14,13,12,20,0,23,0);
738         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
739         /*S5*/
740         nist_set_384(t_d, buf,0,0,0,0,23,22,21,20,0,0,0,0);
741         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
742         /*S6*/
743         nist_set_384(t_d,buf,0,0,0,0,0,0,23,22,21,0,0,20);
744         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
745         /*D1*/
746         nist_set_384(t_d,buf,22,21,20,19,18,17,16,15,14,13,12,23);
747         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP);
748         /*D2*/
749         nist_set_384(t_d,buf,0,0,0,0,0,0,0,23,22,21,20,0);
750         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP);
751         /*D3*/
752         nist_set_384(t_d,buf,0,0,0,0,0,0,0,23,23,0,0,0);
753         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP);
754
755         /* see BN_nist_mod_224 for explanation */
756         u.f = bn_sub_words;
757         if (carry > 0)
758                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_384[carry-1],BN_NIST_384_TOP);
759         else if (carry < 0)
760                 {
761                 carry = (int)bn_add_words(r_d,r_d,_nist_p_384[-carry-1],BN_NIST_384_TOP);
762                 mask = 0-(PTR_SIZE_INT)carry;
763                 u.p = ((PTR_SIZE_INT)bn_sub_words&mask) |
764                  ((PTR_SIZE_INT)bn_add_words&~mask);
765                 }
766         else
767                 carry = 1;
768
769         mask  = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_384[0],BN_NIST_384_TOP);
770         mask &= 0-(PTR_SIZE_INT)carry;
771         res   = (BN_ULONG *)(((PTR_SIZE_INT)c_d&~mask) |
772          ((PTR_SIZE_INT)r_d&mask));
773         nist_cp_bn(r_d, res, BN_NIST_384_TOP);
774         r->top = BN_NIST_384_TOP;
775         bn_correct_top(r);
776
777         return 1;
778         }
779
780 #define BN_NIST_521_RSHIFT      (521%BN_BITS2)
781 #define BN_NIST_521_LSHIFT      (BN_BITS2-BN_NIST_521_RSHIFT)
782 #define BN_NIST_521_TOP_MASK    ((BN_ULONG)BN_MASK2>>BN_NIST_521_LSHIFT)
783
784 int BN_nist_mod_521(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
785         BN_CTX *ctx)
786         {
787         int     top = a->top, i;
788         BN_ULONG *r_d, *a_d = a->d,
789                  t_d[BN_NIST_521_TOP],
790                  val,tmp,*res;
791         PTR_SIZE_INT mask;
792         static const BIGNUM _bignum_nist_p_521_sqr = {
793                 (BN_ULONG *)_nist_p_521_sqr,
794                 sizeof(_nist_p_521_sqr)/sizeof(_nist_p_521_sqr[0]),
795                 sizeof(_nist_p_521_sqr)/sizeof(_nist_p_521_sqr[0]),
796                 0,BN_FLG_STATIC_DATA };
797
798         field = &_bignum_nist_p_521; /* just to make sure */
799
800         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_521_sqr)>=0)
801                 return BN_nnmod(r, a, field, ctx);
802
803         i = BN_ucmp(field, a);
804         if (i == 0)
805                 {
806                 BN_zero(r);
807                 return 1;
808                 }
809         else if (i > 0)
810                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
811
812         if (r != a)
813                 {
814                 if (!bn_wexpand(r,BN_NIST_521_TOP))
815                         return 0;
816                 r_d = r->d;
817                 nist_cp_bn(r_d,a_d, BN_NIST_521_TOP);
818                 }
819         else
820                 r_d = a_d;
821
822         /* upper 521 bits, copy ... */
823         nist_cp_bn_0(t_d,a_d + (BN_NIST_521_TOP-1), top - (BN_NIST_521_TOP-1),BN_NIST_521_TOP);
824         /* ... and right shift */
825         for (val=t_d[0],i=0; i<BN_NIST_521_TOP-1; i++)
826                 {
827                 tmp = val>>BN_NIST_521_RSHIFT;
828                 val = t_d[i+1];
829                 t_d[i] = (tmp | val<<BN_NIST_521_LSHIFT) & BN_MASK2;
830                 }
831         t_d[i] = val>>BN_NIST_521_RSHIFT;
832         /* lower 521 bits */
833         r_d[i] &= BN_NIST_521_TOP_MASK;
834
835         bn_add_words(r_d,r_d,t_d,BN_NIST_521_TOP);
836         mask = 0-(PTR_SIZE_INT)bn_sub_words(t_d,r_d,_nist_p_521,BN_NIST_521_TOP);
837         res  = (BN_ULONG *)(((PTR_SIZE_INT)t_d&~mask) |
838          ((PTR_SIZE_INT)r_d&mask));
839         nist_cp_bn(r_d,res,BN_NIST_521_TOP);
840         r->top = BN_NIST_521_TOP;
841         bn_correct_top(r);
842
843         return 1;
844         }
845
846 int (*BN_nist_mod_func(const BIGNUM *p))(BIGNUM *r, const BIGNUM *a, const BIGNUM *field, BN_CTX *ctx)
847         {
848         if (BN_ucmp(&_bignum_nist_p_192, p) == 0)
849                 return BN_nist_mod_192;
850         if (BN_ucmp(&_bignum_nist_p_224, p) == 0)
851                 return BN_nist_mod_224;
852         if (BN_ucmp(&_bignum_nist_p_256, p) == 0)
853                 return BN_nist_mod_256;
854         if (BN_ucmp(&_bignum_nist_p_384, p) == 0)
855                 return BN_nist_mod_384;
856         if (BN_ucmp(&_bignum_nist_p_521, p) == 0)
857                 return BN_nist_mod_521;
858         return 0;
859         }