cf6013a6099b1f7577ddf7f8f8525d51d69c147d
[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 __fips_constseg
72 static const BN_ULONG _nist_p_192[][BN_NIST_192_TOP] = {
73         {0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFEULL,0xFFFFFFFFFFFFFFFFULL},
74         {0xFFFFFFFFFFFFFFFEULL,0xFFFFFFFFFFFFFFFDULL,0xFFFFFFFFFFFFFFFFULL},
75         {0xFFFFFFFFFFFFFFFDULL,0xFFFFFFFFFFFFFFFCULL,0xFFFFFFFFFFFFFFFFULL}
76         };
77 __fips_constseg
78 static const BN_ULONG _nist_p_192_sqr[] = {
79         0x0000000000000001ULL,0x0000000000000002ULL,0x0000000000000001ULL,
80         0xFFFFFFFFFFFFFFFEULL,0xFFFFFFFFFFFFFFFDULL,0xFFFFFFFFFFFFFFFFULL
81         };
82 __fips_constseg
83 static const BN_ULONG _nist_p_224[][BN_NIST_224_TOP] = {
84         {0x0000000000000001ULL,0xFFFFFFFF00000000ULL,
85          0xFFFFFFFFFFFFFFFFULL,0x00000000FFFFFFFFULL},
86         {0x0000000000000002ULL,0xFFFFFFFE00000000ULL,
87          0xFFFFFFFFFFFFFFFFULL,0x00000001FFFFFFFFULL} /* this one is "carry-full" */
88         };
89 __fips_constseg
90 static const BN_ULONG _nist_p_224_sqr[] = {
91         0x0000000000000001ULL,0xFFFFFFFE00000000ULL,
92         0xFFFFFFFFFFFFFFFFULL,0x0000000200000000ULL,
93         0x0000000000000000ULL,0xFFFFFFFFFFFFFFFEULL,
94         0xFFFFFFFFFFFFFFFFULL
95         };
96 __fips_constseg
97 static const BN_ULONG _nist_p_256[][BN_NIST_256_TOP] = {
98         {0xFFFFFFFFFFFFFFFFULL,0x00000000FFFFFFFFULL,
99          0x0000000000000000ULL,0xFFFFFFFF00000001ULL},
100         {0xFFFFFFFFFFFFFFFEULL,0x00000001FFFFFFFFULL,
101          0x0000000000000000ULL,0xFFFFFFFE00000002ULL},
102         {0xFFFFFFFFFFFFFFFDULL,0x00000002FFFFFFFFULL,
103          0x0000000000000000ULL,0xFFFFFFFD00000003ULL},
104         {0xFFFFFFFFFFFFFFFCULL,0x00000003FFFFFFFFULL,
105          0x0000000000000000ULL,0xFFFFFFFC00000004ULL},
106         {0xFFFFFFFFFFFFFFFBULL,0x00000004FFFFFFFFULL,
107          0x0000000000000000ULL,0xFFFFFFFB00000005ULL},
108         };
109 __fips_constseg
110 static const BN_ULONG _nist_p_256_sqr[] = {
111         0x0000000000000001ULL,0xFFFFFFFE00000000ULL,
112         0xFFFFFFFFFFFFFFFFULL,0x00000001FFFFFFFEULL,
113         0x00000001FFFFFFFEULL,0x00000001FFFFFFFEULL,
114         0xFFFFFFFE00000001ULL,0xFFFFFFFE00000002ULL
115         };
116 __fips_constseg
117 static const BN_ULONG _nist_p_384[][BN_NIST_384_TOP] = {
118         {0x00000000FFFFFFFFULL,0xFFFFFFFF00000000ULL,0xFFFFFFFFFFFFFFFEULL,
119          0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL},
120         {0x00000001FFFFFFFEULL,0xFFFFFFFE00000000ULL,0xFFFFFFFFFFFFFFFDULL,
121          0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL},
122         {0x00000002FFFFFFFDULL,0xFFFFFFFD00000000ULL,0xFFFFFFFFFFFFFFFCULL,
123          0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL},
124         {0x00000003FFFFFFFCULL,0xFFFFFFFC00000000ULL,0xFFFFFFFFFFFFFFFBULL,
125          0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL},
126         {0x00000004FFFFFFFBULL,0xFFFFFFFB00000000ULL,0xFFFFFFFFFFFFFFFAULL,
127          0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL},
128         };
129 __fips_constseg
130 static const BN_ULONG _nist_p_384_sqr[] = {
131         0xFFFFFFFE00000001ULL,0x0000000200000000ULL,0xFFFFFFFE00000000ULL,
132         0x0000000200000000ULL,0x0000000000000001ULL,0x0000000000000000ULL,
133         0x00000001FFFFFFFEULL,0xFFFFFFFE00000000ULL,0xFFFFFFFFFFFFFFFDULL,
134         0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL
135         };
136 __fips_constseg
137 static const BN_ULONG _nist_p_521[] =
138         {0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,
139         0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,
140         0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,
141         0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,
142         0x00000000000001FFULL};
143 __fips_constseg
144 static const BN_ULONG _nist_p_521_sqr[] = {
145         0x0000000000000001ULL,0x0000000000000000ULL,0x0000000000000000ULL,
146         0x0000000000000000ULL,0x0000000000000000ULL,0x0000000000000000ULL,
147         0x0000000000000000ULL,0x0000000000000000ULL,0xFFFFFFFFFFFFFC00ULL,
148         0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,
149         0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,0xFFFFFFFFFFFFFFFFULL,
150         0xFFFFFFFFFFFFFFFFULL,0x000000000003FFFFULL
151         };
152 #elif BN_BITS2 == 32
153 __fips_constseg
154 static const BN_ULONG _nist_p_192[][BN_NIST_192_TOP] = {
155         {0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFE,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
156         {0xFFFFFFFE,0xFFFFFFFF,0xFFFFFFFD,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
157         {0xFFFFFFFD,0xFFFFFFFF,0xFFFFFFFC,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF}
158         };
159 __fips_constseg
160 static const BN_ULONG _nist_p_192_sqr[] = {
161         0x00000001,0x00000000,0x00000002,0x00000000,0x00000001,0x00000000,
162         0xFFFFFFFE,0xFFFFFFFF,0xFFFFFFFD,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF
163         };
164 __fips_constseg
165 static const BN_ULONG _nist_p_224[][BN_NIST_224_TOP] = {
166         {0x00000001,0x00000000,0x00000000,0xFFFFFFFF,
167          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
168         {0x00000002,0x00000000,0x00000000,0xFFFFFFFE,
169          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF}
170         };
171 __fips_constseg
172 static const BN_ULONG _nist_p_224_sqr[] = {
173         0x00000001,0x00000000,0x00000000,0xFFFFFFFE,
174         0xFFFFFFFF,0xFFFFFFFF,0x00000000,0x00000002,
175         0x00000000,0x00000000,0xFFFFFFFE,0xFFFFFFFF,
176         0xFFFFFFFF,0xFFFFFFFF
177         };
178 __fips_constseg
179 static const BN_ULONG _nist_p_256[][BN_NIST_256_TOP] = {
180         {0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0x00000000,
181          0x00000000,0x00000000,0x00000001,0xFFFFFFFF},
182         {0xFFFFFFFE,0xFFFFFFFF,0xFFFFFFFF,0x00000001,
183          0x00000000,0x00000000,0x00000002,0xFFFFFFFE},
184         {0xFFFFFFFD,0xFFFFFFFF,0xFFFFFFFF,0x00000002,
185          0x00000000,0x00000000,0x00000003,0xFFFFFFFD},
186         {0xFFFFFFFC,0xFFFFFFFF,0xFFFFFFFF,0x00000003,
187          0x00000000,0x00000000,0x00000004,0xFFFFFFFC},
188         {0xFFFFFFFB,0xFFFFFFFF,0xFFFFFFFF,0x00000004,
189          0x00000000,0x00000000,0x00000005,0xFFFFFFFB},
190         };
191 __fips_constseg
192 static const BN_ULONG _nist_p_256_sqr[] = {
193         0x00000001,0x00000000,0x00000000,0xFFFFFFFE,
194         0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFE,0x00000001,
195         0xFFFFFFFE,0x00000001,0xFFFFFFFE,0x00000001,
196         0x00000001,0xFFFFFFFE,0x00000002,0xFFFFFFFE
197         };
198 __fips_constseg
199 static const BN_ULONG _nist_p_384[][BN_NIST_384_TOP] = {
200         {0xFFFFFFFF,0x00000000,0x00000000,0xFFFFFFFF,0xFFFFFFFE,0xFFFFFFFF,
201          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
202         {0xFFFFFFFE,0x00000001,0x00000000,0xFFFFFFFE,0xFFFFFFFD,0xFFFFFFFF,
203          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
204         {0xFFFFFFFD,0x00000002,0x00000000,0xFFFFFFFD,0xFFFFFFFC,0xFFFFFFFF,
205          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
206         {0xFFFFFFFC,0x00000003,0x00000000,0xFFFFFFFC,0xFFFFFFFB,0xFFFFFFFF,
207          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
208         {0xFFFFFFFB,0x00000004,0x00000000,0xFFFFFFFB,0xFFFFFFFA,0xFFFFFFFF,
209          0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF},
210         };
211 __fips_constseg
212 static const BN_ULONG _nist_p_384_sqr[] = {
213         0x00000001,0xFFFFFFFE,0x00000000,0x00000002,0x00000000,0xFFFFFFFE,
214         0x00000000,0x00000002,0x00000001,0x00000000,0x00000000,0x00000000,
215         0xFFFFFFFE,0x00000001,0x00000000,0xFFFFFFFE,0xFFFFFFFD,0xFFFFFFFF,
216         0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF
217         };
218 __fips_constseg
219 static const BN_ULONG _nist_p_521[] = {0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,
220         0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,
221         0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,
222         0xFFFFFFFF,0x000001FF};
223 __fips_constseg
224 static const BN_ULONG _nist_p_521_sqr[] = {
225         0x00000001,0x00000000,0x00000000,0x00000000,0x00000000,0x00000000,
226         0x00000000,0x00000000,0x00000000,0x00000000,0x00000000,0x00000000,
227         0x00000000,0x00000000,0x00000000,0x00000000,0xFFFFFC00,0xFFFFFFFF,
228         0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,
229         0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,0xFFFFFFFF,
230         0xFFFFFFFF,0xFFFFFFFF,0x0003FFFF
231         };
232 #else
233 #error "unsupported BN_BITS2"
234 #endif
235
236
237 static const BIGNUM _bignum_nist_p_192 =
238         {
239         (BN_ULONG *)_nist_p_192[0],
240         BN_NIST_192_TOP,
241         BN_NIST_192_TOP,
242         0,
243         BN_FLG_STATIC_DATA
244         };
245
246 static const BIGNUM _bignum_nist_p_224 =
247         {
248         (BN_ULONG *)_nist_p_224[0],
249         BN_NIST_224_TOP,
250         BN_NIST_224_TOP,
251         0,
252         BN_FLG_STATIC_DATA
253         };
254
255 static const BIGNUM _bignum_nist_p_256 =
256         {
257         (BN_ULONG *)_nist_p_256[0],
258         BN_NIST_256_TOP,
259         BN_NIST_256_TOP,
260         0,
261         BN_FLG_STATIC_DATA
262         };
263
264 static const BIGNUM _bignum_nist_p_384 =
265         {
266         (BN_ULONG *)_nist_p_384[0],
267         BN_NIST_384_TOP,
268         BN_NIST_384_TOP,
269         0,
270         BN_FLG_STATIC_DATA
271         };
272
273 static const BIGNUM _bignum_nist_p_521 =
274         {
275         (BN_ULONG *)_nist_p_521,
276         BN_NIST_521_TOP,
277         BN_NIST_521_TOP,
278         0,
279         BN_FLG_STATIC_DATA
280         };
281
282
283 const BIGNUM *BN_get0_nist_prime_192(void)
284         {
285         return &_bignum_nist_p_192;
286         }
287
288 const BIGNUM *BN_get0_nist_prime_224(void)
289         {
290         return &_bignum_nist_p_224;
291         }
292
293 const BIGNUM *BN_get0_nist_prime_256(void)
294         {
295         return &_bignum_nist_p_256;
296         }
297
298 const BIGNUM *BN_get0_nist_prime_384(void)
299         {
300         return &_bignum_nist_p_384;
301         }
302
303 const BIGNUM *BN_get0_nist_prime_521(void)
304         {
305         return &_bignum_nist_p_521;
306         }
307
308
309 static void nist_cp_bn_0(BN_ULONG *buf, BN_ULONG *a, int top, int max)
310         {
311         int i;
312         BN_ULONG *_tmp1 = (buf), *_tmp2 = (a);
313
314 #ifdef BN_DEBUG
315         OPENSSL_assert(top <= max);
316 #endif
317         for (i = (top); i != 0; i--)
318                 *_tmp1++ = *_tmp2++;
319         for (i = (max) - (top); i != 0; i--)
320                 *_tmp1++ = (BN_ULONG) 0;
321         }
322
323 static void nist_cp_bn(BN_ULONG *buf, BN_ULONG *a, int top)
324         { 
325         int i;
326         BN_ULONG *_tmp1 = (buf), *_tmp2 = (a);
327         for (i = (top); i != 0; i--)
328                 *_tmp1++ = *_tmp2++;
329         }
330
331 #if BN_BITS2 == 64
332 #define bn_cp_64(to, n, from, m)        (to)[n] = (m>=0)?((from)[m]):0;
333 #define bn_64_set_0(to, n)              (to)[n] = (BN_ULONG)0;
334 /*
335  * two following macros are implemented under assumption that they
336  * are called in a sequence with *ascending* n, i.e. as they are...
337  */
338 #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))\
339                                                 :(to[(n)/2] =((m)&1)?(from[(m)/2]>>32):(from[(m)/2]&BN_MASK2l)))
340 #define bn_32_set_0(to, n)              (((n)&1)?(to[(n)/2]&=BN_MASK2l):(to[(n)/2]=0));
341 #define bn_cp_32(to,n,from,m)           ((m)>=0)?bn_cp_32_naked(to,n,from,m):bn_32_set_0(to,n)
342 # if defined(L_ENDIAN)
343 #  if defined(__arch64__)
344 #   define NIST_INT64 long
345 #  else
346 #   define NIST_INT64 long long
347 #  endif
348 # endif
349 #else
350 #define bn_cp_64(to, n, from, m) \
351         { \
352         bn_cp_32(to, (n)*2, from, (m)*2); \
353         bn_cp_32(to, (n)*2+1, from, (m)*2+1); \
354         }
355 #define bn_64_set_0(to, n) \
356         { \
357         bn_32_set_0(to, (n)*2); \
358         bn_32_set_0(to, (n)*2+1); \
359         }
360 #define bn_cp_32(to, n, from, m)        (to)[n] = (m>=0)?((from)[m]):0;
361 #define bn_32_set_0(to, n)              (to)[n] = (BN_ULONG)0;
362 # if defined(_WIN32) && !defined(__GNUC__)
363 #  define NIST_INT64 __int64
364 # else
365 #  define NIST_INT64 long long
366 # endif
367 #endif /* BN_BITS2 != 64 */
368
369 #define nist_set_192(to, from, a1, a2, a3) \
370         { \
371         bn_cp_64(to, 0, from, (a3) - 3) \
372         bn_cp_64(to, 1, from, (a2) - 3) \
373         bn_cp_64(to, 2, from, (a1) - 3) \
374         }
375
376 int BN_nist_mod_192(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
377         BN_CTX *ctx)
378         {
379         int      top = a->top, i;
380         int      carry;
381         register BN_ULONG *r_d, *a_d = a->d;
382         union   {
383                 BN_ULONG        bn[BN_NIST_192_TOP];
384                 unsigned int    ui[BN_NIST_192_TOP*sizeof(BN_ULONG)/sizeof(unsigned int)];
385                 } buf;
386         BN_ULONG c_d[BN_NIST_192_TOP],
387                 *res;
388         PTR_SIZE_INT mask;
389         static const BIGNUM _bignum_nist_p_192_sqr = {
390                 (BN_ULONG *)_nist_p_192_sqr,
391                 sizeof(_nist_p_192_sqr)/sizeof(_nist_p_192_sqr[0]),
392                 sizeof(_nist_p_192_sqr)/sizeof(_nist_p_192_sqr[0]),
393                 0,BN_FLG_STATIC_DATA };
394
395         field = &_bignum_nist_p_192; /* just to make sure */
396
397         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_192_sqr)>=0)
398                 return BN_nnmod(r, a, field, ctx);
399
400         i = BN_ucmp(field, a);
401         if (i == 0)
402                 {
403                 BN_zero(r);
404                 return 1;
405                 }
406         else if (i > 0)
407                 return (r == a) ? 1 : (BN_copy(r ,a) != NULL);
408
409         if (r != a)
410                 {
411                 if (!bn_wexpand(r, BN_NIST_192_TOP))
412                         return 0;
413                 r_d = r->d;
414                 nist_cp_bn(r_d, a_d, BN_NIST_192_TOP);
415                 }
416         else
417                 r_d = a_d;
418
419         nist_cp_bn_0(buf.bn, a_d + BN_NIST_192_TOP, top - BN_NIST_192_TOP, BN_NIST_192_TOP);
420
421 #if defined(NIST_INT64)
422         {
423         NIST_INT64              acc;    /* accumulator */
424         unsigned int            *rp=(unsigned int *)r_d;
425         const unsigned int      *bp=(const unsigned int *)buf.ui;
426
427         acc  = rp[0];   acc += bp[3*2-6];
428                         acc += bp[5*2-6]; rp[0] = (unsigned int)acc; acc >>= 32;
429
430         acc += rp[1];   acc += bp[3*2-5];
431                         acc += bp[5*2-5]; rp[1] = (unsigned int)acc; acc >>= 32;
432
433         acc += rp[2];   acc += bp[3*2-6];
434                         acc += bp[4*2-6];
435                         acc += bp[5*2-6]; rp[2] = (unsigned int)acc; acc >>= 32;
436
437         acc += rp[3];   acc += bp[3*2-5];
438                         acc += bp[4*2-5];
439                         acc += bp[5*2-5]; rp[3] = (unsigned int)acc; acc >>= 32;
440
441         acc += rp[4];   acc += bp[4*2-6];
442                         acc += bp[5*2-6]; rp[4] = (unsigned int)acc; acc >>= 32;
443
444         acc += rp[5];   acc += bp[4*2-5];
445                         acc += bp[5*2-5]; rp[5] = (unsigned int)acc;
446
447         carry = (int)(acc>>32);
448         }
449 #else
450         {
451         BN_ULONG t_d[BN_NIST_192_TOP];
452
453         nist_set_192(t_d, buf.bn, 0, 3, 3);
454         carry = (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP);
455         nist_set_192(t_d, buf.bn, 4, 4, 0);
456         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP);
457         nist_set_192(t_d, buf.bn, 5, 5, 5)
458         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP);
459         }
460 #endif
461         if (carry > 0)
462                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_192[carry-1],BN_NIST_192_TOP);
463         else
464                 carry = 1;
465
466         /*
467          * we need 'if (carry==0 || result>=modulus) result-=modulus;'
468          * as comparison implies subtraction, we can write
469          * 'tmp=result-modulus; if (!carry || !borrow) result=tmp;'
470          * this is what happens below, but without explicit if:-) a.
471          */
472         mask  = 0-(PTR_SIZE_INT)bn_sub_words(c_d,r_d,_nist_p_192[0],BN_NIST_192_TOP);
473         mask &= 0-(PTR_SIZE_INT)carry;
474         res   = (BN_ULONG *)
475          (((PTR_SIZE_INT)c_d&~mask) | ((PTR_SIZE_INT)r_d&mask));
476         nist_cp_bn(r_d, res, BN_NIST_192_TOP);
477         r->top = BN_NIST_192_TOP;
478         bn_correct_top(r);
479
480         return 1;
481         }
482
483 typedef BN_ULONG (*bn_addsub_f)(BN_ULONG *,const BN_ULONG *,const BN_ULONG *,int);
484
485 #define nist_set_224(to, from, a1, a2, a3, a4, a5, a6, a7) \
486         { \
487         bn_cp_32(to, 0, from, (a7) - 7) \
488         bn_cp_32(to, 1, from, (a6) - 7) \
489         bn_cp_32(to, 2, from, (a5) - 7) \
490         bn_cp_32(to, 3, from, (a4) - 7) \
491         bn_cp_32(to, 4, from, (a3) - 7) \
492         bn_cp_32(to, 5, from, (a2) - 7) \
493         bn_cp_32(to, 6, from, (a1) - 7) \
494         }
495
496 int BN_nist_mod_224(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
497         BN_CTX *ctx)
498         {
499         int     top = a->top, i;
500         int     carry;
501         BN_ULONG *r_d, *a_d = a->d;
502         BN_ULONG buf[BN_NIST_224_TOP],
503                  c_d[BN_NIST_224_TOP],
504                 *res;
505         PTR_SIZE_INT mask;
506         union { bn_addsub_f f; PTR_SIZE_INT p; } u;
507         static const BIGNUM _bignum_nist_p_224_sqr = {
508                 (BN_ULONG *)_nist_p_224_sqr,
509                 sizeof(_nist_p_224_sqr)/sizeof(_nist_p_224_sqr[0]),
510                 sizeof(_nist_p_224_sqr)/sizeof(_nist_p_224_sqr[0]),
511                 0,BN_FLG_STATIC_DATA };
512
513
514         field = &_bignum_nist_p_224; /* just to make sure */
515
516         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_224_sqr)>=0)
517                 return BN_nnmod(r, a, field, ctx);
518
519         i = BN_ucmp(field, a);
520         if (i == 0)
521                 {
522                 BN_zero(r);
523                 return 1;
524                 }
525         else if (i > 0)
526                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
527
528         if (r != a)
529                 {
530                 if (!bn_wexpand(r, BN_NIST_224_TOP))
531                         return 0;
532                 r_d = r->d;
533                 nist_cp_bn(r_d, a_d, BN_NIST_224_TOP);
534                 }
535         else
536                 r_d = a_d;
537
538 #if BN_BITS2==64
539         /* copy upper 256 bits of 448 bit number ... */
540         nist_cp_bn_0(c_d, a_d + (BN_NIST_224_TOP-1), top - (BN_NIST_224_TOP-1), BN_NIST_224_TOP);
541         /* ... and right shift by 32 to obtain upper 224 bits */
542         nist_set_224(buf, c_d, 14, 13, 12, 11, 10, 9, 8);
543         /* truncate lower part to 224 bits too */
544         r_d[BN_NIST_224_TOP-1] &= BN_MASK2l;
545 #else
546         nist_cp_bn_0(buf, a_d + BN_NIST_224_TOP, top - BN_NIST_224_TOP, BN_NIST_224_TOP);
547 #endif
548
549 #if defined(NIST_INT64) && BN_BITS2!=64
550         {
551         NIST_INT64              acc;    /* accumulator */
552         unsigned int            *rp=(unsigned int *)r_d;
553         const unsigned int      *bp=(const unsigned int *)buf;
554
555         acc  = rp[0];   acc -= bp[7-7];
556                         acc -= bp[11-7]; rp[0] = (unsigned int)acc; acc >>= 32;
557
558         acc += rp[1];   acc -= bp[8-7];
559                         acc -= bp[12-7]; rp[1] = (unsigned int)acc; acc >>= 32;
560
561         acc += rp[2];   acc -= bp[9-7];
562                         acc -= bp[13-7]; rp[2] = (unsigned int)acc; acc >>= 32;
563
564         acc += rp[3];   acc += bp[7-7];
565                         acc += bp[11-7];
566                         acc -= bp[10-7]; rp[3] = (unsigned int)acc; acc>>= 32;
567
568         acc += rp[4];   acc += bp[8-7];
569                         acc += bp[12-7];
570                         acc -= bp[11-7]; rp[4] = (unsigned int)acc; acc >>= 32;
571
572         acc += rp[5];   acc += bp[9-7];
573                         acc += bp[13-7];
574                         acc -= bp[12-7]; rp[5] = (unsigned int)acc; acc >>= 32;
575
576         acc += rp[6];   acc += bp[10-7];
577                         acc -= bp[13-7]; rp[6] = (unsigned int)acc;
578
579         carry = (int)(acc>>32);
580 # if BN_BITS2==64
581         rp[7] = carry;
582 # endif
583         }       
584 #else
585         {
586         BN_ULONG t_d[BN_NIST_224_TOP];
587
588         nist_set_224(t_d, buf, 10, 9, 8, 7, 0, 0, 0);
589         carry = (int)bn_add_words(r_d, r_d, t_d, BN_NIST_224_TOP);
590         nist_set_224(t_d, buf, 0, 13, 12, 11, 0, 0, 0);
591         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_224_TOP);
592         nist_set_224(t_d, buf, 13, 12, 11, 10, 9, 8, 7);
593         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_224_TOP);
594         nist_set_224(t_d, buf, 0, 0, 0, 0, 13, 12, 11);
595         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_224_TOP);
596
597 #if BN_BITS2==64
598         carry = (int)(r_d[BN_NIST_224_TOP-1]>>32);
599 #endif
600         }
601 #endif
602         u.f = bn_sub_words;
603         if (carry > 0)
604                 {
605                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_224[carry-1],BN_NIST_224_TOP);
606 #if BN_BITS2==64
607                 carry=(int)(~(r_d[BN_NIST_224_TOP-1]>>32))&1;
608 #endif
609                 }
610         else if (carry < 0)
611                 {
612                 /* it's a bit more comlicated logic in this case.
613                  * if bn_add_words yields no carry, then result
614                  * has to be adjusted by unconditionally *adding*
615                  * the modulus. but if it does, then result has
616                  * to be compared to the modulus and conditionally
617                  * adjusted by *subtracting* the latter. */
618                 carry = (int)bn_add_words(r_d,r_d,_nist_p_224[-carry-1],BN_NIST_224_TOP);
619                 mask = 0-(PTR_SIZE_INT)carry;
620                 u.p = ((PTR_SIZE_INT)bn_sub_words&mask) |
621                  ((PTR_SIZE_INT)bn_add_words&~mask);
622                 }
623         else
624                 carry = 1;
625
626         /* otherwise it's effectively same as in BN_nist_mod_192... */
627         mask  = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_224[0],BN_NIST_224_TOP);
628         mask &= 0-(PTR_SIZE_INT)carry;
629         res   = (BN_ULONG *)(((PTR_SIZE_INT)c_d&~mask) |
630          ((PTR_SIZE_INT)r_d&mask));
631         nist_cp_bn(r_d, res, BN_NIST_224_TOP);
632         r->top = BN_NIST_224_TOP;
633         bn_correct_top(r);
634
635         return 1;
636         }
637
638 #define nist_set_256(to, from, a1, a2, a3, a4, a5, a6, a7, a8) \
639         { \
640         bn_cp_32(to, 0, from, (a8) - 8) \
641         bn_cp_32(to, 1, from, (a7) - 8) \
642         bn_cp_32(to, 2, from, (a6) - 8) \
643         bn_cp_32(to, 3, from, (a5) - 8) \
644         bn_cp_32(to, 4, from, (a4) - 8) \
645         bn_cp_32(to, 5, from, (a3) - 8) \
646         bn_cp_32(to, 6, from, (a2) - 8) \
647         bn_cp_32(to, 7, from, (a1) - 8) \
648         }
649
650 int BN_nist_mod_256(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
651         BN_CTX *ctx)
652         {
653         int     i, top = a->top;
654         int     carry = 0;
655         register BN_ULONG *a_d = a->d, *r_d;
656         union   {
657                 BN_ULONG bn[BN_NIST_256_TOP];
658                 unsigned int ui[BN_NIST_256_TOP*sizeof(BN_ULONG)/sizeof(unsigned int)];
659                 } buf;
660         BN_ULONG c_d[BN_NIST_256_TOP],
661                 *res;
662         PTR_SIZE_INT mask;
663         union { bn_addsub_f f; PTR_SIZE_INT p; } u;
664         static const BIGNUM _bignum_nist_p_256_sqr = {
665                 (BN_ULONG *)_nist_p_256_sqr,
666                 sizeof(_nist_p_256_sqr)/sizeof(_nist_p_256_sqr[0]),
667                 sizeof(_nist_p_256_sqr)/sizeof(_nist_p_256_sqr[0]),
668                 0,BN_FLG_STATIC_DATA };
669
670         field = &_bignum_nist_p_256; /* just to make sure */
671
672         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_256_sqr)>=0)
673                 return BN_nnmod(r, a, field, ctx);
674
675         i = BN_ucmp(field, a);
676         if (i == 0)
677                 {
678                 BN_zero(r);
679                 return 1;
680                 }
681         else if (i > 0)
682                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
683
684         if (r != a)
685                 {
686                 if (!bn_wexpand(r, BN_NIST_256_TOP))
687                         return 0;
688                 r_d = r->d;
689                 nist_cp_bn(r_d, a_d, BN_NIST_256_TOP);
690                 }
691         else
692                 r_d = a_d;
693
694         nist_cp_bn_0(buf.bn, a_d + BN_NIST_256_TOP, top - BN_NIST_256_TOP, BN_NIST_256_TOP);
695
696 #if defined(NIST_INT64)
697         {
698         NIST_INT64              acc;    /* accumulator */
699         unsigned int            *rp=(unsigned int *)r_d;
700         const unsigned int      *bp=(const unsigned int *)buf.ui;
701
702         acc = rp[0];    acc += bp[8-8];
703                         acc += bp[9-8];
704                         acc -= bp[11-8];
705                         acc -= bp[12-8];
706                         acc -= bp[13-8];
707                         acc -= bp[14-8]; rp[0] = (unsigned int)acc; acc >>= 32;
708
709         acc += rp[1];   acc += bp[9-8];
710                         acc += bp[10-8];
711                         acc -= bp[12-8];
712                         acc -= bp[13-8];
713                         acc -= bp[14-8];
714                         acc -= bp[15-8]; rp[1] = (unsigned int)acc; acc >>= 32;
715
716         acc += rp[2];   acc += bp[10-8];
717                         acc += bp[11-8];
718                         acc -= bp[13-8];
719                         acc -= bp[14-8];
720                         acc -= bp[15-8]; rp[2] = (unsigned int)acc; acc >>= 32;
721
722         acc += rp[3];   acc += bp[11-8];
723                         acc += bp[11-8];
724                         acc += bp[12-8];
725                         acc += bp[12-8];
726                         acc += bp[13-8];
727                         acc -= bp[15-8];
728                         acc -= bp[8-8];
729                         acc -= bp[9-8];  rp[3] = (unsigned int)acc; acc >>= 32;
730
731         acc += rp[4];   acc += bp[12-8];
732                         acc += bp[12-8];
733                         acc += bp[13-8];
734                         acc += bp[13-8];
735                         acc += bp[14-8];
736                         acc -= bp[9-8];
737                         acc -= bp[10-8]; rp[4] = (unsigned int)acc; acc >>= 32;
738
739         acc += rp[5];   acc += bp[13-8];
740                         acc += bp[13-8];
741                         acc += bp[14-8];
742                         acc += bp[14-8];
743                         acc += bp[15-8];
744                         acc -= bp[10-8];
745                         acc -= bp[11-8]; rp[5] = (unsigned int)acc; acc >>= 32;
746
747         acc += rp[6];   acc += bp[14-8];
748                         acc += bp[14-8];
749                         acc += bp[15-8];
750                         acc += bp[15-8];
751                         acc += bp[14-8];
752                         acc += bp[13-8];
753                         acc -= bp[8-8];
754                         acc -= bp[9-8];  rp[6] = (unsigned int)acc; acc >>= 32;
755
756         acc += rp[7];   acc += bp[15-8];
757                         acc += bp[15-8];
758                         acc += bp[15-8];
759                         acc += bp[8 -8];
760                         acc -= bp[10-8];
761                         acc -= bp[11-8];
762                         acc -= bp[12-8];
763                         acc -= bp[13-8]; rp[7] = (unsigned int)acc;
764
765         carry = (int)(acc>>32);
766         }
767 #else
768         {
769         BN_ULONG t_d[BN_NIST_256_TOP];
770
771         /*S1*/
772         nist_set_256(t_d, buf.bn, 15, 14, 13, 12, 11, 0, 0, 0);
773         /*S2*/
774         nist_set_256(c_d, buf.bn, 0, 15, 14, 13, 12, 0, 0, 0);
775         carry = (int)bn_add_words(t_d, t_d, c_d, BN_NIST_256_TOP);
776         /* left shift */
777                 {
778                 register BN_ULONG *ap,t,c;
779                 ap = t_d;
780                 c=0;
781                 for (i = BN_NIST_256_TOP; i != 0; --i)
782                         {
783                         t= *ap;
784                         *(ap++)=((t<<1)|c)&BN_MASK2;
785                         c=(t & BN_TBIT)?1:0;
786                         }
787                 carry <<= 1;
788                 carry  |= c;
789                 }
790         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP);
791         /*S3*/
792         nist_set_256(t_d, buf.bn, 15, 14, 0, 0, 0, 10, 9, 8);
793         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP);
794         /*S4*/
795         nist_set_256(t_d, buf.bn, 8, 13, 15, 14, 13, 11, 10, 9);
796         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP);
797         /*D1*/
798         nist_set_256(t_d, buf.bn, 10, 8, 0, 0, 0, 13, 12, 11);
799         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
800         /*D2*/
801         nist_set_256(t_d, buf.bn, 11, 9, 0, 0, 15, 14, 13, 12);
802         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
803         /*D3*/
804         nist_set_256(t_d, buf.bn, 12, 0, 10, 9, 8, 15, 14, 13);
805         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
806         /*D4*/
807         nist_set_256(t_d, buf.bn, 13, 0, 11, 10, 9, 0, 15, 14);
808         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
809
810         }
811 #endif
812         /* see BN_nist_mod_224 for explanation */
813         u.f = bn_sub_words;
814         if (carry > 0)
815                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_256[carry-1],BN_NIST_256_TOP);
816         else if (carry < 0)
817                 {
818                 carry = (int)bn_add_words(r_d,r_d,_nist_p_256[-carry-1],BN_NIST_256_TOP);
819                 mask = 0-(PTR_SIZE_INT)carry;
820                 u.p = ((PTR_SIZE_INT)bn_sub_words&mask) |
821                  ((PTR_SIZE_INT)bn_add_words&~mask);
822                 }
823         else
824                 carry = 1;
825
826         mask  = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_256[0],BN_NIST_256_TOP);
827         mask &= 0-(PTR_SIZE_INT)carry;
828         res   = (BN_ULONG *)(((PTR_SIZE_INT)c_d&~mask) |
829          ((PTR_SIZE_INT)r_d&mask));
830         nist_cp_bn(r_d, res, BN_NIST_256_TOP);
831         r->top = BN_NIST_256_TOP;
832         bn_correct_top(r);
833
834         return 1;
835         }
836
837 #define nist_set_384(to,from,a1,a2,a3,a4,a5,a6,a7,a8,a9,a10,a11,a12) \
838         { \
839         bn_cp_32(to, 0, from,  (a12) - 12) \
840         bn_cp_32(to, 1, from,  (a11) - 12) \
841         bn_cp_32(to, 2, from,  (a10) - 12) \
842         bn_cp_32(to, 3, from,  (a9) - 12)  \
843         bn_cp_32(to, 4, from,  (a8) - 12)  \
844         bn_cp_32(to, 5, from,  (a7) - 12)  \
845         bn_cp_32(to, 6, from,  (a6) - 12)  \
846         bn_cp_32(to, 7, from,  (a5) - 12)  \
847         bn_cp_32(to, 8, from,  (a4) - 12)  \
848         bn_cp_32(to, 9, from,  (a3) - 12)  \
849         bn_cp_32(to, 10, from, (a2) - 12)  \
850         bn_cp_32(to, 11, from, (a1) - 12)  \
851         }
852
853 int BN_nist_mod_384(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
854         BN_CTX *ctx)
855         {
856         int     i, top = a->top;
857         int     carry = 0;
858         register BN_ULONG *r_d, *a_d = a->d;
859         union   {
860                 BN_ULONG bn[BN_NIST_384_TOP];
861                 unsigned int ui[BN_NIST_384_TOP*sizeof(BN_ULONG)/sizeof(unsigned int)];
862                 } buf;
863         BN_ULONG c_d[BN_NIST_384_TOP],
864                 *res;
865         PTR_SIZE_INT mask;
866         union { bn_addsub_f f; PTR_SIZE_INT p; } u;
867         static const BIGNUM _bignum_nist_p_384_sqr = {
868                 (BN_ULONG *)_nist_p_384_sqr,
869                 sizeof(_nist_p_384_sqr)/sizeof(_nist_p_384_sqr[0]),
870                 sizeof(_nist_p_384_sqr)/sizeof(_nist_p_384_sqr[0]),
871                 0,BN_FLG_STATIC_DATA };
872
873
874         field = &_bignum_nist_p_384; /* just to make sure */
875
876         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_384_sqr)>=0)
877                 return BN_nnmod(r, a, field, ctx);
878
879         i = BN_ucmp(field, a);
880         if (i == 0)
881                 {
882                 BN_zero(r);
883                 return 1;
884                 }
885         else if (i > 0)
886                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
887
888         if (r != a)
889                 {
890                 if (!bn_wexpand(r, BN_NIST_384_TOP))
891                         return 0;
892                 r_d = r->d;
893                 nist_cp_bn(r_d, a_d, BN_NIST_384_TOP);
894                 }
895         else
896                 r_d = a_d;
897
898         nist_cp_bn_0(buf.bn, a_d + BN_NIST_384_TOP, top - BN_NIST_384_TOP, BN_NIST_384_TOP);
899
900 #if defined(NIST_INT64)
901         {
902         NIST_INT64              acc;    /* accumulator */
903         unsigned int            *rp=(unsigned int *)r_d;
904         const unsigned int      *bp=(const unsigned int *)buf.ui;
905
906         acc = rp[0];    acc += bp[12-12];
907                         acc += bp[21-12];
908                         acc += bp[20-12];
909                         acc -= bp[23-12]; rp[0] = (unsigned int)acc; acc >>= 32;
910
911         acc += rp[1];   acc += bp[13-12];
912                         acc += bp[22-12];
913                         acc += bp[23-12];
914                         acc -= bp[12-12];
915                         acc -= bp[20-12]; rp[1] = (unsigned int)acc; acc >>= 32;
916
917         acc += rp[2];   acc += bp[14-12];
918                         acc += bp[23-12];
919                         acc -= bp[13-12];
920                         acc -= bp[21-12]; rp[2] = (unsigned int)acc; acc >>= 32;
921
922         acc += rp[3];   acc += bp[15-12];
923                         acc += bp[12-12];
924                         acc += bp[20-12];
925                         acc += bp[21-12];
926                         acc -= bp[14-12];
927                         acc -= bp[22-12];
928                         acc -= bp[23-12]; rp[3] = (unsigned int)acc; acc >>= 32;
929
930         acc += rp[4];   acc += bp[21-12];
931                         acc += bp[21-12];
932                         acc += bp[16-12];
933                         acc += bp[13-12];
934                         acc += bp[12-12];
935                         acc += bp[20-12];
936                         acc += bp[22-12];
937                         acc -= bp[15-12];
938                         acc -= bp[23-12];
939                         acc -= bp[23-12]; rp[4] = (unsigned int)acc; acc >>= 32;
940
941         acc += rp[5];   acc += bp[22-12];
942                         acc += bp[22-12];
943                         acc += bp[17-12];
944                         acc += bp[14-12];
945                         acc += bp[13-12];
946                         acc += bp[21-12];
947                         acc += bp[23-12];
948                         acc -= bp[16-12]; rp[5] = (unsigned int)acc; acc >>= 32;
949                         
950         acc += rp[6];   acc += bp[23-12];
951                         acc += bp[23-12];
952                         acc += bp[18-12];
953                         acc += bp[15-12];
954                         acc += bp[14-12];
955                         acc += bp[22-12];
956                         acc -= bp[17-12]; rp[6] = (unsigned int)acc; acc >>= 32;
957                         
958         acc += rp[7];   acc += bp[19-12];
959                         acc += bp[16-12];
960                         acc += bp[15-12];
961                         acc += bp[23-12];
962                         acc -= bp[18-12]; rp[7] = (unsigned int)acc; acc >>= 32;
963                         
964         acc += rp[8];   acc += bp[20-12];
965                         acc += bp[17-12];
966                         acc += bp[16-12];
967                         acc -= bp[19-12]; rp[8] = (unsigned int)acc; acc >>= 32;
968                         
969         acc += rp[9];   acc += bp[21-12];
970                         acc += bp[18-12];
971                         acc += bp[17-12];
972                         acc -= bp[20-12]; rp[9] = (unsigned int)acc; acc >>= 32;
973                         
974         acc += rp[10];  acc += bp[22-12];
975                         acc += bp[19-12];
976                         acc += bp[18-12];
977                         acc -= bp[21-12]; rp[10] = (unsigned int)acc; acc >>= 32;
978                         
979         acc += rp[11];  acc += bp[23-12];
980                         acc += bp[20-12];
981                         acc += bp[19-12];
982                         acc -= bp[22-12]; rp[11] = (unsigned int)acc;
983
984         carry = (int)(acc>>32);
985         }
986 #else
987         {
988         BN_ULONG t_d[BN_NIST_384_TOP];
989
990         /*S1*/
991         nist_set_256(t_d, buf.bn, 0, 0, 0, 0, 0, 23-4, 22-4, 21-4);
992                 /* left shift */
993                 {
994                 register BN_ULONG *ap,t,c;
995                 ap = t_d;
996                 c=0;
997                 for (i = 3; i != 0; --i)
998                         {
999                         t= *ap;
1000                         *(ap++)=((t<<1)|c)&BN_MASK2;
1001                         c=(t & BN_TBIT)?1:0;
1002                         }
1003                 *ap=c;
1004                 }
1005         carry = (int)bn_add_words(r_d+(128/BN_BITS2), r_d+(128/BN_BITS2), 
1006                 t_d, BN_NIST_256_TOP);
1007         /*S2 */
1008         carry += (int)bn_add_words(r_d, r_d, buf.bn, BN_NIST_384_TOP);
1009         /*S3*/
1010         nist_set_384(t_d,buf.bn,20,19,18,17,16,15,14,13,12,23,22,21);
1011         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1012         /*S4*/
1013         nist_set_384(t_d,buf.bn,19,18,17,16,15,14,13,12,20,0,23,0);
1014         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1015         /*S5*/
1016         nist_set_384(t_d, buf.bn,0,0,0,0,23,22,21,20,0,0,0,0);
1017         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1018         /*S6*/
1019         nist_set_384(t_d,buf.bn,0,0,0,0,0,0,23,22,21,0,0,20);
1020         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1021         /*D1*/
1022         nist_set_384(t_d,buf.bn,22,21,20,19,18,17,16,15,14,13,12,23);
1023         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1024         /*D2*/
1025         nist_set_384(t_d,buf.bn,0,0,0,0,0,0,0,23,22,21,20,0);
1026         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1027         /*D3*/
1028         nist_set_384(t_d,buf.bn,0,0,0,0,0,0,0,23,23,0,0,0);
1029         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1030
1031         }
1032 #endif
1033         /* see BN_nist_mod_224 for explanation */
1034         u.f = bn_sub_words;
1035         if (carry > 0)
1036                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_384[carry-1],BN_NIST_384_TOP);
1037         else if (carry < 0)
1038                 {
1039                 carry = (int)bn_add_words(r_d,r_d,_nist_p_384[-carry-1],BN_NIST_384_TOP);
1040                 mask = 0-(PTR_SIZE_INT)carry;
1041                 u.p = ((PTR_SIZE_INT)bn_sub_words&mask) |
1042                  ((PTR_SIZE_INT)bn_add_words&~mask);
1043                 }
1044         else
1045                 carry = 1;
1046
1047         mask  = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_384[0],BN_NIST_384_TOP);
1048         mask &= 0-(PTR_SIZE_INT)carry;
1049         res   = (BN_ULONG *)(((PTR_SIZE_INT)c_d&~mask) |
1050          ((PTR_SIZE_INT)r_d&mask));
1051         nist_cp_bn(r_d, res, BN_NIST_384_TOP);
1052         r->top = BN_NIST_384_TOP;
1053         bn_correct_top(r);
1054
1055         return 1;
1056         }
1057
1058 #define BN_NIST_521_RSHIFT      (521%BN_BITS2)
1059 #define BN_NIST_521_LSHIFT      (BN_BITS2-BN_NIST_521_RSHIFT)
1060 #define BN_NIST_521_TOP_MASK    ((BN_ULONG)BN_MASK2>>BN_NIST_521_LSHIFT)
1061
1062 int BN_nist_mod_521(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
1063         BN_CTX *ctx)
1064         {
1065         int     top = a->top, i;
1066         BN_ULONG *r_d, *a_d = a->d,
1067                  t_d[BN_NIST_521_TOP],
1068                  val,tmp,*res;
1069         PTR_SIZE_INT mask;
1070         static const BIGNUM _bignum_nist_p_521_sqr = {
1071                 (BN_ULONG *)_nist_p_521_sqr,
1072                 sizeof(_nist_p_521_sqr)/sizeof(_nist_p_521_sqr[0]),
1073                 sizeof(_nist_p_521_sqr)/sizeof(_nist_p_521_sqr[0]),
1074                 0,BN_FLG_STATIC_DATA };
1075
1076         field = &_bignum_nist_p_521; /* just to make sure */
1077
1078         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_521_sqr)>=0)
1079                 return BN_nnmod(r, a, field, ctx);
1080
1081         i = BN_ucmp(field, a);
1082         if (i == 0)
1083                 {
1084                 BN_zero(r);
1085                 return 1;
1086                 }
1087         else if (i > 0)
1088                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
1089
1090         if (r != a)
1091                 {
1092                 if (!bn_wexpand(r,BN_NIST_521_TOP))
1093                         return 0;
1094                 r_d = r->d;
1095                 nist_cp_bn(r_d,a_d, BN_NIST_521_TOP);
1096                 }
1097         else
1098                 r_d = a_d;
1099
1100         /* upper 521 bits, copy ... */
1101         nist_cp_bn_0(t_d,a_d + (BN_NIST_521_TOP-1), top - (BN_NIST_521_TOP-1),BN_NIST_521_TOP);
1102         /* ... and right shift */
1103         for (val=t_d[0],i=0; i<BN_NIST_521_TOP-1; i++)
1104                 {
1105                 tmp = val>>BN_NIST_521_RSHIFT;
1106                 val = t_d[i+1];
1107                 t_d[i] = (tmp | val<<BN_NIST_521_LSHIFT) & BN_MASK2;
1108                 }
1109         t_d[i] = val>>BN_NIST_521_RSHIFT;
1110         /* lower 521 bits */
1111         r_d[i] &= BN_NIST_521_TOP_MASK;
1112
1113         bn_add_words(r_d,r_d,t_d,BN_NIST_521_TOP);
1114         mask = 0-(PTR_SIZE_INT)bn_sub_words(t_d,r_d,_nist_p_521,BN_NIST_521_TOP);
1115         res  = (BN_ULONG *)(((PTR_SIZE_INT)t_d&~mask) |
1116          ((PTR_SIZE_INT)r_d&mask));
1117         nist_cp_bn(r_d,res,BN_NIST_521_TOP);
1118         r->top = BN_NIST_521_TOP;
1119         bn_correct_top(r);
1120
1121         return 1;
1122         }
1123
1124 int (*BN_nist_mod_func(const BIGNUM *p))(BIGNUM *r, const BIGNUM *a, const BIGNUM *field, BN_CTX *ctx)
1125         {
1126         if (BN_ucmp(&_bignum_nist_p_192, p) == 0)
1127                 return BN_nist_mod_192;
1128         if (BN_ucmp(&_bignum_nist_p_224, p) == 0)
1129                 return BN_nist_mod_224;
1130         if (BN_ucmp(&_bignum_nist_p_256, p) == 0)
1131                 return BN_nist_mod_256;
1132         if (BN_ucmp(&_bignum_nist_p_384, p) == 0)
1133                 return BN_nist_mod_384;
1134         if (BN_ucmp(&_bignum_nist_p_521, p) == 0)
1135                 return BN_nist_mod_521;
1136         return 0;
1137         }