• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  * Minimal code for RSA support from LibTomMath 0.41
3  * http://libtom.org/
4  * http://libtom.org/files/ltm-0.41.tar.bz2
5  * This library was released in public domain by Tom St Denis.
6  *
7  * The combination in this file may not use all of the optimized algorithms
8  * from LibTomMath and may be considerable slower than the LibTomMath with its
9  * default settings. The main purpose of having this version here is to make it
10  * easier to build bignum.c wrapper without having to install and build an
11  * external library.
12  *
13  * If CONFIG_INTERNAL_LIBTOMMATH is defined, bignum.c includes this
14  * libtommath.c file instead of using the external LibTomMath library.
15  */
16 
17 #ifndef CHAR_BIT
18 #define CHAR_BIT 8
19 #endif
20 
21 #define BN_MP_INVMOD_C
22 #define BN_S_MP_EXPTMOD_C /* Note: #undef in tommath_superclass.h; this would
23 			   * require BN_MP_EXPTMOD_FAST_C instead */
24 #define BN_S_MP_MUL_DIGS_C
25 #define BN_MP_INVMOD_SLOW_C
26 #define BN_S_MP_SQR_C
27 #define BN_S_MP_MUL_HIGH_DIGS_C /* Note: #undef in tommath_superclass.h; this
28 				 * would require other than mp_reduce */
29 
30 #ifdef LTM_FAST
31 
32 /* Use faster div at the cost of about 1 kB */
33 #define BN_MP_MUL_D_C
34 
35 /* Include faster exptmod (Montgomery) at the cost of about 2.5 kB in code */
36 #define BN_MP_EXPTMOD_FAST_C
37 #define BN_MP_MONTGOMERY_SETUP_C
38 #define BN_FAST_MP_MONTGOMERY_REDUCE_C
39 #define BN_MP_MONTGOMERY_CALC_NORMALIZATION_C
40 #define BN_MP_MUL_2_C
41 
42 /* Include faster sqr at the cost of about 0.5 kB in code */
43 #define BN_FAST_S_MP_SQR_C
44 
45 #else /* LTM_FAST */
46 
47 #define BN_MP_DIV_SMALL
48 #define BN_MP_INIT_MULTI_C
49 #define BN_MP_CLEAR_MULTI_C
50 #define BN_MP_ABS_C
51 #endif /* LTM_FAST */
52 
53 /* Current uses do not require support for negative exponent in exptmod, so we
54  * can save about 1.5 kB in leaving out invmod. */
55 #define LTM_NO_NEG_EXP
56 
57 /* from tommath.h */
58 
59 #ifndef MIN
60    #define MIN(x,y) ((x)<(y)?(x):(y))
61 #endif
62 
63 #ifndef MAX
64    #define MAX(x,y) ((x)>(y)?(x):(y))
65 #endif
66 
67 #define  OPT_CAST(x)
68 
69 typedef unsigned long mp_digit;
70 typedef u64 mp_word;
71 
72 #define DIGIT_BIT          28
73 #define MP_28BIT
74 
75 
76 #define XMALLOC  os_malloc
77 #define XFREE    os_free
78 #define XREALLOC os_realloc
79 
80 
81 #define MP_MASK          ((((mp_digit)1)<<((mp_digit)DIGIT_BIT))-((mp_digit)1))
82 
83 #define MP_LT        -1   /* less than */
84 #define MP_EQ         0   /* equal to */
85 #define MP_GT         1   /* greater than */
86 
87 #define MP_ZPOS       0   /* positive integer */
88 #define MP_NEG        1   /* negative */
89 
90 #define MP_OKAY       0   /* ok result */
91 #define MP_MEM        -2  /* out of mem */
92 #define MP_VAL        -3  /* invalid input */
93 
94 #define MP_YES        1   /* yes response */
95 #define MP_NO         0   /* no response */
96 
97 typedef int           mp_err;
98 
99 /* define this to use lower memory usage routines (exptmods mostly) */
100 #define MP_LOW_MEM
101 
102 /* default precision */
103 #ifndef MP_PREC
104    #ifndef MP_LOW_MEM
105       #define MP_PREC                 32     /* default digits of precision */
106    #else
107       #define MP_PREC                 8      /* default digits of precision */
108    #endif
109 #endif
110 
111 /* size of comba arrays, should be at least 2 * 2**(BITS_PER_WORD - BITS_PER_DIGIT*2) */
112 #define MP_WARRAY               (1 << (sizeof(mp_word) * CHAR_BIT - 2 * DIGIT_BIT + 1))
113 
114 /* the infamous mp_int structure */
115 typedef struct  {
116     int used, alloc, sign;
117     mp_digit *dp;
118 } mp_int;
119 
120 
121 /* ---> Basic Manipulations <--- */
122 #define mp_iszero(a) (((a)->used == 0) ? MP_YES : MP_NO)
123 #define mp_iseven(a) (((a)->used > 0 && (((a)->dp[0] & 1) == 0)) ? MP_YES : MP_NO)
124 #define mp_isodd(a)  (((a)->used > 0 && (((a)->dp[0] & 1) == 1)) ? MP_YES : MP_NO)
125 
126 
127 /* prototypes for copied functions */
128 #define s_mp_mul(a, b, c) s_mp_mul_digs(a, b, c, (a)->used + (b)->used + 1)
129 static int s_mp_exptmod(mp_int * G, mp_int * X, mp_int * P, mp_int * Y, int redmode);
130 static int s_mp_mul_digs (mp_int * a, mp_int * b, mp_int * c, int digs);
131 static int s_mp_sqr(mp_int * a, mp_int * b);
132 static int s_mp_mul_high_digs(mp_int * a, mp_int * b, mp_int * c, int digs);
133 
134 static int fast_s_mp_mul_digs (mp_int * a, mp_int * b, mp_int * c, int digs);
135 
136 #ifdef BN_MP_INIT_MULTI_C
137 static int mp_init_multi(mp_int *mp, ...);
138 #endif
139 #ifdef BN_MP_CLEAR_MULTI_C
140 static void mp_clear_multi(mp_int *mp, ...);
141 #endif
142 static int mp_lshd(mp_int * a, int b);
143 static void mp_set(mp_int * a, mp_digit b);
144 static void mp_clamp(mp_int * a);
145 static void mp_exch(mp_int * a, mp_int * b);
146 static void mp_rshd(mp_int * a, int b);
147 static void mp_zero(mp_int * a);
148 static int mp_mod_2d(mp_int * a, int b, mp_int * c);
149 static int mp_div_2d(mp_int * a, int b, mp_int * c, mp_int * d);
150 static int mp_init_copy(mp_int * a, mp_int * b);
151 static int mp_mul_2d(mp_int * a, int b, mp_int * c);
152 #ifndef LTM_NO_NEG_EXP
153 static int mp_div_2(mp_int * a, mp_int * b);
154 static int mp_invmod(mp_int * a, mp_int * b, mp_int * c);
155 static int mp_invmod_slow(mp_int * a, mp_int * b, mp_int * c);
156 #endif /* LTM_NO_NEG_EXP */
157 static int mp_copy(mp_int * a, mp_int * b);
158 static int mp_count_bits(mp_int * a);
159 static int mp_div(mp_int * a, mp_int * b, mp_int * c, mp_int * d);
160 static int mp_mod(mp_int * a, mp_int * b, mp_int * c);
161 static int mp_grow(mp_int * a, int size);
162 static int mp_cmp_mag(mp_int * a, mp_int * b);
163 #ifdef BN_MP_ABS_C
164 static int mp_abs(mp_int * a, mp_int * b);
165 #endif
166 static int mp_sqr(mp_int * a, mp_int * b);
167 static int mp_reduce_2k_l(mp_int *a, mp_int *n, mp_int *d);
168 static int mp_reduce_2k_setup_l(mp_int *a, mp_int *d);
169 static int mp_2expt(mp_int * a, int b);
170 static int mp_reduce_setup(mp_int * a, mp_int * b);
171 static int mp_reduce(mp_int * x, mp_int * m, mp_int * mu);
172 static int mp_init_size(mp_int * a, int size);
173 #ifdef BN_MP_EXPTMOD_FAST_C
174 static int mp_exptmod_fast (mp_int * G, mp_int * X, mp_int * P, mp_int * Y, int redmode);
175 #endif /* BN_MP_EXPTMOD_FAST_C */
176 #ifdef BN_FAST_S_MP_SQR_C
177 static int fast_s_mp_sqr (mp_int * a, mp_int * b);
178 #endif /* BN_FAST_S_MP_SQR_C */
179 #ifdef BN_MP_MUL_D_C
180 static int mp_mul_d (mp_int * a, mp_digit b, mp_int * c);
181 #endif /* BN_MP_MUL_D_C */
182 
183 
184 
185 /* functions from bn_<func name>.c */
186 
187 
188 /* reverse an array, used for radix code */
bn_reverse(unsigned char * s,int len)189 static void bn_reverse (unsigned char *s, int len)
190 {
191   int     ix, iy;
192   unsigned char t;
193 
194   ix = 0;
195   iy = len - 1;
196   while (ix < iy) {
197     t     = s[ix];
198     s[ix] = s[iy];
199     s[iy] = t;
200     ++ix;
201     --iy;
202   }
203 }
204 
205 
206 /* low level addition, based on HAC pp.594, Algorithm 14.7 */
s_mp_add(mp_int * a,mp_int * b,mp_int * c)207 static int s_mp_add (mp_int * a, mp_int * b, mp_int * c)
208 {
209   mp_int *x;
210   int     olduse, res, min, max;
211 
212   /* find sizes, we let |a| <= |b| which means we have to sort
213    * them.  "x" will point to the input with the most digits
214    */
215   if (a->used > b->used) {
216     min = b->used;
217     max = a->used;
218     x = a;
219   } else {
220     min = a->used;
221     max = b->used;
222     x = b;
223   }
224 
225   /* init result */
226   if (c->alloc < max + 1) {
227     if ((res = mp_grow (c, max + 1)) != MP_OKAY) {
228       return res;
229     }
230   }
231 
232   /* get old used digit count and set new one */
233   olduse = c->used;
234   c->used = max + 1;
235 
236   {
237     register mp_digit u, *tmpa, *tmpb, *tmpc;
238     register int i;
239 
240     /* alias for digit pointers */
241 
242     /* first input */
243     tmpa = a->dp;
244 
245     /* second input */
246     tmpb = b->dp;
247 
248     /* destination */
249     tmpc = c->dp;
250 
251     /* zero the carry */
252     u = 0;
253     for (i = 0; i < min; i++) {
254       /* Compute the sum at one digit, T[i] = A[i] + B[i] + U */
255       *tmpc = *tmpa++ + *tmpb++ + u;
256 
257       /* U = carry bit of T[i] */
258       u = *tmpc >> ((mp_digit)DIGIT_BIT);
259 
260       /* take away carry bit from T[i] */
261       *tmpc++ &= MP_MASK;
262     }
263 
264     /* now copy higher words if any, that is in A+B
265      * if A or B has more digits add those in
266      */
267     if (min != max) {
268       for (; i < max; i++) {
269         /* T[i] = X[i] + U */
270         *tmpc = x->dp[i] + u;
271 
272         /* U = carry bit of T[i] */
273         u = *tmpc >> ((mp_digit)DIGIT_BIT);
274 
275         /* take away carry bit from T[i] */
276         *tmpc++ &= MP_MASK;
277       }
278     }
279 
280     /* add carry */
281     *tmpc++ = u;
282 
283     /* clear digits above oldused */
284     for (i = c->used; i < olduse; i++) {
285       *tmpc++ = 0;
286     }
287   }
288 
289   mp_clamp (c);
290   return MP_OKAY;
291 }
292 
293 
294 /* low level subtraction (assumes |a| > |b|), HAC pp.595 Algorithm 14.9 */
s_mp_sub(mp_int * a,mp_int * b,mp_int * c)295 static int s_mp_sub (mp_int * a, mp_int * b, mp_int * c)
296 {
297   int     olduse, res, min, max;
298 
299   /* find sizes */
300   min = b->used;
301   max = a->used;
302 
303   /* init result */
304   if (c->alloc < max) {
305     if ((res = mp_grow (c, max)) != MP_OKAY) {
306       return res;
307     }
308   }
309   olduse = c->used;
310   c->used = max;
311 
312   {
313     register mp_digit u, *tmpa, *tmpb, *tmpc;
314     register int i;
315 
316     /* alias for digit pointers */
317     tmpa = a->dp;
318     tmpb = b->dp;
319     tmpc = c->dp;
320 
321     /* set carry to zero */
322     u = 0;
323     for (i = 0; i < min; i++) {
324       /* T[i] = A[i] - B[i] - U */
325       *tmpc = *tmpa++ - *tmpb++ - u;
326 
327       /* U = carry bit of T[i]
328        * Note this saves performing an AND operation since
329        * if a carry does occur it will propagate all the way to the
330        * MSB.  As a result a single shift is enough to get the carry
331        */
332       u = *tmpc >> ((mp_digit)(CHAR_BIT * sizeof (mp_digit) - 1));
333 
334       /* Clear carry from T[i] */
335       *tmpc++ &= MP_MASK;
336     }
337 
338     /* now copy higher words if any, e.g. if A has more digits than B  */
339     for (; i < max; i++) {
340       /* T[i] = A[i] - U */
341       *tmpc = *tmpa++ - u;
342 
343       /* U = carry bit of T[i] */
344       u = *tmpc >> ((mp_digit)(CHAR_BIT * sizeof (mp_digit) - 1));
345 
346       /* Clear carry from T[i] */
347       *tmpc++ &= MP_MASK;
348     }
349 
350     /* clear digits above used (since we may not have grown result above) */
351     for (i = c->used; i < olduse; i++) {
352       *tmpc++ = 0;
353     }
354   }
355 
356   mp_clamp (c);
357   return MP_OKAY;
358 }
359 
360 
361 /* init a new mp_int */
mp_init(mp_int * a)362 static int mp_init (mp_int * a)
363 {
364   int i;
365 
366   /* allocate memory required and clear it */
367   a->dp = OPT_CAST(mp_digit) XMALLOC (sizeof (mp_digit) * MP_PREC);
368   if (a->dp == NULL) {
369     return MP_MEM;
370   }
371 
372   /* set the digits to zero */
373   for (i = 0; i < MP_PREC; i++) {
374       a->dp[i] = 0;
375   }
376 
377   /* set the used to zero, allocated digits to the default precision
378    * and sign to positive */
379   a->used  = 0;
380   a->alloc = MP_PREC;
381   a->sign  = MP_ZPOS;
382 
383   return MP_OKAY;
384 }
385 
386 
387 /* clear one (frees)  */
mp_clear(mp_int * a)388 static void mp_clear (mp_int * a)
389 {
390   int i;
391 
392   /* only do anything if a hasn't been freed previously */
393   if (a->dp != NULL) {
394     /* first zero the digits */
395     for (i = 0; i < a->used; i++) {
396         a->dp[i] = 0;
397     }
398 
399     /* free ram */
400     XFREE(a->dp);
401 
402     /* reset members to make debugging easier */
403     a->dp    = NULL;
404     a->alloc = a->used = 0;
405     a->sign  = MP_ZPOS;
406   }
407 }
408 
409 
410 /* high level addition (handles signs) */
mp_add(mp_int * a,mp_int * b,mp_int * c)411 static int mp_add (mp_int * a, mp_int * b, mp_int * c)
412 {
413   int     sa, sb, res;
414 
415   /* get sign of both inputs */
416   sa = a->sign;
417   sb = b->sign;
418 
419   /* handle two cases, not four */
420   if (sa == sb) {
421     /* both positive or both negative */
422     /* add their magnitudes, copy the sign */
423     c->sign = sa;
424     res = s_mp_add (a, b, c);
425   } else {
426     /* one positive, the other negative */
427     /* subtract the one with the greater magnitude from */
428     /* the one of the lesser magnitude.  The result gets */
429     /* the sign of the one with the greater magnitude. */
430     if (mp_cmp_mag (a, b) == MP_LT) {
431       c->sign = sb;
432       res = s_mp_sub (b, a, c);
433     } else {
434       c->sign = sa;
435       res = s_mp_sub (a, b, c);
436     }
437   }
438   return res;
439 }
440 
441 
442 /* high level subtraction (handles signs) */
mp_sub(mp_int * a,mp_int * b,mp_int * c)443 static int mp_sub (mp_int * a, mp_int * b, mp_int * c)
444 {
445   int     sa, sb, res;
446 
447   sa = a->sign;
448   sb = b->sign;
449 
450   if (sa != sb) {
451     /* subtract a negative from a positive, OR */
452     /* subtract a positive from a negative. */
453     /* In either case, ADD their magnitudes, */
454     /* and use the sign of the first number. */
455     c->sign = sa;
456     res = s_mp_add (a, b, c);
457   } else {
458     /* subtract a positive from a positive, OR */
459     /* subtract a negative from a negative. */
460     /* First, take the difference between their */
461     /* magnitudes, then... */
462     if (mp_cmp_mag (a, b) != MP_LT) {
463       /* Copy the sign from the first */
464       c->sign = sa;
465       /* The first has a larger or equal magnitude */
466       res = s_mp_sub (a, b, c);
467     } else {
468       /* The result has the *opposite* sign from */
469       /* the first number. */
470       c->sign = (sa == MP_ZPOS) ? MP_NEG : MP_ZPOS;
471       /* The second has a larger magnitude */
472       res = s_mp_sub (b, a, c);
473     }
474   }
475   return res;
476 }
477 
478 
479 /* high level multiplication (handles sign) */
mp_mul(mp_int * a,mp_int * b,mp_int * c)480 static int mp_mul (mp_int * a, mp_int * b, mp_int * c)
481 {
482   int     res, neg;
483   neg = (a->sign == b->sign) ? MP_ZPOS : MP_NEG;
484 
485   /* use Toom-Cook? */
486 #ifdef BN_MP_TOOM_MUL_C
487   if (MIN (a->used, b->used) >= TOOM_MUL_CUTOFF) {
488     res = mp_toom_mul(a, b, c);
489   } else
490 #endif
491 #ifdef BN_MP_KARATSUBA_MUL_C
492   /* use Karatsuba? */
493   if (MIN (a->used, b->used) >= KARATSUBA_MUL_CUTOFF) {
494     res = mp_karatsuba_mul (a, b, c);
495   } else
496 #endif
497   {
498     /* can we use the fast multiplier?
499      *
500      * The fast multiplier can be used if the output will
501      * have less than MP_WARRAY digits and the number of
502      * digits won't affect carry propagation
503      */
504 #ifdef BN_FAST_S_MP_MUL_DIGS_C
505     int     digs = a->used + b->used + 1;
506 
507     if ((digs < MP_WARRAY) &&
508         MIN(a->used, b->used) <=
509         (1 << ((CHAR_BIT * sizeof (mp_word)) - (2 * DIGIT_BIT)))) {
510       res = fast_s_mp_mul_digs (a, b, c, digs);
511     } else
512 #endif
513 #ifdef BN_S_MP_MUL_DIGS_C
514       res = s_mp_mul (a, b, c); /* uses s_mp_mul_digs */
515 #else
516 #error mp_mul could fail
517       res = MP_VAL;
518 #endif
519 
520   }
521   c->sign = (c->used > 0) ? neg : MP_ZPOS;
522   return res;
523 }
524 
525 
526 /* d = a * b (mod c) */
mp_mulmod(mp_int * a,mp_int * b,mp_int * c,mp_int * d)527 static int mp_mulmod (mp_int * a, mp_int * b, mp_int * c, mp_int * d)
528 {
529   int     res;
530   mp_int  t;
531 
532   if ((res = mp_init (&t)) != MP_OKAY) {
533     return res;
534   }
535 
536   if ((res = mp_mul (a, b, &t)) != MP_OKAY) {
537     mp_clear (&t);
538     return res;
539   }
540   res = mp_mod (&t, c, d);
541   mp_clear (&t);
542   return res;
543 }
544 
545 
546 /* c = a mod b, 0 <= c < b */
mp_mod(mp_int * a,mp_int * b,mp_int * c)547 static int mp_mod (mp_int * a, mp_int * b, mp_int * c)
548 {
549   mp_int  t;
550   int     res;
551 
552   if ((res = mp_init (&t)) != MP_OKAY) {
553     return res;
554   }
555 
556   if ((res = mp_div (a, b, NULL, &t)) != MP_OKAY) {
557     mp_clear (&t);
558     return res;
559   }
560 
561   if (t.sign != b->sign) {
562     res = mp_add (b, &t, c);
563   } else {
564     res = MP_OKAY;
565     mp_exch (&t, c);
566   }
567 
568   mp_clear (&t);
569   return res;
570 }
571 
572 
573 /* this is a shell function that calls either the normal or Montgomery
574  * exptmod functions.  Originally the call to the montgomery code was
575  * embedded in the normal function but that wasted alot of stack space
576  * for nothing (since 99% of the time the Montgomery code would be called)
577  */
mp_exptmod(mp_int * G,mp_int * X,mp_int * P,mp_int * Y)578 static int mp_exptmod (mp_int * G, mp_int * X, mp_int * P, mp_int * Y)
579 {
580   int dr;
581 
582   /* modulus P must be positive */
583   if (P->sign == MP_NEG) {
584      return MP_VAL;
585   }
586 
587   /* if exponent X is negative we have to recurse */
588   if (X->sign == MP_NEG) {
589 #ifdef LTM_NO_NEG_EXP
590         return MP_VAL;
591 #else /* LTM_NO_NEG_EXP */
592 #ifdef BN_MP_INVMOD_C
593      mp_int tmpG, tmpX;
594      int err;
595 
596      /* first compute 1/G mod P */
597      if ((err = mp_init(&tmpG)) != MP_OKAY) {
598         return err;
599      }
600      if ((err = mp_invmod(G, P, &tmpG)) != MP_OKAY) {
601         mp_clear(&tmpG);
602         return err;
603      }
604 
605      /* now get |X| */
606      if ((err = mp_init(&tmpX)) != MP_OKAY) {
607         mp_clear(&tmpG);
608         return err;
609      }
610      if ((err = mp_abs(X, &tmpX)) != MP_OKAY) {
611         mp_clear_multi(&tmpG, &tmpX, NULL);
612         return err;
613      }
614 
615      /* and now compute (1/G)**|X| instead of G**X [X < 0] */
616      err = mp_exptmod(&tmpG, &tmpX, P, Y);
617      mp_clear_multi(&tmpG, &tmpX, NULL);
618      return err;
619 #else
620 #error mp_exptmod would always fail
621      /* no invmod */
622      return MP_VAL;
623 #endif
624 #endif /* LTM_NO_NEG_EXP */
625   }
626 
627 /* modified diminished radix reduction */
628 #if defined(BN_MP_REDUCE_IS_2K_L_C) && defined(BN_MP_REDUCE_2K_L_C) && defined(BN_S_MP_EXPTMOD_C)
629   if (mp_reduce_is_2k_l(P) == MP_YES) {
630      return s_mp_exptmod(G, X, P, Y, 1);
631   }
632 #endif
633 
634 #ifdef BN_MP_DR_IS_MODULUS_C
635   /* is it a DR modulus? */
636   dr = mp_dr_is_modulus(P);
637 #else
638   /* default to no */
639   dr = 0;
640 #endif
641 
642 #ifdef BN_MP_REDUCE_IS_2K_C
643   /* if not, is it a unrestricted DR modulus? */
644   if (dr == 0) {
645      dr = mp_reduce_is_2k(P) << 1;
646   }
647 #endif
648 
649   /* if the modulus is odd or dr != 0 use the montgomery method */
650 #ifdef BN_MP_EXPTMOD_FAST_C
651   if (mp_isodd (P) == 1 || dr !=  0) {
652     return mp_exptmod_fast (G, X, P, Y, dr);
653   } else {
654 #endif
655 #ifdef BN_S_MP_EXPTMOD_C
656     /* otherwise use the generic Barrett reduction technique */
657     return s_mp_exptmod (G, X, P, Y, 0);
658 #else
659 #error mp_exptmod could fail
660     /* no exptmod for evens */
661     return MP_VAL;
662 #endif
663 #ifdef BN_MP_EXPTMOD_FAST_C
664   }
665 #endif
666 }
667 
668 
669 /* compare two ints (signed)*/
mp_cmp(mp_int * a,mp_int * b)670 static int mp_cmp (mp_int * a, mp_int * b)
671 {
672   /* compare based on sign */
673   if (a->sign != b->sign) {
674      if (a->sign == MP_NEG) {
675         return MP_LT;
676      } else {
677         return MP_GT;
678      }
679   }
680 
681   /* compare digits */
682   if (a->sign == MP_NEG) {
683      /* if negative compare opposite direction */
684      return mp_cmp_mag(b, a);
685   } else {
686      return mp_cmp_mag(a, b);
687   }
688 }
689 
690 
691 /* compare a digit */
mp_cmp_d(mp_int * a,mp_digit b)692 static int mp_cmp_d(mp_int * a, mp_digit b)
693 {
694   /* compare based on sign */
695   if (a->sign == MP_NEG) {
696     return MP_LT;
697   }
698 
699   /* compare based on magnitude */
700   if (a->used > 1) {
701     return MP_GT;
702   }
703 
704   /* compare the only digit of a to b */
705   if (a->dp[0] > b) {
706     return MP_GT;
707   } else if (a->dp[0] < b) {
708     return MP_LT;
709   } else {
710     return MP_EQ;
711   }
712 }
713 
714 
715 #ifndef LTM_NO_NEG_EXP
716 /* hac 14.61, pp608 */
mp_invmod(mp_int * a,mp_int * b,mp_int * c)717 static int mp_invmod (mp_int * a, mp_int * b, mp_int * c)
718 {
719   /* b cannot be negative */
720   if (b->sign == MP_NEG || mp_iszero(b) == 1) {
721     return MP_VAL;
722   }
723 
724 #ifdef BN_FAST_MP_INVMOD_C
725   /* if the modulus is odd we can use a faster routine instead */
726   if (mp_isodd (b) == 1) {
727     return fast_mp_invmod (a, b, c);
728   }
729 #endif
730 
731 #ifdef BN_MP_INVMOD_SLOW_C
732   return mp_invmod_slow(a, b, c);
733 #endif
734 
735 #ifndef BN_FAST_MP_INVMOD_C
736 #ifndef BN_MP_INVMOD_SLOW_C
737 #error mp_invmod would always fail
738 #endif
739 #endif
740   return MP_VAL;
741 }
742 #endif /* LTM_NO_NEG_EXP */
743 
744 
745 /* get the size for an unsigned equivalent */
mp_unsigned_bin_size(mp_int * a)746 static int mp_unsigned_bin_size (mp_int * a)
747 {
748   int     size = mp_count_bits (a);
749   return (size / 8 + ((size & 7) != 0 ? 1 : 0));
750 }
751 
752 
753 #ifndef LTM_NO_NEG_EXP
754 /* hac 14.61, pp608 */
mp_invmod_slow(mp_int * a,mp_int * b,mp_int * c)755 static int mp_invmod_slow (mp_int * a, mp_int * b, mp_int * c)
756 {
757   mp_int  x, y, u, v, A, B, C, D;
758   int     res;
759 
760   /* b cannot be negative */
761   if (b->sign == MP_NEG || mp_iszero(b) == 1) {
762     return MP_VAL;
763   }
764 
765   /* init temps */
766   if ((res = mp_init_multi(&x, &y, &u, &v,
767                            &A, &B, &C, &D, NULL)) != MP_OKAY) {
768      return res;
769   }
770 
771   /* x = a, y = b */
772   if ((res = mp_mod(a, b, &x)) != MP_OKAY) {
773       goto LBL_ERR;
774   }
775   if ((res = mp_copy (b, &y)) != MP_OKAY) {
776     goto LBL_ERR;
777   }
778 
779   /* 2. [modified] if x,y are both even then return an error! */
780   if (mp_iseven (&x) == 1 && mp_iseven (&y) == 1) {
781     res = MP_VAL;
782     goto LBL_ERR;
783   }
784 
785   /* 3. u=x, v=y, A=1, B=0, C=0,D=1 */
786   if ((res = mp_copy (&x, &u)) != MP_OKAY) {
787     goto LBL_ERR;
788   }
789   if ((res = mp_copy (&y, &v)) != MP_OKAY) {
790     goto LBL_ERR;
791   }
792   mp_set (&A, 1);
793   mp_set (&D, 1);
794 
795 top:
796   /* 4.  while u is even do */
797   while (mp_iseven (&u) == 1) {
798     /* 4.1 u = u/2 */
799     if ((res = mp_div_2 (&u, &u)) != MP_OKAY) {
800       goto LBL_ERR;
801     }
802     /* 4.2 if A or B is odd then */
803     if (mp_isodd (&A) == 1 || mp_isodd (&B) == 1) {
804       /* A = (A+y)/2, B = (B-x)/2 */
805       if ((res = mp_add (&A, &y, &A)) != MP_OKAY) {
806          goto LBL_ERR;
807       }
808       if ((res = mp_sub (&B, &x, &B)) != MP_OKAY) {
809          goto LBL_ERR;
810       }
811     }
812     /* A = A/2, B = B/2 */
813     if ((res = mp_div_2 (&A, &A)) != MP_OKAY) {
814       goto LBL_ERR;
815     }
816     if ((res = mp_div_2 (&B, &B)) != MP_OKAY) {
817       goto LBL_ERR;
818     }
819   }
820 
821   /* 5.  while v is even do */
822   while (mp_iseven (&v) == 1) {
823     /* 5.1 v = v/2 */
824     if ((res = mp_div_2 (&v, &v)) != MP_OKAY) {
825       goto LBL_ERR;
826     }
827     /* 5.2 if C or D is odd then */
828     if (mp_isodd (&C) == 1 || mp_isodd (&D) == 1) {
829       /* C = (C+y)/2, D = (D-x)/2 */
830       if ((res = mp_add (&C, &y, &C)) != MP_OKAY) {
831          goto LBL_ERR;
832       }
833       if ((res = mp_sub (&D, &x, &D)) != MP_OKAY) {
834          goto LBL_ERR;
835       }
836     }
837     /* C = C/2, D = D/2 */
838     if ((res = mp_div_2 (&C, &C)) != MP_OKAY) {
839       goto LBL_ERR;
840     }
841     if ((res = mp_div_2 (&D, &D)) != MP_OKAY) {
842       goto LBL_ERR;
843     }
844   }
845 
846   /* 6.  if u >= v then */
847   if (mp_cmp (&u, &v) != MP_LT) {
848     /* u = u - v, A = A - C, B = B - D */
849     if ((res = mp_sub (&u, &v, &u)) != MP_OKAY) {
850       goto LBL_ERR;
851     }
852 
853     if ((res = mp_sub (&A, &C, &A)) != MP_OKAY) {
854       goto LBL_ERR;
855     }
856 
857     if ((res = mp_sub (&B, &D, &B)) != MP_OKAY) {
858       goto LBL_ERR;
859     }
860   } else {
861     /* v - v - u, C = C - A, D = D - B */
862     if ((res = mp_sub (&v, &u, &v)) != MP_OKAY) {
863       goto LBL_ERR;
864     }
865 
866     if ((res = mp_sub (&C, &A, &C)) != MP_OKAY) {
867       goto LBL_ERR;
868     }
869 
870     if ((res = mp_sub (&D, &B, &D)) != MP_OKAY) {
871       goto LBL_ERR;
872     }
873   }
874 
875   /* if not zero goto step 4 */
876   if (mp_iszero (&u) == 0)
877     goto top;
878 
879   /* now a = C, b = D, gcd == g*v */
880 
881   /* if v != 1 then there is no inverse */
882   if (mp_cmp_d (&v, 1) != MP_EQ) {
883     res = MP_VAL;
884     goto LBL_ERR;
885   }
886 
887   /* if its too low */
888   while (mp_cmp_d(&C, 0) == MP_LT) {
889       if ((res = mp_add(&C, b, &C)) != MP_OKAY) {
890          goto LBL_ERR;
891       }
892   }
893 
894   /* too big */
895   while (mp_cmp_mag(&C, b) != MP_LT) {
896       if ((res = mp_sub(&C, b, &C)) != MP_OKAY) {
897          goto LBL_ERR;
898       }
899   }
900 
901   /* C is now the inverse */
902   mp_exch (&C, c);
903   res = MP_OKAY;
904 LBL_ERR:mp_clear_multi (&x, &y, &u, &v, &A, &B, &C, &D, NULL);
905   return res;
906 }
907 #endif /* LTM_NO_NEG_EXP */
908 
909 
910 /* compare maginitude of two ints (unsigned) */
mp_cmp_mag(mp_int * a,mp_int * b)911 static int mp_cmp_mag (mp_int * a, mp_int * b)
912 {
913   int     n;
914   mp_digit *tmpa, *tmpb;
915 
916   /* compare based on # of non-zero digits */
917   if (a->used > b->used) {
918     return MP_GT;
919   }
920 
921   if (a->used < b->used) {
922     return MP_LT;
923   }
924 
925   /* alias for a */
926   tmpa = a->dp + (a->used - 1);
927 
928   /* alias for b */
929   tmpb = b->dp + (a->used - 1);
930 
931   /* compare based on digits  */
932   for (n = 0; n < a->used; ++n, --tmpa, --tmpb) {
933     if (*tmpa > *tmpb) {
934       return MP_GT;
935     }
936 
937     if (*tmpa < *tmpb) {
938       return MP_LT;
939     }
940   }
941   return MP_EQ;
942 }
943 
944 
945 /* reads a unsigned char array, assumes the msb is stored first [big endian] */
mp_read_unsigned_bin(mp_int * a,const unsigned char * b,int c)946 static int mp_read_unsigned_bin (mp_int * a, const unsigned char *b, int c)
947 {
948   int     res;
949 
950   /* make sure there are at least two digits */
951   if (a->alloc < 2) {
952      if ((res = mp_grow(a, 2)) != MP_OKAY) {
953         return res;
954      }
955   }
956 
957   /* zero the int */
958   mp_zero (a);
959 
960   /* read the bytes in */
961   while (c-- > 0) {
962     if ((res = mp_mul_2d (a, 8, a)) != MP_OKAY) {
963       return res;
964     }
965 
966 #ifndef MP_8BIT
967       a->dp[0] |= *b++;
968       a->used += 1;
969 #else
970       a->dp[0] = (*b & MP_MASK);
971       a->dp[1] |= ((*b++ >> 7U) & 1);
972       a->used += 2;
973 #endif
974   }
975   mp_clamp (a);
976   return MP_OKAY;
977 }
978 
979 
980 /* store in unsigned [big endian] format */
mp_to_unsigned_bin(mp_int * a,unsigned char * b)981 static int mp_to_unsigned_bin (mp_int * a, unsigned char *b)
982 {
983   int     x, res;
984   mp_int  t;
985 
986   if ((res = mp_init_copy (&t, a)) != MP_OKAY) {
987     return res;
988   }
989 
990   x = 0;
991   while (mp_iszero (&t) == 0) {
992 #ifndef MP_8BIT
993       b[x++] = (unsigned char) (t.dp[0] & 255);
994 #else
995       b[x++] = (unsigned char) (t.dp[0] | ((t.dp[1] & 0x01) << 7));
996 #endif
997     if ((res = mp_div_2d (&t, 8, &t, NULL)) != MP_OKAY) {
998       mp_clear (&t);
999       return res;
1000     }
1001   }
1002   bn_reverse (b, x);
1003   mp_clear (&t);
1004   return MP_OKAY;
1005 }
1006 
1007 
1008 /* shift right by a certain bit count (store quotient in c, optional remainder in d) */
mp_div_2d(mp_int * a,int b,mp_int * c,mp_int * d)1009 static int mp_div_2d (mp_int * a, int b, mp_int * c, mp_int * d)
1010 {
1011   mp_digit D, r, rr;
1012   int     x, res;
1013   mp_int  t;
1014 
1015 
1016   /* if the shift count is <= 0 then we do no work */
1017   if (b <= 0) {
1018     res = mp_copy (a, c);
1019     if (d != NULL) {
1020       mp_zero (d);
1021     }
1022     return res;
1023   }
1024 
1025   if ((res = mp_init (&t)) != MP_OKAY) {
1026     return res;
1027   }
1028 
1029   /* get the remainder */
1030   if (d != NULL) {
1031     if ((res = mp_mod_2d (a, b, &t)) != MP_OKAY) {
1032       mp_clear (&t);
1033       return res;
1034     }
1035   }
1036 
1037   /* copy */
1038   if ((res = mp_copy (a, c)) != MP_OKAY) {
1039     mp_clear (&t);
1040     return res;
1041   }
1042 
1043   /* shift by as many digits in the bit count */
1044   if (b >= (int)DIGIT_BIT) {
1045     mp_rshd (c, b / DIGIT_BIT);
1046   }
1047 
1048   /* shift any bit count < DIGIT_BIT */
1049   D = (mp_digit) (b % DIGIT_BIT);
1050   if (D != 0) {
1051     register mp_digit *tmpc, mask, shift;
1052 
1053     /* mask */
1054     mask = (((mp_digit)1) << D) - 1;
1055 
1056     /* shift for lsb */
1057     shift = DIGIT_BIT - D;
1058 
1059     /* alias */
1060     tmpc = c->dp + (c->used - 1);
1061 
1062     /* carry */
1063     r = 0;
1064     for (x = c->used - 1; x >= 0; x--) {
1065       /* get the lower  bits of this word in a temp */
1066       rr = *tmpc & mask;
1067 
1068       /* shift the current word and mix in the carry bits from the previous word */
1069       *tmpc = (*tmpc >> D) | (r << shift);
1070       --tmpc;
1071 
1072       /* set the carry to the carry bits of the current word found above */
1073       r = rr;
1074     }
1075   }
1076   mp_clamp (c);
1077   if (d != NULL) {
1078     mp_exch (&t, d);
1079   }
1080   mp_clear (&t);
1081   return MP_OKAY;
1082 }
1083 
1084 
mp_init_copy(mp_int * a,mp_int * b)1085 static int mp_init_copy (mp_int * a, mp_int * b)
1086 {
1087   int     res;
1088 
1089   if ((res = mp_init (a)) != MP_OKAY) {
1090     return res;
1091   }
1092   return mp_copy (b, a);
1093 }
1094 
1095 
1096 /* set to zero */
mp_zero(mp_int * a)1097 static void mp_zero (mp_int * a)
1098 {
1099   int       n;
1100   mp_digit *tmp;
1101 
1102   a->sign = MP_ZPOS;
1103   a->used = 0;
1104 
1105   tmp = a->dp;
1106   for (n = 0; n < a->alloc; n++) {
1107      *tmp++ = 0;
1108   }
1109 }
1110 
1111 
1112 /* copy, b = a */
mp_copy(mp_int * a,mp_int * b)1113 static int mp_copy (mp_int * a, mp_int * b)
1114 {
1115   int     res, n;
1116 
1117   /* if dst == src do nothing */
1118   if (a == b) {
1119     return MP_OKAY;
1120   }
1121 
1122   /* grow dest */
1123   if (b->alloc < a->used) {
1124      if ((res = mp_grow (b, a->used)) != MP_OKAY) {
1125         return res;
1126      }
1127   }
1128 
1129   /* zero b and copy the parameters over */
1130   {
1131     register mp_digit *tmpa, *tmpb;
1132 
1133     /* pointer aliases */
1134 
1135     /* source */
1136     tmpa = a->dp;
1137 
1138     /* destination */
1139     tmpb = b->dp;
1140 
1141     /* copy all the digits */
1142     for (n = 0; n < a->used; n++) {
1143       *tmpb++ = *tmpa++;
1144     }
1145 
1146     /* clear high digits */
1147     for (; n < b->used; n++) {
1148       *tmpb++ = 0;
1149     }
1150   }
1151 
1152   /* copy used count and sign */
1153   b->used = a->used;
1154   b->sign = a->sign;
1155   return MP_OKAY;
1156 }
1157 
1158 
1159 /* shift right a certain amount of digits */
mp_rshd(mp_int * a,int b)1160 static void mp_rshd (mp_int * a, int b)
1161 {
1162   int     x;
1163 
1164   /* if b <= 0 then ignore it */
1165   if (b <= 0) {
1166     return;
1167   }
1168 
1169   /* if b > used then simply zero it and return */
1170   if (a->used <= b) {
1171     mp_zero (a);
1172     return;
1173   }
1174 
1175   {
1176     register mp_digit *bottom, *top;
1177 
1178     /* shift the digits down */
1179 
1180     /* bottom */
1181     bottom = a->dp;
1182 
1183     /* top [offset into digits] */
1184     top = a->dp + b;
1185 
1186     /* this is implemented as a sliding window where
1187      * the window is b-digits long and digits from
1188      * the top of the window are copied to the bottom
1189      *
1190      * e.g.
1191 
1192      b-2 | b-1 | b0 | b1 | b2 | ... | bb |   ---->
1193                  /\                   |      ---->
1194                   \-------------------/      ---->
1195      */
1196     for (x = 0; x < (a->used - b); x++) {
1197       *bottom++ = *top++;
1198     }
1199 
1200     /* zero the top digits */
1201     for (; x < a->used; x++) {
1202       *bottom++ = 0;
1203     }
1204   }
1205 
1206   /* remove excess digits */
1207   a->used -= b;
1208 }
1209 
1210 
1211 /* swap the elements of two integers, for cases where you can't simply swap the
1212  * mp_int pointers around
1213  */
mp_exch(mp_int * a,mp_int * b)1214 static void mp_exch (mp_int * a, mp_int * b)
1215 {
1216   mp_int  t;
1217 
1218   t  = *a;
1219   *a = *b;
1220   *b = t;
1221 }
1222 
1223 
1224 /* trim unused digits
1225  *
1226  * This is used to ensure that leading zero digits are
1227  * trimed and the leading "used" digit will be non-zero
1228  * Typically very fast.  Also fixes the sign if there
1229  * are no more leading digits
1230  */
mp_clamp(mp_int * a)1231 static void mp_clamp (mp_int * a)
1232 {
1233   /* decrease used while the most significant digit is
1234    * zero.
1235    */
1236   while (a->used > 0 && a->dp[a->used - 1] == 0) {
1237     --(a->used);
1238   }
1239 
1240   /* reset the sign flag if used == 0 */
1241   if (a->used == 0) {
1242     a->sign = MP_ZPOS;
1243   }
1244 }
1245 
1246 
1247 /* grow as required */
mp_grow(mp_int * a,int size)1248 static int mp_grow (mp_int * a, int size)
1249 {
1250   int     i;
1251   mp_digit *tmp;
1252 
1253   /* if the alloc size is smaller alloc more ram */
1254   if (a->alloc < size) {
1255     /* ensure there are always at least MP_PREC digits extra on top */
1256     size += (MP_PREC * 2) - (size % MP_PREC);
1257 
1258     /* reallocate the array a->dp
1259      *
1260      * We store the return in a temporary variable
1261      * in case the operation failed we don't want
1262      * to overwrite the dp member of a.
1263      */
1264     tmp = OPT_CAST(mp_digit) XREALLOC (a->dp, sizeof (mp_digit) * size);
1265     if (tmp == NULL) {
1266       /* reallocation failed but "a" is still valid [can be freed] */
1267       return MP_MEM;
1268     }
1269 
1270     /* reallocation succeeded so set a->dp */
1271     a->dp = tmp;
1272 
1273     /* zero excess digits */
1274     i        = a->alloc;
1275     a->alloc = size;
1276     for (; i < a->alloc; i++) {
1277       a->dp[i] = 0;
1278     }
1279   }
1280   return MP_OKAY;
1281 }
1282 
1283 
1284 #ifdef BN_MP_ABS_C
1285 /* b = |a|
1286  *
1287  * Simple function copies the input and fixes the sign to positive
1288  */
mp_abs(mp_int * a,mp_int * b)1289 static int mp_abs (mp_int * a, mp_int * b)
1290 {
1291   int     res;
1292 
1293   /* copy a to b */
1294   if (a != b) {
1295      if ((res = mp_copy (a, b)) != MP_OKAY) {
1296        return res;
1297      }
1298   }
1299 
1300   /* force the sign of b to positive */
1301   b->sign = MP_ZPOS;
1302 
1303   return MP_OKAY;
1304 }
1305 #endif
1306 
1307 
1308 /* set to a digit */
mp_set(mp_int * a,mp_digit b)1309 static void mp_set (mp_int * a, mp_digit b)
1310 {
1311   mp_zero (a);
1312   a->dp[0] = b & MP_MASK;
1313   a->used  = (a->dp[0] != 0) ? 1 : 0;
1314 }
1315 
1316 
1317 #ifndef LTM_NO_NEG_EXP
1318 /* b = a/2 */
mp_div_2(mp_int * a,mp_int * b)1319 static int mp_div_2(mp_int * a, mp_int * b)
1320 {
1321   int     x, res, oldused;
1322 
1323   /* copy */
1324   if (b->alloc < a->used) {
1325     if ((res = mp_grow (b, a->used)) != MP_OKAY) {
1326       return res;
1327     }
1328   }
1329 
1330   oldused = b->used;
1331   b->used = a->used;
1332   {
1333     register mp_digit r, rr, *tmpa, *tmpb;
1334 
1335     /* source alias */
1336     tmpa = a->dp + b->used - 1;
1337 
1338     /* dest alias */
1339     tmpb = b->dp + b->used - 1;
1340 
1341     /* carry */
1342     r = 0;
1343     for (x = b->used - 1; x >= 0; x--) {
1344       /* get the carry for the next iteration */
1345       rr = *tmpa & 1;
1346 
1347       /* shift the current digit, add in carry and store */
1348       *tmpb-- = (*tmpa-- >> 1) | (r << (DIGIT_BIT - 1));
1349 
1350       /* forward carry to next iteration */
1351       r = rr;
1352     }
1353 
1354     /* zero excess digits */
1355     tmpb = b->dp + b->used;
1356     for (x = b->used; x < oldused; x++) {
1357       *tmpb++ = 0;
1358     }
1359   }
1360   b->sign = a->sign;
1361   mp_clamp (b);
1362   return MP_OKAY;
1363 }
1364 #endif /* LTM_NO_NEG_EXP */
1365 
1366 
1367 /* shift left by a certain bit count */
mp_mul_2d(mp_int * a,int b,mp_int * c)1368 static int mp_mul_2d (mp_int * a, int b, mp_int * c)
1369 {
1370   mp_digit d;
1371   int      res;
1372 
1373   /* copy */
1374   if (a != c) {
1375      if ((res = mp_copy (a, c)) != MP_OKAY) {
1376        return res;
1377      }
1378   }
1379 
1380   if (c->alloc < (int)(c->used + b/DIGIT_BIT + 1)) {
1381      if ((res = mp_grow (c, c->used + b / DIGIT_BIT + 1)) != MP_OKAY) {
1382        return res;
1383      }
1384   }
1385 
1386   /* shift by as many digits in the bit count */
1387   if (b >= (int)DIGIT_BIT) {
1388     if ((res = mp_lshd (c, b / DIGIT_BIT)) != MP_OKAY) {
1389       return res;
1390     }
1391   }
1392 
1393   /* shift any bit count < DIGIT_BIT */
1394   d = (mp_digit) (b % DIGIT_BIT);
1395   if (d != 0) {
1396     register mp_digit *tmpc, shift, mask, r, rr;
1397     register int x;
1398 
1399     /* bitmask for carries */
1400     mask = (((mp_digit)1) << d) - 1;
1401 
1402     /* shift for msbs */
1403     shift = DIGIT_BIT - d;
1404 
1405     /* alias */
1406     tmpc = c->dp;
1407 
1408     /* carry */
1409     r    = 0;
1410     for (x = 0; x < c->used; x++) {
1411       /* get the higher bits of the current word */
1412       rr = (*tmpc >> shift) & mask;
1413 
1414       /* shift the current word and OR in the carry */
1415       *tmpc = ((*tmpc << d) | r) & MP_MASK;
1416       ++tmpc;
1417 
1418       /* set the carry to the carry bits of the current word */
1419       r = rr;
1420     }
1421 
1422     /* set final carry */
1423     if (r != 0) {
1424        c->dp[(c->used)++] = r;
1425     }
1426   }
1427   mp_clamp (c);
1428   return MP_OKAY;
1429 }
1430 
1431 
1432 #ifdef BN_MP_INIT_MULTI_C
mp_init_multi(mp_int * mp,...)1433 static int mp_init_multi(mp_int *mp, ...)
1434 {
1435     mp_err res = MP_OKAY;      /* Assume ok until proven otherwise */
1436     int n = 0;                 /* Number of ok inits */
1437     mp_int* cur_arg = mp;
1438     va_list args;
1439 
1440     va_start(args, mp);        /* init args to next argument from caller */
1441     while (cur_arg != NULL) {
1442         if (mp_init(cur_arg) != MP_OKAY) {
1443             /* Oops - error! Back-track and mp_clear what we already
1444                succeeded in init-ing, then return error.
1445             */
1446             va_list clean_args;
1447 
1448             /* end the current list */
1449             va_end(args);
1450 
1451             /* now start cleaning up */
1452             cur_arg = mp;
1453             va_start(clean_args, mp);
1454             while (n--) {
1455                 mp_clear(cur_arg);
1456                 cur_arg = va_arg(clean_args, mp_int*);
1457             }
1458             va_end(clean_args);
1459             res = MP_MEM;
1460             break;
1461         }
1462         n++;
1463         cur_arg = va_arg(args, mp_int*);
1464     }
1465     va_end(args);
1466     return res;                /* Assumed ok, if error flagged above. */
1467 }
1468 #endif
1469 
1470 
1471 #ifdef BN_MP_CLEAR_MULTI_C
mp_clear_multi(mp_int * mp,...)1472 static void mp_clear_multi(mp_int *mp, ...)
1473 {
1474     mp_int* next_mp = mp;
1475     va_list args;
1476     va_start(args, mp);
1477     while (next_mp != NULL) {
1478         mp_clear(next_mp);
1479         next_mp = va_arg(args, mp_int*);
1480     }
1481     va_end(args);
1482 }
1483 #endif
1484 
1485 
1486 /* shift left a certain amount of digits */
mp_lshd(mp_int * a,int b)1487 static int mp_lshd (mp_int * a, int b)
1488 {
1489   int     x, res;
1490 
1491   /* if its less than zero return */
1492   if (b <= 0) {
1493     return MP_OKAY;
1494   }
1495 
1496   /* grow to fit the new digits */
1497   if (a->alloc < a->used + b) {
1498      if ((res = mp_grow (a, a->used + b)) != MP_OKAY) {
1499        return res;
1500      }
1501   }
1502 
1503   {
1504     register mp_digit *top, *bottom;
1505 
1506     /* increment the used by the shift amount then copy upwards */
1507     a->used += b;
1508 
1509     /* top */
1510     top = a->dp + a->used - 1;
1511 
1512     /* base */
1513     bottom = a->dp + a->used - 1 - b;
1514 
1515     /* much like mp_rshd this is implemented using a sliding window
1516      * except the window goes the otherway around.  Copying from
1517      * the bottom to the top.  see bn_mp_rshd.c for more info.
1518      */
1519     for (x = a->used - 1; x >= b; x--) {
1520       *top-- = *bottom--;
1521     }
1522 
1523     /* zero the lower digits */
1524     top = a->dp;
1525     for (x = 0; x < b; x++) {
1526       *top++ = 0;
1527     }
1528   }
1529   return MP_OKAY;
1530 }
1531 
1532 
1533 /* returns the number of bits in an int */
mp_count_bits(mp_int * a)1534 static int mp_count_bits (mp_int * a)
1535 {
1536   int     r;
1537   mp_digit q;
1538 
1539   /* shortcut */
1540   if (a->used == 0) {
1541     return 0;
1542   }
1543 
1544   /* get number of digits and add that */
1545   r = (a->used - 1) * DIGIT_BIT;
1546 
1547   /* take the last digit and count the bits in it */
1548   q = a->dp[a->used - 1];
1549   while (q > ((mp_digit) 0)) {
1550     ++r;
1551     q >>= ((mp_digit) 1);
1552   }
1553   return r;
1554 }
1555 
1556 
1557 /* calc a value mod 2**b */
mp_mod_2d(mp_int * a,int b,mp_int * c)1558 static int mp_mod_2d (mp_int * a, int b, mp_int * c)
1559 {
1560   int     x, res;
1561 
1562   /* if b is <= 0 then zero the int */
1563   if (b <= 0) {
1564     mp_zero (c);
1565     return MP_OKAY;
1566   }
1567 
1568   /* if the modulus is larger than the value than return */
1569   if (b >= (int) (a->used * DIGIT_BIT)) {
1570     res = mp_copy (a, c);
1571     return res;
1572   }
1573 
1574   /* copy */
1575   if ((res = mp_copy (a, c)) != MP_OKAY) {
1576     return res;
1577   }
1578 
1579   /* zero digits above the last digit of the modulus */
1580   for (x = (b / DIGIT_BIT) + ((b % DIGIT_BIT) == 0 ? 0 : 1); x < c->used; x++) {
1581     c->dp[x] = 0;
1582   }
1583   /* clear the digit that is not completely outside/inside the modulus */
1584   c->dp[b / DIGIT_BIT] &=
1585     (mp_digit) ((((mp_digit) 1) << (((mp_digit) b) % DIGIT_BIT)) - ((mp_digit) 1));
1586   mp_clamp (c);
1587   return MP_OKAY;
1588 }
1589 
1590 
1591 #ifdef BN_MP_DIV_SMALL
1592 
1593 /* slower bit-bang division... also smaller */
mp_div(mp_int * a,mp_int * b,mp_int * c,mp_int * d)1594 static int mp_div(mp_int * a, mp_int * b, mp_int * c, mp_int * d)
1595 {
1596    mp_int ta, tb, tq, q;
1597    int    res, n, n2;
1598 
1599   /* is divisor zero ? */
1600   if (mp_iszero (b) == 1) {
1601     return MP_VAL;
1602   }
1603 
1604   /* if a < b then q=0, r = a */
1605   if (mp_cmp_mag (a, b) == MP_LT) {
1606     if (d != NULL) {
1607       res = mp_copy (a, d);
1608     } else {
1609       res = MP_OKAY;
1610     }
1611     if (c != NULL) {
1612       mp_zero (c);
1613     }
1614     return res;
1615   }
1616 
1617   /* init our temps */
1618   if ((res = mp_init_multi(&ta, &tb, &tq, &q, NULL) != MP_OKAY)) {
1619      return res;
1620   }
1621 
1622 
1623   mp_set(&tq, 1);
1624   n = mp_count_bits(a) - mp_count_bits(b);
1625   if (((res = mp_abs(a, &ta)) != MP_OKAY) ||
1626       ((res = mp_abs(b, &tb)) != MP_OKAY) ||
1627       ((res = mp_mul_2d(&tb, n, &tb)) != MP_OKAY) ||
1628       ((res = mp_mul_2d(&tq, n, &tq)) != MP_OKAY)) {
1629       goto LBL_ERR;
1630   }
1631 
1632   while (n-- >= 0) {
1633      if (mp_cmp(&tb, &ta) != MP_GT) {
1634         if (((res = mp_sub(&ta, &tb, &ta)) != MP_OKAY) ||
1635             ((res = mp_add(&q, &tq, &q)) != MP_OKAY)) {
1636            goto LBL_ERR;
1637         }
1638      }
1639      if (((res = mp_div_2d(&tb, 1, &tb, NULL)) != MP_OKAY) ||
1640          ((res = mp_div_2d(&tq, 1, &tq, NULL)) != MP_OKAY)) {
1641            goto LBL_ERR;
1642      }
1643   }
1644 
1645   /* now q == quotient and ta == remainder */
1646   n  = a->sign;
1647   n2 = (a->sign == b->sign ? MP_ZPOS : MP_NEG);
1648   if (c != NULL) {
1649      mp_exch(c, &q);
1650      c->sign  = (mp_iszero(c) == MP_YES) ? MP_ZPOS : n2;
1651   }
1652   if (d != NULL) {
1653      mp_exch(d, &ta);
1654      d->sign = (mp_iszero(d) == MP_YES) ? MP_ZPOS : n;
1655   }
1656 LBL_ERR:
1657    mp_clear_multi(&ta, &tb, &tq, &q, NULL);
1658    return res;
1659 }
1660 
1661 #else
1662 
1663 /* integer signed division.
1664  * c*b + d == a [e.g. a/b, c=quotient, d=remainder]
1665  * HAC pp.598 Algorithm 14.20
1666  *
1667  * Note that the description in HAC is horribly
1668  * incomplete.  For example, it doesn't consider
1669  * the case where digits are removed from 'x' in
1670  * the inner loop.  It also doesn't consider the
1671  * case that y has fewer than three digits, etc..
1672  *
1673  * The overall algorithm is as described as
1674  * 14.20 from HAC but fixed to treat these cases.
1675 */
mp_div(mp_int * a,mp_int * b,mp_int * c,mp_int * d)1676 static int mp_div (mp_int * a, mp_int * b, mp_int * c, mp_int * d)
1677 {
1678   mp_int  q, x, y, t1, t2;
1679   int     res, n, t, i, norm, neg;
1680 
1681   /* is divisor zero ? */
1682   if (mp_iszero (b) == 1) {
1683     return MP_VAL;
1684   }
1685 
1686   /* if a < b then q=0, r = a */
1687   if (mp_cmp_mag (a, b) == MP_LT) {
1688     if (d != NULL) {
1689       res = mp_copy (a, d);
1690     } else {
1691       res = MP_OKAY;
1692     }
1693     if (c != NULL) {
1694       mp_zero (c);
1695     }
1696     return res;
1697   }
1698 
1699   if ((res = mp_init_size (&q, a->used + 2)) != MP_OKAY) {
1700     return res;
1701   }
1702   q.used = a->used + 2;
1703 
1704   if ((res = mp_init (&t1)) != MP_OKAY) {
1705     goto LBL_Q;
1706   }
1707 
1708   if ((res = mp_init (&t2)) != MP_OKAY) {
1709     goto LBL_T1;
1710   }
1711 
1712   if ((res = mp_init_copy (&x, a)) != MP_OKAY) {
1713     goto LBL_T2;
1714   }
1715 
1716   if ((res = mp_init_copy (&y, b)) != MP_OKAY) {
1717     goto LBL_X;
1718   }
1719 
1720   /* fix the sign */
1721   neg = (a->sign == b->sign) ? MP_ZPOS : MP_NEG;
1722   x.sign = y.sign = MP_ZPOS;
1723 
1724   /* normalize both x and y, ensure that y >= b/2, [b == 2**DIGIT_BIT] */
1725   norm = mp_count_bits(&y) % DIGIT_BIT;
1726   if (norm < (int)(DIGIT_BIT-1)) {
1727      norm = (DIGIT_BIT-1) - norm;
1728      if ((res = mp_mul_2d (&x, norm, &x)) != MP_OKAY) {
1729        goto LBL_Y;
1730      }
1731      if ((res = mp_mul_2d (&y, norm, &y)) != MP_OKAY) {
1732        goto LBL_Y;
1733      }
1734   } else {
1735      norm = 0;
1736   }
1737 
1738   /* note hac does 0 based, so if used==5 then its 0,1,2,3,4, e.g. use 4 */
1739   n = x.used - 1;
1740   t = y.used - 1;
1741 
1742   /* while (x >= y*b**n-t) do { q[n-t] += 1; x -= y*b**{n-t} } */
1743   if ((res = mp_lshd (&y, n - t)) != MP_OKAY) { /* y = y*b**{n-t} */
1744     goto LBL_Y;
1745   }
1746 
1747   while (mp_cmp (&x, &y) != MP_LT) {
1748     ++(q.dp[n - t]);
1749     if ((res = mp_sub (&x, &y, &x)) != MP_OKAY) {
1750       goto LBL_Y;
1751     }
1752   }
1753 
1754   /* reset y by shifting it back down */
1755   mp_rshd (&y, n - t);
1756 
1757   /* step 3. for i from n down to (t + 1) */
1758   for (i = n; i >= (t + 1); i--) {
1759     if (i > x.used) {
1760       continue;
1761     }
1762 
1763     /* step 3.1 if xi == yt then set q{i-t-1} to b-1,
1764      * otherwise set q{i-t-1} to (xi*b + x{i-1})/yt */
1765     if (x.dp[i] == y.dp[t]) {
1766       q.dp[i - t - 1] = ((((mp_digit)1) << DIGIT_BIT) - 1);
1767     } else {
1768       mp_word tmp;
1769       tmp = ((mp_word) x.dp[i]) << ((mp_word) DIGIT_BIT);
1770       tmp |= ((mp_word) x.dp[i - 1]);
1771       tmp /= ((mp_word) y.dp[t]);
1772       if (tmp > (mp_word) MP_MASK)
1773         tmp = MP_MASK;
1774       q.dp[i - t - 1] = (mp_digit) (tmp & (mp_word) (MP_MASK));
1775     }
1776 
1777     /* while (q{i-t-1} * (yt * b + y{t-1})) >
1778              xi * b**2 + xi-1 * b + xi-2
1779 
1780        do q{i-t-1} -= 1;
1781     */
1782     q.dp[i - t - 1] = (q.dp[i - t - 1] + 1) & MP_MASK;
1783     do {
1784       q.dp[i - t - 1] = (q.dp[i - t - 1] - 1) & MP_MASK;
1785 
1786       /* find left hand */
1787       mp_zero (&t1);
1788       t1.dp[0] = (t - 1 < 0) ? 0 : y.dp[t - 1];
1789       t1.dp[1] = y.dp[t];
1790       t1.used = 2;
1791       if ((res = mp_mul_d (&t1, q.dp[i - t - 1], &t1)) != MP_OKAY) {
1792         goto LBL_Y;
1793       }
1794 
1795       /* find right hand */
1796       t2.dp[0] = (i - 2 < 0) ? 0 : x.dp[i - 2];
1797       t2.dp[1] = (i - 1 < 0) ? 0 : x.dp[i - 1];
1798       t2.dp[2] = x.dp[i];
1799       t2.used = 3;
1800     } while (mp_cmp_mag(&t1, &t2) == MP_GT);
1801 
1802     /* step 3.3 x = x - q{i-t-1} * y * b**{i-t-1} */
1803     if ((res = mp_mul_d (&y, q.dp[i - t - 1], &t1)) != MP_OKAY) {
1804       goto LBL_Y;
1805     }
1806 
1807     if ((res = mp_lshd (&t1, i - t - 1)) != MP_OKAY) {
1808       goto LBL_Y;
1809     }
1810 
1811     if ((res = mp_sub (&x, &t1, &x)) != MP_OKAY) {
1812       goto LBL_Y;
1813     }
1814 
1815     /* if x < 0 then { x = x + y*b**{i-t-1}; q{i-t-1} -= 1; } */
1816     if (x.sign == MP_NEG) {
1817       if ((res = mp_copy (&y, &t1)) != MP_OKAY) {
1818         goto LBL_Y;
1819       }
1820       if ((res = mp_lshd (&t1, i - t - 1)) != MP_OKAY) {
1821         goto LBL_Y;
1822       }
1823       if ((res = mp_add (&x, &t1, &x)) != MP_OKAY) {
1824         goto LBL_Y;
1825       }
1826 
1827       q.dp[i - t - 1] = (q.dp[i - t - 1] - 1UL) & MP_MASK;
1828     }
1829   }
1830 
1831   /* now q is the quotient and x is the remainder
1832    * [which we have to normalize]
1833    */
1834 
1835   /* get sign before writing to c */
1836   x.sign = x.used == 0 ? MP_ZPOS : a->sign;
1837 
1838   if (c != NULL) {
1839     mp_clamp (&q);
1840     mp_exch (&q, c);
1841     c->sign = neg;
1842   }
1843 
1844   if (d != NULL) {
1845     mp_div_2d (&x, norm, &x, NULL);
1846     mp_exch (&x, d);
1847   }
1848 
1849   res = MP_OKAY;
1850 
1851 LBL_Y:mp_clear (&y);
1852 LBL_X:mp_clear (&x);
1853 LBL_T2:mp_clear (&t2);
1854 LBL_T1:mp_clear (&t1);
1855 LBL_Q:mp_clear (&q);
1856   return res;
1857 }
1858 
1859 #endif
1860 
1861 
1862 #ifdef MP_LOW_MEM
1863    #define TAB_SIZE 32
1864 #else
1865    #define TAB_SIZE 256
1866 #endif
1867 
s_mp_exptmod(mp_int * G,mp_int * X,mp_int * P,mp_int * Y,int redmode)1868 static int s_mp_exptmod (mp_int * G, mp_int * X, mp_int * P, mp_int * Y, int redmode)
1869 {
1870   mp_int  M[TAB_SIZE], res, mu;
1871   mp_digit buf;
1872   int     err, bitbuf, bitcpy, bitcnt, mode, digidx, x, y, winsize;
1873   int (*redux)(mp_int*,mp_int*,mp_int*);
1874 
1875   /* find window size */
1876   x = mp_count_bits (X);
1877   if (x <= 7) {
1878     winsize = 2;
1879   } else if (x <= 36) {
1880     winsize = 3;
1881   } else if (x <= 140) {
1882     winsize = 4;
1883   } else if (x <= 450) {
1884     winsize = 5;
1885   } else if (x <= 1303) {
1886     winsize = 6;
1887   } else if (x <= 3529) {
1888     winsize = 7;
1889   } else {
1890     winsize = 8;
1891   }
1892 
1893 #ifdef MP_LOW_MEM
1894     if (winsize > 5) {
1895        winsize = 5;
1896     }
1897 #endif
1898 
1899   /* init M array */
1900   /* init first cell */
1901   if ((err = mp_init(&M[1])) != MP_OKAY) {
1902      return err;
1903   }
1904 
1905   /* now init the second half of the array */
1906   for (x = 1<<(winsize-1); x < (1 << winsize); x++) {
1907     if ((err = mp_init(&M[x])) != MP_OKAY) {
1908       for (y = 1<<(winsize-1); y < x; y++) {
1909         mp_clear (&M[y]);
1910       }
1911       mp_clear(&M[1]);
1912       return err;
1913     }
1914   }
1915 
1916   /* create mu, used for Barrett reduction */
1917   if ((err = mp_init (&mu)) != MP_OKAY) {
1918     goto LBL_M;
1919   }
1920 
1921   if (redmode == 0) {
1922      if ((err = mp_reduce_setup (&mu, P)) != MP_OKAY) {
1923         goto LBL_MU;
1924      }
1925      redux = mp_reduce;
1926   } else {
1927      if ((err = mp_reduce_2k_setup_l (P, &mu)) != MP_OKAY) {
1928         goto LBL_MU;
1929      }
1930      redux = mp_reduce_2k_l;
1931   }
1932 
1933   /* create M table
1934    *
1935    * The M table contains powers of the base,
1936    * e.g. M[x] = G**x mod P
1937    *
1938    * The first half of the table is not
1939    * computed though accept for M[0] and M[1]
1940    */
1941   if ((err = mp_mod (G, P, &M[1])) != MP_OKAY) {
1942     goto LBL_MU;
1943   }
1944 
1945   /* compute the value at M[1<<(winsize-1)] by squaring
1946    * M[1] (winsize-1) times
1947    */
1948   if ((err = mp_copy (&M[1], &M[1 << (winsize - 1)])) != MP_OKAY) {
1949     goto LBL_MU;
1950   }
1951 
1952   for (x = 0; x < (winsize - 1); x++) {
1953     /* square it */
1954     if ((err = mp_sqr (&M[1 << (winsize - 1)],
1955                        &M[1 << (winsize - 1)])) != MP_OKAY) {
1956       goto LBL_MU;
1957     }
1958 
1959     /* reduce modulo P */
1960     if ((err = redux (&M[1 << (winsize - 1)], P, &mu)) != MP_OKAY) {
1961       goto LBL_MU;
1962     }
1963   }
1964 
1965   /* create upper table, that is M[x] = M[x-1] * M[1] (mod P)
1966    * for x = (2**(winsize - 1) + 1) to (2**winsize - 1)
1967    */
1968   for (x = (1 << (winsize - 1)) + 1; x < (1 << winsize); x++) {
1969     if ((err = mp_mul (&M[x - 1], &M[1], &M[x])) != MP_OKAY) {
1970       goto LBL_MU;
1971     }
1972     if ((err = redux (&M[x], P, &mu)) != MP_OKAY) {
1973       goto LBL_MU;
1974     }
1975   }
1976 
1977   /* setup result */
1978   if ((err = mp_init (&res)) != MP_OKAY) {
1979     goto LBL_MU;
1980   }
1981   mp_set (&res, 1);
1982 
1983   /* set initial mode and bit cnt */
1984   mode   = 0;
1985   bitcnt = 1;
1986   buf    = 0;
1987   digidx = X->used - 1;
1988   bitcpy = 0;
1989   bitbuf = 0;
1990 
1991   for (;;) {
1992     /* grab next digit as required */
1993     if (--bitcnt == 0) {
1994       /* if digidx == -1 we are out of digits */
1995       if (digidx == -1) {
1996         break;
1997       }
1998       /* read next digit and reset the bitcnt */
1999       buf    = X->dp[digidx--];
2000       bitcnt = (int) DIGIT_BIT;
2001     }
2002 
2003     /* grab the next msb from the exponent */
2004     y     = (buf >> (mp_digit)(DIGIT_BIT - 1)) & 1;
2005     buf <<= (mp_digit)1;
2006 
2007     /* if the bit is zero and mode == 0 then we ignore it
2008      * These represent the leading zero bits before the first 1 bit
2009      * in the exponent.  Technically this opt is not required but it
2010      * does lower the # of trivial squaring/reductions used
2011      */
2012     if (mode == 0 && y == 0) {
2013       continue;
2014     }
2015 
2016     /* if the bit is zero and mode == 1 then we square */
2017     if (mode == 1 && y == 0) {
2018       if ((err = mp_sqr (&res, &res)) != MP_OKAY) {
2019         goto LBL_RES;
2020       }
2021       if ((err = redux (&res, P, &mu)) != MP_OKAY) {
2022         goto LBL_RES;
2023       }
2024       continue;
2025     }
2026 
2027     /* else we add it to the window */
2028     bitbuf |= (y << (winsize - ++bitcpy));
2029     mode    = 2;
2030 
2031     if (bitcpy == winsize) {
2032       /* ok window is filled so square as required and multiply  */
2033       /* square first */
2034       for (x = 0; x < winsize; x++) {
2035         if ((err = mp_sqr (&res, &res)) != MP_OKAY) {
2036           goto LBL_RES;
2037         }
2038         if ((err = redux (&res, P, &mu)) != MP_OKAY) {
2039           goto LBL_RES;
2040         }
2041       }
2042 
2043       /* then multiply */
2044       if ((err = mp_mul (&res, &M[bitbuf], &res)) != MP_OKAY) {
2045         goto LBL_RES;
2046       }
2047       if ((err = redux (&res, P, &mu)) != MP_OKAY) {
2048         goto LBL_RES;
2049       }
2050 
2051       /* empty window and reset */
2052       bitcpy = 0;
2053       bitbuf = 0;
2054       mode   = 1;
2055     }
2056   }
2057 
2058   /* if bits remain then square/multiply */
2059   if (mode == 2 && bitcpy > 0) {
2060     /* square then multiply if the bit is set */
2061     for (x = 0; x < bitcpy; x++) {
2062       if ((err = mp_sqr (&res, &res)) != MP_OKAY) {
2063         goto LBL_RES;
2064       }
2065       if ((err = redux (&res, P, &mu)) != MP_OKAY) {
2066         goto LBL_RES;
2067       }
2068 
2069       bitbuf <<= 1;
2070       if ((bitbuf & (1 << winsize)) != 0) {
2071         /* then multiply */
2072         if ((err = mp_mul (&res, &M[1], &res)) != MP_OKAY) {
2073           goto LBL_RES;
2074         }
2075         if ((err = redux (&res, P, &mu)) != MP_OKAY) {
2076           goto LBL_RES;
2077         }
2078       }
2079     }
2080   }
2081 
2082   mp_exch (&res, Y);
2083   err = MP_OKAY;
2084 LBL_RES:mp_clear (&res);
2085 LBL_MU:mp_clear (&mu);
2086 LBL_M:
2087   mp_clear(&M[1]);
2088   for (x = 1<<(winsize-1); x < (1 << winsize); x++) {
2089     mp_clear (&M[x]);
2090   }
2091   return err;
2092 }
2093 
2094 
2095 /* computes b = a*a */
mp_sqr(mp_int * a,mp_int * b)2096 static int mp_sqr (mp_int * a, mp_int * b)
2097 {
2098   int     res;
2099 
2100 #ifdef BN_MP_TOOM_SQR_C
2101   /* use Toom-Cook? */
2102   if (a->used >= TOOM_SQR_CUTOFF) {
2103     res = mp_toom_sqr(a, b);
2104   /* Karatsuba? */
2105   } else
2106 #endif
2107 #ifdef BN_MP_KARATSUBA_SQR_C
2108 if (a->used >= KARATSUBA_SQR_CUTOFF) {
2109     res = mp_karatsuba_sqr (a, b);
2110   } else
2111 #endif
2112   {
2113 #ifdef BN_FAST_S_MP_SQR_C
2114     /* can we use the fast comba multiplier? */
2115     if ((a->used * 2 + 1) < MP_WARRAY &&
2116          a->used <
2117          (1 << (sizeof(mp_word) * CHAR_BIT - 2*DIGIT_BIT - 1))) {
2118       res = fast_s_mp_sqr (a, b);
2119     } else
2120 #endif
2121 #ifdef BN_S_MP_SQR_C
2122       res = s_mp_sqr (a, b);
2123 #else
2124 #error mp_sqr could fail
2125       res = MP_VAL;
2126 #endif
2127   }
2128   b->sign = MP_ZPOS;
2129   return res;
2130 }
2131 
2132 
2133 /* reduces a modulo n where n is of the form 2**p - d
2134    This differs from reduce_2k since "d" can be larger
2135    than a single digit.
2136 */
mp_reduce_2k_l(mp_int * a,mp_int * n,mp_int * d)2137 static int mp_reduce_2k_l(mp_int *a, mp_int *n, mp_int *d)
2138 {
2139    mp_int q;
2140    int    p, res;
2141 
2142    if ((res = mp_init(&q)) != MP_OKAY) {
2143       return res;
2144    }
2145 
2146    p = mp_count_bits(n);
2147 top:
2148    /* q = a/2**p, a = a mod 2**p */
2149    if ((res = mp_div_2d(a, p, &q, a)) != MP_OKAY) {
2150       goto ERR;
2151    }
2152 
2153    /* q = q * d */
2154    if ((res = mp_mul(&q, d, &q)) != MP_OKAY) {
2155       goto ERR;
2156    }
2157 
2158    /* a = a + q */
2159    if ((res = s_mp_add(a, &q, a)) != MP_OKAY) {
2160       goto ERR;
2161    }
2162 
2163    if (mp_cmp_mag(a, n) != MP_LT) {
2164       s_mp_sub(a, n, a);
2165       goto top;
2166    }
2167 
2168 ERR:
2169    mp_clear(&q);
2170    return res;
2171 }
2172 
2173 
2174 /* determines the setup value */
mp_reduce_2k_setup_l(mp_int * a,mp_int * d)2175 static int mp_reduce_2k_setup_l(mp_int *a, mp_int *d)
2176 {
2177    int    res;
2178    mp_int tmp;
2179 
2180    if ((res = mp_init(&tmp)) != MP_OKAY) {
2181       return res;
2182    }
2183 
2184    if ((res = mp_2expt(&tmp, mp_count_bits(a))) != MP_OKAY) {
2185       goto ERR;
2186    }
2187 
2188    if ((res = s_mp_sub(&tmp, a, d)) != MP_OKAY) {
2189       goto ERR;
2190    }
2191 
2192 ERR:
2193    mp_clear(&tmp);
2194    return res;
2195 }
2196 
2197 
2198 /* computes a = 2**b
2199  *
2200  * Simple algorithm which zeroes the int, grows it then just sets one bit
2201  * as required.
2202  */
mp_2expt(mp_int * a,int b)2203 static int mp_2expt (mp_int * a, int b)
2204 {
2205   int     res;
2206 
2207   /* zero a as per default */
2208   mp_zero (a);
2209 
2210   /* grow a to accomodate the single bit */
2211   if ((res = mp_grow (a, b / DIGIT_BIT + 1)) != MP_OKAY) {
2212     return res;
2213   }
2214 
2215   /* set the used count of where the bit will go */
2216   a->used = b / DIGIT_BIT + 1;
2217 
2218   /* put the single bit in its place */
2219   a->dp[b / DIGIT_BIT] = ((mp_digit)1) << (b % DIGIT_BIT);
2220 
2221   return MP_OKAY;
2222 }
2223 
2224 
2225 /* pre-calculate the value required for Barrett reduction
2226  * For a given modulus "b" it calulates the value required in "a"
2227  */
mp_reduce_setup(mp_int * a,mp_int * b)2228 static int mp_reduce_setup (mp_int * a, mp_int * b)
2229 {
2230   int     res;
2231 
2232   if ((res = mp_2expt (a, b->used * 2 * DIGIT_BIT)) != MP_OKAY) {
2233     return res;
2234   }
2235   return mp_div (a, b, a, NULL);
2236 }
2237 
2238 
2239 /* reduces x mod m, assumes 0 < x < m**2, mu is
2240  * precomputed via mp_reduce_setup.
2241  * From HAC pp.604 Algorithm 14.42
2242  */
mp_reduce(mp_int * x,mp_int * m,mp_int * mu)2243 static int mp_reduce (mp_int * x, mp_int * m, mp_int * mu)
2244 {
2245   mp_int  q;
2246   int     res, um = m->used;
2247 
2248   /* q = x */
2249   if ((res = mp_init_copy (&q, x)) != MP_OKAY) {
2250     return res;
2251   }
2252 
2253   /* q1 = x / b**(k-1)  */
2254   mp_rshd (&q, um - 1);
2255 
2256   /* according to HAC this optimization is ok */
2257   if (((unsigned long) um) > (((mp_digit)1) << (DIGIT_BIT - 1))) {
2258     if ((res = mp_mul (&q, mu, &q)) != MP_OKAY) {
2259       goto CLEANUP;
2260     }
2261   } else {
2262 #ifdef BN_S_MP_MUL_HIGH_DIGS_C
2263     if ((res = s_mp_mul_high_digs (&q, mu, &q, um)) != MP_OKAY) {
2264       goto CLEANUP;
2265     }
2266 #elif defined(BN_FAST_S_MP_MUL_HIGH_DIGS_C)
2267     if ((res = fast_s_mp_mul_high_digs (&q, mu, &q, um)) != MP_OKAY) {
2268       goto CLEANUP;
2269     }
2270 #else
2271     {
2272 #error mp_reduce would always fail
2273       res = MP_VAL;
2274       goto CLEANUP;
2275     }
2276 #endif
2277   }
2278 
2279   /* q3 = q2 / b**(k+1) */
2280   mp_rshd (&q, um + 1);
2281 
2282   /* x = x mod b**(k+1), quick (no division) */
2283   if ((res = mp_mod_2d (x, DIGIT_BIT * (um + 1), x)) != MP_OKAY) {
2284     goto CLEANUP;
2285   }
2286 
2287   /* q = q * m mod b**(k+1), quick (no division) */
2288   if ((res = s_mp_mul_digs (&q, m, &q, um + 1)) != MP_OKAY) {
2289     goto CLEANUP;
2290   }
2291 
2292   /* x = x - q */
2293   if ((res = mp_sub (x, &q, x)) != MP_OKAY) {
2294     goto CLEANUP;
2295   }
2296 
2297   /* If x < 0, add b**(k+1) to it */
2298   if (mp_cmp_d (x, 0) == MP_LT) {
2299     mp_set (&q, 1);
2300     if ((res = mp_lshd (&q, um + 1)) != MP_OKAY) {
2301       goto CLEANUP;
2302     }
2303     if ((res = mp_add (x, &q, x)) != MP_OKAY) {
2304       goto CLEANUP;
2305     }
2306   }
2307 
2308   /* Back off if it's too big */
2309   while (mp_cmp (x, m) != MP_LT) {
2310     if ((res = s_mp_sub (x, m, x)) != MP_OKAY) {
2311       goto CLEANUP;
2312     }
2313   }
2314 
2315 CLEANUP:
2316   mp_clear (&q);
2317 
2318   return res;
2319 }
2320 
2321 
2322 /* multiplies |a| * |b| and only computes upto digs digits of result
2323  * HAC pp. 595, Algorithm 14.12  Modified so you can control how
2324  * many digits of output are created.
2325  */
s_mp_mul_digs(mp_int * a,mp_int * b,mp_int * c,int digs)2326 static int s_mp_mul_digs (mp_int * a, mp_int * b, mp_int * c, int digs)
2327 {
2328   mp_int  t;
2329   int     res, pa, pb, ix, iy;
2330   mp_digit u;
2331   mp_word r;
2332   mp_digit tmpx, *tmpt, *tmpy;
2333 
2334   /* can we use the fast multiplier? */
2335   if (((digs) < MP_WARRAY) &&
2336       MIN (a->used, b->used) <
2337           (1 << ((CHAR_BIT * sizeof (mp_word)) - (2 * DIGIT_BIT)))) {
2338     return fast_s_mp_mul_digs (a, b, c, digs);
2339   }
2340 
2341   if ((res = mp_init_size (&t, digs)) != MP_OKAY) {
2342     return res;
2343   }
2344   t.used = digs;
2345 
2346   /* compute the digits of the product directly */
2347   pa = a->used;
2348   for (ix = 0; ix < pa; ix++) {
2349     /* set the carry to zero */
2350     u = 0;
2351 
2352     /* limit ourselves to making digs digits of output */
2353     pb = MIN (b->used, digs - ix);
2354 
2355     /* setup some aliases */
2356     /* copy of the digit from a used within the nested loop */
2357     tmpx = a->dp[ix];
2358 
2359     /* an alias for the destination shifted ix places */
2360     tmpt = t.dp + ix;
2361 
2362     /* an alias for the digits of b */
2363     tmpy = b->dp;
2364 
2365     /* compute the columns of the output and propagate the carry */
2366     for (iy = 0; iy < pb; iy++) {
2367       /* compute the column as a mp_word */
2368       r       = ((mp_word)*tmpt) +
2369                 ((mp_word)tmpx) * ((mp_word)*tmpy++) +
2370                 ((mp_word) u);
2371 
2372       /* the new column is the lower part of the result */
2373       *tmpt++ = (mp_digit) (r & ((mp_word) MP_MASK));
2374 
2375       /* get the carry word from the result */
2376       u       = (mp_digit) (r >> ((mp_word) DIGIT_BIT));
2377     }
2378     /* set carry if it is placed below digs */
2379     if (ix + iy < digs) {
2380       *tmpt = u;
2381     }
2382   }
2383 
2384   mp_clamp (&t);
2385   mp_exch (&t, c);
2386 
2387   mp_clear (&t);
2388   return MP_OKAY;
2389 }
2390 
2391 
2392 /* Fast (comba) multiplier
2393  *
2394  * This is the fast column-array [comba] multiplier.  It is
2395  * designed to compute the columns of the product first
2396  * then handle the carries afterwards.  This has the effect
2397  * of making the nested loops that compute the columns very
2398  * simple and schedulable on super-scalar processors.
2399  *
2400  * This has been modified to produce a variable number of
2401  * digits of output so if say only a half-product is required
2402  * you don't have to compute the upper half (a feature
2403  * required for fast Barrett reduction).
2404  *
2405  * Based on Algorithm 14.12 on pp.595 of HAC.
2406  *
2407  */
fast_s_mp_mul_digs(mp_int * a,mp_int * b,mp_int * c,int digs)2408 static int fast_s_mp_mul_digs (mp_int * a, mp_int * b, mp_int * c, int digs)
2409 {
2410   int     olduse, res, pa, ix, iz;
2411   mp_digit W[MP_WARRAY];
2412   register mp_word  _W;
2413 
2414   /* grow the destination as required */
2415   if (c->alloc < digs) {
2416     if ((res = mp_grow (c, digs)) != MP_OKAY) {
2417       return res;
2418     }
2419   }
2420 
2421   /* number of output digits to produce */
2422   pa = MIN(digs, a->used + b->used);
2423 
2424   /* clear the carry */
2425   _W = 0;
2426   for (ix = 0; ix < pa; ix++) {
2427       int      tx, ty;
2428       int      iy;
2429       mp_digit *tmpx, *tmpy;
2430 
2431       /* get offsets into the two bignums */
2432       ty = MIN(b->used-1, ix);
2433       tx = ix - ty;
2434 
2435       /* setup temp aliases */
2436       tmpx = a->dp + tx;
2437       tmpy = b->dp + ty;
2438 
2439       /* this is the number of times the loop will iterrate, essentially
2440          while (tx++ < a->used && ty-- >= 0) { ... }
2441        */
2442       iy = MIN(a->used-tx, ty+1);
2443 
2444       /* execute loop */
2445       for (iz = 0; iz < iy; ++iz) {
2446          _W += ((mp_word)*tmpx++)*((mp_word)*tmpy--);
2447 
2448       }
2449 
2450       /* store term */
2451       W[ix] = ((mp_digit)_W) & MP_MASK;
2452 
2453       /* make next carry */
2454       _W = _W >> ((mp_word)DIGIT_BIT);
2455  }
2456 
2457   /* setup dest */
2458   olduse  = c->used;
2459   c->used = pa;
2460 
2461   {
2462     register mp_digit *tmpc;
2463     tmpc = c->dp;
2464     for (ix = 0; ix < pa+1; ix++) {
2465       /* now extract the previous digit [below the carry] */
2466       *tmpc++ = W[ix];
2467     }
2468 
2469     /* clear unused digits [that existed in the old copy of c] */
2470     for (; ix < olduse; ix++) {
2471       *tmpc++ = 0;
2472     }
2473   }
2474   mp_clamp (c);
2475   return MP_OKAY;
2476 }
2477 
2478 
2479 /* init an mp_init for a given size */
mp_init_size(mp_int * a,int size)2480 static int mp_init_size (mp_int * a, int size)
2481 {
2482   int x;
2483 
2484   /* pad size so there are always extra digits */
2485   size += (MP_PREC * 2) - (size % MP_PREC);
2486 
2487   /* alloc mem */
2488   a->dp = OPT_CAST(mp_digit) XMALLOC (sizeof (mp_digit) * size);
2489   if (a->dp == NULL) {
2490     return MP_MEM;
2491   }
2492 
2493   /* set the members */
2494   a->used  = 0;
2495   a->alloc = size;
2496   a->sign  = MP_ZPOS;
2497 
2498   /* zero the digits */
2499   for (x = 0; x < size; x++) {
2500       a->dp[x] = 0;
2501   }
2502 
2503   return MP_OKAY;
2504 }
2505 
2506 
2507 /* low level squaring, b = a*a, HAC pp.596-597, Algorithm 14.16 */
s_mp_sqr(mp_int * a,mp_int * b)2508 static int s_mp_sqr (mp_int * a, mp_int * b)
2509 {
2510   mp_int  t;
2511   int     res, ix, iy, pa;
2512   mp_word r;
2513   mp_digit u, tmpx, *tmpt;
2514 
2515   pa = a->used;
2516   if ((res = mp_init_size (&t, 2*pa + 1)) != MP_OKAY) {
2517     return res;
2518   }
2519 
2520   /* default used is maximum possible size */
2521   t.used = 2*pa + 1;
2522 
2523   for (ix = 0; ix < pa; ix++) {
2524     /* first calculate the digit at 2*ix */
2525     /* calculate double precision result */
2526     r = ((mp_word) t.dp[2*ix]) +
2527         ((mp_word)a->dp[ix])*((mp_word)a->dp[ix]);
2528 
2529     /* store lower part in result */
2530     t.dp[ix+ix] = (mp_digit) (r & ((mp_word) MP_MASK));
2531 
2532     /* get the carry */
2533     u           = (mp_digit)(r >> ((mp_word) DIGIT_BIT));
2534 
2535     /* left hand side of A[ix] * A[iy] */
2536     tmpx        = a->dp[ix];
2537 
2538     /* alias for where to store the results */
2539     tmpt        = t.dp + (2*ix + 1);
2540 
2541     for (iy = ix + 1; iy < pa; iy++) {
2542       /* first calculate the product */
2543       r       = ((mp_word)tmpx) * ((mp_word)a->dp[iy]);
2544 
2545       /* now calculate the double precision result, note we use
2546        * addition instead of *2 since it's easier to optimize
2547        */
2548       r       = ((mp_word) *tmpt) + r + r + ((mp_word) u);
2549 
2550       /* store lower part */
2551       *tmpt++ = (mp_digit) (r & ((mp_word) MP_MASK));
2552 
2553       /* get carry */
2554       u       = (mp_digit)(r >> ((mp_word) DIGIT_BIT));
2555     }
2556     /* propagate upwards */
2557     while (u != ((mp_digit) 0)) {
2558       r       = ((mp_word) *tmpt) + ((mp_word) u);
2559       *tmpt++ = (mp_digit) (r & ((mp_word) MP_MASK));
2560       u       = (mp_digit)(r >> ((mp_word) DIGIT_BIT));
2561     }
2562   }
2563 
2564   mp_clamp (&t);
2565   mp_exch (&t, b);
2566   mp_clear (&t);
2567   return MP_OKAY;
2568 }
2569 
2570 
2571 /* multiplies |a| * |b| and does not compute the lower digs digits
2572  * [meant to get the higher part of the product]
2573  */
s_mp_mul_high_digs(mp_int * a,mp_int * b,mp_int * c,int digs)2574 static int s_mp_mul_high_digs (mp_int * a, mp_int * b, mp_int * c, int digs)
2575 {
2576   mp_int  t;
2577   int     res, pa, pb, ix, iy;
2578   mp_digit u;
2579   mp_word r;
2580   mp_digit tmpx, *tmpt, *tmpy;
2581 
2582   /* can we use the fast multiplier? */
2583 #ifdef BN_FAST_S_MP_MUL_HIGH_DIGS_C
2584   if (((a->used + b->used + 1) < MP_WARRAY)
2585       && MIN (a->used, b->used) < (1 << ((CHAR_BIT * sizeof (mp_word)) - (2 * DIGIT_BIT)))) {
2586     return fast_s_mp_mul_high_digs (a, b, c, digs);
2587   }
2588 #endif
2589 
2590   if ((res = mp_init_size (&t, a->used + b->used + 1)) != MP_OKAY) {
2591     return res;
2592   }
2593   t.used = a->used + b->used + 1;
2594 
2595   pa = a->used;
2596   pb = b->used;
2597   for (ix = 0; ix < pa; ix++) {
2598     /* clear the carry */
2599     u = 0;
2600 
2601     /* left hand side of A[ix] * B[iy] */
2602     tmpx = a->dp[ix];
2603 
2604     /* alias to the address of where the digits will be stored */
2605     tmpt = &(t.dp[digs]);
2606 
2607     /* alias for where to read the right hand side from */
2608     tmpy = b->dp + (digs - ix);
2609 
2610     for (iy = digs - ix; iy < pb; iy++) {
2611       /* calculate the double precision result */
2612       r       = ((mp_word)*tmpt) +
2613                 ((mp_word)tmpx) * ((mp_word)*tmpy++) +
2614                 ((mp_word) u);
2615 
2616       /* get the lower part */
2617       *tmpt++ = (mp_digit) (r & ((mp_word) MP_MASK));
2618 
2619       /* carry the carry */
2620       u       = (mp_digit) (r >> ((mp_word) DIGIT_BIT));
2621     }
2622     *tmpt = u;
2623   }
2624   mp_clamp (&t);
2625   mp_exch (&t, c);
2626   mp_clear (&t);
2627   return MP_OKAY;
2628 }
2629 
2630 
2631 #ifdef BN_MP_MONTGOMERY_SETUP_C
2632 /* setups the montgomery reduction stuff */
2633 static int
mp_montgomery_setup(mp_int * n,mp_digit * rho)2634 mp_montgomery_setup (mp_int * n, mp_digit * rho)
2635 {
2636   mp_digit x, b;
2637 
2638 /* fast inversion mod 2**k
2639  *
2640  * Based on the fact that
2641  *
2642  * XA = 1 (mod 2**n)  =>  (X(2-XA)) A = 1 (mod 2**2n)
2643  *                    =>  2*X*A - X*X*A*A = 1
2644  *                    =>  2*(1) - (1)     = 1
2645  */
2646   b = n->dp[0];
2647 
2648   if ((b & 1) == 0) {
2649     return MP_VAL;
2650   }
2651 
2652   x = (((b + 2) & 4) << 1) + b; /* here x*a==1 mod 2**4 */
2653   x *= 2 - b * x;               /* here x*a==1 mod 2**8 */
2654 #if !defined(MP_8BIT)
2655   x *= 2 - b * x;               /* here x*a==1 mod 2**16 */
2656 #endif
2657 #if defined(MP_64BIT) || !(defined(MP_8BIT) || defined(MP_16BIT))
2658   x *= 2 - b * x;               /* here x*a==1 mod 2**32 */
2659 #endif
2660 #ifdef MP_64BIT
2661   x *= 2 - b * x;               /* here x*a==1 mod 2**64 */
2662 #endif
2663 
2664   /* rho = -1/m mod b */
2665   *rho = (unsigned long)(((mp_word)1 << ((mp_word) DIGIT_BIT)) - x) & MP_MASK;
2666 
2667   return MP_OKAY;
2668 }
2669 #endif
2670 
2671 
2672 #ifdef BN_FAST_MP_MONTGOMERY_REDUCE_C
2673 /* computes xR**-1 == x (mod N) via Montgomery Reduction
2674  *
2675  * This is an optimized implementation of montgomery_reduce
2676  * which uses the comba method to quickly calculate the columns of the
2677  * reduction.
2678  *
2679  * Based on Algorithm 14.32 on pp.601 of HAC.
2680 */
fast_mp_montgomery_reduce(mp_int * x,mp_int * n,mp_digit rho)2681 int fast_mp_montgomery_reduce (mp_int * x, mp_int * n, mp_digit rho)
2682 {
2683   int     ix, res, olduse;
2684   mp_word W[MP_WARRAY];
2685 
2686   /* get old used count */
2687   olduse = x->used;
2688 
2689   /* grow a as required */
2690   if (x->alloc < n->used + 1) {
2691     if ((res = mp_grow (x, n->used + 1)) != MP_OKAY) {
2692       return res;
2693     }
2694   }
2695 
2696   /* first we have to get the digits of the input into
2697    * an array of double precision words W[...]
2698    */
2699   {
2700     register mp_word *_W;
2701     register mp_digit *tmpx;
2702 
2703     /* alias for the W[] array */
2704     _W   = W;
2705 
2706     /* alias for the digits of  x*/
2707     tmpx = x->dp;
2708 
2709     /* copy the digits of a into W[0..a->used-1] */
2710     for (ix = 0; ix < x->used; ix++) {
2711       *_W++ = *tmpx++;
2712     }
2713 
2714     /* zero the high words of W[a->used..m->used*2] */
2715     for (; ix < n->used * 2 + 1; ix++) {
2716       *_W++ = 0;
2717     }
2718   }
2719 
2720   /* now we proceed to zero successive digits
2721    * from the least significant upwards
2722    */
2723   for (ix = 0; ix < n->used; ix++) {
2724     /* mu = ai * m' mod b
2725      *
2726      * We avoid a double precision multiplication (which isn't required)
2727      * by casting the value down to a mp_digit.  Note this requires
2728      * that W[ix-1] have  the carry cleared (see after the inner loop)
2729      */
2730     register mp_digit mu;
2731     mu = (mp_digit) (((W[ix] & MP_MASK) * rho) & MP_MASK);
2732 
2733     /* a = a + mu * m * b**i
2734      *
2735      * This is computed in place and on the fly.  The multiplication
2736      * by b**i is handled by offseting which columns the results
2737      * are added to.
2738      *
2739      * Note the comba method normally doesn't handle carries in the
2740      * inner loop In this case we fix the carry from the previous
2741      * column since the Montgomery reduction requires digits of the
2742      * result (so far) [see above] to work.  This is
2743      * handled by fixing up one carry after the inner loop.  The
2744      * carry fixups are done in order so after these loops the
2745      * first m->used words of W[] have the carries fixed
2746      */
2747     {
2748       register int iy;
2749       register mp_digit *tmpn;
2750       register mp_word *_W;
2751 
2752       /* alias for the digits of the modulus */
2753       tmpn = n->dp;
2754 
2755       /* Alias for the columns set by an offset of ix */
2756       _W = W + ix;
2757 
2758       /* inner loop */
2759       for (iy = 0; iy < n->used; iy++) {
2760           *_W++ += ((mp_word)mu) * ((mp_word)*tmpn++);
2761       }
2762     }
2763 
2764     /* now fix carry for next digit, W[ix+1] */
2765     W[ix + 1] += W[ix] >> ((mp_word) DIGIT_BIT);
2766   }
2767 
2768   /* now we have to propagate the carries and
2769    * shift the words downward [all those least
2770    * significant digits we zeroed].
2771    */
2772   {
2773     register mp_digit *tmpx;
2774     register mp_word *_W, *_W1;
2775 
2776     /* nox fix rest of carries */
2777 
2778     /* alias for current word */
2779     _W1 = W + ix;
2780 
2781     /* alias for next word, where the carry goes */
2782     _W = W + ++ix;
2783 
2784     for (; ix <= n->used * 2 + 1; ix++) {
2785       *_W++ += *_W1++ >> ((mp_word) DIGIT_BIT);
2786     }
2787 
2788     /* copy out, A = A/b**n
2789      *
2790      * The result is A/b**n but instead of converting from an
2791      * array of mp_word to mp_digit than calling mp_rshd
2792      * we just copy them in the right order
2793      */
2794 
2795     /* alias for destination word */
2796     tmpx = x->dp;
2797 
2798     /* alias for shifted double precision result */
2799     _W = W + n->used;
2800 
2801     for (ix = 0; ix < n->used + 1; ix++) {
2802       *tmpx++ = (mp_digit)(*_W++ & ((mp_word) MP_MASK));
2803     }
2804 
2805     /* zero oldused digits, if the input a was larger than
2806      * m->used+1 we'll have to clear the digits
2807      */
2808     for (; ix < olduse; ix++) {
2809       *tmpx++ = 0;
2810     }
2811   }
2812 
2813   /* set the max used and clamp */
2814   x->used = n->used + 1;
2815   mp_clamp (x);
2816 
2817   /* if A >= m then A = A - m */
2818   if (mp_cmp_mag (x, n) != MP_LT) {
2819     return s_mp_sub (x, n, x);
2820   }
2821   return MP_OKAY;
2822 }
2823 #endif
2824 
2825 
2826 #ifdef BN_MP_MUL_2_C
2827 /* b = a*2 */
mp_mul_2(mp_int * a,mp_int * b)2828 static int mp_mul_2(mp_int * a, mp_int * b)
2829 {
2830   int     x, res, oldused;
2831 
2832   /* grow to accomodate result */
2833   if (b->alloc < a->used + 1) {
2834     if ((res = mp_grow (b, a->used + 1)) != MP_OKAY) {
2835       return res;
2836     }
2837   }
2838 
2839   oldused = b->used;
2840   b->used = a->used;
2841 
2842   {
2843     register mp_digit r, rr, *tmpa, *tmpb;
2844 
2845     /* alias for source */
2846     tmpa = a->dp;
2847 
2848     /* alias for dest */
2849     tmpb = b->dp;
2850 
2851     /* carry */
2852     r = 0;
2853     for (x = 0; x < a->used; x++) {
2854 
2855       /* get what will be the *next* carry bit from the
2856        * MSB of the current digit
2857        */
2858       rr = *tmpa >> ((mp_digit)(DIGIT_BIT - 1));
2859 
2860       /* now shift up this digit, add in the carry [from the previous] */
2861       *tmpb++ = ((*tmpa++ << ((mp_digit)1)) | r) & MP_MASK;
2862 
2863       /* copy the carry that would be from the source
2864        * digit into the next iteration
2865        */
2866       r = rr;
2867     }
2868 
2869     /* new leading digit? */
2870     if (r != 0) {
2871       /* add a MSB which is always 1 at this point */
2872       *tmpb = 1;
2873       ++(b->used);
2874     }
2875 
2876     /* now zero any excess digits on the destination
2877      * that we didn't write to
2878      */
2879     tmpb = b->dp + b->used;
2880     for (x = b->used; x < oldused; x++) {
2881       *tmpb++ = 0;
2882     }
2883   }
2884   b->sign = a->sign;
2885   return MP_OKAY;
2886 }
2887 #endif
2888 
2889 
2890 #ifdef BN_MP_MONTGOMERY_CALC_NORMALIZATION_C
2891 /*
2892  * shifts with subtractions when the result is greater than b.
2893  *
2894  * The method is slightly modified to shift B unconditionally upto just under
2895  * the leading bit of b.  This saves alot of multiple precision shifting.
2896  */
mp_montgomery_calc_normalization(mp_int * a,mp_int * b)2897 static int mp_montgomery_calc_normalization (mp_int * a, mp_int * b)
2898 {
2899   int     x, bits, res;
2900 
2901   /* how many bits of last digit does b use */
2902   bits = mp_count_bits (b) % DIGIT_BIT;
2903 
2904   if (b->used > 1) {
2905      if ((res = mp_2expt (a, (b->used - 1) * DIGIT_BIT + bits - 1)) != MP_OKAY) {
2906         return res;
2907      }
2908   } else {
2909      mp_set(a, 1);
2910      bits = 1;
2911   }
2912 
2913 
2914   /* now compute C = A * B mod b */
2915   for (x = bits - 1; x < (int)DIGIT_BIT; x++) {
2916     if ((res = mp_mul_2 (a, a)) != MP_OKAY) {
2917       return res;
2918     }
2919     if (mp_cmp_mag (a, b) != MP_LT) {
2920       if ((res = s_mp_sub (a, b, a)) != MP_OKAY) {
2921         return res;
2922       }
2923     }
2924   }
2925 
2926   return MP_OKAY;
2927 }
2928 #endif
2929 
2930 
2931 #ifdef BN_MP_EXPTMOD_FAST_C
2932 /* computes Y == G**X mod P, HAC pp.616, Algorithm 14.85
2933  *
2934  * Uses a left-to-right k-ary sliding window to compute the modular exponentiation.
2935  * The value of k changes based on the size of the exponent.
2936  *
2937  * Uses Montgomery or Diminished Radix reduction [whichever appropriate]
2938  */
2939 
mp_exptmod_fast(mp_int * G,mp_int * X,mp_int * P,mp_int * Y,int redmode)2940 static int mp_exptmod_fast (mp_int * G, mp_int * X, mp_int * P, mp_int * Y, int redmode)
2941 {
2942   mp_int  M[TAB_SIZE], res;
2943   mp_digit buf, mp;
2944   int     err, bitbuf, bitcpy, bitcnt, mode, digidx, x, y, winsize;
2945 
2946   /* use a pointer to the reduction algorithm.  This allows us to use
2947    * one of many reduction algorithms without modding the guts of
2948    * the code with if statements everywhere.
2949    */
2950   int     (*redux)(mp_int*,mp_int*,mp_digit);
2951 
2952   /* find window size */
2953   x = mp_count_bits (X);
2954   if (x <= 7) {
2955     winsize = 2;
2956   } else if (x <= 36) {
2957     winsize = 3;
2958   } else if (x <= 140) {
2959     winsize = 4;
2960   } else if (x <= 450) {
2961     winsize = 5;
2962   } else if (x <= 1303) {
2963     winsize = 6;
2964   } else if (x <= 3529) {
2965     winsize = 7;
2966   } else {
2967     winsize = 8;
2968   }
2969 
2970 #ifdef MP_LOW_MEM
2971   if (winsize > 5) {
2972      winsize = 5;
2973   }
2974 #endif
2975 
2976   /* init M array */
2977   /* init first cell */
2978   if ((err = mp_init(&M[1])) != MP_OKAY) {
2979      return err;
2980   }
2981 
2982   /* now init the second half of the array */
2983   for (x = 1<<(winsize-1); x < (1 << winsize); x++) {
2984     if ((err = mp_init(&M[x])) != MP_OKAY) {
2985       for (y = 1<<(winsize-1); y < x; y++) {
2986         mp_clear (&M[y]);
2987       }
2988       mp_clear(&M[1]);
2989       return err;
2990     }
2991   }
2992 
2993   /* determine and setup reduction code */
2994   if (redmode == 0) {
2995 #ifdef BN_MP_MONTGOMERY_SETUP_C
2996      /* now setup montgomery  */
2997      if ((err = mp_montgomery_setup (P, &mp)) != MP_OKAY) {
2998         goto LBL_M;
2999      }
3000 #else
3001      err = MP_VAL;
3002      goto LBL_M;
3003 #endif
3004 
3005      /* automatically pick the comba one if available (saves quite a few calls/ifs) */
3006 #ifdef BN_FAST_MP_MONTGOMERY_REDUCE_C
3007      if (((P->used * 2 + 1) < MP_WARRAY) &&
3008           P->used < (1 << ((CHAR_BIT * sizeof (mp_word)) - (2 * DIGIT_BIT)))) {
3009         redux = fast_mp_montgomery_reduce;
3010      } else
3011 #endif
3012      {
3013 #ifdef BN_MP_MONTGOMERY_REDUCE_C
3014         /* use slower baseline Montgomery method */
3015         redux = mp_montgomery_reduce;
3016 #else
3017         err = MP_VAL;
3018         goto LBL_M;
3019 #endif
3020      }
3021   } else if (redmode == 1) {
3022 #if defined(BN_MP_DR_SETUP_C) && defined(BN_MP_DR_REDUCE_C)
3023      /* setup DR reduction for moduli of the form B**k - b */
3024      mp_dr_setup(P, &mp);
3025      redux = mp_dr_reduce;
3026 #else
3027      err = MP_VAL;
3028      goto LBL_M;
3029 #endif
3030   } else {
3031 #if defined(BN_MP_REDUCE_2K_SETUP_C) && defined(BN_MP_REDUCE_2K_C)
3032      /* setup DR reduction for moduli of the form 2**k - b */
3033      if ((err = mp_reduce_2k_setup(P, &mp)) != MP_OKAY) {
3034         goto LBL_M;
3035      }
3036      redux = mp_reduce_2k;
3037 #else
3038      err = MP_VAL;
3039      goto LBL_M;
3040 #endif
3041   }
3042 
3043   /* setup result */
3044   if ((err = mp_init (&res)) != MP_OKAY) {
3045     goto LBL_M;
3046   }
3047 
3048   /* create M table
3049    *
3050 
3051    *
3052    * The first half of the table is not computed though accept for M[0] and M[1]
3053    */
3054 
3055   if (redmode == 0) {
3056 #ifdef BN_MP_MONTGOMERY_CALC_NORMALIZATION_C
3057      /* now we need R mod m */
3058      if ((err = mp_montgomery_calc_normalization (&res, P)) != MP_OKAY) {
3059        goto LBL_RES;
3060      }
3061 #else
3062      err = MP_VAL;
3063      goto LBL_RES;
3064 #endif
3065 
3066      /* now set M[1] to G * R mod m */
3067      if ((err = mp_mulmod (G, &res, P, &M[1])) != MP_OKAY) {
3068        goto LBL_RES;
3069      }
3070   } else {
3071      mp_set(&res, 1);
3072      if ((err = mp_mod(G, P, &M[1])) != MP_OKAY) {
3073         goto LBL_RES;
3074      }
3075   }
3076 
3077   /* compute the value at M[1<<(winsize-1)] by squaring M[1] (winsize-1) times */
3078   if ((err = mp_copy (&M[1], &M[1 << (winsize - 1)])) != MP_OKAY) {
3079     goto LBL_RES;
3080   }
3081 
3082   for (x = 0; x < (winsize - 1); x++) {
3083     if ((err = mp_sqr (&M[1 << (winsize - 1)], &M[1 << (winsize - 1)])) != MP_OKAY) {
3084       goto LBL_RES;
3085     }
3086     if ((err = redux (&M[1 << (winsize - 1)], P, mp)) != MP_OKAY) {
3087       goto LBL_RES;
3088     }
3089   }
3090 
3091   /* create upper table */
3092   for (x = (1 << (winsize - 1)) + 1; x < (1 << winsize); x++) {
3093     if ((err = mp_mul (&M[x - 1], &M[1], &M[x])) != MP_OKAY) {
3094       goto LBL_RES;
3095     }
3096     if ((err = redux (&M[x], P, mp)) != MP_OKAY) {
3097       goto LBL_RES;
3098     }
3099   }
3100 
3101   /* set initial mode and bit cnt */
3102   mode   = 0;
3103   bitcnt = 1;
3104   buf    = 0;
3105   digidx = X->used - 1;
3106   bitcpy = 0;
3107   bitbuf = 0;
3108 
3109   for (;;) {
3110     /* grab next digit as required */
3111     if (--bitcnt == 0) {
3112       /* if digidx == -1 we are out of digits so break */
3113       if (digidx == -1) {
3114         break;
3115       }
3116       /* read next digit and reset bitcnt */
3117       buf    = X->dp[digidx--];
3118       bitcnt = (int)DIGIT_BIT;
3119     }
3120 
3121     /* grab the next msb from the exponent */
3122     y     = (mp_digit)(buf >> (DIGIT_BIT - 1)) & 1;
3123     buf <<= (mp_digit)1;
3124 
3125     /* if the bit is zero and mode == 0 then we ignore it
3126      * These represent the leading zero bits before the first 1 bit
3127      * in the exponent.  Technically this opt is not required but it
3128      * does lower the # of trivial squaring/reductions used
3129      */
3130     if (mode == 0 && y == 0) {
3131       continue;
3132     }
3133 
3134     /* if the bit is zero and mode == 1 then we square */
3135     if (mode == 1 && y == 0) {
3136       if ((err = mp_sqr (&res, &res)) != MP_OKAY) {
3137         goto LBL_RES;
3138       }
3139       if ((err = redux (&res, P, mp)) != MP_OKAY) {
3140         goto LBL_RES;
3141       }
3142       continue;
3143     }
3144 
3145     /* else we add it to the window */
3146     bitbuf |= (y << (winsize - ++bitcpy));
3147     mode    = 2;
3148 
3149     if (bitcpy == winsize) {
3150       /* ok window is filled so square as required and multiply  */
3151       /* square first */
3152       for (x = 0; x < winsize; x++) {
3153         if ((err = mp_sqr (&res, &res)) != MP_OKAY) {
3154           goto LBL_RES;
3155         }
3156         if ((err = redux (&res, P, mp)) != MP_OKAY) {
3157           goto LBL_RES;
3158         }
3159       }
3160 
3161       /* then multiply */
3162       if ((err = mp_mul (&res, &M[bitbuf], &res)) != MP_OKAY) {
3163         goto LBL_RES;
3164       }
3165       if ((err = redux (&res, P, mp)) != MP_OKAY) {
3166         goto LBL_RES;
3167       }
3168 
3169       /* empty window and reset */
3170       bitcpy = 0;
3171       bitbuf = 0;
3172       mode   = 1;
3173     }
3174   }
3175 
3176   /* if bits remain then square/multiply */
3177   if (mode == 2 && bitcpy > 0) {
3178     /* square then multiply if the bit is set */
3179     for (x = 0; x < bitcpy; x++) {
3180       if ((err = mp_sqr (&res, &res)) != MP_OKAY) {
3181         goto LBL_RES;
3182       }
3183       if ((err = redux (&res, P, mp)) != MP_OKAY) {
3184         goto LBL_RES;
3185       }
3186 
3187       /* get next bit of the window */
3188       bitbuf <<= 1;
3189       if ((bitbuf & (1 << winsize)) != 0) {
3190         /* then multiply */
3191         if ((err = mp_mul (&res, &M[1], &res)) != MP_OKAY) {
3192           goto LBL_RES;
3193         }
3194         if ((err = redux (&res, P, mp)) != MP_OKAY) {
3195           goto LBL_RES;
3196         }
3197       }
3198     }
3199   }
3200 
3201   if (redmode == 0) {
3202      /* fixup result if Montgomery reduction is used
3203       * recall that any value in a Montgomery system is
3204       * actually multiplied by R mod n.  So we have
3205       * to reduce one more time to cancel out the factor
3206       * of R.
3207       */
3208      if ((err = redux(&res, P, mp)) != MP_OKAY) {
3209        goto LBL_RES;
3210      }
3211   }
3212 
3213   /* swap res with Y */
3214   mp_exch (&res, Y);
3215   err = MP_OKAY;
3216 LBL_RES:mp_clear (&res);
3217 LBL_M:
3218   mp_clear(&M[1]);
3219   for (x = 1<<(winsize-1); x < (1 << winsize); x++) {
3220     mp_clear (&M[x]);
3221   }
3222   return err;
3223 }
3224 #endif
3225 
3226 
3227 #ifdef BN_FAST_S_MP_SQR_C
3228 /* the jist of squaring...
3229  * you do like mult except the offset of the tmpx [one that
3230  * starts closer to zero] can't equal the offset of tmpy.
3231  * So basically you set up iy like before then you min it with
3232  * (ty-tx) so that it never happens.  You double all those
3233  * you add in the inner loop
3234 
3235 After that loop you do the squares and add them in.
3236 */
3237 
fast_s_mp_sqr(mp_int * a,mp_int * b)3238 static int fast_s_mp_sqr (mp_int * a, mp_int * b)
3239 {
3240   int       olduse, res, pa, ix, iz;
3241   mp_digit   W[MP_WARRAY], *tmpx;
3242   mp_word   W1;
3243 
3244   /* grow the destination as required */
3245   pa = a->used + a->used;
3246   if (b->alloc < pa) {
3247     if ((res = mp_grow (b, pa)) != MP_OKAY) {
3248       return res;
3249     }
3250   }
3251 
3252   /* number of output digits to produce */
3253   W1 = 0;
3254   for (ix = 0; ix < pa; ix++) {
3255       int      tx, ty, iy;
3256       mp_word  _W;
3257       mp_digit *tmpy;
3258 
3259       /* clear counter */
3260       _W = 0;
3261 
3262       /* get offsets into the two bignums */
3263       ty = MIN(a->used-1, ix);
3264       tx = ix - ty;
3265 
3266       /* setup temp aliases */
3267       tmpx = a->dp + tx;
3268       tmpy = a->dp + ty;
3269 
3270       /* this is the number of times the loop will iterrate, essentially
3271          while (tx++ < a->used && ty-- >= 0) { ... }
3272        */
3273       iy = MIN(a->used-tx, ty+1);
3274 
3275       /* now for squaring tx can never equal ty
3276        * we halve the distance since they approach at a rate of 2x
3277        * and we have to round because odd cases need to be executed
3278        */
3279       iy = MIN(iy, (ty-tx+1)>>1);
3280 
3281       /* execute loop */
3282       for (iz = 0; iz < iy; iz++) {
3283          _W += ((mp_word)*tmpx++)*((mp_word)*tmpy--);
3284       }
3285 
3286       /* double the inner product and add carry */
3287       _W = _W + _W + W1;
3288 
3289       /* even columns have the square term in them */
3290       if ((ix&1) == 0) {
3291          _W += ((mp_word)a->dp[ix>>1])*((mp_word)a->dp[ix>>1]);
3292       }
3293 
3294       /* store it */
3295       W[ix] = (mp_digit)(_W & MP_MASK);
3296 
3297       /* make next carry */
3298       W1 = _W >> ((mp_word)DIGIT_BIT);
3299   }
3300 
3301   /* setup dest */
3302   olduse  = b->used;
3303   b->used = a->used+a->used;
3304 
3305   {
3306     mp_digit *tmpb;
3307     tmpb = b->dp;
3308     for (ix = 0; ix < pa; ix++) {
3309       *tmpb++ = W[ix] & MP_MASK;
3310     }
3311 
3312     /* clear unused digits [that existed in the old copy of c] */
3313     for (; ix < olduse; ix++) {
3314       *tmpb++ = 0;
3315     }
3316   }
3317   mp_clamp (b);
3318   return MP_OKAY;
3319 }
3320 #endif
3321 
3322 
3323 #ifdef BN_MP_MUL_D_C
3324 /* multiply by a digit */
3325 static int
mp_mul_d(mp_int * a,mp_digit b,mp_int * c)3326 mp_mul_d (mp_int * a, mp_digit b, mp_int * c)
3327 {
3328   mp_digit u, *tmpa, *tmpc;
3329   mp_word  r;
3330   int      ix, res, olduse;
3331 
3332   /* make sure c is big enough to hold a*b */
3333   if (c->alloc < a->used + 1) {
3334     if ((res = mp_grow (c, a->used + 1)) != MP_OKAY) {
3335       return res;
3336     }
3337   }
3338 
3339   /* get the original destinations used count */
3340   olduse = c->used;
3341 
3342   /* set the sign */
3343   c->sign = a->sign;
3344 
3345   /* alias for a->dp [source] */
3346   tmpa = a->dp;
3347 
3348   /* alias for c->dp [dest] */
3349   tmpc = c->dp;
3350 
3351   /* zero carry */
3352   u = 0;
3353 
3354   /* compute columns */
3355   for (ix = 0; ix < a->used; ix++) {
3356     /* compute product and carry sum for this term */
3357     r       = ((mp_word) u) + ((mp_word)*tmpa++) * ((mp_word)b);
3358 
3359     /* mask off higher bits to get a single digit */
3360     *tmpc++ = (mp_digit) (r & ((mp_word) MP_MASK));
3361 
3362     /* send carry into next iteration */
3363     u       = (mp_digit) (r >> ((mp_word) DIGIT_BIT));
3364   }
3365 
3366   /* store final carry [if any] and increment ix offset  */
3367   *tmpc++ = u;
3368   ++ix;
3369 
3370   /* now zero digits above the top */
3371   while (ix++ < olduse) {
3372      *tmpc++ = 0;
3373   }
3374 
3375   /* set used count */
3376   c->used = a->used + 1;
3377   mp_clamp(c);
3378 
3379   return MP_OKAY;
3380 }
3381 #endif
3382