• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  * Minimal code for RSA support from LibTomMath 0.3.9
3  * http://math.libtomcrypt.com/
4  * http://math.libtomcrypt.com/files/ltm-0.39.tar.bz2
5  * This library was released in public domain by Tom St Denis.
6  *
7  * The combination in this file is not using many of the optimized algorithms
8  * (e.g., Montgomery reduction) and is considerable slower than the LibTomMath
9  * with its default of SC_RSA_1 settins. The main purpose of having this
10  * version here is to make it easier to build bignum.c wrapper without having
11  * to install and build an external library. However, it is likely worth the
12  * effort to use the full library with SC_RSA_1 instead of this minimized copy.
13  * Including the optimized algorithms may increase the size requirements by
14  * 15 kB or so (measured with x86 build).
15  *
16  * If CONFIG_INTERNAL_LIBTOMMATH is defined, bignum.c includes this
17  * libtommath.c file instead of using the external LibTomMath library.
18  */
19 
20 #ifndef CHAR_BIT
21 #define CHAR_BIT 8
22 #endif
23 
24 #define BN_MP_INVMOD_C
25 #define BN_S_MP_EXPTMOD_C /* Note: #undef in tommath_superclass.h; this would
26 			   * require BN_MP_EXPTMOD_FAST_C instead */
27 #define BN_S_MP_MUL_DIGS_C
28 #define BN_MP_INVMOD_SLOW_C
29 #define BN_S_MP_SQR_C
30 #define BN_S_MP_MUL_HIGH_DIGS_C /* Note: #undef in tommath_superclass.h; this
31 				 * would require other than mp_reduce */
32 
33 
34 /* from tommath.h */
35 
36 #ifndef MIN
37    #define MIN(x,y) ((x)<(y)?(x):(y))
38 #endif
39 
40 #ifndef MAX
41    #define MAX(x,y) ((x)>(y)?(x):(y))
42 #endif
43 
44 #define  OPT_CAST(x)
45 
46 typedef unsigned long mp_digit;
47 typedef u64 mp_word;
48 
49 #define DIGIT_BIT          28
50 #define MP_28BIT
51 
52 
53 #define XMALLOC  os_malloc
54 #define XFREE    os_free
55 #define XREALLOC os_realloc
56 
57 
58 #define MP_MASK          ((((mp_digit)1)<<((mp_digit)DIGIT_BIT))-((mp_digit)1))
59 
60 #define MP_LT        -1   /* less than */
61 #define MP_EQ         0   /* equal to */
62 #define MP_GT         1   /* greater than */
63 
64 #define MP_ZPOS       0   /* positive integer */
65 #define MP_NEG        1   /* negative */
66 
67 #define MP_OKAY       0   /* ok result */
68 #define MP_MEM        -2  /* out of mem */
69 #define MP_VAL        -3  /* invalid input */
70 
71 #define MP_YES        1   /* yes response */
72 #define MP_NO         0   /* no response */
73 
74 typedef int           mp_err;
75 
76 /* define this to use lower memory usage routines (exptmods mostly) */
77 #define MP_LOW_MEM
78 
79 /* default precision */
80 #ifndef MP_PREC
81    #ifndef MP_LOW_MEM
82       #define MP_PREC                 32     /* default digits of precision */
83    #else
84       #define MP_PREC                 8      /* default digits of precision */
85    #endif
86 #endif
87 
88 /* size of comba arrays, should be at least 2 * 2**(BITS_PER_WORD - BITS_PER_DIGIT*2) */
89 #define MP_WARRAY               (1 << (sizeof(mp_word) * CHAR_BIT - 2 * DIGIT_BIT + 1))
90 
91 /* the infamous mp_int structure */
92 typedef struct  {
93     int used, alloc, sign;
94     mp_digit *dp;
95 } mp_int;
96 
97 
98 /* ---> Basic Manipulations <--- */
99 #define mp_iszero(a) (((a)->used == 0) ? MP_YES : MP_NO)
100 #define mp_iseven(a) (((a)->used > 0 && (((a)->dp[0] & 1) == 0)) ? MP_YES : MP_NO)
101 #define mp_isodd(a)  (((a)->used > 0 && (((a)->dp[0] & 1) == 1)) ? MP_YES : MP_NO)
102 
103 
104 /* prototypes for copied functions */
105 #define s_mp_mul(a, b, c) s_mp_mul_digs(a, b, c, (a)->used + (b)->used + 1)
106 static int s_mp_exptmod(mp_int * G, mp_int * X, mp_int * P, mp_int * Y, int redmode);
107 static int s_mp_mul_digs (mp_int * a, mp_int * b, mp_int * c, int digs);
108 static int s_mp_sqr(mp_int * a, mp_int * b);
109 static int s_mp_mul_high_digs(mp_int * a, mp_int * b, mp_int * c, int digs);
110 
111 static int fast_s_mp_mul_digs (mp_int * a, mp_int * b, mp_int * c, int digs);
112 
113 static int mp_init_multi(mp_int *mp, ...);
114 static void mp_clear_multi(mp_int *mp, ...);
115 static int mp_lshd(mp_int * a, int b);
116 static void mp_set(mp_int * a, mp_digit b);
117 static void mp_clamp(mp_int * a);
118 static void mp_exch(mp_int * a, mp_int * b);
119 static void mp_rshd(mp_int * a, int b);
120 static void mp_zero(mp_int * a);
121 static int mp_mod_2d(mp_int * a, int b, mp_int * c);
122 static int mp_div_2d(mp_int * a, int b, mp_int * c, mp_int * d);
123 static int mp_init_copy(mp_int * a, mp_int * b);
124 static int mp_mul_2d(mp_int * a, int b, mp_int * c);
125 static int mp_div_2(mp_int * a, mp_int * b);
126 static int mp_copy(mp_int * a, mp_int * b);
127 static int mp_count_bits(mp_int * a);
128 static int mp_div(mp_int * a, mp_int * b, mp_int * c, mp_int * d);
129 static int mp_mod(mp_int * a, mp_int * b, mp_int * c);
130 static int mp_grow(mp_int * a, int size);
131 static int mp_cmp_mag(mp_int * a, mp_int * b);
132 static int mp_invmod(mp_int * a, mp_int * b, mp_int * c);
133 static int mp_abs(mp_int * a, mp_int * b);
134 static int mp_invmod_slow(mp_int * a, mp_int * b, mp_int * c);
135 static int mp_sqr(mp_int * a, mp_int * b);
136 static int mp_reduce_2k_l(mp_int *a, mp_int *n, mp_int *d);
137 static int mp_reduce_2k_setup_l(mp_int *a, mp_int *d);
138 static int mp_2expt(mp_int * a, int b);
139 static int mp_reduce_setup(mp_int * a, mp_int * b);
140 static int mp_reduce(mp_int * x, mp_int * m, mp_int * mu);
141 static int mp_init_size(mp_int * a, int size);
142 
143 
144 
145 /* functions from bn_<func name>.c */
146 
147 
148 /* reverse an array, used for radix code */
bn_reverse(unsigned char * s,int len)149 static void bn_reverse (unsigned char *s, int len)
150 {
151   int     ix, iy;
152   unsigned char t;
153 
154   ix = 0;
155   iy = len - 1;
156   while (ix < iy) {
157     t     = s[ix];
158     s[ix] = s[iy];
159     s[iy] = t;
160     ++ix;
161     --iy;
162   }
163 }
164 
165 
166 /* low level addition, based on HAC pp.594, Algorithm 14.7 */
s_mp_add(mp_int * a,mp_int * b,mp_int * c)167 static int s_mp_add (mp_int * a, mp_int * b, mp_int * c)
168 {
169   mp_int *x;
170   int     olduse, res, min, max;
171 
172   /* find sizes, we let |a| <= |b| which means we have to sort
173    * them.  "x" will point to the input with the most digits
174    */
175   if (a->used > b->used) {
176     min = b->used;
177     max = a->used;
178     x = a;
179   } else {
180     min = a->used;
181     max = b->used;
182     x = b;
183   }
184 
185   /* init result */
186   if (c->alloc < max + 1) {
187     if ((res = mp_grow (c, max + 1)) != MP_OKAY) {
188       return res;
189     }
190   }
191 
192   /* get old used digit count and set new one */
193   olduse = c->used;
194   c->used = max + 1;
195 
196   {
197     register mp_digit u, *tmpa, *tmpb, *tmpc;
198     register int i;
199 
200     /* alias for digit pointers */
201 
202     /* first input */
203     tmpa = a->dp;
204 
205     /* second input */
206     tmpb = b->dp;
207 
208     /* destination */
209     tmpc = c->dp;
210 
211     /* zero the carry */
212     u = 0;
213     for (i = 0; i < min; i++) {
214       /* Compute the sum at one digit, T[i] = A[i] + B[i] + U */
215       *tmpc = *tmpa++ + *tmpb++ + u;
216 
217       /* U = carry bit of T[i] */
218       u = *tmpc >> ((mp_digit)DIGIT_BIT);
219 
220       /* take away carry bit from T[i] */
221       *tmpc++ &= MP_MASK;
222     }
223 
224     /* now copy higher words if any, that is in A+B
225      * if A or B has more digits add those in
226      */
227     if (min != max) {
228       for (; i < max; i++) {
229         /* T[i] = X[i] + U */
230         *tmpc = x->dp[i] + u;
231 
232         /* U = carry bit of T[i] */
233         u = *tmpc >> ((mp_digit)DIGIT_BIT);
234 
235         /* take away carry bit from T[i] */
236         *tmpc++ &= MP_MASK;
237       }
238     }
239 
240     /* add carry */
241     *tmpc++ = u;
242 
243     /* clear digits above oldused */
244     for (i = c->used; i < olduse; i++) {
245       *tmpc++ = 0;
246     }
247   }
248 
249   mp_clamp (c);
250   return MP_OKAY;
251 }
252 
253 
254 /* low level subtraction (assumes |a| > |b|), HAC pp.595 Algorithm 14.9 */
s_mp_sub(mp_int * a,mp_int * b,mp_int * c)255 static int s_mp_sub (mp_int * a, mp_int * b, mp_int * c)
256 {
257   int     olduse, res, min, max;
258 
259   /* find sizes */
260   min = b->used;
261   max = a->used;
262 
263   /* init result */
264   if (c->alloc < max) {
265     if ((res = mp_grow (c, max)) != MP_OKAY) {
266       return res;
267     }
268   }
269   olduse = c->used;
270   c->used = max;
271 
272   {
273     register mp_digit u, *tmpa, *tmpb, *tmpc;
274     register int i;
275 
276     /* alias for digit pointers */
277     tmpa = a->dp;
278     tmpb = b->dp;
279     tmpc = c->dp;
280 
281     /* set carry to zero */
282     u = 0;
283     for (i = 0; i < min; i++) {
284       /* T[i] = A[i] - B[i] - U */
285       *tmpc = *tmpa++ - *tmpb++ - u;
286 
287       /* U = carry bit of T[i]
288        * Note this saves performing an AND operation since
289        * if a carry does occur it will propagate all the way to the
290        * MSB.  As a result a single shift is enough to get the carry
291        */
292       u = *tmpc >> ((mp_digit)(CHAR_BIT * sizeof (mp_digit) - 1));
293 
294       /* Clear carry from T[i] */
295       *tmpc++ &= MP_MASK;
296     }
297 
298     /* now copy higher words if any, e.g. if A has more digits than B  */
299     for (; i < max; i++) {
300       /* T[i] = A[i] - U */
301       *tmpc = *tmpa++ - u;
302 
303       /* U = carry bit of T[i] */
304       u = *tmpc >> ((mp_digit)(CHAR_BIT * sizeof (mp_digit) - 1));
305 
306       /* Clear carry from T[i] */
307       *tmpc++ &= MP_MASK;
308     }
309 
310     /* clear digits above used (since we may not have grown result above) */
311     for (i = c->used; i < olduse; i++) {
312       *tmpc++ = 0;
313     }
314   }
315 
316   mp_clamp (c);
317   return MP_OKAY;
318 }
319 
320 
321 /* init a new mp_int */
mp_init(mp_int * a)322 static int mp_init (mp_int * a)
323 {
324   int i;
325 
326   /* allocate memory required and clear it */
327   a->dp = OPT_CAST(mp_digit) XMALLOC (sizeof (mp_digit) * MP_PREC);
328   if (a->dp == NULL) {
329     return MP_MEM;
330   }
331 
332   /* set the digits to zero */
333   for (i = 0; i < MP_PREC; i++) {
334       a->dp[i] = 0;
335   }
336 
337   /* set the used to zero, allocated digits to the default precision
338    * and sign to positive */
339   a->used  = 0;
340   a->alloc = MP_PREC;
341   a->sign  = MP_ZPOS;
342 
343   return MP_OKAY;
344 }
345 
346 
347 /* clear one (frees)  */
mp_clear(mp_int * a)348 static void mp_clear (mp_int * a)
349 {
350   int i;
351 
352   /* only do anything if a hasn't been freed previously */
353   if (a->dp != NULL) {
354     /* first zero the digits */
355     for (i = 0; i < a->used; i++) {
356         a->dp[i] = 0;
357     }
358 
359     /* free ram */
360     XFREE(a->dp);
361 
362     /* reset members to make debugging easier */
363     a->dp    = NULL;
364     a->alloc = a->used = 0;
365     a->sign  = MP_ZPOS;
366   }
367 }
368 
369 
370 /* high level addition (handles signs) */
mp_add(mp_int * a,mp_int * b,mp_int * c)371 static int mp_add (mp_int * a, mp_int * b, mp_int * c)
372 {
373   int     sa, sb, res;
374 
375   /* get sign of both inputs */
376   sa = a->sign;
377   sb = b->sign;
378 
379   /* handle two cases, not four */
380   if (sa == sb) {
381     /* both positive or both negative */
382     /* add their magnitudes, copy the sign */
383     c->sign = sa;
384     res = s_mp_add (a, b, c);
385   } else {
386     /* one positive, the other negative */
387     /* subtract the one with the greater magnitude from */
388     /* the one of the lesser magnitude.  The result gets */
389     /* the sign of the one with the greater magnitude. */
390     if (mp_cmp_mag (a, b) == MP_LT) {
391       c->sign = sb;
392       res = s_mp_sub (b, a, c);
393     } else {
394       c->sign = sa;
395       res = s_mp_sub (a, b, c);
396     }
397   }
398   return res;
399 }
400 
401 
402 /* high level subtraction (handles signs) */
mp_sub(mp_int * a,mp_int * b,mp_int * c)403 static int mp_sub (mp_int * a, mp_int * b, mp_int * c)
404 {
405   int     sa, sb, res;
406 
407   sa = a->sign;
408   sb = b->sign;
409 
410   if (sa != sb) {
411     /* subtract a negative from a positive, OR */
412     /* subtract a positive from a negative. */
413     /* In either case, ADD their magnitudes, */
414     /* and use the sign of the first number. */
415     c->sign = sa;
416     res = s_mp_add (a, b, c);
417   } else {
418     /* subtract a positive from a positive, OR */
419     /* subtract a negative from a negative. */
420     /* First, take the difference between their */
421     /* magnitudes, then... */
422     if (mp_cmp_mag (a, b) != MP_LT) {
423       /* Copy the sign from the first */
424       c->sign = sa;
425       /* The first has a larger or equal magnitude */
426       res = s_mp_sub (a, b, c);
427     } else {
428       /* The result has the *opposite* sign from */
429       /* the first number. */
430       c->sign = (sa == MP_ZPOS) ? MP_NEG : MP_ZPOS;
431       /* The second has a larger magnitude */
432       res = s_mp_sub (b, a, c);
433     }
434   }
435   return res;
436 }
437 
438 
439 /* high level multiplication (handles sign) */
mp_mul(mp_int * a,mp_int * b,mp_int * c)440 static int mp_mul (mp_int * a, mp_int * b, mp_int * c)
441 {
442   int     res, neg;
443   neg = (a->sign == b->sign) ? MP_ZPOS : MP_NEG;
444 
445   /* use Toom-Cook? */
446 #ifdef BN_MP_TOOM_MUL_C
447   if (MIN (a->used, b->used) >= TOOM_MUL_CUTOFF) {
448     res = mp_toom_mul(a, b, c);
449   } else
450 #endif
451 #ifdef BN_MP_KARATSUBA_MUL_C
452   /* use Karatsuba? */
453   if (MIN (a->used, b->used) >= KARATSUBA_MUL_CUTOFF) {
454     res = mp_karatsuba_mul (a, b, c);
455   } else
456 #endif
457   {
458     /* can we use the fast multiplier?
459      *
460      * The fast multiplier can be used if the output will
461      * have less than MP_WARRAY digits and the number of
462      * digits won't affect carry propagation
463      */
464 #ifdef BN_FAST_S_MP_MUL_DIGS_C
465     int     digs = a->used + b->used + 1;
466 
467     if ((digs < MP_WARRAY) &&
468         MIN(a->used, b->used) <=
469         (1 << ((CHAR_BIT * sizeof (mp_word)) - (2 * DIGIT_BIT)))) {
470       res = fast_s_mp_mul_digs (a, b, c, digs);
471     } else
472 #endif
473 #ifdef BN_S_MP_MUL_DIGS_C
474       res = s_mp_mul (a, b, c); /* uses s_mp_mul_digs */
475 #else
476 #error mp_mul could fail
477       res = MP_VAL;
478 #endif
479 
480   }
481   c->sign = (c->used > 0) ? neg : MP_ZPOS;
482   return res;
483 }
484 
485 
486 /* d = a * b (mod c) */
mp_mulmod(mp_int * a,mp_int * b,mp_int * c,mp_int * d)487 static int mp_mulmod (mp_int * a, mp_int * b, mp_int * c, mp_int * d)
488 {
489   int     res;
490   mp_int  t;
491 
492   if ((res = mp_init (&t)) != MP_OKAY) {
493     return res;
494   }
495 
496   if ((res = mp_mul (a, b, &t)) != MP_OKAY) {
497     mp_clear (&t);
498     return res;
499   }
500   res = mp_mod (&t, c, d);
501   mp_clear (&t);
502   return res;
503 }
504 
505 
506 /* c = a mod b, 0 <= c < b */
mp_mod(mp_int * a,mp_int * b,mp_int * c)507 static int mp_mod (mp_int * a, mp_int * b, mp_int * c)
508 {
509   mp_int  t;
510   int     res;
511 
512   if ((res = mp_init (&t)) != MP_OKAY) {
513     return res;
514   }
515 
516   if ((res = mp_div (a, b, NULL, &t)) != MP_OKAY) {
517     mp_clear (&t);
518     return res;
519   }
520 
521   if (t.sign != b->sign) {
522     res = mp_add (b, &t, c);
523   } else {
524     res = MP_OKAY;
525     mp_exch (&t, c);
526   }
527 
528   mp_clear (&t);
529   return res;
530 }
531 
532 
533 /* this is a shell function that calls either the normal or Montgomery
534  * exptmod functions.  Originally the call to the montgomery code was
535  * embedded in the normal function but that wasted alot of stack space
536  * for nothing (since 99% of the time the Montgomery code would be called)
537  */
mp_exptmod(mp_int * G,mp_int * X,mp_int * P,mp_int * Y)538 static int mp_exptmod (mp_int * G, mp_int * X, mp_int * P, mp_int * Y)
539 {
540   int dr;
541 
542   /* modulus P must be positive */
543   if (P->sign == MP_NEG) {
544      return MP_VAL;
545   }
546 
547   /* if exponent X is negative we have to recurse */
548   if (X->sign == MP_NEG) {
549 #ifdef BN_MP_INVMOD_C
550      mp_int tmpG, tmpX;
551      int err;
552 
553      /* first compute 1/G mod P */
554      if ((err = mp_init(&tmpG)) != MP_OKAY) {
555         return err;
556      }
557      if ((err = mp_invmod(G, P, &tmpG)) != MP_OKAY) {
558         mp_clear(&tmpG);
559         return err;
560      }
561 
562      /* now get |X| */
563      if ((err = mp_init(&tmpX)) != MP_OKAY) {
564         mp_clear(&tmpG);
565         return err;
566      }
567      if ((err = mp_abs(X, &tmpX)) != MP_OKAY) {
568         mp_clear_multi(&tmpG, &tmpX, NULL);
569         return err;
570      }
571 
572      /* and now compute (1/G)**|X| instead of G**X [X < 0] */
573      err = mp_exptmod(&tmpG, &tmpX, P, Y);
574      mp_clear_multi(&tmpG, &tmpX, NULL);
575      return err;
576 #else
577 #error mp_exptmod would always fail
578      /* no invmod */
579      return MP_VAL;
580 #endif
581   }
582 
583 /* modified diminished radix reduction */
584 #if defined(BN_MP_REDUCE_IS_2K_L_C) && defined(BN_MP_REDUCE_2K_L_C) && defined(BN_S_MP_EXPTMOD_C)
585   if (mp_reduce_is_2k_l(P) == MP_YES) {
586      return s_mp_exptmod(G, X, P, Y, 1);
587   }
588 #endif
589 
590 #ifdef BN_MP_DR_IS_MODULUS_C
591   /* is it a DR modulus? */
592   dr = mp_dr_is_modulus(P);
593 #else
594   /* default to no */
595   dr = 0;
596 #endif
597 
598 #ifdef BN_MP_REDUCE_IS_2K_C
599   /* if not, is it a unrestricted DR modulus? */
600   if (dr == 0) {
601      dr = mp_reduce_is_2k(P) << 1;
602   }
603 #endif
604 
605   /* if the modulus is odd or dr != 0 use the montgomery method */
606 #ifdef BN_MP_EXPTMOD_FAST_C
607   if (mp_isodd (P) == 1 || dr !=  0) {
608     return mp_exptmod_fast (G, X, P, Y, dr);
609   } else {
610 #endif
611 #ifdef BN_S_MP_EXPTMOD_C
612     /* otherwise use the generic Barrett reduction technique */
613     return s_mp_exptmod (G, X, P, Y, 0);
614 #else
615 #error mp_exptmod could fail
616     /* no exptmod for evens */
617     return MP_VAL;
618 #endif
619 #ifdef BN_MP_EXPTMOD_FAST_C
620   }
621 #endif
622 }
623 
624 
625 /* compare two ints (signed)*/
mp_cmp(mp_int * a,mp_int * b)626 static int mp_cmp (mp_int * a, mp_int * b)
627 {
628   /* compare based on sign */
629   if (a->sign != b->sign) {
630      if (a->sign == MP_NEG) {
631         return MP_LT;
632      } else {
633         return MP_GT;
634      }
635   }
636 
637   /* compare digits */
638   if (a->sign == MP_NEG) {
639      /* if negative compare opposite direction */
640      return mp_cmp_mag(b, a);
641   } else {
642      return mp_cmp_mag(a, b);
643   }
644 }
645 
646 
647 /* compare a digit */
mp_cmp_d(mp_int * a,mp_digit b)648 static int mp_cmp_d(mp_int * a, mp_digit b)
649 {
650   /* compare based on sign */
651   if (a->sign == MP_NEG) {
652     return MP_LT;
653   }
654 
655   /* compare based on magnitude */
656   if (a->used > 1) {
657     return MP_GT;
658   }
659 
660   /* compare the only digit of a to b */
661   if (a->dp[0] > b) {
662     return MP_GT;
663   } else if (a->dp[0] < b) {
664     return MP_LT;
665   } else {
666     return MP_EQ;
667   }
668 }
669 
670 
671 /* hac 14.61, pp608 */
mp_invmod(mp_int * a,mp_int * b,mp_int * c)672 static int mp_invmod (mp_int * a, mp_int * b, mp_int * c)
673 {
674   /* b cannot be negative */
675   if (b->sign == MP_NEG || mp_iszero(b) == 1) {
676     return MP_VAL;
677   }
678 
679 #ifdef BN_FAST_MP_INVMOD_C
680   /* if the modulus is odd we can use a faster routine instead */
681   if (mp_isodd (b) == 1) {
682     return fast_mp_invmod (a, b, c);
683   }
684 #endif
685 
686 #ifdef BN_MP_INVMOD_SLOW_C
687   return mp_invmod_slow(a, b, c);
688 #endif
689 
690 #ifndef BN_FAST_MP_INVMOD_C
691 #ifndef BN_MP_INVMOD_SLOW_C
692 #error mp_invmod would always fail
693 #endif
694 #endif
695   return MP_VAL;
696 }
697 
698 
699 /* get the size for an unsigned equivalent */
mp_unsigned_bin_size(mp_int * a)700 static int mp_unsigned_bin_size (mp_int * a)
701 {
702   int     size = mp_count_bits (a);
703   return (size / 8 + ((size & 7) != 0 ? 1 : 0));
704 }
705 
706 
707 /* hac 14.61, pp608 */
mp_invmod_slow(mp_int * a,mp_int * b,mp_int * c)708 static int mp_invmod_slow (mp_int * a, mp_int * b, mp_int * c)
709 {
710   mp_int  x, y, u, v, A, B, C, D;
711   int     res;
712 
713   /* b cannot be negative */
714   if (b->sign == MP_NEG || mp_iszero(b) == 1) {
715     return MP_VAL;
716   }
717 
718   /* init temps */
719   if ((res = mp_init_multi(&x, &y, &u, &v,
720                            &A, &B, &C, &D, NULL)) != MP_OKAY) {
721      return res;
722   }
723 
724   /* x = a, y = b */
725   if ((res = mp_mod(a, b, &x)) != MP_OKAY) {
726       goto LBL_ERR;
727   }
728   if ((res = mp_copy (b, &y)) != MP_OKAY) {
729     goto LBL_ERR;
730   }
731 
732   /* 2. [modified] if x,y are both even then return an error! */
733   if (mp_iseven (&x) == 1 && mp_iseven (&y) == 1) {
734     res = MP_VAL;
735     goto LBL_ERR;
736   }
737 
738   /* 3. u=x, v=y, A=1, B=0, C=0,D=1 */
739   if ((res = mp_copy (&x, &u)) != MP_OKAY) {
740     goto LBL_ERR;
741   }
742   if ((res = mp_copy (&y, &v)) != MP_OKAY) {
743     goto LBL_ERR;
744   }
745   mp_set (&A, 1);
746   mp_set (&D, 1);
747 
748 top:
749   /* 4.  while u is even do */
750   while (mp_iseven (&u) == 1) {
751     /* 4.1 u = u/2 */
752     if ((res = mp_div_2 (&u, &u)) != MP_OKAY) {
753       goto LBL_ERR;
754     }
755     /* 4.2 if A or B is odd then */
756     if (mp_isodd (&A) == 1 || mp_isodd (&B) == 1) {
757       /* A = (A+y)/2, B = (B-x)/2 */
758       if ((res = mp_add (&A, &y, &A)) != MP_OKAY) {
759          goto LBL_ERR;
760       }
761       if ((res = mp_sub (&B, &x, &B)) != MP_OKAY) {
762          goto LBL_ERR;
763       }
764     }
765     /* A = A/2, B = B/2 */
766     if ((res = mp_div_2 (&A, &A)) != MP_OKAY) {
767       goto LBL_ERR;
768     }
769     if ((res = mp_div_2 (&B, &B)) != MP_OKAY) {
770       goto LBL_ERR;
771     }
772   }
773 
774   /* 5.  while v is even do */
775   while (mp_iseven (&v) == 1) {
776     /* 5.1 v = v/2 */
777     if ((res = mp_div_2 (&v, &v)) != MP_OKAY) {
778       goto LBL_ERR;
779     }
780     /* 5.2 if C or D is odd then */
781     if (mp_isodd (&C) == 1 || mp_isodd (&D) == 1) {
782       /* C = (C+y)/2, D = (D-x)/2 */
783       if ((res = mp_add (&C, &y, &C)) != MP_OKAY) {
784          goto LBL_ERR;
785       }
786       if ((res = mp_sub (&D, &x, &D)) != MP_OKAY) {
787          goto LBL_ERR;
788       }
789     }
790     /* C = C/2, D = D/2 */
791     if ((res = mp_div_2 (&C, &C)) != MP_OKAY) {
792       goto LBL_ERR;
793     }
794     if ((res = mp_div_2 (&D, &D)) != MP_OKAY) {
795       goto LBL_ERR;
796     }
797   }
798 
799   /* 6.  if u >= v then */
800   if (mp_cmp (&u, &v) != MP_LT) {
801     /* u = u - v, A = A - C, B = B - D */
802     if ((res = mp_sub (&u, &v, &u)) != MP_OKAY) {
803       goto LBL_ERR;
804     }
805 
806     if ((res = mp_sub (&A, &C, &A)) != MP_OKAY) {
807       goto LBL_ERR;
808     }
809 
810     if ((res = mp_sub (&B, &D, &B)) != MP_OKAY) {
811       goto LBL_ERR;
812     }
813   } else {
814     /* v - v - u, C = C - A, D = D - B */
815     if ((res = mp_sub (&v, &u, &v)) != MP_OKAY) {
816       goto LBL_ERR;
817     }
818 
819     if ((res = mp_sub (&C, &A, &C)) != MP_OKAY) {
820       goto LBL_ERR;
821     }
822 
823     if ((res = mp_sub (&D, &B, &D)) != MP_OKAY) {
824       goto LBL_ERR;
825     }
826   }
827 
828   /* if not zero goto step 4 */
829   if (mp_iszero (&u) == 0)
830     goto top;
831 
832   /* now a = C, b = D, gcd == g*v */
833 
834   /* if v != 1 then there is no inverse */
835   if (mp_cmp_d (&v, 1) != MP_EQ) {
836     res = MP_VAL;
837     goto LBL_ERR;
838   }
839 
840   /* if its too low */
841   while (mp_cmp_d(&C, 0) == MP_LT) {
842       if ((res = mp_add(&C, b, &C)) != MP_OKAY) {
843          goto LBL_ERR;
844       }
845   }
846 
847   /* too big */
848   while (mp_cmp_mag(&C, b) != MP_LT) {
849       if ((res = mp_sub(&C, b, &C)) != MP_OKAY) {
850          goto LBL_ERR;
851       }
852   }
853 
854   /* C is now the inverse */
855   mp_exch (&C, c);
856   res = MP_OKAY;
857 LBL_ERR:mp_clear_multi (&x, &y, &u, &v, &A, &B, &C, &D, NULL);
858   return res;
859 }
860 
861 
862 /* compare maginitude of two ints (unsigned) */
mp_cmp_mag(mp_int * a,mp_int * b)863 static int mp_cmp_mag (mp_int * a, mp_int * b)
864 {
865   int     n;
866   mp_digit *tmpa, *tmpb;
867 
868   /* compare based on # of non-zero digits */
869   if (a->used > b->used) {
870     return MP_GT;
871   }
872 
873   if (a->used < b->used) {
874     return MP_LT;
875   }
876 
877   /* alias for a */
878   tmpa = a->dp + (a->used - 1);
879 
880   /* alias for b */
881   tmpb = b->dp + (a->used - 1);
882 
883   /* compare based on digits  */
884   for (n = 0; n < a->used; ++n, --tmpa, --tmpb) {
885     if (*tmpa > *tmpb) {
886       return MP_GT;
887     }
888 
889     if (*tmpa < *tmpb) {
890       return MP_LT;
891     }
892   }
893   return MP_EQ;
894 }
895 
896 
897 /* 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)898 static int mp_read_unsigned_bin (mp_int * a, const unsigned char *b, int c)
899 {
900   int     res;
901 
902   /* make sure there are at least two digits */
903   if (a->alloc < 2) {
904      if ((res = mp_grow(a, 2)) != MP_OKAY) {
905         return res;
906      }
907   }
908 
909   /* zero the int */
910   mp_zero (a);
911 
912   /* read the bytes in */
913   while (c-- > 0) {
914     if ((res = mp_mul_2d (a, 8, a)) != MP_OKAY) {
915       return res;
916     }
917 
918 #ifndef MP_8BIT
919       a->dp[0] |= *b++;
920       a->used += 1;
921 #else
922       a->dp[0] = (*b & MP_MASK);
923       a->dp[1] |= ((*b++ >> 7U) & 1);
924       a->used += 2;
925 #endif
926   }
927   mp_clamp (a);
928   return MP_OKAY;
929 }
930 
931 
932 /* store in unsigned [big endian] format */
mp_to_unsigned_bin(mp_int * a,unsigned char * b)933 static int mp_to_unsigned_bin (mp_int * a, unsigned char *b)
934 {
935   int     x, res;
936   mp_int  t;
937 
938   if ((res = mp_init_copy (&t, a)) != MP_OKAY) {
939     return res;
940   }
941 
942   x = 0;
943   while (mp_iszero (&t) == 0) {
944 #ifndef MP_8BIT
945       b[x++] = (unsigned char) (t.dp[0] & 255);
946 #else
947       b[x++] = (unsigned char) (t.dp[0] | ((t.dp[1] & 0x01) << 7));
948 #endif
949     if ((res = mp_div_2d (&t, 8, &t, NULL)) != MP_OKAY) {
950       mp_clear (&t);
951       return res;
952     }
953   }
954   bn_reverse (b, x);
955   mp_clear (&t);
956   return MP_OKAY;
957 }
958 
959 
960 /* 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)961 static int mp_div_2d (mp_int * a, int b, mp_int * c, mp_int * d)
962 {
963   mp_digit D, r, rr;
964   int     x, res;
965   mp_int  t;
966 
967 
968   /* if the shift count is <= 0 then we do no work */
969   if (b <= 0) {
970     res = mp_copy (a, c);
971     if (d != NULL) {
972       mp_zero (d);
973     }
974     return res;
975   }
976 
977   if ((res = mp_init (&t)) != MP_OKAY) {
978     return res;
979   }
980 
981   /* get the remainder */
982   if (d != NULL) {
983     if ((res = mp_mod_2d (a, b, &t)) != MP_OKAY) {
984       mp_clear (&t);
985       return res;
986     }
987   }
988 
989   /* copy */
990   if ((res = mp_copy (a, c)) != MP_OKAY) {
991     mp_clear (&t);
992     return res;
993   }
994 
995   /* shift by as many digits in the bit count */
996   if (b >= (int)DIGIT_BIT) {
997     mp_rshd (c, b / DIGIT_BIT);
998   }
999 
1000   /* shift any bit count < DIGIT_BIT */
1001   D = (mp_digit) (b % DIGIT_BIT);
1002   if (D != 0) {
1003     register mp_digit *tmpc, mask, shift;
1004 
1005     /* mask */
1006     mask = (((mp_digit)1) << D) - 1;
1007 
1008     /* shift for lsb */
1009     shift = DIGIT_BIT - D;
1010 
1011     /* alias */
1012     tmpc = c->dp + (c->used - 1);
1013 
1014     /* carry */
1015     r = 0;
1016     for (x = c->used - 1; x >= 0; x--) {
1017       /* get the lower  bits of this word in a temp */
1018       rr = *tmpc & mask;
1019 
1020       /* shift the current word and mix in the carry bits from the previous word */
1021       *tmpc = (*tmpc >> D) | (r << shift);
1022       --tmpc;
1023 
1024       /* set the carry to the carry bits of the current word found above */
1025       r = rr;
1026     }
1027   }
1028   mp_clamp (c);
1029   if (d != NULL) {
1030     mp_exch (&t, d);
1031   }
1032   mp_clear (&t);
1033   return MP_OKAY;
1034 }
1035 
1036 
mp_init_copy(mp_int * a,mp_int * b)1037 static int mp_init_copy (mp_int * a, mp_int * b)
1038 {
1039   int     res;
1040 
1041   if ((res = mp_init (a)) != MP_OKAY) {
1042     return res;
1043   }
1044   return mp_copy (b, a);
1045 }
1046 
1047 
1048 /* set to zero */
mp_zero(mp_int * a)1049 static void mp_zero (mp_int * a)
1050 {
1051   int       n;
1052   mp_digit *tmp;
1053 
1054   a->sign = MP_ZPOS;
1055   a->used = 0;
1056 
1057   tmp = a->dp;
1058   for (n = 0; n < a->alloc; n++) {
1059      *tmp++ = 0;
1060   }
1061 }
1062 
1063 
1064 /* copy, b = a */
mp_copy(mp_int * a,mp_int * b)1065 static int mp_copy (mp_int * a, mp_int * b)
1066 {
1067   int     res, n;
1068 
1069   /* if dst == src do nothing */
1070   if (a == b) {
1071     return MP_OKAY;
1072   }
1073 
1074   /* grow dest */
1075   if (b->alloc < a->used) {
1076      if ((res = mp_grow (b, a->used)) != MP_OKAY) {
1077         return res;
1078      }
1079   }
1080 
1081   /* zero b and copy the parameters over */
1082   {
1083     register mp_digit *tmpa, *tmpb;
1084 
1085     /* pointer aliases */
1086 
1087     /* source */
1088     tmpa = a->dp;
1089 
1090     /* destination */
1091     tmpb = b->dp;
1092 
1093     /* copy all the digits */
1094     for (n = 0; n < a->used; n++) {
1095       *tmpb++ = *tmpa++;
1096     }
1097 
1098     /* clear high digits */
1099     for (; n < b->used; n++) {
1100       *tmpb++ = 0;
1101     }
1102   }
1103 
1104   /* copy used count and sign */
1105   b->used = a->used;
1106   b->sign = a->sign;
1107   return MP_OKAY;
1108 }
1109 
1110 
1111 /* shift right a certain amount of digits */
mp_rshd(mp_int * a,int b)1112 static void mp_rshd (mp_int * a, int b)
1113 {
1114   int     x;
1115 
1116   /* if b <= 0 then ignore it */
1117   if (b <= 0) {
1118     return;
1119   }
1120 
1121   /* if b > used then simply zero it and return */
1122   if (a->used <= b) {
1123     mp_zero (a);
1124     return;
1125   }
1126 
1127   {
1128     register mp_digit *bottom, *top;
1129 
1130     /* shift the digits down */
1131 
1132     /* bottom */
1133     bottom = a->dp;
1134 
1135     /* top [offset into digits] */
1136     top = a->dp + b;
1137 
1138     /* this is implemented as a sliding window where
1139      * the window is b-digits long and digits from
1140      * the top of the window are copied to the bottom
1141      *
1142      * e.g.
1143 
1144      b-2 | b-1 | b0 | b1 | b2 | ... | bb |   ---->
1145                  /\                   |      ---->
1146                   \-------------------/      ---->
1147      */
1148     for (x = 0; x < (a->used - b); x++) {
1149       *bottom++ = *top++;
1150     }
1151 
1152     /* zero the top digits */
1153     for (; x < a->used; x++) {
1154       *bottom++ = 0;
1155     }
1156   }
1157 
1158   /* remove excess digits */
1159   a->used -= b;
1160 }
1161 
1162 
1163 /* swap the elements of two integers, for cases where you can't simply swap the
1164  * mp_int pointers around
1165  */
mp_exch(mp_int * a,mp_int * b)1166 static void mp_exch (mp_int * a, mp_int * b)
1167 {
1168   mp_int  t;
1169 
1170   t  = *a;
1171   *a = *b;
1172   *b = t;
1173 }
1174 
1175 
1176 /* trim unused digits
1177  *
1178  * This is used to ensure that leading zero digits are
1179  * trimed and the leading "used" digit will be non-zero
1180  * Typically very fast.  Also fixes the sign if there
1181  * are no more leading digits
1182  */
mp_clamp(mp_int * a)1183 static void mp_clamp (mp_int * a)
1184 {
1185   /* decrease used while the most significant digit is
1186    * zero.
1187    */
1188   while (a->used > 0 && a->dp[a->used - 1] == 0) {
1189     --(a->used);
1190   }
1191 
1192   /* reset the sign flag if used == 0 */
1193   if (a->used == 0) {
1194     a->sign = MP_ZPOS;
1195   }
1196 }
1197 
1198 
1199 /* grow as required */
mp_grow(mp_int * a,int size)1200 static int mp_grow (mp_int * a, int size)
1201 {
1202   int     i;
1203   mp_digit *tmp;
1204 
1205   /* if the alloc size is smaller alloc more ram */
1206   if (a->alloc < size) {
1207     /* ensure there are always at least MP_PREC digits extra on top */
1208     size += (MP_PREC * 2) - (size % MP_PREC);
1209 
1210     /* reallocate the array a->dp
1211      *
1212      * We store the return in a temporary variable
1213      * in case the operation failed we don't want
1214      * to overwrite the dp member of a.
1215      */
1216     tmp = OPT_CAST(mp_digit) XREALLOC (a->dp, sizeof (mp_digit) * size);
1217     if (tmp == NULL) {
1218       /* reallocation failed but "a" is still valid [can be freed] */
1219       return MP_MEM;
1220     }
1221 
1222     /* reallocation succeeded so set a->dp */
1223     a->dp = tmp;
1224 
1225     /* zero excess digits */
1226     i        = a->alloc;
1227     a->alloc = size;
1228     for (; i < a->alloc; i++) {
1229       a->dp[i] = 0;
1230     }
1231   }
1232   return MP_OKAY;
1233 }
1234 
1235 
1236 /* b = |a|
1237  *
1238  * Simple function copies the input and fixes the sign to positive
1239  */
mp_abs(mp_int * a,mp_int * b)1240 static int mp_abs (mp_int * a, mp_int * b)
1241 {
1242   int     res;
1243 
1244   /* copy a to b */
1245   if (a != b) {
1246      if ((res = mp_copy (a, b)) != MP_OKAY) {
1247        return res;
1248      }
1249   }
1250 
1251   /* force the sign of b to positive */
1252   b->sign = MP_ZPOS;
1253 
1254   return MP_OKAY;
1255 }
1256 
1257 
1258 /* set to a digit */
mp_set(mp_int * a,mp_digit b)1259 static void mp_set (mp_int * a, mp_digit b)
1260 {
1261   mp_zero (a);
1262   a->dp[0] = b & MP_MASK;
1263   a->used  = (a->dp[0] != 0) ? 1 : 0;
1264 }
1265 
1266 
1267 /* b = a/2 */
mp_div_2(mp_int * a,mp_int * b)1268 static int mp_div_2(mp_int * a, mp_int * b)
1269 {
1270   int     x, res, oldused;
1271 
1272   /* copy */
1273   if (b->alloc < a->used) {
1274     if ((res = mp_grow (b, a->used)) != MP_OKAY) {
1275       return res;
1276     }
1277   }
1278 
1279   oldused = b->used;
1280   b->used = a->used;
1281   {
1282     register mp_digit r, rr, *tmpa, *tmpb;
1283 
1284     /* source alias */
1285     tmpa = a->dp + b->used - 1;
1286 
1287     /* dest alias */
1288     tmpb = b->dp + b->used - 1;
1289 
1290     /* carry */
1291     r = 0;
1292     for (x = b->used - 1; x >= 0; x--) {
1293       /* get the carry for the next iteration */
1294       rr = *tmpa & 1;
1295 
1296       /* shift the current digit, add in carry and store */
1297       *tmpb-- = (*tmpa-- >> 1) | (r << (DIGIT_BIT - 1));
1298 
1299       /* forward carry to next iteration */
1300       r = rr;
1301     }
1302 
1303     /* zero excess digits */
1304     tmpb = b->dp + b->used;
1305     for (x = b->used; x < oldused; x++) {
1306       *tmpb++ = 0;
1307     }
1308   }
1309   b->sign = a->sign;
1310   mp_clamp (b);
1311   return MP_OKAY;
1312 }
1313 
1314 
1315 /* shift left by a certain bit count */
mp_mul_2d(mp_int * a,int b,mp_int * c)1316 static int mp_mul_2d (mp_int * a, int b, mp_int * c)
1317 {
1318   mp_digit d;
1319   int      res;
1320 
1321   /* copy */
1322   if (a != c) {
1323      if ((res = mp_copy (a, c)) != MP_OKAY) {
1324        return res;
1325      }
1326   }
1327 
1328   if (c->alloc < (int)(c->used + b/DIGIT_BIT + 1)) {
1329      if ((res = mp_grow (c, c->used + b / DIGIT_BIT + 1)) != MP_OKAY) {
1330        return res;
1331      }
1332   }
1333 
1334   /* shift by as many digits in the bit count */
1335   if (b >= (int)DIGIT_BIT) {
1336     if ((res = mp_lshd (c, b / DIGIT_BIT)) != MP_OKAY) {
1337       return res;
1338     }
1339   }
1340 
1341   /* shift any bit count < DIGIT_BIT */
1342   d = (mp_digit) (b % DIGIT_BIT);
1343   if (d != 0) {
1344     register mp_digit *tmpc, shift, mask, r, rr;
1345     register int x;
1346 
1347     /* bitmask for carries */
1348     mask = (((mp_digit)1) << d) - 1;
1349 
1350     /* shift for msbs */
1351     shift = DIGIT_BIT - d;
1352 
1353     /* alias */
1354     tmpc = c->dp;
1355 
1356     /* carry */
1357     r    = 0;
1358     for (x = 0; x < c->used; x++) {
1359       /* get the higher bits of the current word */
1360       rr = (*tmpc >> shift) & mask;
1361 
1362       /* shift the current word and OR in the carry */
1363       *tmpc = ((*tmpc << d) | r) & MP_MASK;
1364       ++tmpc;
1365 
1366       /* set the carry to the carry bits of the current word */
1367       r = rr;
1368     }
1369 
1370     /* set final carry */
1371     if (r != 0) {
1372        c->dp[(c->used)++] = r;
1373     }
1374   }
1375   mp_clamp (c);
1376   return MP_OKAY;
1377 }
1378 
1379 
mp_init_multi(mp_int * mp,...)1380 static int mp_init_multi(mp_int *mp, ...)
1381 {
1382     mp_err res = MP_OKAY;      /* Assume ok until proven otherwise */
1383     int n = 0;                 /* Number of ok inits */
1384     mp_int* cur_arg = mp;
1385     va_list args;
1386 
1387     va_start(args, mp);        /* init args to next argument from caller */
1388     while (cur_arg != NULL) {
1389         if (mp_init(cur_arg) != MP_OKAY) {
1390             /* Oops - error! Back-track and mp_clear what we already
1391                succeeded in init-ing, then return error.
1392             */
1393             va_list clean_args;
1394 
1395             /* end the current list */
1396             va_end(args);
1397 
1398             /* now start cleaning up */
1399             cur_arg = mp;
1400             va_start(clean_args, mp);
1401             while (n--) {
1402                 mp_clear(cur_arg);
1403                 cur_arg = va_arg(clean_args, mp_int*);
1404             }
1405             va_end(clean_args);
1406             res = MP_MEM;
1407             break;
1408         }
1409         n++;
1410         cur_arg = va_arg(args, mp_int*);
1411     }
1412     va_end(args);
1413     return res;                /* Assumed ok, if error flagged above. */
1414 }
1415 
1416 
mp_clear_multi(mp_int * mp,...)1417 static void mp_clear_multi(mp_int *mp, ...)
1418 {
1419     mp_int* next_mp = mp;
1420     va_list args;
1421     va_start(args, mp);
1422     while (next_mp != NULL) {
1423         mp_clear(next_mp);
1424         next_mp = va_arg(args, mp_int*);
1425     }
1426     va_end(args);
1427 }
1428 
1429 
1430 /* shift left a certain amount of digits */
mp_lshd(mp_int * a,int b)1431 static int mp_lshd (mp_int * a, int b)
1432 {
1433   int     x, res;
1434 
1435   /* if its less than zero return */
1436   if (b <= 0) {
1437     return MP_OKAY;
1438   }
1439 
1440   /* grow to fit the new digits */
1441   if (a->alloc < a->used + b) {
1442      if ((res = mp_grow (a, a->used + b)) != MP_OKAY) {
1443        return res;
1444      }
1445   }
1446 
1447   {
1448     register mp_digit *top, *bottom;
1449 
1450     /* increment the used by the shift amount then copy upwards */
1451     a->used += b;
1452 
1453     /* top */
1454     top = a->dp + a->used - 1;
1455 
1456     /* base */
1457     bottom = a->dp + a->used - 1 - b;
1458 
1459     /* much like mp_rshd this is implemented using a sliding window
1460      * except the window goes the otherway around.  Copying from
1461      * the bottom to the top.  see bn_mp_rshd.c for more info.
1462      */
1463     for (x = a->used - 1; x >= b; x--) {
1464       *top-- = *bottom--;
1465     }
1466 
1467     /* zero the lower digits */
1468     top = a->dp;
1469     for (x = 0; x < b; x++) {
1470       *top++ = 0;
1471     }
1472   }
1473   return MP_OKAY;
1474 }
1475 
1476 
1477 /* returns the number of bits in an int */
mp_count_bits(mp_int * a)1478 static int mp_count_bits (mp_int * a)
1479 {
1480   int     r;
1481   mp_digit q;
1482 
1483   /* shortcut */
1484   if (a->used == 0) {
1485     return 0;
1486   }
1487 
1488   /* get number of digits and add that */
1489   r = (a->used - 1) * DIGIT_BIT;
1490 
1491   /* take the last digit and count the bits in it */
1492   q = a->dp[a->used - 1];
1493   while (q > ((mp_digit) 0)) {
1494     ++r;
1495     q >>= ((mp_digit) 1);
1496   }
1497   return r;
1498 }
1499 
1500 
1501 /* calc a value mod 2**b */
mp_mod_2d(mp_int * a,int b,mp_int * c)1502 static int mp_mod_2d (mp_int * a, int b, mp_int * c)
1503 {
1504   int     x, res;
1505 
1506   /* if b is <= 0 then zero the int */
1507   if (b <= 0) {
1508     mp_zero (c);
1509     return MP_OKAY;
1510   }
1511 
1512   /* if the modulus is larger than the value than return */
1513   if (b >= (int) (a->used * DIGIT_BIT)) {
1514     res = mp_copy (a, c);
1515     return res;
1516   }
1517 
1518   /* copy */
1519   if ((res = mp_copy (a, c)) != MP_OKAY) {
1520     return res;
1521   }
1522 
1523   /* zero digits above the last digit of the modulus */
1524   for (x = (b / DIGIT_BIT) + ((b % DIGIT_BIT) == 0 ? 0 : 1); x < c->used; x++) {
1525     c->dp[x] = 0;
1526   }
1527   /* clear the digit that is not completely outside/inside the modulus */
1528   c->dp[b / DIGIT_BIT] &=
1529     (mp_digit) ((((mp_digit) 1) << (((mp_digit) b) % DIGIT_BIT)) - ((mp_digit) 1));
1530   mp_clamp (c);
1531   return MP_OKAY;
1532 }
1533 
1534 
1535 /* slower bit-bang division... also smaller */
mp_div(mp_int * a,mp_int * b,mp_int * c,mp_int * d)1536 static int mp_div(mp_int * a, mp_int * b, mp_int * c, mp_int * d)
1537 {
1538    mp_int ta, tb, tq, q;
1539    int    res, n, n2;
1540 
1541   /* is divisor zero ? */
1542   if (mp_iszero (b) == 1) {
1543     return MP_VAL;
1544   }
1545 
1546   /* if a < b then q=0, r = a */
1547   if (mp_cmp_mag (a, b) == MP_LT) {
1548     if (d != NULL) {
1549       res = mp_copy (a, d);
1550     } else {
1551       res = MP_OKAY;
1552     }
1553     if (c != NULL) {
1554       mp_zero (c);
1555     }
1556     return res;
1557   }
1558 
1559   /* init our temps */
1560   if ((res = mp_init_multi(&ta, &tb, &tq, &q, NULL) != MP_OKAY)) {
1561      return res;
1562   }
1563 
1564 
1565   mp_set(&tq, 1);
1566   n = mp_count_bits(a) - mp_count_bits(b);
1567   if (((res = mp_abs(a, &ta)) != MP_OKAY) ||
1568       ((res = mp_abs(b, &tb)) != MP_OKAY) ||
1569       ((res = mp_mul_2d(&tb, n, &tb)) != MP_OKAY) ||
1570       ((res = mp_mul_2d(&tq, n, &tq)) != MP_OKAY)) {
1571       goto LBL_ERR;
1572   }
1573 
1574   while (n-- >= 0) {
1575      if (mp_cmp(&tb, &ta) != MP_GT) {
1576         if (((res = mp_sub(&ta, &tb, &ta)) != MP_OKAY) ||
1577             ((res = mp_add(&q, &tq, &q)) != MP_OKAY)) {
1578            goto LBL_ERR;
1579         }
1580      }
1581      if (((res = mp_div_2d(&tb, 1, &tb, NULL)) != MP_OKAY) ||
1582          ((res = mp_div_2d(&tq, 1, &tq, NULL)) != MP_OKAY)) {
1583            goto LBL_ERR;
1584      }
1585   }
1586 
1587   /* now q == quotient and ta == remainder */
1588   n  = a->sign;
1589   n2 = (a->sign == b->sign ? MP_ZPOS : MP_NEG);
1590   if (c != NULL) {
1591      mp_exch(c, &q);
1592      c->sign  = (mp_iszero(c) == MP_YES) ? MP_ZPOS : n2;
1593   }
1594   if (d != NULL) {
1595      mp_exch(d, &ta);
1596      d->sign = (mp_iszero(d) == MP_YES) ? MP_ZPOS : n;
1597   }
1598 LBL_ERR:
1599    mp_clear_multi(&ta, &tb, &tq, &q, NULL);
1600    return res;
1601 }
1602 
1603 
1604 #ifdef MP_LOW_MEM
1605    #define TAB_SIZE 32
1606 #else
1607    #define TAB_SIZE 256
1608 #endif
1609 
s_mp_exptmod(mp_int * G,mp_int * X,mp_int * P,mp_int * Y,int redmode)1610 static int s_mp_exptmod (mp_int * G, mp_int * X, mp_int * P, mp_int * Y, int redmode)
1611 {
1612   mp_int  M[TAB_SIZE], res, mu;
1613   mp_digit buf;
1614   int     err, bitbuf, bitcpy, bitcnt, mode, digidx, x, y, winsize;
1615   int (*redux)(mp_int*,mp_int*,mp_int*);
1616 
1617   /* find window size */
1618   x = mp_count_bits (X);
1619   if (x <= 7) {
1620     winsize = 2;
1621   } else if (x <= 36) {
1622     winsize = 3;
1623   } else if (x <= 140) {
1624     winsize = 4;
1625   } else if (x <= 450) {
1626     winsize = 5;
1627   } else if (x <= 1303) {
1628     winsize = 6;
1629   } else if (x <= 3529) {
1630     winsize = 7;
1631   } else {
1632     winsize = 8;
1633   }
1634 
1635 #ifdef MP_LOW_MEM
1636     if (winsize > 5) {
1637        winsize = 5;
1638     }
1639 #endif
1640 
1641   /* init M array */
1642   /* init first cell */
1643   if ((err = mp_init(&M[1])) != MP_OKAY) {
1644      return err;
1645   }
1646 
1647   /* now init the second half of the array */
1648   for (x = 1<<(winsize-1); x < (1 << winsize); x++) {
1649     if ((err = mp_init(&M[x])) != MP_OKAY) {
1650       for (y = 1<<(winsize-1); y < x; y++) {
1651         mp_clear (&M[y]);
1652       }
1653       mp_clear(&M[1]);
1654       return err;
1655     }
1656   }
1657 
1658   /* create mu, used for Barrett reduction */
1659   if ((err = mp_init (&mu)) != MP_OKAY) {
1660     goto LBL_M;
1661   }
1662 
1663   if (redmode == 0) {
1664      if ((err = mp_reduce_setup (&mu, P)) != MP_OKAY) {
1665         goto LBL_MU;
1666      }
1667      redux = mp_reduce;
1668   } else {
1669      if ((err = mp_reduce_2k_setup_l (P, &mu)) != MP_OKAY) {
1670         goto LBL_MU;
1671      }
1672      redux = mp_reduce_2k_l;
1673   }
1674 
1675   /* create M table
1676    *
1677    * The M table contains powers of the base,
1678    * e.g. M[x] = G**x mod P
1679    *
1680    * The first half of the table is not
1681    * computed though accept for M[0] and M[1]
1682    */
1683   if ((err = mp_mod (G, P, &M[1])) != MP_OKAY) {
1684     goto LBL_MU;
1685   }
1686 
1687   /* compute the value at M[1<<(winsize-1)] by squaring
1688    * M[1] (winsize-1) times
1689    */
1690   if ((err = mp_copy (&M[1], &M[1 << (winsize - 1)])) != MP_OKAY) {
1691     goto LBL_MU;
1692   }
1693 
1694   for (x = 0; x < (winsize - 1); x++) {
1695     /* square it */
1696     if ((err = mp_sqr (&M[1 << (winsize - 1)],
1697                        &M[1 << (winsize - 1)])) != MP_OKAY) {
1698       goto LBL_MU;
1699     }
1700 
1701     /* reduce modulo P */
1702     if ((err = redux (&M[1 << (winsize - 1)], P, &mu)) != MP_OKAY) {
1703       goto LBL_MU;
1704     }
1705   }
1706 
1707   /* create upper table, that is M[x] = M[x-1] * M[1] (mod P)
1708    * for x = (2**(winsize - 1) + 1) to (2**winsize - 1)
1709    */
1710   for (x = (1 << (winsize - 1)) + 1; x < (1 << winsize); x++) {
1711     if ((err = mp_mul (&M[x - 1], &M[1], &M[x])) != MP_OKAY) {
1712       goto LBL_MU;
1713     }
1714     if ((err = redux (&M[x], P, &mu)) != MP_OKAY) {
1715       goto LBL_MU;
1716     }
1717   }
1718 
1719   /* setup result */
1720   if ((err = mp_init (&res)) != MP_OKAY) {
1721     goto LBL_MU;
1722   }
1723   mp_set (&res, 1);
1724 
1725   /* set initial mode and bit cnt */
1726   mode   = 0;
1727   bitcnt = 1;
1728   buf    = 0;
1729   digidx = X->used - 1;
1730   bitcpy = 0;
1731   bitbuf = 0;
1732 
1733   for (;;) {
1734     /* grab next digit as required */
1735     if (--bitcnt == 0) {
1736       /* if digidx == -1 we are out of digits */
1737       if (digidx == -1) {
1738         break;
1739       }
1740       /* read next digit and reset the bitcnt */
1741       buf    = X->dp[digidx--];
1742       bitcnt = (int) DIGIT_BIT;
1743     }
1744 
1745     /* grab the next msb from the exponent */
1746     y     = (buf >> (mp_digit)(DIGIT_BIT - 1)) & 1;
1747     buf <<= (mp_digit)1;
1748 
1749     /* if the bit is zero and mode == 0 then we ignore it
1750      * These represent the leading zero bits before the first 1 bit
1751      * in the exponent.  Technically this opt is not required but it
1752      * does lower the # of trivial squaring/reductions used
1753      */
1754     if (mode == 0 && y == 0) {
1755       continue;
1756     }
1757 
1758     /* if the bit is zero and mode == 1 then we square */
1759     if (mode == 1 && y == 0) {
1760       if ((err = mp_sqr (&res, &res)) != MP_OKAY) {
1761         goto LBL_RES;
1762       }
1763       if ((err = redux (&res, P, &mu)) != MP_OKAY) {
1764         goto LBL_RES;
1765       }
1766       continue;
1767     }
1768 
1769     /* else we add it to the window */
1770     bitbuf |= (y << (winsize - ++bitcpy));
1771     mode    = 2;
1772 
1773     if (bitcpy == winsize) {
1774       /* ok window is filled so square as required and multiply  */
1775       /* square first */
1776       for (x = 0; x < winsize; x++) {
1777         if ((err = mp_sqr (&res, &res)) != MP_OKAY) {
1778           goto LBL_RES;
1779         }
1780         if ((err = redux (&res, P, &mu)) != MP_OKAY) {
1781           goto LBL_RES;
1782         }
1783       }
1784 
1785       /* then multiply */
1786       if ((err = mp_mul (&res, &M[bitbuf], &res)) != MP_OKAY) {
1787         goto LBL_RES;
1788       }
1789       if ((err = redux (&res, P, &mu)) != MP_OKAY) {
1790         goto LBL_RES;
1791       }
1792 
1793       /* empty window and reset */
1794       bitcpy = 0;
1795       bitbuf = 0;
1796       mode   = 1;
1797     }
1798   }
1799 
1800   /* if bits remain then square/multiply */
1801   if (mode == 2 && bitcpy > 0) {
1802     /* square then multiply if the bit is set */
1803     for (x = 0; x < bitcpy; x++) {
1804       if ((err = mp_sqr (&res, &res)) != MP_OKAY) {
1805         goto LBL_RES;
1806       }
1807       if ((err = redux (&res, P, &mu)) != MP_OKAY) {
1808         goto LBL_RES;
1809       }
1810 
1811       bitbuf <<= 1;
1812       if ((bitbuf & (1 << winsize)) != 0) {
1813         /* then multiply */
1814         if ((err = mp_mul (&res, &M[1], &res)) != MP_OKAY) {
1815           goto LBL_RES;
1816         }
1817         if ((err = redux (&res, P, &mu)) != MP_OKAY) {
1818           goto LBL_RES;
1819         }
1820       }
1821     }
1822   }
1823 
1824   mp_exch (&res, Y);
1825   err = MP_OKAY;
1826 LBL_RES:mp_clear (&res);
1827 LBL_MU:mp_clear (&mu);
1828 LBL_M:
1829   mp_clear(&M[1]);
1830   for (x = 1<<(winsize-1); x < (1 << winsize); x++) {
1831     mp_clear (&M[x]);
1832   }
1833   return err;
1834 }
1835 
1836 
1837 /* computes b = a*a */
mp_sqr(mp_int * a,mp_int * b)1838 static int mp_sqr (mp_int * a, mp_int * b)
1839 {
1840   int     res;
1841 
1842 #ifdef BN_MP_TOOM_SQR_C
1843   /* use Toom-Cook? */
1844   if (a->used >= TOOM_SQR_CUTOFF) {
1845     res = mp_toom_sqr(a, b);
1846   /* Karatsuba? */
1847   } else
1848 #endif
1849 #ifdef BN_MP_KARATSUBA_SQR_C
1850 if (a->used >= KARATSUBA_SQR_CUTOFF) {
1851     res = mp_karatsuba_sqr (a, b);
1852   } else
1853 #endif
1854   {
1855 #ifdef BN_FAST_S_MP_SQR_C
1856     /* can we use the fast comba multiplier? */
1857     if ((a->used * 2 + 1) < MP_WARRAY &&
1858          a->used <
1859          (1 << (sizeof(mp_word) * CHAR_BIT - 2*DIGIT_BIT - 1))) {
1860       res = fast_s_mp_sqr (a, b);
1861     } else
1862 #endif
1863 #ifdef BN_S_MP_SQR_C
1864       res = s_mp_sqr (a, b);
1865 #else
1866 #error mp_sqr could fail
1867       res = MP_VAL;
1868 #endif
1869   }
1870   b->sign = MP_ZPOS;
1871   return res;
1872 }
1873 
1874 
1875 /* reduces a modulo n where n is of the form 2**p - d
1876    This differs from reduce_2k since "d" can be larger
1877    than a single digit.
1878 */
mp_reduce_2k_l(mp_int * a,mp_int * n,mp_int * d)1879 static int mp_reduce_2k_l(mp_int *a, mp_int *n, mp_int *d)
1880 {
1881    mp_int q;
1882    int    p, res;
1883 
1884    if ((res = mp_init(&q)) != MP_OKAY) {
1885       return res;
1886    }
1887 
1888    p = mp_count_bits(n);
1889 top:
1890    /* q = a/2**p, a = a mod 2**p */
1891    if ((res = mp_div_2d(a, p, &q, a)) != MP_OKAY) {
1892       goto ERR;
1893    }
1894 
1895    /* q = q * d */
1896    if ((res = mp_mul(&q, d, &q)) != MP_OKAY) {
1897       goto ERR;
1898    }
1899 
1900    /* a = a + q */
1901    if ((res = s_mp_add(a, &q, a)) != MP_OKAY) {
1902       goto ERR;
1903    }
1904 
1905    if (mp_cmp_mag(a, n) != MP_LT) {
1906       s_mp_sub(a, n, a);
1907       goto top;
1908    }
1909 
1910 ERR:
1911    mp_clear(&q);
1912    return res;
1913 }
1914 
1915 
1916 /* determines the setup value */
mp_reduce_2k_setup_l(mp_int * a,mp_int * d)1917 static int mp_reduce_2k_setup_l(mp_int *a, mp_int *d)
1918 {
1919    int    res;
1920    mp_int tmp;
1921 
1922    if ((res = mp_init(&tmp)) != MP_OKAY) {
1923       return res;
1924    }
1925 
1926    if ((res = mp_2expt(&tmp, mp_count_bits(a))) != MP_OKAY) {
1927       goto ERR;
1928    }
1929 
1930    if ((res = s_mp_sub(&tmp, a, d)) != MP_OKAY) {
1931       goto ERR;
1932    }
1933 
1934 ERR:
1935    mp_clear(&tmp);
1936    return res;
1937 }
1938 
1939 
1940 /* computes a = 2**b
1941  *
1942  * Simple algorithm which zeroes the int, grows it then just sets one bit
1943  * as required.
1944  */
mp_2expt(mp_int * a,int b)1945 static int mp_2expt (mp_int * a, int b)
1946 {
1947   int     res;
1948 
1949   /* zero a as per default */
1950   mp_zero (a);
1951 
1952   /* grow a to accomodate the single bit */
1953   if ((res = mp_grow (a, b / DIGIT_BIT + 1)) != MP_OKAY) {
1954     return res;
1955   }
1956 
1957   /* set the used count of where the bit will go */
1958   a->used = b / DIGIT_BIT + 1;
1959 
1960   /* put the single bit in its place */
1961   a->dp[b / DIGIT_BIT] = ((mp_digit)1) << (b % DIGIT_BIT);
1962 
1963   return MP_OKAY;
1964 }
1965 
1966 
1967 /* pre-calculate the value required for Barrett reduction
1968  * For a given modulus "b" it calulates the value required in "a"
1969  */
mp_reduce_setup(mp_int * a,mp_int * b)1970 static int mp_reduce_setup (mp_int * a, mp_int * b)
1971 {
1972   int     res;
1973 
1974   if ((res = mp_2expt (a, b->used * 2 * DIGIT_BIT)) != MP_OKAY) {
1975     return res;
1976   }
1977   return mp_div (a, b, a, NULL);
1978 }
1979 
1980 
1981 /* reduces x mod m, assumes 0 < x < m**2, mu is
1982  * precomputed via mp_reduce_setup.
1983  * From HAC pp.604 Algorithm 14.42
1984  */
mp_reduce(mp_int * x,mp_int * m,mp_int * mu)1985 static int mp_reduce (mp_int * x, mp_int * m, mp_int * mu)
1986 {
1987   mp_int  q;
1988   int     res, um = m->used;
1989 
1990   /* q = x */
1991   if ((res = mp_init_copy (&q, x)) != MP_OKAY) {
1992     return res;
1993   }
1994 
1995   /* q1 = x / b**(k-1)  */
1996   mp_rshd (&q, um - 1);
1997 
1998   /* according to HAC this optimization is ok */
1999   if (((unsigned long) um) > (((mp_digit)1) << (DIGIT_BIT - 1))) {
2000     if ((res = mp_mul (&q, mu, &q)) != MP_OKAY) {
2001       goto CLEANUP;
2002     }
2003   } else {
2004 #ifdef BN_S_MP_MUL_HIGH_DIGS_C
2005     if ((res = s_mp_mul_high_digs (&q, mu, &q, um)) != MP_OKAY) {
2006       goto CLEANUP;
2007     }
2008 #elif defined(BN_FAST_S_MP_MUL_HIGH_DIGS_C)
2009     if ((res = fast_s_mp_mul_high_digs (&q, mu, &q, um)) != MP_OKAY) {
2010       goto CLEANUP;
2011     }
2012 #else
2013     {
2014 #error mp_reduce would always fail
2015       res = MP_VAL;
2016       goto CLEANUP;
2017     }
2018 #endif
2019   }
2020 
2021   /* q3 = q2 / b**(k+1) */
2022   mp_rshd (&q, um + 1);
2023 
2024   /* x = x mod b**(k+1), quick (no division) */
2025   if ((res = mp_mod_2d (x, DIGIT_BIT * (um + 1), x)) != MP_OKAY) {
2026     goto CLEANUP;
2027   }
2028 
2029   /* q = q * m mod b**(k+1), quick (no division) */
2030   if ((res = s_mp_mul_digs (&q, m, &q, um + 1)) != MP_OKAY) {
2031     goto CLEANUP;
2032   }
2033 
2034   /* x = x - q */
2035   if ((res = mp_sub (x, &q, x)) != MP_OKAY) {
2036     goto CLEANUP;
2037   }
2038 
2039   /* If x < 0, add b**(k+1) to it */
2040   if (mp_cmp_d (x, 0) == MP_LT) {
2041     mp_set (&q, 1);
2042     if ((res = mp_lshd (&q, um + 1)) != MP_OKAY) {
2043       goto CLEANUP;
2044     }
2045     if ((res = mp_add (x, &q, x)) != MP_OKAY) {
2046       goto CLEANUP;
2047     }
2048   }
2049 
2050   /* Back off if it's too big */
2051   while (mp_cmp (x, m) != MP_LT) {
2052     if ((res = s_mp_sub (x, m, x)) != MP_OKAY) {
2053       goto CLEANUP;
2054     }
2055   }
2056 
2057 CLEANUP:
2058   mp_clear (&q);
2059 
2060   return res;
2061 }
2062 
2063 
2064 /* multiplies |a| * |b| and only computes upto digs digits of result
2065  * HAC pp. 595, Algorithm 14.12  Modified so you can control how
2066  * many digits of output are created.
2067  */
s_mp_mul_digs(mp_int * a,mp_int * b,mp_int * c,int digs)2068 static int s_mp_mul_digs (mp_int * a, mp_int * b, mp_int * c, int digs)
2069 {
2070   mp_int  t;
2071   int     res, pa, pb, ix, iy;
2072   mp_digit u;
2073   mp_word r;
2074   mp_digit tmpx, *tmpt, *tmpy;
2075 
2076   /* can we use the fast multiplier? */
2077   if (((digs) < MP_WARRAY) &&
2078       MIN (a->used, b->used) <
2079           (1 << ((CHAR_BIT * sizeof (mp_word)) - (2 * DIGIT_BIT)))) {
2080     return fast_s_mp_mul_digs (a, b, c, digs);
2081   }
2082 
2083   if ((res = mp_init_size (&t, digs)) != MP_OKAY) {
2084     return res;
2085   }
2086   t.used = digs;
2087 
2088   /* compute the digits of the product directly */
2089   pa = a->used;
2090   for (ix = 0; ix < pa; ix++) {
2091     /* set the carry to zero */
2092     u = 0;
2093 
2094     /* limit ourselves to making digs digits of output */
2095     pb = MIN (b->used, digs - ix);
2096 
2097     /* setup some aliases */
2098     /* copy of the digit from a used within the nested loop */
2099     tmpx = a->dp[ix];
2100 
2101     /* an alias for the destination shifted ix places */
2102     tmpt = t.dp + ix;
2103 
2104     /* an alias for the digits of b */
2105     tmpy = b->dp;
2106 
2107     /* compute the columns of the output and propagate the carry */
2108     for (iy = 0; iy < pb; iy++) {
2109       /* compute the column as a mp_word */
2110       r       = ((mp_word)*tmpt) +
2111                 ((mp_word)tmpx) * ((mp_word)*tmpy++) +
2112                 ((mp_word) u);
2113 
2114       /* the new column is the lower part of the result */
2115       *tmpt++ = (mp_digit) (r & ((mp_word) MP_MASK));
2116 
2117       /* get the carry word from the result */
2118       u       = (mp_digit) (r >> ((mp_word) DIGIT_BIT));
2119     }
2120     /* set carry if it is placed below digs */
2121     if (ix + iy < digs) {
2122       *tmpt = u;
2123     }
2124   }
2125 
2126   mp_clamp (&t);
2127   mp_exch (&t, c);
2128 
2129   mp_clear (&t);
2130   return MP_OKAY;
2131 }
2132 
2133 
2134 /* Fast (comba) multiplier
2135  *
2136  * This is the fast column-array [comba] multiplier.  It is
2137  * designed to compute the columns of the product first
2138  * then handle the carries afterwards.  This has the effect
2139  * of making the nested loops that compute the columns very
2140  * simple and schedulable on super-scalar processors.
2141  *
2142  * This has been modified to produce a variable number of
2143  * digits of output so if say only a half-product is required
2144  * you don't have to compute the upper half (a feature
2145  * required for fast Barrett reduction).
2146  *
2147  * Based on Algorithm 14.12 on pp.595 of HAC.
2148  *
2149  */
fast_s_mp_mul_digs(mp_int * a,mp_int * b,mp_int * c,int digs)2150 static int fast_s_mp_mul_digs (mp_int * a, mp_int * b, mp_int * c, int digs)
2151 {
2152   int     olduse, res, pa, ix, iz;
2153   mp_digit W[MP_WARRAY];
2154   register mp_word  _W;
2155 
2156   /* grow the destination as required */
2157   if (c->alloc < digs) {
2158     if ((res = mp_grow (c, digs)) != MP_OKAY) {
2159       return res;
2160     }
2161   }
2162 
2163   /* number of output digits to produce */
2164   pa = MIN(digs, a->used + b->used);
2165 
2166   /* clear the carry */
2167   _W = 0;
2168   for (ix = 0; ix < pa; ix++) {
2169       int      tx, ty;
2170       int      iy;
2171       mp_digit *tmpx, *tmpy;
2172 
2173       /* get offsets into the two bignums */
2174       ty = MIN(b->used-1, ix);
2175       tx = ix - ty;
2176 
2177       /* setup temp aliases */
2178       tmpx = a->dp + tx;
2179       tmpy = b->dp + ty;
2180 
2181       /* this is the number of times the loop will iterrate, essentially
2182          while (tx++ < a->used && ty-- >= 0) { ... }
2183        */
2184       iy = MIN(a->used-tx, ty+1);
2185 
2186       /* execute loop */
2187       for (iz = 0; iz < iy; ++iz) {
2188          _W += ((mp_word)*tmpx++)*((mp_word)*tmpy--);
2189 
2190       }
2191 
2192       /* store term */
2193       W[ix] = ((mp_digit)_W) & MP_MASK;
2194 
2195       /* make next carry */
2196       _W = _W >> ((mp_word)DIGIT_BIT);
2197  }
2198 
2199   /* setup dest */
2200   olduse  = c->used;
2201   c->used = pa;
2202 
2203   {
2204     register mp_digit *tmpc;
2205     tmpc = c->dp;
2206     for (ix = 0; ix < pa+1; ix++) {
2207       /* now extract the previous digit [below the carry] */
2208       *tmpc++ = W[ix];
2209     }
2210 
2211     /* clear unused digits [that existed in the old copy of c] */
2212     for (; ix < olduse; ix++) {
2213       *tmpc++ = 0;
2214     }
2215   }
2216   mp_clamp (c);
2217   return MP_OKAY;
2218 }
2219 
2220 
2221 /* init an mp_init for a given size */
mp_init_size(mp_int * a,int size)2222 static int mp_init_size (mp_int * a, int size)
2223 {
2224   int x;
2225 
2226   /* pad size so there are always extra digits */
2227   size += (MP_PREC * 2) - (size % MP_PREC);
2228 
2229   /* alloc mem */
2230   a->dp = OPT_CAST(mp_digit) XMALLOC (sizeof (mp_digit) * size);
2231   if (a->dp == NULL) {
2232     return MP_MEM;
2233   }
2234 
2235   /* set the members */
2236   a->used  = 0;
2237   a->alloc = size;
2238   a->sign  = MP_ZPOS;
2239 
2240   /* zero the digits */
2241   for (x = 0; x < size; x++) {
2242       a->dp[x] = 0;
2243   }
2244 
2245   return MP_OKAY;
2246 }
2247 
2248 
2249 /* low level squaring, b = a*a, HAC pp.596-597, Algorithm 14.16 */
s_mp_sqr(mp_int * a,mp_int * b)2250 static int s_mp_sqr (mp_int * a, mp_int * b)
2251 {
2252   mp_int  t;
2253   int     res, ix, iy, pa;
2254   mp_word r;
2255   mp_digit u, tmpx, *tmpt;
2256 
2257   pa = a->used;
2258   if ((res = mp_init_size (&t, 2*pa + 1)) != MP_OKAY) {
2259     return res;
2260   }
2261 
2262   /* default used is maximum possible size */
2263   t.used = 2*pa + 1;
2264 
2265   for (ix = 0; ix < pa; ix++) {
2266     /* first calculate the digit at 2*ix */
2267     /* calculate double precision result */
2268     r = ((mp_word) t.dp[2*ix]) +
2269         ((mp_word)a->dp[ix])*((mp_word)a->dp[ix]);
2270 
2271     /* store lower part in result */
2272     t.dp[ix+ix] = (mp_digit) (r & ((mp_word) MP_MASK));
2273 
2274     /* get the carry */
2275     u           = (mp_digit)(r >> ((mp_word) DIGIT_BIT));
2276 
2277     /* left hand side of A[ix] * A[iy] */
2278     tmpx        = a->dp[ix];
2279 
2280     /* alias for where to store the results */
2281     tmpt        = t.dp + (2*ix + 1);
2282 
2283     for (iy = ix + 1; iy < pa; iy++) {
2284       /* first calculate the product */
2285       r       = ((mp_word)tmpx) * ((mp_word)a->dp[iy]);
2286 
2287       /* now calculate the double precision result, note we use
2288        * addition instead of *2 since it's easier to optimize
2289        */
2290       r       = ((mp_word) *tmpt) + r + r + ((mp_word) u);
2291 
2292       /* store lower part */
2293       *tmpt++ = (mp_digit) (r & ((mp_word) MP_MASK));
2294 
2295       /* get carry */
2296       u       = (mp_digit)(r >> ((mp_word) DIGIT_BIT));
2297     }
2298     /* propagate upwards */
2299     while (u != ((mp_digit) 0)) {
2300       r       = ((mp_word) *tmpt) + ((mp_word) u);
2301       *tmpt++ = (mp_digit) (r & ((mp_word) MP_MASK));
2302       u       = (mp_digit)(r >> ((mp_word) DIGIT_BIT));
2303     }
2304   }
2305 
2306   mp_clamp (&t);
2307   mp_exch (&t, b);
2308   mp_clear (&t);
2309   return MP_OKAY;
2310 }
2311 
2312 
2313 /* multiplies |a| * |b| and does not compute the lower digs digits
2314  * [meant to get the higher part of the product]
2315  */
s_mp_mul_high_digs(mp_int * a,mp_int * b,mp_int * c,int digs)2316 static int s_mp_mul_high_digs (mp_int * a, mp_int * b, mp_int * c, int digs)
2317 {
2318   mp_int  t;
2319   int     res, pa, pb, ix, iy;
2320   mp_digit u;
2321   mp_word r;
2322   mp_digit tmpx, *tmpt, *tmpy;
2323 
2324   /* can we use the fast multiplier? */
2325 #ifdef BN_FAST_S_MP_MUL_HIGH_DIGS_C
2326   if (((a->used + b->used + 1) < MP_WARRAY)
2327       && MIN (a->used, b->used) < (1 << ((CHAR_BIT * sizeof (mp_word)) - (2 * DIGIT_BIT)))) {
2328     return fast_s_mp_mul_high_digs (a, b, c, digs);
2329   }
2330 #endif
2331 
2332   if ((res = mp_init_size (&t, a->used + b->used + 1)) != MP_OKAY) {
2333     return res;
2334   }
2335   t.used = a->used + b->used + 1;
2336 
2337   pa = a->used;
2338   pb = b->used;
2339   for (ix = 0; ix < pa; ix++) {
2340     /* clear the carry */
2341     u = 0;
2342 
2343     /* left hand side of A[ix] * B[iy] */
2344     tmpx = a->dp[ix];
2345 
2346     /* alias to the address of where the digits will be stored */
2347     tmpt = &(t.dp[digs]);
2348 
2349     /* alias for where to read the right hand side from */
2350     tmpy = b->dp + (digs - ix);
2351 
2352     for (iy = digs - ix; iy < pb; iy++) {
2353       /* calculate the double precision result */
2354       r       = ((mp_word)*tmpt) +
2355                 ((mp_word)tmpx) * ((mp_word)*tmpy++) +
2356                 ((mp_word) u);
2357 
2358       /* get the lower part */
2359       *tmpt++ = (mp_digit) (r & ((mp_word) MP_MASK));
2360 
2361       /* carry the carry */
2362       u       = (mp_digit) (r >> ((mp_word) DIGIT_BIT));
2363     }
2364     *tmpt = u;
2365   }
2366   mp_clamp (&t);
2367   mp_exch (&t, c);
2368   mp_clear (&t);
2369   return MP_OKAY;
2370 }
2371