modexp512-x86_64.pl: make it work with ml64.
[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 # if defined(L_ENDIAN)
323 #  if defined(__arch64__)
324 #   define NIST_INT64 long
325 #  else
326 #   define NIST_INT64 long long
327 #  endif
328 # endif
329 #else
330 #define bn_cp_64(to, n, from, m) \
331         { \
332         bn_cp_32(to, (n)*2, from, (m)*2); \
333         bn_cp_32(to, (n)*2+1, from, (m)*2+1); \
334         }
335 #define bn_64_set_0(to, n) \
336         { \
337         bn_32_set_0(to, (n)*2); \
338         bn_32_set_0(to, (n)*2+1); \
339         }
340 #define bn_cp_32(to, n, from, m)        (to)[n] = (m>=0)?((from)[m]):0;
341 #define bn_32_set_0(to, n)              (to)[n] = (BN_ULONG)0;
342 # if defined(_WIN32) && !defined(__GNUC__)
343 #  define NIST_INT64 __int64
344 # else
345 #  define NIST_INT64 long long
346 # endif
347 #endif /* BN_BITS2 != 64 */
348
349
350 #define nist_set_192(to, from, a1, a2, a3) \
351         { \
352         bn_cp_64(to, 0, from, (a3) - 3) \
353         bn_cp_64(to, 1, from, (a2) - 3) \
354         bn_cp_64(to, 2, from, (a1) - 3) \
355         }
356
357 int BN_nist_mod_192(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
358         BN_CTX *ctx)
359         {
360         int      top = a->top, i;
361         int      carry;
362         register BN_ULONG *r_d, *a_d = a->d;
363         BN_ULONG buf[BN_NIST_192_TOP],
364                  c_d[BN_NIST_192_TOP],
365                 *res;
366         PTR_SIZE_INT mask;
367         static const BIGNUM _bignum_nist_p_192_sqr = {
368                 (BN_ULONG *)_nist_p_192_sqr,
369                 sizeof(_nist_p_192_sqr)/sizeof(_nist_p_192_sqr[0]),
370                 sizeof(_nist_p_192_sqr)/sizeof(_nist_p_192_sqr[0]),
371                 0,BN_FLG_STATIC_DATA };
372
373         field = &_bignum_nist_p_192; /* just to make sure */
374
375         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_192_sqr)>=0)
376                 return BN_nnmod(r, a, field, ctx);
377
378         i = BN_ucmp(field, a);
379         if (i == 0)
380                 {
381                 BN_zero(r);
382                 return 1;
383                 }
384         else if (i > 0)
385                 return (r == a) ? 1 : (BN_copy(r ,a) != NULL);
386
387         if (r != a)
388                 {
389                 if (!bn_wexpand(r, BN_NIST_192_TOP))
390                         return 0;
391                 r_d = r->d;
392                 nist_cp_bn(r_d, a_d, BN_NIST_192_TOP);
393                 }
394         else
395                 r_d = a_d;
396
397         nist_cp_bn_0(buf, a_d + BN_NIST_192_TOP, top - BN_NIST_192_TOP, BN_NIST_192_TOP);
398
399 #if defined(NIST_INT64)
400         {
401         NIST_INT64              acc;    /* accumulator */
402         unsigned int            *rp=(unsigned int *)r_d;
403         const unsigned int      *bp=(const unsigned int *)buf;
404
405         acc  = rp[0];   acc += bp[3*2-6];
406                         acc += bp[5*2-6]; rp[0] = (unsigned int)acc; acc >>= 32;
407
408         acc += rp[1];   acc += bp[3*2-5];
409                         acc += bp[5*2-5]; rp[1] = (unsigned int)acc; acc >>= 32;
410
411         acc += rp[2];   acc += bp[3*2-6];
412                         acc += bp[4*2-6];
413                         acc += bp[5*2-6]; rp[2] = (unsigned int)acc; acc >>= 32;
414
415         acc += rp[3];   acc += bp[3*2-5];
416                         acc += bp[4*2-5];
417                         acc += bp[5*2-5]; rp[3] = (unsigned int)acc; acc >>= 32;
418
419         acc += rp[4];   acc += bp[4*2-6];
420                         acc += bp[5*2-6]; rp[4] = (unsigned int)acc; acc >>= 32;
421
422         acc += rp[5];   acc += bp[4*2-5];
423                         acc += bp[5*2-5]; rp[5] = (unsigned int)acc;
424
425         carry = (int)(acc>>32);
426         }
427 #else
428         {
429         BN_ULONG t_d[BN_NIST_192_TOP];
430
431         nist_set_192(t_d, buf, 0, 3, 3);
432         carry = (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP);
433         nist_set_192(t_d, buf, 4, 4, 0);
434         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP);
435         nist_set_192(t_d, buf, 5, 5, 5)
436         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP);
437         }
438 #endif
439         if (carry > 0)
440                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_192[carry-1],BN_NIST_192_TOP);
441         else
442                 carry = 1;
443
444         /*
445          * we need 'if (carry==0 || result>=modulus) result-=modulus;'
446          * as comparison implies subtraction, we can write
447          * 'tmp=result-modulus; if (!carry || !borrow) result=tmp;'
448          * this is what happens below, but without explicit if:-) a.
449          */
450         mask  = 0-(PTR_SIZE_INT)bn_sub_words(c_d,r_d,_nist_p_192[0],BN_NIST_192_TOP);
451         mask &= 0-(PTR_SIZE_INT)carry;
452         res   = (BN_ULONG *)
453          (((PTR_SIZE_INT)c_d&~mask) | ((PTR_SIZE_INT)r_d&mask));
454         nist_cp_bn(r_d, res, BN_NIST_192_TOP);
455         r->top = BN_NIST_192_TOP;
456         bn_correct_top(r);
457
458         return 1;
459         }
460
461 typedef BN_ULONG (*bn_addsub_f)(BN_ULONG *,const BN_ULONG *,const BN_ULONG *,int);
462
463 #define nist_set_224(to, from, a1, a2, a3, a4, a5, a6, a7) \
464         { \
465         bn_cp_32(to, 0, from, (a7) - 7) \
466         bn_cp_32(to, 1, from, (a6) - 7) \
467         bn_cp_32(to, 2, from, (a5) - 7) \
468         bn_cp_32(to, 3, from, (a4) - 7) \
469         bn_cp_32(to, 4, from, (a3) - 7) \
470         bn_cp_32(to, 5, from, (a2) - 7) \
471         bn_cp_32(to, 6, from, (a1) - 7) \
472         }
473
474 int BN_nist_mod_224(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
475         BN_CTX *ctx)
476         {
477         int     top = a->top, i;
478         int     carry;
479         BN_ULONG *r_d, *a_d = a->d;
480         BN_ULONG buf[BN_NIST_224_TOP],
481                  c_d[BN_NIST_224_TOP],
482                 *res;
483         PTR_SIZE_INT mask;
484         union { bn_addsub_f f; PTR_SIZE_INT p; } u;
485         static const BIGNUM _bignum_nist_p_224_sqr = {
486                 (BN_ULONG *)_nist_p_224_sqr,
487                 sizeof(_nist_p_224_sqr)/sizeof(_nist_p_224_sqr[0]),
488                 sizeof(_nist_p_224_sqr)/sizeof(_nist_p_224_sqr[0]),
489                 0,BN_FLG_STATIC_DATA };
490
491
492         field = &_bignum_nist_p_224; /* just to make sure */
493
494         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_224_sqr)>=0)
495                 return BN_nnmod(r, a, field, ctx);
496
497         i = BN_ucmp(field, a);
498         if (i == 0)
499                 {
500                 BN_zero(r);
501                 return 1;
502                 }
503         else if (i > 0)
504                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
505
506         if (r != a)
507                 {
508                 if (!bn_wexpand(r, BN_NIST_224_TOP))
509                         return 0;
510                 r_d = r->d;
511                 nist_cp_bn(r_d, a_d, BN_NIST_224_TOP);
512                 }
513         else
514                 r_d = a_d;
515
516 #if BN_BITS2==64
517         /* copy upper 256 bits of 448 bit number ... */
518         nist_cp_bn_0(c_d, a_d + (BN_NIST_224_TOP-1), top - (BN_NIST_224_TOP-1), BN_NIST_224_TOP);
519         /* ... and right shift by 32 to obtain upper 224 bits */
520         nist_set_224(buf, c_d, 14, 13, 12, 11, 10, 9, 8);
521         /* truncate lower part to 224 bits too */
522         r_d[BN_NIST_224_TOP-1] &= BN_MASK2l;
523 #else
524         nist_cp_bn_0(buf, a_d + BN_NIST_224_TOP, top - BN_NIST_224_TOP, BN_NIST_224_TOP);
525 #endif
526
527 #if defined(NIST_INT64) && BN_BITS2!=64
528         {
529         NIST_INT64              acc;    /* accumulator */
530         unsigned int            *rp=(unsigned int *)r_d;
531         const unsigned int      *bp=(const unsigned int *)buf;
532
533         acc  = rp[0];   acc -= bp[7-7];
534                         acc -= bp[11-7]; rp[0] = (unsigned int)acc; acc >>= 32;
535
536         acc += rp[1];   acc -= bp[8-7];
537                         acc -= bp[12-7]; rp[1] = (unsigned int)acc; acc >>= 32;
538
539         acc += rp[2];   acc -= bp[9-7];
540                         acc -= bp[13-7]; rp[2] = (unsigned int)acc; acc >>= 32;
541
542         acc += rp[3];   acc += bp[7-7];
543                         acc += bp[11-7];
544                         acc -= bp[10-7]; rp[3] = (unsigned int)acc; acc>>= 32;
545
546         acc += rp[4];   acc += bp[8-7];
547                         acc += bp[12-7];
548                         acc -= bp[11-7]; rp[4] = (unsigned int)acc; acc >>= 32;
549
550         acc += rp[5];   acc += bp[9-7];
551                         acc += bp[13-7];
552                         acc -= bp[12-7]; rp[5] = (unsigned int)acc; acc >>= 32;
553
554         acc += rp[6];   acc += bp[10-7];
555                         acc -= bp[13-7]; rp[6] = (unsigned int)acc;
556
557         carry = (int)(acc>>32);
558 # if BN_BITS2==64
559         rp[7] = carry;
560 # endif
561         }       
562 #else
563         {
564         BN_ULONG t_d[BN_NIST_224_TOP];
565
566         nist_set_224(t_d, buf, 10, 9, 8, 7, 0, 0, 0);
567         carry = (int)bn_add_words(r_d, r_d, t_d, BN_NIST_224_TOP);
568         nist_set_224(t_d, buf, 0, 13, 12, 11, 0, 0, 0);
569         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_224_TOP);
570         nist_set_224(t_d, buf, 13, 12, 11, 10, 9, 8, 7);
571         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_224_TOP);
572         nist_set_224(t_d, buf, 0, 0, 0, 0, 13, 12, 11);
573         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_224_TOP);
574
575 #if BN_BITS2==64
576         carry = (int)(r_d[BN_NIST_224_TOP-1]>>32);
577 #endif
578         }
579 #endif
580         u.f = bn_sub_words;
581         if (carry > 0)
582                 {
583                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_224[carry-1],BN_NIST_224_TOP);
584 #if BN_BITS2==64
585                 carry=(int)(~(r_d[BN_NIST_224_TOP-1]>>32))&1;
586 #endif
587                 }
588         else if (carry < 0)
589                 {
590                 /* it's a bit more comlicated logic in this case.
591                  * if bn_add_words yields no carry, then result
592                  * has to be adjusted by unconditionally *adding*
593                  * the modulus. but if it does, then result has
594                  * to be compared to the modulus and conditionally
595                  * adjusted by *subtracting* the latter. */
596                 carry = (int)bn_add_words(r_d,r_d,_nist_p_224[-carry-1],BN_NIST_224_TOP);
597                 mask = 0-(PTR_SIZE_INT)carry;
598                 u.p = ((PTR_SIZE_INT)bn_sub_words&mask) |
599                  ((PTR_SIZE_INT)bn_add_words&~mask);
600                 }
601         else
602                 carry = 1;
603
604         /* otherwise it's effectively same as in BN_nist_mod_192... */
605         mask  = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_224[0],BN_NIST_224_TOP);
606         mask &= 0-(PTR_SIZE_INT)carry;
607         res   = (BN_ULONG *)(((PTR_SIZE_INT)c_d&~mask) |
608          ((PTR_SIZE_INT)r_d&mask));
609         nist_cp_bn(r_d, res, BN_NIST_224_TOP);
610         r->top = BN_NIST_224_TOP;
611         bn_correct_top(r);
612
613         return 1;
614         }
615
616 #define nist_set_256(to, from, a1, a2, a3, a4, a5, a6, a7, a8) \
617         { \
618         bn_cp_32(to, 0, from, (a8) - 8) \
619         bn_cp_32(to, 1, from, (a7) - 8) \
620         bn_cp_32(to, 2, from, (a6) - 8) \
621         bn_cp_32(to, 3, from, (a5) - 8) \
622         bn_cp_32(to, 4, from, (a4) - 8) \
623         bn_cp_32(to, 5, from, (a3) - 8) \
624         bn_cp_32(to, 6, from, (a2) - 8) \
625         bn_cp_32(to, 7, from, (a1) - 8) \
626         }
627
628 int BN_nist_mod_256(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
629         BN_CTX *ctx)
630         {
631         int     i, top = a->top;
632         int     carry = 0;
633         register BN_ULONG *a_d = a->d, *r_d;
634         BN_ULONG buf[BN_NIST_256_TOP],
635                  c_d[BN_NIST_256_TOP],
636                 *res;
637         PTR_SIZE_INT mask;
638         union { bn_addsub_f f; PTR_SIZE_INT p; } u;
639         static const BIGNUM _bignum_nist_p_256_sqr = {
640                 (BN_ULONG *)_nist_p_256_sqr,
641                 sizeof(_nist_p_256_sqr)/sizeof(_nist_p_256_sqr[0]),
642                 sizeof(_nist_p_256_sqr)/sizeof(_nist_p_256_sqr[0]),
643                 0,BN_FLG_STATIC_DATA };
644
645         field = &_bignum_nist_p_256; /* just to make sure */
646
647         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_256_sqr)>=0)
648                 return BN_nnmod(r, a, field, ctx);
649
650         i = BN_ucmp(field, a);
651         if (i == 0)
652                 {
653                 BN_zero(r);
654                 return 1;
655                 }
656         else if (i > 0)
657                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
658
659         if (r != a)
660                 {
661                 if (!bn_wexpand(r, BN_NIST_256_TOP))
662                         return 0;
663                 r_d = r->d;
664                 nist_cp_bn(r_d, a_d, BN_NIST_256_TOP);
665                 }
666         else
667                 r_d = a_d;
668
669         nist_cp_bn_0(buf, a_d + BN_NIST_256_TOP, top - BN_NIST_256_TOP, BN_NIST_256_TOP);
670
671 #if defined(NIST_INT64)
672         {
673         NIST_INT64              acc;    /* accumulator */
674         unsigned int            *rp=(unsigned int *)r_d;
675         const unsigned int      *bp=(const unsigned int *)buf;
676
677         acc = rp[0];    acc += bp[8-8];
678                         acc += bp[9-8];
679                         acc -= bp[11-8];
680                         acc -= bp[12-8];
681                         acc -= bp[13-8];
682                         acc -= bp[14-8]; rp[0] = (unsigned int)acc; acc >>= 32;
683
684         acc += rp[1];   acc += bp[9-8];
685                         acc += bp[10-8];
686                         acc -= bp[12-8];
687                         acc -= bp[13-8];
688                         acc -= bp[14-8];
689                         acc -= bp[15-8]; rp[1] = (unsigned int)acc; acc >>= 32;
690
691         acc += rp[2];   acc += bp[10-8];
692                         acc += bp[11-8];
693                         acc -= bp[13-8];
694                         acc -= bp[14-8];
695                         acc -= bp[15-8]; rp[2] = (unsigned int)acc; acc >>= 32;
696
697         acc += rp[3];   acc += bp[11-8];
698                         acc += bp[11-8];
699                         acc += bp[12-8];
700                         acc += bp[12-8];
701                         acc += bp[13-8];
702                         acc -= bp[15-8];
703                         acc -= bp[8-8];
704                         acc -= bp[9-8];  rp[3] = (unsigned int)acc; acc >>= 32;
705
706         acc += rp[4];   acc += bp[12-8];
707                         acc += bp[12-8];
708                         acc += bp[13-8];
709                         acc += bp[13-8];
710                         acc += bp[14-8];
711                         acc -= bp[9-8];
712                         acc -= bp[10-8]; rp[4] = (unsigned int)acc; acc >>= 32;
713
714         acc += rp[5];   acc += bp[13-8];
715                         acc += bp[13-8];
716                         acc += bp[14-8];
717                         acc += bp[14-8];
718                         acc += bp[15-8];
719                         acc -= bp[10-8];
720                         acc -= bp[11-8]; rp[5] = (unsigned int)acc; acc >>= 32;
721
722         acc += rp[6];   acc += bp[14-8];
723                         acc += bp[14-8];
724                         acc += bp[15-8];
725                         acc += bp[15-8];
726                         acc += bp[14-8];
727                         acc += bp[13-8];
728                         acc -= bp[8-8];
729                         acc -= bp[9-8];  rp[6] = (unsigned int)acc; acc >>= 32;
730
731         acc += rp[7];   acc += bp[15-8];
732                         acc += bp[15-8];
733                         acc += bp[15-8];
734                         acc += bp[8 -8];
735                         acc -= bp[10-8];
736                         acc -= bp[11-8];
737                         acc -= bp[12-8];
738                         acc -= bp[13-8]; rp[7] = (unsigned int)acc;
739
740         carry = (int)(acc>>32);
741         }
742 #else
743         {
744         BN_ULONG t_d[BN_NIST_256_TOP];
745
746         /*S1*/
747         nist_set_256(t_d, buf, 15, 14, 13, 12, 11, 0, 0, 0);
748         /*S2*/
749         nist_set_256(c_d, buf, 0, 15, 14, 13, 12, 0, 0, 0);
750         carry = (int)bn_add_words(t_d, t_d, c_d, BN_NIST_256_TOP);
751         /* left shift */
752                 {
753                 register BN_ULONG *ap,t,c;
754                 ap = t_d;
755                 c=0;
756                 for (i = BN_NIST_256_TOP; i != 0; --i)
757                         {
758                         t= *ap;
759                         *(ap++)=((t<<1)|c)&BN_MASK2;
760                         c=(t & BN_TBIT)?1:0;
761                         }
762                 carry <<= 1;
763                 carry  |= c;
764                 }
765         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP);
766         /*S3*/
767         nist_set_256(t_d, buf, 15, 14, 0, 0, 0, 10, 9, 8);
768         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP);
769         /*S4*/
770         nist_set_256(t_d, buf, 8, 13, 15, 14, 13, 11, 10, 9);
771         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP);
772         /*D1*/
773         nist_set_256(t_d, buf, 10, 8, 0, 0, 0, 13, 12, 11);
774         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
775         /*D2*/
776         nist_set_256(t_d, buf, 11, 9, 0, 0, 15, 14, 13, 12);
777         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
778         /*D3*/
779         nist_set_256(t_d, buf, 12, 0, 10, 9, 8, 15, 14, 13);
780         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
781         /*D4*/
782         nist_set_256(t_d, buf, 13, 0, 11, 10, 9, 0, 15, 14);
783         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
784
785         }
786 #endif
787         /* see BN_nist_mod_224 for explanation */
788         u.f = bn_sub_words;
789         if (carry > 0)
790                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_256[carry-1],BN_NIST_256_TOP);
791         else if (carry < 0)
792                 {
793                 carry = (int)bn_add_words(r_d,r_d,_nist_p_256[-carry-1],BN_NIST_256_TOP);
794                 mask = 0-(PTR_SIZE_INT)carry;
795                 u.p = ((PTR_SIZE_INT)bn_sub_words&mask) |
796                  ((PTR_SIZE_INT)bn_add_words&~mask);
797                 }
798         else
799                 carry = 1;
800
801         mask  = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_256[0],BN_NIST_256_TOP);
802         mask &= 0-(PTR_SIZE_INT)carry;
803         res   = (BN_ULONG *)(((PTR_SIZE_INT)c_d&~mask) |
804          ((PTR_SIZE_INT)r_d&mask));
805         nist_cp_bn(r_d, res, BN_NIST_256_TOP);
806         r->top = BN_NIST_256_TOP;
807         bn_correct_top(r);
808
809         return 1;
810         }
811
812 #define nist_set_384(to,from,a1,a2,a3,a4,a5,a6,a7,a8,a9,a10,a11,a12) \
813         { \
814         bn_cp_32(to, 0, from,  (a12) - 12) \
815         bn_cp_32(to, 1, from,  (a11) - 12) \
816         bn_cp_32(to, 2, from,  (a10) - 12) \
817         bn_cp_32(to, 3, from,  (a9) - 12)  \
818         bn_cp_32(to, 4, from,  (a8) - 12)  \
819         bn_cp_32(to, 5, from,  (a7) - 12)  \
820         bn_cp_32(to, 6, from,  (a6) - 12)  \
821         bn_cp_32(to, 7, from,  (a5) - 12)  \
822         bn_cp_32(to, 8, from,  (a4) - 12)  \
823         bn_cp_32(to, 9, from,  (a3) - 12)  \
824         bn_cp_32(to, 10, from, (a2) - 12)  \
825         bn_cp_32(to, 11, from, (a1) - 12)  \
826         }
827
828 int BN_nist_mod_384(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
829         BN_CTX *ctx)
830         {
831         int     i, top = a->top;
832         int     carry = 0;
833         register BN_ULONG *r_d, *a_d = a->d;
834         BN_ULONG buf[BN_NIST_384_TOP],
835                  c_d[BN_NIST_384_TOP],
836                 *res;
837         PTR_SIZE_INT mask;
838         union { bn_addsub_f f; PTR_SIZE_INT p; } u;
839         static const BIGNUM _bignum_nist_p_384_sqr = {
840                 (BN_ULONG *)_nist_p_384_sqr,
841                 sizeof(_nist_p_384_sqr)/sizeof(_nist_p_384_sqr[0]),
842                 sizeof(_nist_p_384_sqr)/sizeof(_nist_p_384_sqr[0]),
843                 0,BN_FLG_STATIC_DATA };
844
845
846         field = &_bignum_nist_p_384; /* just to make sure */
847
848         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_384_sqr)>=0)
849                 return BN_nnmod(r, a, field, ctx);
850
851         i = BN_ucmp(field, a);
852         if (i == 0)
853                 {
854                 BN_zero(r);
855                 return 1;
856                 }
857         else if (i > 0)
858                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
859
860         if (r != a)
861                 {
862                 if (!bn_wexpand(r, BN_NIST_384_TOP))
863                         return 0;
864                 r_d = r->d;
865                 nist_cp_bn(r_d, a_d, BN_NIST_384_TOP);
866                 }
867         else
868                 r_d = a_d;
869
870         nist_cp_bn_0(buf, a_d + BN_NIST_384_TOP, top - BN_NIST_384_TOP, BN_NIST_384_TOP);
871
872 #if defined(NIST_INT64)
873         {
874         NIST_INT64              acc;    /* accumulator */
875         unsigned int            *rp=(unsigned int *)r_d;
876         const unsigned int      *bp=(const unsigned int *)buf;
877
878         acc = rp[0];    acc += bp[12-12];
879                         acc += bp[21-12];
880                         acc += bp[20-12];
881                         acc -= bp[23-12]; rp[0] = (unsigned int)acc; acc >>= 32;
882
883         acc += rp[1];   acc += bp[13-12];
884                         acc += bp[22-12];
885                         acc += bp[23-12];
886                         acc -= bp[12-12];
887                         acc -= bp[20-12]; rp[1] = (unsigned int)acc; acc >>= 32;
888
889         acc += rp[2];   acc += bp[14-12];
890                         acc += bp[23-12];
891                         acc -= bp[13-12];
892                         acc -= bp[21-12]; rp[2] = (unsigned int)acc; acc >>= 32;
893
894         acc += rp[3];   acc += bp[15-12];
895                         acc += bp[12-12];
896                         acc += bp[20-12];
897                         acc += bp[21-12];
898                         acc -= bp[14-12];
899                         acc -= bp[22-12];
900                         acc -= bp[23-12]; rp[3] = (unsigned int)acc; acc >>= 32;
901
902         acc += rp[4];   acc += bp[21-12];
903                         acc += bp[21-12];
904                         acc += bp[16-12];
905                         acc += bp[13-12];
906                         acc += bp[12-12];
907                         acc += bp[20-12];
908                         acc += bp[22-12];
909                         acc -= bp[15-12];
910                         acc -= bp[23-12];
911                         acc -= bp[23-12]; rp[4] = (unsigned int)acc; acc >>= 32;
912
913         acc += rp[5];   acc += bp[22-12];
914                         acc += bp[22-12];
915                         acc += bp[17-12];
916                         acc += bp[14-12];
917                         acc += bp[13-12];
918                         acc += bp[21-12];
919                         acc += bp[23-12];
920                         acc -= bp[16-12]; rp[5] = (unsigned int)acc; acc >>= 32;
921                         
922         acc += rp[6];   acc += bp[23-12];
923                         acc += bp[23-12];
924                         acc += bp[18-12];
925                         acc += bp[15-12];
926                         acc += bp[14-12];
927                         acc += bp[22-12];
928                         acc -= bp[17-12]; rp[6] = (unsigned int)acc; acc >>= 32;
929                         
930         acc += rp[7];   acc += bp[19-12];
931                         acc += bp[16-12];
932                         acc += bp[15-12];
933                         acc += bp[23-12];
934                         acc -= bp[18-12]; rp[7] = (unsigned int)acc; acc >>= 32;
935                         
936         acc += rp[8];   acc += bp[20-12];
937                         acc += bp[17-12];
938                         acc += bp[16-12];
939                         acc -= bp[19-12]; rp[8] = (unsigned int)acc; acc >>= 32;
940                         
941         acc += rp[9];   acc += bp[21-12];
942                         acc += bp[18-12];
943                         acc += bp[17-12];
944                         acc -= bp[20-12]; rp[9] = (unsigned int)acc; acc >>= 32;
945                         
946         acc += rp[10];  acc += bp[22-12];
947                         acc += bp[19-12];
948                         acc += bp[18-12];
949                         acc -= bp[21-12]; rp[10] = (unsigned int)acc; acc >>= 32;
950                         
951         acc += rp[11];  acc += bp[23-12];
952                         acc += bp[20-12];
953                         acc += bp[19-12];
954                         acc -= bp[22-12]; rp[11] = (unsigned int)acc;
955
956         carry = (int)(acc>>32);
957         }
958 #else
959         {
960         BN_ULONG t_d[BN_NIST_384_TOP];
961
962         /*S1*/
963         nist_set_256(t_d, buf, 0, 0, 0, 0, 0, 23-4, 22-4, 21-4);
964                 /* left shift */
965                 {
966                 register BN_ULONG *ap,t,c;
967                 ap = t_d;
968                 c=0;
969                 for (i = 3; i != 0; --i)
970                         {
971                         t= *ap;
972                         *(ap++)=((t<<1)|c)&BN_MASK2;
973                         c=(t & BN_TBIT)?1:0;
974                         }
975                 *ap=c;
976                 }
977         carry = (int)bn_add_words(r_d+(128/BN_BITS2), r_d+(128/BN_BITS2), 
978                 t_d, BN_NIST_256_TOP);
979         /*S2 */
980         carry += (int)bn_add_words(r_d, r_d, buf, BN_NIST_384_TOP);
981         /*S3*/
982         nist_set_384(t_d,buf,20,19,18,17,16,15,14,13,12,23,22,21);
983         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
984         /*S4*/
985         nist_set_384(t_d,buf,19,18,17,16,15,14,13,12,20,0,23,0);
986         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
987         /*S5*/
988         nist_set_384(t_d, buf,0,0,0,0,23,22,21,20,0,0,0,0);
989         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
990         /*S6*/
991         nist_set_384(t_d,buf,0,0,0,0,0,0,23,22,21,0,0,20);
992         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
993         /*D1*/
994         nist_set_384(t_d,buf,22,21,20,19,18,17,16,15,14,13,12,23);
995         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP);
996         /*D2*/
997         nist_set_384(t_d,buf,0,0,0,0,0,0,0,23,22,21,20,0);
998         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP);
999         /*D3*/
1000         nist_set_384(t_d,buf,0,0,0,0,0,0,0,23,23,0,0,0);
1001         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1002
1003         }
1004 #endif
1005         /* see BN_nist_mod_224 for explanation */
1006         u.f = bn_sub_words;
1007         if (carry > 0)
1008                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_384[carry-1],BN_NIST_384_TOP);
1009         else if (carry < 0)
1010                 {
1011                 carry = (int)bn_add_words(r_d,r_d,_nist_p_384[-carry-1],BN_NIST_384_TOP);
1012                 mask = 0-(PTR_SIZE_INT)carry;
1013                 u.p = ((PTR_SIZE_INT)bn_sub_words&mask) |
1014                  ((PTR_SIZE_INT)bn_add_words&~mask);
1015                 }
1016         else
1017                 carry = 1;
1018
1019         mask  = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_384[0],BN_NIST_384_TOP);
1020         mask &= 0-(PTR_SIZE_INT)carry;
1021         res   = (BN_ULONG *)(((PTR_SIZE_INT)c_d&~mask) |
1022          ((PTR_SIZE_INT)r_d&mask));
1023         nist_cp_bn(r_d, res, BN_NIST_384_TOP);
1024         r->top = BN_NIST_384_TOP;
1025         bn_correct_top(r);
1026
1027         return 1;
1028         }
1029
1030 #define BN_NIST_521_RSHIFT      (521%BN_BITS2)
1031 #define BN_NIST_521_LSHIFT      (BN_BITS2-BN_NIST_521_RSHIFT)
1032 #define BN_NIST_521_TOP_MASK    ((BN_ULONG)BN_MASK2>>BN_NIST_521_LSHIFT)
1033
1034 int BN_nist_mod_521(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
1035         BN_CTX *ctx)
1036         {
1037         int     top = a->top, i;
1038         BN_ULONG *r_d, *a_d = a->d,
1039                  t_d[BN_NIST_521_TOP],
1040                  val,tmp,*res;
1041         PTR_SIZE_INT mask;
1042         static const BIGNUM _bignum_nist_p_521_sqr = {
1043                 (BN_ULONG *)_nist_p_521_sqr,
1044                 sizeof(_nist_p_521_sqr)/sizeof(_nist_p_521_sqr[0]),
1045                 sizeof(_nist_p_521_sqr)/sizeof(_nist_p_521_sqr[0]),
1046                 0,BN_FLG_STATIC_DATA };
1047
1048         field = &_bignum_nist_p_521; /* just to make sure */
1049
1050         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_521_sqr)>=0)
1051                 return BN_nnmod(r, a, field, ctx);
1052
1053         i = BN_ucmp(field, a);
1054         if (i == 0)
1055                 {
1056                 BN_zero(r);
1057                 return 1;
1058                 }
1059         else if (i > 0)
1060                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
1061
1062         if (r != a)
1063                 {
1064                 if (!bn_wexpand(r,BN_NIST_521_TOP))
1065                         return 0;
1066                 r_d = r->d;
1067                 nist_cp_bn(r_d,a_d, BN_NIST_521_TOP);
1068                 }
1069         else
1070                 r_d = a_d;
1071
1072         /* upper 521 bits, copy ... */
1073         nist_cp_bn_0(t_d,a_d + (BN_NIST_521_TOP-1), top - (BN_NIST_521_TOP-1),BN_NIST_521_TOP);
1074         /* ... and right shift */
1075         for (val=t_d[0],i=0; i<BN_NIST_521_TOP-1; i++)
1076                 {
1077                 tmp = val>>BN_NIST_521_RSHIFT;
1078                 val = t_d[i+1];
1079                 t_d[i] = (tmp | val<<BN_NIST_521_LSHIFT) & BN_MASK2;
1080                 }
1081         t_d[i] = val>>BN_NIST_521_RSHIFT;
1082         /* lower 521 bits */
1083         r_d[i] &= BN_NIST_521_TOP_MASK;
1084
1085         bn_add_words(r_d,r_d,t_d,BN_NIST_521_TOP);
1086         mask = 0-(PTR_SIZE_INT)bn_sub_words(t_d,r_d,_nist_p_521,BN_NIST_521_TOP);
1087         res  = (BN_ULONG *)(((PTR_SIZE_INT)t_d&~mask) |
1088          ((PTR_SIZE_INT)r_d&mask));
1089         nist_cp_bn(r_d,res,BN_NIST_521_TOP);
1090         r->top = BN_NIST_521_TOP;
1091         bn_correct_top(r);
1092
1093         return 1;
1094         }
1095
1096 int (*BN_nist_mod_func(const BIGNUM *p))(BIGNUM *r, const BIGNUM *a, const BIGNUM *field, BN_CTX *ctx)
1097         {
1098         if (BN_ucmp(&_bignum_nist_p_192, p) == 0)
1099                 return BN_nist_mod_192;
1100         if (BN_ucmp(&_bignum_nist_p_224, p) == 0)
1101                 return BN_nist_mod_224;
1102         if (BN_ucmp(&_bignum_nist_p_256, p) == 0)
1103                 return BN_nist_mod_256;
1104         if (BN_ucmp(&_bignum_nist_p_384, p) == 0)
1105                 return BN_nist_mod_384;
1106         if (BN_ucmp(&_bignum_nist_p_521, p) == 0)
1107                 return BN_nist_mod_521;
1108         return 0;
1109         }