bn_shift.c: minimize reallocations, which allows BN_FLG_STATIC_DATA to
[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
370 #define nist_set_192(to, from, a1, a2, a3) \
371         { \
372         bn_cp_64(to, 0, from, (a3) - 3) \
373         bn_cp_64(to, 1, from, (a2) - 3) \
374         bn_cp_64(to, 2, from, (a1) - 3) \
375         }
376
377 int BN_nist_mod_192(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
378         BN_CTX *ctx)
379         {
380         int      top = a->top, i;
381         int      carry;
382         register BN_ULONG *r_d, *a_d = a->d;
383         BN_ULONG buf[BN_NIST_192_TOP],
384                  c_d[BN_NIST_192_TOP],
385                 *res;
386         PTR_SIZE_INT mask;
387         static const BIGNUM _bignum_nist_p_192_sqr = {
388                 (BN_ULONG *)_nist_p_192_sqr,
389                 sizeof(_nist_p_192_sqr)/sizeof(_nist_p_192_sqr[0]),
390                 sizeof(_nist_p_192_sqr)/sizeof(_nist_p_192_sqr[0]),
391                 0,BN_FLG_STATIC_DATA };
392
393         field = &_bignum_nist_p_192; /* just to make sure */
394
395         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_192_sqr)>=0)
396                 return BN_nnmod(r, a, field, ctx);
397
398         i = BN_ucmp(field, a);
399         if (i == 0)
400                 {
401                 BN_zero(r);
402                 return 1;
403                 }
404         else if (i > 0)
405                 return (r == a) ? 1 : (BN_copy(r ,a) != NULL);
406
407         if (r != a)
408                 {
409                 if (!bn_wexpand(r, BN_NIST_192_TOP))
410                         return 0;
411                 r_d = r->d;
412                 nist_cp_bn(r_d, a_d, BN_NIST_192_TOP);
413                 }
414         else
415                 r_d = a_d;
416
417         nist_cp_bn_0(buf, a_d + BN_NIST_192_TOP, top - BN_NIST_192_TOP, BN_NIST_192_TOP);
418
419 #if defined(NIST_INT64)
420         {
421         NIST_INT64              acc;    /* accumulator */
422         unsigned int            *rp=(unsigned int *)r_d;
423         const unsigned int      *bp=(const unsigned int *)buf;
424
425         acc  = rp[0];   acc += bp[3*2-6];
426                         acc += bp[5*2-6]; rp[0] = (unsigned int)acc; acc >>= 32;
427
428         acc += rp[1];   acc += bp[3*2-5];
429                         acc += bp[5*2-5]; rp[1] = (unsigned int)acc; acc >>= 32;
430
431         acc += rp[2];   acc += bp[3*2-6];
432                         acc += bp[4*2-6];
433                         acc += bp[5*2-6]; rp[2] = (unsigned int)acc; acc >>= 32;
434
435         acc += rp[3];   acc += bp[3*2-5];
436                         acc += bp[4*2-5];
437                         acc += bp[5*2-5]; rp[3] = (unsigned int)acc; acc >>= 32;
438
439         acc += rp[4];   acc += bp[4*2-6];
440                         acc += bp[5*2-6]; rp[4] = (unsigned int)acc; acc >>= 32;
441
442         acc += rp[5];   acc += bp[4*2-5];
443                         acc += bp[5*2-5]; rp[5] = (unsigned int)acc;
444
445         carry = (int)(acc>>32);
446         }
447 #else
448         {
449         BN_ULONG t_d[BN_NIST_192_TOP];
450
451         nist_set_192(t_d, buf, 0, 3, 3);
452         carry = (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP);
453         nist_set_192(t_d, buf, 4, 4, 0);
454         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP);
455         nist_set_192(t_d, buf, 5, 5, 5)
456         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_192_TOP);
457         }
458 #endif
459         if (carry > 0)
460                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_192[carry-1],BN_NIST_192_TOP);
461         else
462                 carry = 1;
463
464         /*
465          * we need 'if (carry==0 || result>=modulus) result-=modulus;'
466          * as comparison implies subtraction, we can write
467          * 'tmp=result-modulus; if (!carry || !borrow) result=tmp;'
468          * this is what happens below, but without explicit if:-) a.
469          */
470         mask  = 0-(PTR_SIZE_INT)bn_sub_words(c_d,r_d,_nist_p_192[0],BN_NIST_192_TOP);
471         mask &= 0-(PTR_SIZE_INT)carry;
472         res   = (BN_ULONG *)
473          (((PTR_SIZE_INT)c_d&~mask) | ((PTR_SIZE_INT)r_d&mask));
474         nist_cp_bn(r_d, res, BN_NIST_192_TOP);
475         r->top = BN_NIST_192_TOP;
476         bn_correct_top(r);
477
478         return 1;
479         }
480
481 typedef BN_ULONG (*bn_addsub_f)(BN_ULONG *,const BN_ULONG *,const BN_ULONG *,int);
482
483 #define nist_set_224(to, from, a1, a2, a3, a4, a5, a6, a7) \
484         { \
485         bn_cp_32(to, 0, from, (a7) - 7) \
486         bn_cp_32(to, 1, from, (a6) - 7) \
487         bn_cp_32(to, 2, from, (a5) - 7) \
488         bn_cp_32(to, 3, from, (a4) - 7) \
489         bn_cp_32(to, 4, from, (a3) - 7) \
490         bn_cp_32(to, 5, from, (a2) - 7) \
491         bn_cp_32(to, 6, from, (a1) - 7) \
492         }
493
494 int BN_nist_mod_224(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
495         BN_CTX *ctx)
496         {
497         int     top = a->top, i;
498         int     carry;
499         BN_ULONG *r_d, *a_d = a->d;
500         BN_ULONG buf[BN_NIST_224_TOP],
501                  c_d[BN_NIST_224_TOP],
502                 *res;
503         PTR_SIZE_INT mask;
504         union { bn_addsub_f f; PTR_SIZE_INT p; } u;
505         static const BIGNUM _bignum_nist_p_224_sqr = {
506                 (BN_ULONG *)_nist_p_224_sqr,
507                 sizeof(_nist_p_224_sqr)/sizeof(_nist_p_224_sqr[0]),
508                 sizeof(_nist_p_224_sqr)/sizeof(_nist_p_224_sqr[0]),
509                 0,BN_FLG_STATIC_DATA };
510
511
512         field = &_bignum_nist_p_224; /* just to make sure */
513
514         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_224_sqr)>=0)
515                 return BN_nnmod(r, a, field, ctx);
516
517         i = BN_ucmp(field, a);
518         if (i == 0)
519                 {
520                 BN_zero(r);
521                 return 1;
522                 }
523         else if (i > 0)
524                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
525
526         if (r != a)
527                 {
528                 if (!bn_wexpand(r, BN_NIST_224_TOP))
529                         return 0;
530                 r_d = r->d;
531                 nist_cp_bn(r_d, a_d, BN_NIST_224_TOP);
532                 }
533         else
534                 r_d = a_d;
535
536 #if BN_BITS2==64
537         /* copy upper 256 bits of 448 bit number ... */
538         nist_cp_bn_0(c_d, a_d + (BN_NIST_224_TOP-1), top - (BN_NIST_224_TOP-1), BN_NIST_224_TOP);
539         /* ... and right shift by 32 to obtain upper 224 bits */
540         nist_set_224(buf, c_d, 14, 13, 12, 11, 10, 9, 8);
541         /* truncate lower part to 224 bits too */
542         r_d[BN_NIST_224_TOP-1] &= BN_MASK2l;
543 #else
544         nist_cp_bn_0(buf, a_d + BN_NIST_224_TOP, top - BN_NIST_224_TOP, BN_NIST_224_TOP);
545 #endif
546
547 #if defined(NIST_INT64) && BN_BITS2!=64
548         {
549         NIST_INT64              acc;    /* accumulator */
550         unsigned int            *rp=(unsigned int *)r_d;
551         const unsigned int      *bp=(const unsigned int *)buf;
552
553         acc  = rp[0];   acc -= bp[7-7];
554                         acc -= bp[11-7]; rp[0] = (unsigned int)acc; acc >>= 32;
555
556         acc += rp[1];   acc -= bp[8-7];
557                         acc -= bp[12-7]; rp[1] = (unsigned int)acc; acc >>= 32;
558
559         acc += rp[2];   acc -= bp[9-7];
560                         acc -= bp[13-7]; rp[2] = (unsigned int)acc; acc >>= 32;
561
562         acc += rp[3];   acc += bp[7-7];
563                         acc += bp[11-7];
564                         acc -= bp[10-7]; rp[3] = (unsigned int)acc; acc>>= 32;
565
566         acc += rp[4];   acc += bp[8-7];
567                         acc += bp[12-7];
568                         acc -= bp[11-7]; rp[4] = (unsigned int)acc; acc >>= 32;
569
570         acc += rp[5];   acc += bp[9-7];
571                         acc += bp[13-7];
572                         acc -= bp[12-7]; rp[5] = (unsigned int)acc; acc >>= 32;
573
574         acc += rp[6];   acc += bp[10-7];
575                         acc -= bp[13-7]; rp[6] = (unsigned int)acc;
576
577         carry = (int)(acc>>32);
578 # if BN_BITS2==64
579         rp[7] = carry;
580 # endif
581         }       
582 #else
583         {
584         BN_ULONG t_d[BN_NIST_224_TOP];
585
586         nist_set_224(t_d, buf, 10, 9, 8, 7, 0, 0, 0);
587         carry = (int)bn_add_words(r_d, r_d, t_d, BN_NIST_224_TOP);
588         nist_set_224(t_d, buf, 0, 13, 12, 11, 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, 13, 12, 11, 10, 9, 8, 7);
591         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_224_TOP);
592         nist_set_224(t_d, buf, 0, 0, 0, 0, 13, 12, 11);
593         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_224_TOP);
594
595 #if BN_BITS2==64
596         carry = (int)(r_d[BN_NIST_224_TOP-1]>>32);
597 #endif
598         }
599 #endif
600         u.f = bn_sub_words;
601         if (carry > 0)
602                 {
603                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_224[carry-1],BN_NIST_224_TOP);
604 #if BN_BITS2==64
605                 carry=(int)(~(r_d[BN_NIST_224_TOP-1]>>32))&1;
606 #endif
607                 }
608         else if (carry < 0)
609                 {
610                 /* it's a bit more comlicated logic in this case.
611                  * if bn_add_words yields no carry, then result
612                  * has to be adjusted by unconditionally *adding*
613                  * the modulus. but if it does, then result has
614                  * to be compared to the modulus and conditionally
615                  * adjusted by *subtracting* the latter. */
616                 carry = (int)bn_add_words(r_d,r_d,_nist_p_224[-carry-1],BN_NIST_224_TOP);
617                 mask = 0-(PTR_SIZE_INT)carry;
618                 u.p = ((PTR_SIZE_INT)bn_sub_words&mask) |
619                  ((PTR_SIZE_INT)bn_add_words&~mask);
620                 }
621         else
622                 carry = 1;
623
624         /* otherwise it's effectively same as in BN_nist_mod_192... */
625         mask  = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_224[0],BN_NIST_224_TOP);
626         mask &= 0-(PTR_SIZE_INT)carry;
627         res   = (BN_ULONG *)(((PTR_SIZE_INT)c_d&~mask) |
628          ((PTR_SIZE_INT)r_d&mask));
629         nist_cp_bn(r_d, res, BN_NIST_224_TOP);
630         r->top = BN_NIST_224_TOP;
631         bn_correct_top(r);
632
633         return 1;
634         }
635
636 #define nist_set_256(to, from, a1, a2, a3, a4, a5, a6, a7, a8) \
637         { \
638         bn_cp_32(to, 0, from, (a8) - 8) \
639         bn_cp_32(to, 1, from, (a7) - 8) \
640         bn_cp_32(to, 2, from, (a6) - 8) \
641         bn_cp_32(to, 3, from, (a5) - 8) \
642         bn_cp_32(to, 4, from, (a4) - 8) \
643         bn_cp_32(to, 5, from, (a3) - 8) \
644         bn_cp_32(to, 6, from, (a2) - 8) \
645         bn_cp_32(to, 7, from, (a1) - 8) \
646         }
647
648 int BN_nist_mod_256(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
649         BN_CTX *ctx)
650         {
651         int     i, top = a->top;
652         int     carry = 0;
653         register BN_ULONG *a_d = a->d, *r_d;
654         BN_ULONG buf[BN_NIST_256_TOP],
655                  c_d[BN_NIST_256_TOP],
656                 *res;
657         PTR_SIZE_INT mask;
658         union { bn_addsub_f f; PTR_SIZE_INT p; } u;
659         static const BIGNUM _bignum_nist_p_256_sqr = {
660                 (BN_ULONG *)_nist_p_256_sqr,
661                 sizeof(_nist_p_256_sqr)/sizeof(_nist_p_256_sqr[0]),
662                 sizeof(_nist_p_256_sqr)/sizeof(_nist_p_256_sqr[0]),
663                 0,BN_FLG_STATIC_DATA };
664
665         field = &_bignum_nist_p_256; /* just to make sure */
666
667         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_256_sqr)>=0)
668                 return BN_nnmod(r, a, field, ctx);
669
670         i = BN_ucmp(field, a);
671         if (i == 0)
672                 {
673                 BN_zero(r);
674                 return 1;
675                 }
676         else if (i > 0)
677                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
678
679         if (r != a)
680                 {
681                 if (!bn_wexpand(r, BN_NIST_256_TOP))
682                         return 0;
683                 r_d = r->d;
684                 nist_cp_bn(r_d, a_d, BN_NIST_256_TOP);
685                 }
686         else
687                 r_d = a_d;
688
689         nist_cp_bn_0(buf, a_d + BN_NIST_256_TOP, top - BN_NIST_256_TOP, BN_NIST_256_TOP);
690
691 #if defined(NIST_INT64)
692         {
693         NIST_INT64              acc;    /* accumulator */
694         unsigned int            *rp=(unsigned int *)r_d;
695         const unsigned int      *bp=(const unsigned int *)buf;
696
697         acc = rp[0];    acc += bp[8-8];
698                         acc += bp[9-8];
699                         acc -= bp[11-8];
700                         acc -= bp[12-8];
701                         acc -= bp[13-8];
702                         acc -= bp[14-8]; rp[0] = (unsigned int)acc; acc >>= 32;
703
704         acc += rp[1];   acc += bp[9-8];
705                         acc += bp[10-8];
706                         acc -= bp[12-8];
707                         acc -= bp[13-8];
708                         acc -= bp[14-8];
709                         acc -= bp[15-8]; rp[1] = (unsigned int)acc; acc >>= 32;
710
711         acc += rp[2];   acc += bp[10-8];
712                         acc += bp[11-8];
713                         acc -= bp[13-8];
714                         acc -= bp[14-8];
715                         acc -= bp[15-8]; rp[2] = (unsigned int)acc; acc >>= 32;
716
717         acc += rp[3];   acc += bp[11-8];
718                         acc += bp[11-8];
719                         acc += bp[12-8];
720                         acc += bp[12-8];
721                         acc += bp[13-8];
722                         acc -= bp[15-8];
723                         acc -= bp[8-8];
724                         acc -= bp[9-8];  rp[3] = (unsigned int)acc; acc >>= 32;
725
726         acc += rp[4];   acc += bp[12-8];
727                         acc += bp[12-8];
728                         acc += bp[13-8];
729                         acc += bp[13-8];
730                         acc += bp[14-8];
731                         acc -= bp[9-8];
732                         acc -= bp[10-8]; rp[4] = (unsigned int)acc; acc >>= 32;
733
734         acc += rp[5];   acc += bp[13-8];
735                         acc += bp[13-8];
736                         acc += bp[14-8];
737                         acc += bp[14-8];
738                         acc += bp[15-8];
739                         acc -= bp[10-8];
740                         acc -= bp[11-8]; rp[5] = (unsigned int)acc; acc >>= 32;
741
742         acc += rp[6];   acc += bp[14-8];
743                         acc += bp[14-8];
744                         acc += bp[15-8];
745                         acc += bp[15-8];
746                         acc += bp[14-8];
747                         acc += bp[13-8];
748                         acc -= bp[8-8];
749                         acc -= bp[9-8];  rp[6] = (unsigned int)acc; acc >>= 32;
750
751         acc += rp[7];   acc += bp[15-8];
752                         acc += bp[15-8];
753                         acc += bp[15-8];
754                         acc += bp[8 -8];
755                         acc -= bp[10-8];
756                         acc -= bp[11-8];
757                         acc -= bp[12-8];
758                         acc -= bp[13-8]; rp[7] = (unsigned int)acc;
759
760         carry = (int)(acc>>32);
761         }
762 #else
763         {
764         BN_ULONG t_d[BN_NIST_256_TOP];
765
766         /*S1*/
767         nist_set_256(t_d, buf, 15, 14, 13, 12, 11, 0, 0, 0);
768         /*S2*/
769         nist_set_256(c_d, buf, 0, 15, 14, 13, 12, 0, 0, 0);
770         carry = (int)bn_add_words(t_d, t_d, c_d, BN_NIST_256_TOP);
771         /* left shift */
772                 {
773                 register BN_ULONG *ap,t,c;
774                 ap = t_d;
775                 c=0;
776                 for (i = BN_NIST_256_TOP; i != 0; --i)
777                         {
778                         t= *ap;
779                         *(ap++)=((t<<1)|c)&BN_MASK2;
780                         c=(t & BN_TBIT)?1:0;
781                         }
782                 carry <<= 1;
783                 carry  |= c;
784                 }
785         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP);
786         /*S3*/
787         nist_set_256(t_d, buf, 15, 14, 0, 0, 0, 10, 9, 8);
788         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP);
789         /*S4*/
790         nist_set_256(t_d, buf, 8, 13, 15, 14, 13, 11, 10, 9);
791         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_256_TOP);
792         /*D1*/
793         nist_set_256(t_d, buf, 10, 8, 0, 0, 0, 13, 12, 11);
794         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
795         /*D2*/
796         nist_set_256(t_d, buf, 11, 9, 0, 0, 15, 14, 13, 12);
797         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
798         /*D3*/
799         nist_set_256(t_d, buf, 12, 0, 10, 9, 8, 15, 14, 13);
800         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
801         /*D4*/
802         nist_set_256(t_d, buf, 13, 0, 11, 10, 9, 0, 15, 14);
803         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_256_TOP);
804
805         }
806 #endif
807         /* see BN_nist_mod_224 for explanation */
808         u.f = bn_sub_words;
809         if (carry > 0)
810                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_256[carry-1],BN_NIST_256_TOP);
811         else if (carry < 0)
812                 {
813                 carry = (int)bn_add_words(r_d,r_d,_nist_p_256[-carry-1],BN_NIST_256_TOP);
814                 mask = 0-(PTR_SIZE_INT)carry;
815                 u.p = ((PTR_SIZE_INT)bn_sub_words&mask) |
816                  ((PTR_SIZE_INT)bn_add_words&~mask);
817                 }
818         else
819                 carry = 1;
820
821         mask  = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_256[0],BN_NIST_256_TOP);
822         mask &= 0-(PTR_SIZE_INT)carry;
823         res   = (BN_ULONG *)(((PTR_SIZE_INT)c_d&~mask) |
824          ((PTR_SIZE_INT)r_d&mask));
825         nist_cp_bn(r_d, res, BN_NIST_256_TOP);
826         r->top = BN_NIST_256_TOP;
827         bn_correct_top(r);
828
829         return 1;
830         }
831
832 #define nist_set_384(to,from,a1,a2,a3,a4,a5,a6,a7,a8,a9,a10,a11,a12) \
833         { \
834         bn_cp_32(to, 0, from,  (a12) - 12) \
835         bn_cp_32(to, 1, from,  (a11) - 12) \
836         bn_cp_32(to, 2, from,  (a10) - 12) \
837         bn_cp_32(to, 3, from,  (a9) - 12)  \
838         bn_cp_32(to, 4, from,  (a8) - 12)  \
839         bn_cp_32(to, 5, from,  (a7) - 12)  \
840         bn_cp_32(to, 6, from,  (a6) - 12)  \
841         bn_cp_32(to, 7, from,  (a5) - 12)  \
842         bn_cp_32(to, 8, from,  (a4) - 12)  \
843         bn_cp_32(to, 9, from,  (a3) - 12)  \
844         bn_cp_32(to, 10, from, (a2) - 12)  \
845         bn_cp_32(to, 11, from, (a1) - 12)  \
846         }
847
848 int BN_nist_mod_384(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
849         BN_CTX *ctx)
850         {
851         int     i, top = a->top;
852         int     carry = 0;
853         register BN_ULONG *r_d, *a_d = a->d;
854         BN_ULONG buf[BN_NIST_384_TOP],
855                  c_d[BN_NIST_384_TOP],
856                 *res;
857         PTR_SIZE_INT mask;
858         union { bn_addsub_f f; PTR_SIZE_INT p; } u;
859         static const BIGNUM _bignum_nist_p_384_sqr = {
860                 (BN_ULONG *)_nist_p_384_sqr,
861                 sizeof(_nist_p_384_sqr)/sizeof(_nist_p_384_sqr[0]),
862                 sizeof(_nist_p_384_sqr)/sizeof(_nist_p_384_sqr[0]),
863                 0,BN_FLG_STATIC_DATA };
864
865
866         field = &_bignum_nist_p_384; /* just to make sure */
867
868         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_384_sqr)>=0)
869                 return BN_nnmod(r, a, field, ctx);
870
871         i = BN_ucmp(field, a);
872         if (i == 0)
873                 {
874                 BN_zero(r);
875                 return 1;
876                 }
877         else if (i > 0)
878                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
879
880         if (r != a)
881                 {
882                 if (!bn_wexpand(r, BN_NIST_384_TOP))
883                         return 0;
884                 r_d = r->d;
885                 nist_cp_bn(r_d, a_d, BN_NIST_384_TOP);
886                 }
887         else
888                 r_d = a_d;
889
890         nist_cp_bn_0(buf, a_d + BN_NIST_384_TOP, top - BN_NIST_384_TOP, BN_NIST_384_TOP);
891
892 #if defined(NIST_INT64)
893         {
894         NIST_INT64              acc;    /* accumulator */
895         unsigned int            *rp=(unsigned int *)r_d;
896         const unsigned int      *bp=(const unsigned int *)buf;
897
898         acc = rp[0];    acc += bp[12-12];
899                         acc += bp[21-12];
900                         acc += bp[20-12];
901                         acc -= bp[23-12]; rp[0] = (unsigned int)acc; acc >>= 32;
902
903         acc += rp[1];   acc += bp[13-12];
904                         acc += bp[22-12];
905                         acc += bp[23-12];
906                         acc -= bp[12-12];
907                         acc -= bp[20-12]; rp[1] = (unsigned int)acc; acc >>= 32;
908
909         acc += rp[2];   acc += bp[14-12];
910                         acc += bp[23-12];
911                         acc -= bp[13-12];
912                         acc -= bp[21-12]; rp[2] = (unsigned int)acc; acc >>= 32;
913
914         acc += rp[3];   acc += bp[15-12];
915                         acc += bp[12-12];
916                         acc += bp[20-12];
917                         acc += bp[21-12];
918                         acc -= bp[14-12];
919                         acc -= bp[22-12];
920                         acc -= bp[23-12]; rp[3] = (unsigned int)acc; acc >>= 32;
921
922         acc += rp[4];   acc += bp[21-12];
923                         acc += bp[21-12];
924                         acc += bp[16-12];
925                         acc += bp[13-12];
926                         acc += bp[12-12];
927                         acc += bp[20-12];
928                         acc += bp[22-12];
929                         acc -= bp[15-12];
930                         acc -= bp[23-12];
931                         acc -= bp[23-12]; rp[4] = (unsigned int)acc; acc >>= 32;
932
933         acc += rp[5];   acc += bp[22-12];
934                         acc += bp[22-12];
935                         acc += bp[17-12];
936                         acc += bp[14-12];
937                         acc += bp[13-12];
938                         acc += bp[21-12];
939                         acc += bp[23-12];
940                         acc -= bp[16-12]; rp[5] = (unsigned int)acc; acc >>= 32;
941                         
942         acc += rp[6];   acc += bp[23-12];
943                         acc += bp[23-12];
944                         acc += bp[18-12];
945                         acc += bp[15-12];
946                         acc += bp[14-12];
947                         acc += bp[22-12];
948                         acc -= bp[17-12]; rp[6] = (unsigned int)acc; acc >>= 32;
949                         
950         acc += rp[7];   acc += bp[19-12];
951                         acc += bp[16-12];
952                         acc += bp[15-12];
953                         acc += bp[23-12];
954                         acc -= bp[18-12]; rp[7] = (unsigned int)acc; acc >>= 32;
955                         
956         acc += rp[8];   acc += bp[20-12];
957                         acc += bp[17-12];
958                         acc += bp[16-12];
959                         acc -= bp[19-12]; rp[8] = (unsigned int)acc; acc >>= 32;
960                         
961         acc += rp[9];   acc += bp[21-12];
962                         acc += bp[18-12];
963                         acc += bp[17-12];
964                         acc -= bp[20-12]; rp[9] = (unsigned int)acc; acc >>= 32;
965                         
966         acc += rp[10];  acc += bp[22-12];
967                         acc += bp[19-12];
968                         acc += bp[18-12];
969                         acc -= bp[21-12]; rp[10] = (unsigned int)acc; acc >>= 32;
970                         
971         acc += rp[11];  acc += bp[23-12];
972                         acc += bp[20-12];
973                         acc += bp[19-12];
974                         acc -= bp[22-12]; rp[11] = (unsigned int)acc;
975
976         carry = (int)(acc>>32);
977         }
978 #else
979         {
980         BN_ULONG t_d[BN_NIST_384_TOP];
981
982         /*S1*/
983         nist_set_256(t_d, buf, 0, 0, 0, 0, 0, 23-4, 22-4, 21-4);
984                 /* left shift */
985                 {
986                 register BN_ULONG *ap,t,c;
987                 ap = t_d;
988                 c=0;
989                 for (i = 3; i != 0; --i)
990                         {
991                         t= *ap;
992                         *(ap++)=((t<<1)|c)&BN_MASK2;
993                         c=(t & BN_TBIT)?1:0;
994                         }
995                 *ap=c;
996                 }
997         carry = (int)bn_add_words(r_d+(128/BN_BITS2), r_d+(128/BN_BITS2), 
998                 t_d, BN_NIST_256_TOP);
999         /*S2 */
1000         carry += (int)bn_add_words(r_d, r_d, buf, BN_NIST_384_TOP);
1001         /*S3*/
1002         nist_set_384(t_d,buf,20,19,18,17,16,15,14,13,12,23,22,21);
1003         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1004         /*S4*/
1005         nist_set_384(t_d,buf,19,18,17,16,15,14,13,12,20,0,23,0);
1006         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1007         /*S5*/
1008         nist_set_384(t_d, buf,0,0,0,0,23,22,21,20,0,0,0,0);
1009         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1010         /*S6*/
1011         nist_set_384(t_d,buf,0,0,0,0,0,0,23,22,21,0,0,20);
1012         carry += (int)bn_add_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1013         /*D1*/
1014         nist_set_384(t_d,buf,22,21,20,19,18,17,16,15,14,13,12,23);
1015         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1016         /*D2*/
1017         nist_set_384(t_d,buf,0,0,0,0,0,0,0,23,22,21,20,0);
1018         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1019         /*D3*/
1020         nist_set_384(t_d,buf,0,0,0,0,0,0,0,23,23,0,0,0);
1021         carry -= (int)bn_sub_words(r_d, r_d, t_d, BN_NIST_384_TOP);
1022
1023         }
1024 #endif
1025         /* see BN_nist_mod_224 for explanation */
1026         u.f = bn_sub_words;
1027         if (carry > 0)
1028                 carry = (int)bn_sub_words(r_d,r_d,_nist_p_384[carry-1],BN_NIST_384_TOP);
1029         else if (carry < 0)
1030                 {
1031                 carry = (int)bn_add_words(r_d,r_d,_nist_p_384[-carry-1],BN_NIST_384_TOP);
1032                 mask = 0-(PTR_SIZE_INT)carry;
1033                 u.p = ((PTR_SIZE_INT)bn_sub_words&mask) |
1034                  ((PTR_SIZE_INT)bn_add_words&~mask);
1035                 }
1036         else
1037                 carry = 1;
1038
1039         mask  = 0-(PTR_SIZE_INT)(*u.f)(c_d,r_d,_nist_p_384[0],BN_NIST_384_TOP);
1040         mask &= 0-(PTR_SIZE_INT)carry;
1041         res   = (BN_ULONG *)(((PTR_SIZE_INT)c_d&~mask) |
1042          ((PTR_SIZE_INT)r_d&mask));
1043         nist_cp_bn(r_d, res, BN_NIST_384_TOP);
1044         r->top = BN_NIST_384_TOP;
1045         bn_correct_top(r);
1046
1047         return 1;
1048         }
1049
1050 #define BN_NIST_521_RSHIFT      (521%BN_BITS2)
1051 #define BN_NIST_521_LSHIFT      (BN_BITS2-BN_NIST_521_RSHIFT)
1052 #define BN_NIST_521_TOP_MASK    ((BN_ULONG)BN_MASK2>>BN_NIST_521_LSHIFT)
1053
1054 int BN_nist_mod_521(BIGNUM *r, const BIGNUM *a, const BIGNUM *field,
1055         BN_CTX *ctx)
1056         {
1057         int     top = a->top, i;
1058         BN_ULONG *r_d, *a_d = a->d,
1059                  t_d[BN_NIST_521_TOP],
1060                  val,tmp,*res;
1061         PTR_SIZE_INT mask;
1062         static const BIGNUM _bignum_nist_p_521_sqr = {
1063                 (BN_ULONG *)_nist_p_521_sqr,
1064                 sizeof(_nist_p_521_sqr)/sizeof(_nist_p_521_sqr[0]),
1065                 sizeof(_nist_p_521_sqr)/sizeof(_nist_p_521_sqr[0]),
1066                 0,BN_FLG_STATIC_DATA };
1067
1068         field = &_bignum_nist_p_521; /* just to make sure */
1069
1070         if (BN_is_negative(a) || BN_ucmp(a,&_bignum_nist_p_521_sqr)>=0)
1071                 return BN_nnmod(r, a, field, ctx);
1072
1073         i = BN_ucmp(field, a);
1074         if (i == 0)
1075                 {
1076                 BN_zero(r);
1077                 return 1;
1078                 }
1079         else if (i > 0)
1080                 return (r == a)? 1 : (BN_copy(r ,a) != NULL);
1081
1082         if (r != a)
1083                 {
1084                 if (!bn_wexpand(r,BN_NIST_521_TOP))
1085                         return 0;
1086                 r_d = r->d;
1087                 nist_cp_bn(r_d,a_d, BN_NIST_521_TOP);
1088                 }
1089         else
1090                 r_d = a_d;
1091
1092         /* upper 521 bits, copy ... */
1093         nist_cp_bn_0(t_d,a_d + (BN_NIST_521_TOP-1), top - (BN_NIST_521_TOP-1),BN_NIST_521_TOP);
1094         /* ... and right shift */
1095         for (val=t_d[0],i=0; i<BN_NIST_521_TOP-1; i++)
1096                 {
1097                 tmp = val>>BN_NIST_521_RSHIFT;
1098                 val = t_d[i+1];
1099                 t_d[i] = (tmp | val<<BN_NIST_521_LSHIFT) & BN_MASK2;
1100                 }
1101         t_d[i] = val>>BN_NIST_521_RSHIFT;
1102         /* lower 521 bits */
1103         r_d[i] &= BN_NIST_521_TOP_MASK;
1104
1105         bn_add_words(r_d,r_d,t_d,BN_NIST_521_TOP);
1106         mask = 0-(PTR_SIZE_INT)bn_sub_words(t_d,r_d,_nist_p_521,BN_NIST_521_TOP);
1107         res  = (BN_ULONG *)(((PTR_SIZE_INT)t_d&~mask) |
1108          ((PTR_SIZE_INT)r_d&mask));
1109         nist_cp_bn(r_d,res,BN_NIST_521_TOP);
1110         r->top = BN_NIST_521_TOP;
1111         bn_correct_top(r);
1112
1113         return 1;
1114         }
1115
1116 int (*BN_nist_mod_func(const BIGNUM *p))(BIGNUM *r, const BIGNUM *a, const BIGNUM *field, BN_CTX *ctx)
1117         {
1118         if (BN_ucmp(&_bignum_nist_p_192, p) == 0)
1119                 return BN_nist_mod_192;
1120         if (BN_ucmp(&_bignum_nist_p_224, p) == 0)
1121                 return BN_nist_mod_224;
1122         if (BN_ucmp(&_bignum_nist_p_256, p) == 0)
1123                 return BN_nist_mod_256;
1124         if (BN_ucmp(&_bignum_nist_p_384, p) == 0)
1125                 return BN_nist_mod_384;
1126         if (BN_ucmp(&_bignum_nist_p_521, p) == 0)
1127                 return BN_nist_mod_521;
1128         return 0;
1129         }