1 /*
2 * Copyright 1995-2023 The OpenSSL Project Authors. All Rights Reserved.
3 *
4 * Licensed under the Apache License 2.0 (the "License"). You may not use
5 * this file except in compliance with the License. You can obtain a copy
6 * in the file LICENSE in the source distribution or at
7 * https://www.openssl.org/source/license.html
8 */
9
10 #ifndef OSSL_CRYPTO_BN_LOCAL_H
11 # define OSSL_CRYPTO_BN_LOCAL_H
12
13 /*
14 * The EDK2 build doesn't use bn_conf.h; it sets THIRTY_TWO_BIT or
15 * SIXTY_FOUR_BIT in its own environment since it doesn't re-run our
16 * Configure script and needs to support both 32-bit and 64-bit.
17 */
18 # include <openssl/opensslconf.h>
19
20 # if !defined(OPENSSL_SYS_UEFI)
21 # include "crypto/bn_conf.h"
22 # endif
23
24 # include "crypto/bn.h"
25 # include "internal/cryptlib.h"
26 # include "internal/numbers.h"
27
28 /*
29 * These preprocessor symbols control various aspects of the bignum headers
30 * and library code. They're not defined by any "normal" configuration, as
31 * they are intended for development and testing purposes. NB: defining
32 * them can be useful for debugging application code as well as openssl
33 * itself. BN_DEBUG - turn on various debugging alterations to the bignum
34 * code BN_RAND_DEBUG - uses random poisoning of unused words to trip up
35 * mismanagement of bignum internals. Enable BN_RAND_DEBUG is known to
36 * break some of the OpenSSL tests.
37 */
38 # if defined(BN_RAND_DEBUG) && !defined(BN_DEBUG)
39 # define BN_DEBUG
40 # endif
41 # if defined(BN_RAND_DEBUG)
42 # include <openssl/rand.h>
43 # endif
44
45 /*
46 * This should limit the stack usage due to alloca to about 4K.
47 * BN_SOFT_LIMIT is a soft limit equivalent to 2*OPENSSL_RSA_MAX_MODULUS_BITS.
48 * Beyond that size bn_mul_mont is no longer used, and the constant time
49 * assembler code is disabled, due to the blatant alloca and bn_mul_mont usage.
50 * Note that bn_mul_mont does an alloca that is hidden away in assembly.
51 * It is not recommended to do computations with numbers exceeding this limit,
52 * since the result will be highly version dependent:
53 * While the current OpenSSL version will use non-optimized, but safe code,
54 * previous versions will use optimized code, that may crash due to unexpected
55 * stack overflow, and future versions may very well turn this into a hard
56 * limit.
57 * Note however, that it is possible to override the size limit using
58 * "./config -DBN_SOFT_LIMIT=<limit>" if necessary, and the O/S specific
59 * stack limit is known and taken into consideration.
60 */
61 # ifndef BN_SOFT_LIMIT
62 # define BN_SOFT_LIMIT (4096 / BN_BYTES)
63 # endif
64
65 # ifndef OPENSSL_SMALL_FOOTPRINT
66 # define BN_MUL_COMBA
67 # define BN_SQR_COMBA
68 # define BN_RECURSION
69 # endif
70
71 /*
72 * This next option uses the C libraries (2 word)/(1 word) function. If it is
73 * not defined, I use my C version (which is slower). The reason for this
74 * flag is that when the particular C compiler library routine is used, and
75 * the library is linked with a different compiler, the library is missing.
76 * This mostly happens when the library is built with gcc and then linked
77 * using normal cc. This would be a common occurrence because gcc normally
78 * produces code that is 2 times faster than system compilers for the big
79 * number stuff. For machines with only one compiler (or shared libraries),
80 * this should be on. Again this in only really a problem on machines using
81 * "long long's", are 32bit, and are not using my assembler code.
82 */
83 # if defined(OPENSSL_SYS_MSDOS) || defined(OPENSSL_SYS_WINDOWS) || \
84 defined(OPENSSL_SYS_WIN32) || defined(linux)
85 # define BN_DIV2W
86 # endif
87
88 /*
89 * 64-bit processor with LP64 ABI
90 */
91 # ifdef SIXTY_FOUR_BIT_LONG
92 # define BN_ULLONG unsigned long long
93 # define BN_BITS4 32
94 # define BN_MASK2 (0xffffffffffffffffL)
95 # define BN_MASK2l (0xffffffffL)
96 # define BN_MASK2h (0xffffffff00000000L)
97 # define BN_MASK2h1 (0xffffffff80000000L)
98 # define BN_DEC_CONV (10000000000000000000UL)
99 # define BN_DEC_NUM 19
100 # define BN_DEC_FMT1 "%lu"
101 # define BN_DEC_FMT2 "%019lu"
102 # endif
103
104 /*
105 * 64-bit processor other than LP64 ABI
106 */
107 # ifdef SIXTY_FOUR_BIT
108 # undef BN_LLONG
109 # undef BN_ULLONG
110 # define BN_BITS4 32
111 # define BN_MASK2 (0xffffffffffffffffLL)
112 # define BN_MASK2l (0xffffffffL)
113 # define BN_MASK2h (0xffffffff00000000LL)
114 # define BN_MASK2h1 (0xffffffff80000000LL)
115 # define BN_DEC_CONV (10000000000000000000ULL)
116 # define BN_DEC_NUM 19
117 # define BN_DEC_FMT1 "%llu"
118 # define BN_DEC_FMT2 "%019llu"
119 # endif
120
121 # ifdef THIRTY_TWO_BIT
122 # ifdef BN_LLONG
123 # if defined(_WIN32) && !defined(__GNUC__)
124 # define BN_ULLONG unsigned __int64
125 # else
126 # define BN_ULLONG unsigned long long
127 # endif
128 # endif
129 # define BN_BITS4 16
130 # define BN_MASK2 (0xffffffffL)
131 # define BN_MASK2l (0xffff)
132 # define BN_MASK2h1 (0xffff8000L)
133 # define BN_MASK2h (0xffff0000L)
134 # define BN_DEC_CONV (1000000000L)
135 # define BN_DEC_NUM 9
136 # define BN_DEC_FMT1 "%u"
137 # define BN_DEC_FMT2 "%09u"
138 # endif
139
140
141 /*-
142 * Bignum consistency macros
143 * There is one "API" macro, bn_fix_top(), for stripping leading zeroes from
144 * bignum data after direct manipulations on the data. There is also an
145 * "internal" macro, bn_check_top(), for verifying that there are no leading
146 * zeroes. Unfortunately, some auditing is required due to the fact that
147 * bn_fix_top() has become an overabused duct-tape because bignum data is
148 * occasionally passed around in an inconsistent state. So the following
149 * changes have been made to sort this out;
150 * - bn_fix_top()s implementation has been moved to bn_correct_top()
151 * - if BN_DEBUG isn't defined, bn_fix_top() maps to bn_correct_top(), and
152 * bn_check_top() is as before.
153 * - if BN_DEBUG *is* defined;
154 * - bn_check_top() tries to pollute unused words even if the bignum 'top' is
155 * consistent. (ed: only if BN_RAND_DEBUG is defined)
156 * - bn_fix_top() maps to bn_check_top() rather than "fixing" anything.
157 * The idea is to have debug builds flag up inconsistent bignums when they
158 * occur. If that occurs in a bn_fix_top(), we examine the code in question; if
159 * the use of bn_fix_top() was appropriate (ie. it follows directly after code
160 * that manipulates the bignum) it is converted to bn_correct_top(), and if it
161 * was not appropriate, we convert it permanently to bn_check_top() and track
162 * down the cause of the bug. Eventually, no internal code should be using the
163 * bn_fix_top() macro. External applications and libraries should try this with
164 * their own code too, both in terms of building against the openssl headers
165 * with BN_DEBUG defined *and* linking with a version of OpenSSL built with it
166 * defined. This not only improves external code, it provides more test
167 * coverage for openssl's own code.
168 */
169
170 # ifdef BN_DEBUG
171 /*
172 * The new BN_FLG_FIXED_TOP flag marks vectors that were not treated with
173 * bn_correct_top, in other words such vectors are permitted to have zeros
174 * in most significant limbs. Such vectors are used internally to achieve
175 * execution time invariance for critical operations with private keys.
176 * It's BN_DEBUG-only flag, because user application is not supposed to
177 * observe it anyway. Moreover, optimizing compiler would actually remove
178 * all operations manipulating the bit in question in non-BN_DEBUG build.
179 */
180 # define BN_FLG_FIXED_TOP 0x10000
181 # ifdef BN_RAND_DEBUG
182 # define bn_pollute(a) \
183 do { \
184 const BIGNUM *_bnum1 = (a); \
185 if (_bnum1->top < _bnum1->dmax) { \
186 unsigned char _tmp_char; \
187 /* We cast away const without the compiler knowing, any \
188 * *genuinely* constant variables that aren't mutable \
189 * wouldn't be constructed with top!=dmax. */ \
190 BN_ULONG *_not_const; \
191 memcpy(&_not_const, &_bnum1->d, sizeof(_not_const)); \
192 (void)RAND_bytes(&_tmp_char, 1); /* Debug only - safe to ignore error return */\
193 memset(_not_const + _bnum1->top, _tmp_char, \
194 sizeof(*_not_const) * (_bnum1->dmax - _bnum1->top)); \
195 } \
196 } while(0)
197 # else
198 # define bn_pollute(a)
199 # endif
200 # define bn_check_top(a) \
201 do { \
202 const BIGNUM *_bnum2 = (a); \
203 if (_bnum2 != NULL) { \
204 int _top = _bnum2->top; \
205 (void)ossl_assert((_top == 0 && !_bnum2->neg) || \
206 (_top && ((_bnum2->flags & BN_FLG_FIXED_TOP) \
207 || _bnum2->d[_top - 1] != 0))); \
208 bn_pollute(_bnum2); \
209 } \
210 } while(0)
211
212 # define bn_fix_top(a) bn_check_top(a)
213
214 # define bn_check_size(bn, bits) bn_wcheck_size(bn, ((bits+BN_BITS2-1))/BN_BITS2)
215 # define bn_wcheck_size(bn, words) \
216 do { \
217 const BIGNUM *_bnum2 = (bn); \
218 assert((words) <= (_bnum2)->dmax && \
219 (words) >= (_bnum2)->top); \
220 /* avoid unused variable warning with NDEBUG */ \
221 (void)(_bnum2); \
222 } while(0)
223
224 # else /* !BN_DEBUG */
225
226 # define BN_FLG_FIXED_TOP 0
227 # define bn_pollute(a)
228 # define bn_check_top(a)
229 # define bn_fix_top(a) bn_correct_top(a)
230 # define bn_check_size(bn, bits)
231 # define bn_wcheck_size(bn, words)
232
233 # endif
234
235 BN_ULONG bn_mul_add_words(BN_ULONG *rp, const BN_ULONG *ap, int num,
236 BN_ULONG w);
237 BN_ULONG bn_mul_words(BN_ULONG *rp, const BN_ULONG *ap, int num, BN_ULONG w);
238 void bn_sqr_words(BN_ULONG *rp, const BN_ULONG *ap, int num);
239 BN_ULONG bn_div_words(BN_ULONG h, BN_ULONG l, BN_ULONG d);
240 BN_ULONG bn_add_words(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,
241 int num);
242 BN_ULONG bn_sub_words(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,
243 int num);
244
245 struct bignum_st {
246 BN_ULONG *d; /* Pointer to an array of 'BN_BITS2' bit
247 * chunks. */
248 int top; /* Index of last used d +1. */
249 /* The next are internal book keeping for bn_expand. */
250 int dmax; /* Size of the d array. */
251 int neg; /* one if the number is negative */
252 int flags;
253 };
254
255 /* Used for montgomery multiplication */
256 struct bn_mont_ctx_st {
257 int ri; /* number of bits in R */
258 BIGNUM RR; /* used to convert to montgomery form,
259 possibly zero-padded */
260 BIGNUM N; /* The modulus */
261 BIGNUM Ni; /* R*(1/R mod N) - N*Ni = 1 (Ni is only
262 * stored for bignum algorithm) */
263 BN_ULONG n0[2]; /* least significant word(s) of Ni; (type
264 * changed with 0.9.9, was "BN_ULONG n0;"
265 * before) */
266 int flags;
267 };
268
269 /*
270 * Used for reciprocal division/mod functions It cannot be shared between
271 * threads
272 */
273 struct bn_recp_ctx_st {
274 BIGNUM N; /* the divisor */
275 BIGNUM Nr; /* the reciprocal */
276 int num_bits;
277 int shift;
278 int flags;
279 };
280
281 /* Used for slow "generation" functions. */
282 struct bn_gencb_st {
283 unsigned int ver; /* To handle binary (in)compatibility */
284 void *arg; /* callback-specific data */
285 union {
286 /* if (ver==1) - handles old style callbacks */
287 void (*cb_1) (int, int, void *);
288 /* if (ver==2) - new callback style */
289 int (*cb_2) (int, int, BN_GENCB *);
290 } cb;
291 };
292
293 /*-
294 * BN_window_bits_for_exponent_size -- macro for sliding window mod_exp functions
295 *
296 *
297 * For window size 'w' (w >= 2) and a random 'b' bits exponent,
298 * the number of multiplications is a constant plus on average
299 *
300 * 2^(w-1) + (b-w)/(w+1);
301 *
302 * here 2^(w-1) is for precomputing the table (we actually need
303 * entries only for windows that have the lowest bit set), and
304 * (b-w)/(w+1) is an approximation for the expected number of
305 * w-bit windows, not counting the first one.
306 *
307 * Thus we should use
308 *
309 * w >= 6 if b > 671
310 * w = 5 if 671 > b > 239
311 * w = 4 if 239 > b > 79
312 * w = 3 if 79 > b > 23
313 * w <= 2 if 23 > b
314 *
315 * (with draws in between). Very small exponents are often selected
316 * with low Hamming weight, so we use w = 1 for b <= 23.
317 */
318 # define BN_window_bits_for_exponent_size(b) \
319 ((b) > 671 ? 6 : \
320 (b) > 239 ? 5 : \
321 (b) > 79 ? 4 : \
322 (b) > 23 ? 3 : 1)
323
324 /*
325 * BN_mod_exp_mont_consttime is based on the assumption that the L1 data cache
326 * line width of the target processor is at least the following value.
327 */
328 # define MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH ( 64 )
329 # define MOD_EXP_CTIME_MIN_CACHE_LINE_MASK (MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH - 1)
330
331 /*
332 * Window sizes optimized for fixed window size modular exponentiation
333 * algorithm (BN_mod_exp_mont_consttime). To achieve the security goals of
334 * BN_mode_exp_mont_consttime, the maximum size of the window must not exceed
335 * log_2(MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH). Window size thresholds are
336 * defined for cache line sizes of 32 and 64, cache line sizes where
337 * log_2(32)=5 and log_2(64)=6 respectively. A window size of 7 should only be
338 * used on processors that have a 128 byte or greater cache line size.
339 */
340 # if MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH == 64
341
342 # define BN_window_bits_for_ctime_exponent_size(b) \
343 ((b) > 937 ? 6 : \
344 (b) > 306 ? 5 : \
345 (b) > 89 ? 4 : \
346 (b) > 22 ? 3 : 1)
347 # define BN_MAX_WINDOW_BITS_FOR_CTIME_EXPONENT_SIZE (6)
348
349 # elif MOD_EXP_CTIME_MIN_CACHE_LINE_WIDTH == 32
350
351 # define BN_window_bits_for_ctime_exponent_size(b) \
352 ((b) > 306 ? 5 : \
353 (b) > 89 ? 4 : \
354 (b) > 22 ? 3 : 1)
355 # define BN_MAX_WINDOW_BITS_FOR_CTIME_EXPONENT_SIZE (5)
356
357 # endif
358
359 /* Pentium pro 16,16,16,32,64 */
360 /* Alpha 16,16,16,16.64 */
361 # define BN_MULL_SIZE_NORMAL (16)/* 32 */
362 # define BN_MUL_RECURSIVE_SIZE_NORMAL (16)/* 32 less than */
363 # define BN_SQR_RECURSIVE_SIZE_NORMAL (16)/* 32 */
364 # define BN_MUL_LOW_RECURSIVE_SIZE_NORMAL (32)/* 32 */
365 # define BN_MONT_CTX_SET_SIZE_WORD (64)/* 32 */
366
367 # if !defined(OPENSSL_NO_ASM) && !defined(OPENSSL_NO_INLINE_ASM) && !defined(PEDANTIC)
368 /*
369 * BN_UMULT_HIGH section.
370 * If the compiler doesn't support 2*N integer type, then you have to
371 * replace every N*N multiplication with 4 (N/2)*(N/2) accompanied by some
372 * shifts and additions which unavoidably results in severe performance
373 * penalties. Of course provided that the hardware is capable of producing
374 * 2*N result... That's when you normally start considering assembler
375 * implementation. However! It should be pointed out that some CPUs (e.g.,
376 * PowerPC, Alpha, and IA-64) provide *separate* instruction calculating
377 * the upper half of the product placing the result into a general
378 * purpose register. Now *if* the compiler supports inline assembler,
379 * then it's not impossible to implement the "bignum" routines (and have
380 * the compiler optimize 'em) exhibiting "native" performance in C. That's
381 * what BN_UMULT_HIGH macro is about:-) Note that more recent compilers do
382 * support 2*64 integer type, which is also used here.
383 */
384 # if defined(__SIZEOF_INT128__) && __SIZEOF_INT128__==16 && \
385 (defined(SIXTY_FOUR_BIT) || defined(SIXTY_FOUR_BIT_LONG))
386 # define BN_UMULT_HIGH(a,b) (((uint128_t)(a)*(b))>>64)
387 # define BN_UMULT_LOHI(low,high,a,b) ({ \
388 uint128_t ret=(uint128_t)(a)*(b); \
389 (high)=ret>>64; (low)=ret; })
390 # elif defined(__alpha) && (defined(SIXTY_FOUR_BIT_LONG) || defined(SIXTY_FOUR_BIT))
391 # if defined(__DECC)
392 # include <c_asm.h>
393 # define BN_UMULT_HIGH(a,b) (BN_ULONG)asm("umulh %a0,%a1,%v0",(a),(b))
394 # elif defined(__GNUC__) && __GNUC__>=2
395 # define BN_UMULT_HIGH(a,b) ({ \
396 register BN_ULONG ret; \
397 asm ("umulh %1,%2,%0" \
398 : "=r"(ret) \
399 : "r"(a), "r"(b)); \
400 ret; })
401 # endif /* compiler */
402 # elif defined(_ARCH_PPC64) && defined(SIXTY_FOUR_BIT_LONG)
403 # if defined(__GNUC__) && __GNUC__>=2
404 # define BN_UMULT_HIGH(a,b) ({ \
405 register BN_ULONG ret; \
406 asm ("mulhdu %0,%1,%2" \
407 : "=r"(ret) \
408 : "r"(a), "r"(b)); \
409 ret; })
410 # endif /* compiler */
411 # elif (defined(__x86_64) || defined(__x86_64__)) && \
412 (defined(SIXTY_FOUR_BIT_LONG) || defined(SIXTY_FOUR_BIT))
413 # if defined(__GNUC__) && __GNUC__>=2
414 # define BN_UMULT_HIGH(a,b) ({ \
415 register BN_ULONG ret,discard; \
416 asm ("mulq %3" \
417 : "=a"(discard),"=d"(ret) \
418 : "a"(a), "g"(b) \
419 : "cc"); \
420 ret; })
421 # define BN_UMULT_LOHI(low,high,a,b) \
422 asm ("mulq %3" \
423 : "=a"(low),"=d"(high) \
424 : "a"(a),"g"(b) \
425 : "cc");
426 # endif
427 # elif (defined(_M_AMD64) || defined(_M_X64)) && defined(SIXTY_FOUR_BIT)
428 # if defined(_MSC_VER) && _MSC_VER>=1400
429 unsigned __int64 __umulh(unsigned __int64 a, unsigned __int64 b);
430 unsigned __int64 _umul128(unsigned __int64 a, unsigned __int64 b,
431 unsigned __int64 *h);
432 # pragma intrinsic(__umulh,_umul128)
433 # define BN_UMULT_HIGH(a,b) __umulh((a),(b))
434 # define BN_UMULT_LOHI(low,high,a,b) ((low)=_umul128((a),(b),&(high)))
435 # endif
436 # elif defined(__mips) && (defined(SIXTY_FOUR_BIT) || defined(SIXTY_FOUR_BIT_LONG))
437 # if defined(__GNUC__) && __GNUC__>=2
438 # define BN_UMULT_HIGH(a,b) ({ \
439 register BN_ULONG ret; \
440 asm ("dmultu %1,%2" \
441 : "=h"(ret) \
442 : "r"(a), "r"(b) : "l"); \
443 ret; })
444 # define BN_UMULT_LOHI(low,high,a,b) \
445 asm ("dmultu %2,%3" \
446 : "=l"(low),"=h"(high) \
447 : "r"(a), "r"(b));
448 # endif
449 # elif defined(__aarch64__) && defined(SIXTY_FOUR_BIT_LONG)
450 # if defined(__GNUC__) && __GNUC__>=2
451 # define BN_UMULT_HIGH(a,b) ({ \
452 register BN_ULONG ret; \
453 asm ("umulh %0,%1,%2" \
454 : "=r"(ret) \
455 : "r"(a), "r"(b)); \
456 ret; })
457 # endif
458 # endif /* cpu */
459 # endif /* OPENSSL_NO_ASM */
460
461 # ifdef BN_RAND_DEBUG
462 # define bn_clear_top2max(a) \
463 { \
464 int ind = (a)->dmax - (a)->top; \
465 BN_ULONG *ftl = &(a)->d[(a)->top-1]; \
466 for (; ind != 0; ind--) \
467 *(++ftl) = 0x0; \
468 }
469 # else
470 # define bn_clear_top2max(a)
471 # endif
472
473 # ifdef BN_LLONG
474 /*******************************************************************
475 * Using the long long type, has to be twice as wide as BN_ULONG...
476 */
477 # define Lw(t) (((BN_ULONG)(t))&BN_MASK2)
478 # define Hw(t) (((BN_ULONG)((t)>>BN_BITS2))&BN_MASK2)
479
480 # define mul_add(r,a,w,c) { \
481 BN_ULLONG t; \
482 t=(BN_ULLONG)w * (a) + (r) + (c); \
483 (r)= Lw(t); \
484 (c)= Hw(t); \
485 }
486
487 # define mul(r,a,w,c) { \
488 BN_ULLONG t; \
489 t=(BN_ULLONG)w * (a) + (c); \
490 (r)= Lw(t); \
491 (c)= Hw(t); \
492 }
493
494 # define sqr(r0,r1,a) { \
495 BN_ULLONG t; \
496 t=(BN_ULLONG)(a)*(a); \
497 (r0)=Lw(t); \
498 (r1)=Hw(t); \
499 }
500
501 # elif defined(BN_UMULT_LOHI)
502 # define mul_add(r,a,w,c) { \
503 BN_ULONG high,low,ret,tmp=(a); \
504 ret = (r); \
505 BN_UMULT_LOHI(low,high,w,tmp); \
506 ret += (c); \
507 (c) = (ret<(c)); \
508 (c) += high; \
509 ret += low; \
510 (c) += (ret<low); \
511 (r) = ret; \
512 }
513
514 # define mul(r,a,w,c) { \
515 BN_ULONG high,low,ret,ta=(a); \
516 BN_UMULT_LOHI(low,high,w,ta); \
517 ret = low + (c); \
518 (c) = high; \
519 (c) += (ret<low); \
520 (r) = ret; \
521 }
522
523 # define sqr(r0,r1,a) { \
524 BN_ULONG tmp=(a); \
525 BN_UMULT_LOHI(r0,r1,tmp,tmp); \
526 }
527
528 # elif defined(BN_UMULT_HIGH)
529 # define mul_add(r,a,w,c) { \
530 BN_ULONG high,low,ret,tmp=(a); \
531 ret = (r); \
532 high= BN_UMULT_HIGH(w,tmp); \
533 ret += (c); \
534 low = (w) * tmp; \
535 (c) = (ret<(c)); \
536 (c) += high; \
537 ret += low; \
538 (c) += (ret<low); \
539 (r) = ret; \
540 }
541
542 # define mul(r,a,w,c) { \
543 BN_ULONG high,low,ret,ta=(a); \
544 low = (w) * ta; \
545 high= BN_UMULT_HIGH(w,ta); \
546 ret = low + (c); \
547 (c) = high; \
548 (c) += (ret<low); \
549 (r) = ret; \
550 }
551
552 # define sqr(r0,r1,a) { \
553 BN_ULONG tmp=(a); \
554 (r0) = tmp * tmp; \
555 (r1) = BN_UMULT_HIGH(tmp,tmp); \
556 }
557
558 # else
559 /*************************************************************
560 * No long long type
561 */
562
563 # define LBITS(a) ((a)&BN_MASK2l)
564 # define HBITS(a) (((a)>>BN_BITS4)&BN_MASK2l)
565 # define L2HBITS(a) (((a)<<BN_BITS4)&BN_MASK2)
566
567 # define LLBITS(a) ((a)&BN_MASKl)
568 # define LHBITS(a) (((a)>>BN_BITS2)&BN_MASKl)
569 # define LL2HBITS(a) ((BN_ULLONG)((a)&BN_MASKl)<<BN_BITS2)
570
571 # define mul64(l,h,bl,bh) \
572 { \
573 BN_ULONG m,m1,lt,ht; \
574 \
575 lt=l; \
576 ht=h; \
577 m =(bh)*(lt); \
578 lt=(bl)*(lt); \
579 m1=(bl)*(ht); \
580 ht =(bh)*(ht); \
581 m=(m+m1)&BN_MASK2; ht += L2HBITS((BN_ULONG)(m < m1)); \
582 ht+=HBITS(m); \
583 m1=L2HBITS(m); \
584 lt=(lt+m1)&BN_MASK2; ht += (lt < m1); \
585 (l)=lt; \
586 (h)=ht; \
587 }
588
589 # define sqr64(lo,ho,in) \
590 { \
591 BN_ULONG l,h,m; \
592 \
593 h=(in); \
594 l=LBITS(h); \
595 h=HBITS(h); \
596 m =(l)*(h); \
597 l*=l; \
598 h*=h; \
599 h+=(m&BN_MASK2h1)>>(BN_BITS4-1); \
600 m =(m&BN_MASK2l)<<(BN_BITS4+1); \
601 l=(l+m)&BN_MASK2; h += (l < m); \
602 (lo)=l; \
603 (ho)=h; \
604 }
605
606 # define mul_add(r,a,bl,bh,c) { \
607 BN_ULONG l,h; \
608 \
609 h= (a); \
610 l=LBITS(h); \
611 h=HBITS(h); \
612 mul64(l,h,(bl),(bh)); \
613 \
614 /* non-multiply part */ \
615 l=(l+(c))&BN_MASK2; h += (l < (c)); \
616 (c)=(r); \
617 l=(l+(c))&BN_MASK2; h += (l < (c)); \
618 (c)=h&BN_MASK2; \
619 (r)=l; \
620 }
621
622 # define mul(r,a,bl,bh,c) { \
623 BN_ULONG l,h; \
624 \
625 h= (a); \
626 l=LBITS(h); \
627 h=HBITS(h); \
628 mul64(l,h,(bl),(bh)); \
629 \
630 /* non-multiply part */ \
631 l+=(c); h += ((l&BN_MASK2) < (c)); \
632 (c)=h&BN_MASK2; \
633 (r)=l&BN_MASK2; \
634 }
635 # endif /* !BN_LLONG */
636
637 void BN_RECP_CTX_init(BN_RECP_CTX *recp);
638 void BN_MONT_CTX_init(BN_MONT_CTX *ctx);
639
640 void bn_init(BIGNUM *a);
641 void bn_mul_normal(BN_ULONG *r, BN_ULONG *a, int na, BN_ULONG *b, int nb);
642 void bn_mul_comba8(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b);
643 void bn_mul_comba4(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b);
644 void bn_sqr_normal(BN_ULONG *r, const BN_ULONG *a, int n, BN_ULONG *tmp);
645 void bn_sqr_comba8(BN_ULONG *r, const BN_ULONG *a);
646 void bn_sqr_comba4(BN_ULONG *r, const BN_ULONG *a);
647 int bn_cmp_words(const BN_ULONG *a, const BN_ULONG *b, int n);
648 int bn_cmp_part_words(const BN_ULONG *a, const BN_ULONG *b, int cl, int dl);
649 void bn_mul_recursive(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n2,
650 int dna, int dnb, BN_ULONG *t);
651 void bn_mul_part_recursive(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b,
652 int n, int tna, int tnb, BN_ULONG *t);
653 void bn_sqr_recursive(BN_ULONG *r, const BN_ULONG *a, int n2, BN_ULONG *t);
654 void bn_mul_low_normal(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n);
655 void bn_mul_low_recursive(BN_ULONG *r, BN_ULONG *a, BN_ULONG *b, int n2,
656 BN_ULONG *t);
657 BN_ULONG bn_sub_part_words(BN_ULONG *r, const BN_ULONG *a, const BN_ULONG *b,
658 int cl, int dl);
659 int bn_mul_mont(BN_ULONG *rp, const BN_ULONG *ap, const BN_ULONG *bp,
660 const BN_ULONG *np, const BN_ULONG *n0, int num);
661 void bn_correct_top_consttime(BIGNUM *a);
662 BIGNUM *int_bn_mod_inverse(BIGNUM *in,
663 const BIGNUM *a, const BIGNUM *n, BN_CTX *ctx,
664 int *noinv);
665
bn_expand(BIGNUM * a,int bits)666 static ossl_inline BIGNUM *bn_expand(BIGNUM *a, int bits)
667 {
668 if (bits > (INT_MAX - BN_BITS2 + 1))
669 return NULL;
670
671 if (((bits+BN_BITS2-1)/BN_BITS2) <= (a)->dmax)
672 return a;
673
674 return bn_expand2((a),(bits+BN_BITS2-1)/BN_BITS2);
675 }
676
677 int ossl_bn_check_prime(const BIGNUM *w, int checks, BN_CTX *ctx,
678 int do_trial_division, BN_GENCB *cb);
679
680 #endif
681