예제 #1
0
/* reads a unsigned char array, assumes the msb is stored first [big endian] */
int mp_read_unsigned_bin(mp_int *a, const unsigned char *b, int c)
{
   int     res;

   /* make sure there are at least two digits */
   if (a->alloc < 2) {
      if ((res = mp_grow(a, 2)) != MP_OKAY) {
         return res;
      }
   }

   /* zero the int */
   mp_zero(a);

   /* read the bytes in */
   while (c-- > 0) {
      if ((res = mp_mul_2d(a, 8, a)) != MP_OKAY) {
         return res;
      }

#ifndef MP_8BIT
      a->dp[0] |= *b++;
      a->used += 1;
#else
      a->dp[0] = (*b & MP_MASK);
      a->dp[1] |= ((*b++ >> 7) & 1u);
      a->used += 2;
#endif
   }
   mp_clamp(a);
   return MP_OKAY;
}
예제 #2
0
/* b = a*2 */
int mp_mul_2(mp_int * a, mp_int * b)
{
	int x, res, oldused;

	/* grow to accomodate result */
	if (b->alloc < a->used + 1) {
		if ((res = mp_grow(b, a->used + 1)) != MP_OKAY) {
			return res;
		}
	}

	oldused = b->used;
	b->used = a->used;

	{
		register mp_digit r, rr, *tmpa, *tmpb;

		/* alias for source */
		tmpa = a->dp;

		/* alias for dest */
		tmpb = b->dp;

		/* carry */
		r = 0;
		for (x = 0; x < a->used; x++) {

			/* get what will be the *next* carry bit from the 
			 * MSB of the current digit 
			 */
			rr = *tmpa >> ((mp_digit) (DIGIT_BIT - 1));

			/* now shift up this digit, add in the carry [from the previous] */
			*tmpb++ =
			    ((*tmpa++ << ((mp_digit) 1)) | r) & MP_MASK;

			/* copy the carry that would be from the source 
			 * digit into the next iteration 
			 */
			r = rr;
		}

		/* new leading digit? */
		if (r != 0) {
			/* add a MSB which is always 1 at this point */
			*tmpb = 1;
			++(b->used);
		}

		/* now zero any excess digits on the destination 
		 * that we didn't write to 
		 */
		tmpb = b->dp + b->used;
		for (x = b->used; x < oldused; x++) {
			*tmpb++ = 0;
		}
	}
	b->sign = a->sign;
	return MP_OKAY;
}
/* reduce "x" in place modulo "n" using the Diminished Radix algorithm.
 *
 * Based on algorithm from the paper
 *
 * "Generating Efficient Primes for Discrete Log Cryptosystems"
 *                 Chae Hoon Lim, Pil Joong Lee,
 *          POSTECH Information Research Laboratories
 *
 * The modulus must be of a special format [see manual]
 *
 * Has been modified to use algorithm 7.10 from the LTM book instead
 *
 * Input x must be in the range 0 <= x <= (n-1)**2
 */
int
mp_dr_reduce(mp_int *x, mp_int *n, mp_digit k) {
    int      err, i, m;
    mp_word  r;
    mp_digit mu, *tmpx1, *tmpx2;

    /* m = digits in modulus */
    m = n->used;

    /* ensure that "x" has at least 2m digits */
    if (x->alloc < (m + m)) {
        if ((err = mp_grow(x, m + m)) != MP_OKAY) {
            return err;
        }
    }

/* top of loop, this is where the code resumes if
 * another reduction pass is required.
 */
top:
    /* aliases for digits */
    /* alias for lower half of x */
    tmpx1 = x->dp;

    /* alias for upper half of x, or x/B**m */
    tmpx2 = x->dp + m;

    /* set carry to zero */
    mu = 0;

    /* compute (x mod B**m) + k * [x/B**m] inline and inplace */
    for (i = 0; i < m; i++) {
        r        = (((mp_word) * tmpx2++) * (mp_word)k) + *tmpx1 + mu;
        *tmpx1++ = (mp_digit)(r & MP_MASK);
        mu       = (mp_digit)(r >> ((mp_word)DIGIT_BIT));
    }

    /* set final carry */
    *tmpx1++ = mu;

    /* zero words above m */
    for (i = m + 1; i < x->used; i++) {
        *tmpx1++ = 0;
    }

    /* clamp, sub and return */
    mp_clamp(x);

    /* if x >= n then subtract and reduce again
     * Each successive "recursion" makes the input smaller and smaller.
     */
    if (mp_cmp_mag(x, n) != MP_LT) {
        if ((err = s_mp_sub(x, n, x)) != MP_OKAY) {
            return err;
        }
        goto top;
    }
    return MP_OKAY;
}
예제 #4
0
int main(void)
{
   int res, x, y;
   char buf[4096];
   FILE *out;
   mp_int a, b;
   
   mp_init(&a);
   mp_init(&b);
   
   out = fopen("drprimes.txt", "w");
   for (x = 0; x < (int)(sizeof(sizes)/sizeof(sizes[0])); x++) {
   top:
       printf("Seeking a %d-bit safe prime\n", sizes[x] * DIGIT_BIT);
       mp_grow(&a, sizes[x]);
       mp_zero(&a);
       for (y = 1; y < sizes[x]; y++) {
           a.dp[y] = MP_MASK;
       }
       
       /* make a DR modulus */
       a.dp[0] = -1;
       a.used = sizes[x];
       
       /* now loop */
       res = 0;
       for (;;) { 
          a.dp[0] += 4;
          if (a.dp[0] >= MP_MASK) break;
          mp_prime_is_prime(&a, 1, &res);
          if (res == 0) continue;
          printf("."); fflush(stdout);
          mp_sub_d(&a, 1, &b);
          mp_div_2(&b, &b);
          mp_prime_is_prime(&b, 3, &res);  
          if (res == 0) continue;
          mp_prime_is_prime(&a, 3, &res);
          if (res == 1) break;
	}
        
        if (res != 1) {
           printf("Error not DR modulus\n"); sizes[x] += 1; goto top;
        } else {
           mp_toradix(&a, buf, 10);
           printf("\n\np == %s\n\n", buf);
           fprintf(out, "%d-bit prime:\np == %s\n\n", mp_count_bits(&a), buf); fflush(out);
        }           
   }
   fclose(out);
   
   mp_clear(&a);
   mp_clear(&b);
   
   return 0;
}
예제 #5
0
파일: bigintops.c 프로젝트: mj41/MoarVM
static void grow_and_negate(mp_int *a, int size, mp_int *b) {
    int i;
    int actual_size = MAX(size, USED(a));
    mp_zero(b);
    mp_grow(b, actual_size);
    USED(b) = actual_size;
    for (i = 0; i < actual_size; i++) {
        DIGIT(b, i) = (~DIGIT(a, i)) & MP_MASK;
    }
    mp_add_d(b, 1, b);
}
예제 #6
0
/* b = a/2 */
int
mp_div_2 (mp_int * a, mp_int * b)
{
    int     x, res, oldused;

    /* copy */
    if (b->alloc < a->used) {
        if ((res = mp_grow (b, a->used)) != MP_OKAY) {
            return res;
        }
    }

    oldused = b->used;
    b->used = a->used;
    {
        register mp_digit r, rr, *tmpa, *tmpb;

        /* source alias */
        tmpa = a->dp + b->used - 1;

        /* dest alias */
        tmpb = b->dp + b->used - 1;

        /* carry */
        r = 0;
        for (x = b->used - 1; x >= 0; x--) {
            /* get the carry for the next iteration */
            rr = *tmpa & 1;

            /* shift the current digit, add in carry and store */
            *tmpb-- = (*tmpa-- >> 1) | (r << (DIGIT_BIT - 1));

            /* forward carry to next iteration */
            r = rr;
        }

        /* zero excess digits */
        tmpb = b->dp + b->used;
        for (x = b->used; x < oldused; x++) {
            *tmpb++ = 0;
        }
    }
    b->sign = a->sign;
    mp_clamp (b);
    return MP_OKAY;
}
예제 #7
0
/* shift left a certain amount of digits */
int
mp_lshd (mp_int * a, int b)
{
  int     x, res;

  /* if its less than zero return */
  if (b <= 0) {
    return MP_OKAY;
  }

  /* grow to fit the new digits */
  if (a->alloc < a->used + b) {
     if ((res = mp_grow (a, a->used + b)) != MP_OKAY) {
       return res;
     }
  }

  {
    register mp_digit *top, *bottom;

    /* increment the used by the shift amount then copy upwards */
    a->used += b;

    /* top */
    top = a->dp + a->used - 1;

    /* base */
    bottom = a->dp + a->used - 1 - b;

    /* much like mp_rshd this is implemented using a sliding window
     * except the window goes the otherway around.  Copying from
     * the bottom to the top.  see bn_mp_rshd.c for more info.
     */
    for (x = a->used - 1; x >= b; x--) {
      *top-- = *bottom--;
    }

    /* zero the lower digits */
    top = a->dp;
    for (x = 0; x < b; x++) {
      *top++ = 0;
    }
  }
  return MP_OKAY;
}
예제 #8
0
파일: bn_mp_div_2.c 프로젝트: asr/uhc
/* b = a/2 */
int mp_div_2(mp_int * a, mp_int * b)
{
  int     x, res, oldused;

  /* copy */
  if (ALLOC(b) < USED(a)) {
    if ((res = mp_grow (b, USED(a))) != MP_OKAY) {
      return res;
    }
  }

  oldused = USED(b);
  SET_USED(b,USED(a));
  {
    register mp_digit r, rr, *tmpa, *tmpb;

    /* source alias */
    tmpa = DIGITS(a) + USED(b) - 1;

    /* dest alias */
    tmpb = DIGITS(b) + USED(b) - 1;

    /* carry */
    r = 0;
    for (x = USED(b) - 1; x >= 0; x--) {
      /* get the carry for the next iteration */
      rr = *tmpa & 1;

      /* shift the current digit, add in carry and store */
      *tmpb-- = (*tmpa-- >> 1) | (r << (DIGIT_BIT - 1));

      /* forward carry to next iteration */
      r = rr;
    }

    /* zero excess digits */
    tmpb = DIGITS(b) + USED(b);
    for (x = USED(b); x < oldused; x++) {
      *tmpb++ = 0;
    }
  }
  SET_SIGN(b,SIGN(a));
  mp_clamp (b);
  return MP_OKAY;
}
예제 #9
0
/* copy, b = a */
int
mp_copy MPA(mp_int * a, mp_int * b)
{
  int     res, n;

  /* if dst == src do nothing */
  if (a == b) {
    return MP_OKAY;
  }

  /* grow dest */
  if (b->alloc < a->used) {
     if ((res = mp_grow (MPST, b, a->used)) != MP_OKAY) {
        return res;
     }
  }

  /* zero b and copy the parameters over */
  {
    register mp_digit *tmpa, *tmpb;

    /* pointer aliases */

    /* source */
    tmpa = a->dp;

    /* destination */
    tmpb = b->dp;

    /* copy all the digits */
    for (n = 0; n < a->used; n++) {
      *tmpb++ = *tmpa++;
    }

    /* clear high digits */
    for (; n < b->used; n++) {
      *tmpb++ = 0;
    }
  }

  /* copy used count and sign */
  b->used = a->used;
  b->sign = a->sign;
  return MP_OKAY;
}
예제 #10
0
파일: bigintops.c 프로젝트: baby-gnu/MoarVM
static void from_num(MVMnum64 d, mp_int *a) {
    MVMnum64 d_digit = pow(2, DIGIT_BIT);
    MVMnum64 da      = fabs(d);
    MVMnum64 upper;
    MVMnum64 lower;
    MVMnum64 lowest;
    MVMnum64 rest;
    int      digits  = 0;

    mp_zero(a);

    while (da > d_digit * d_digit * d_digit) {;
        da /= d_digit;
        digits++;
    }
    mp_grow(a, digits + 3);

    /* populate the top 3 digits */
    upper = da / (d_digit*d_digit);
    rest = fmod(da, d_digit*d_digit);
    lower = rest / d_digit;
    lowest = fmod(rest,d_digit );
    if (upper >= 1) {
        mp_set_long(a, (unsigned long) upper);
        mp_mul_2d(a, DIGIT_BIT , a);
        DIGIT(a, 0) = (mp_digit) lower;
        mp_mul_2d(a, DIGIT_BIT , a);
    } else {
        if (lower >= 1) {
            mp_set_long(a, (unsigned long) lower);
            mp_mul_2d(a, DIGIT_BIT , a);
            a->used = 2;
        } else {
            a->used = 1;
        }
    }
    DIGIT(a, 0) = (mp_digit) lowest;

    /* shift the rest */
    mp_mul_2d(a, DIGIT_BIT * digits, a);
    if (d < 0)
        mp_neg(a, a);
    mp_clamp(a);
    mp_shrink(a);
}
예제 #11
0
/* computes a = 2**b 
 *
 * Simple algorithm which zeroes the int, grows it then just sets one bit
 * as required.
 */
int mp_2expt(mp_int * a, int b)
{
	int res;

	/* zero a as per default */
	mp_zero(a);

	/* grow a to accomodate the single bit */
	if ((res = mp_grow(a, b / DIGIT_BIT + 1)) != MP_OKAY) {
		return res;
	}

	/* set the used count of where the bit will go */
	a->used = b / DIGIT_BIT + 1;

	/* put the single bit in its place */
	a->dp[b / DIGIT_BIT] = 1 << (b % DIGIT_BIT);

	return MP_OKAY;
}
예제 #12
0
파일: bigintops.c 프로젝트: baby-gnu/MoarVM
/* Bitops on libtomath (no 2s compliment API) are horrendously inefficient and
 * really should be hand-coded to work DIGIT-by-DIGIT with in-loop carry
 * handling.  For now we have these fixups.
 *
 * The following inverts the bits of a negative bigint, adds 1 to that, and
 * appends sign-bit extension DIGITs to it to give us a 2s compliment
 * representation in memory.  Do not call it on positive bigints.
 */
static void grow_and_negate(const mp_int *a, int size, mp_int *b) {
    int i;
    /* Always add an extra DIGIT so we can tell positive values
     * with a one in the highest bit apart from negative values.
     */
    int actual_size = MAX(size, USED(a)) + 1;

    SIGN(b) = MP_ZPOS;
    mp_grow(b, actual_size);
    USED(b) = actual_size;
    for (i = 0; i < USED(a); i++) {
        DIGIT(b, i) = (~DIGIT(a, i)) & MP_MASK;
    }
    for (; i < actual_size; i++) {
        DIGIT(b, i) = MP_MASK;
    }
    /* Note: This add cannot cause another grow assuming nobody ever
     * tries to use tommath -0 for anything, and nobody tries to use
     * this on positive bigints.
     */
    mp_add_d(b, 1, b);
}
예제 #13
0
/* low level addition, based on HAC pp.594, Algorithm 14.7 */
int
s_mp_add (mp_int * a, mp_int * b, mp_int * c)
{
  mp_int *x;
  int     olduse, res, min, max;

  /* find sizes, we let |a| <= |b| which means we have to sort
   * them.  "x" will point to the input with the most digits
   */
  if (a->used > b->used) {
    min = b->used;
    max = a->used;
    x = a;
  } else {
    min = a->used;
    max = b->used;
    x = b;
  }

  /* init result */
  if (c->alloc < max + 1) {
    if ((res = mp_grow (c, max + 1)) != MP_OKAY) {
      return res;
    }
  }

  /* get old used digit count and set new one */
  olduse = c->used;
  c->used = max + 1;

  {
    register mp_digit u, *tmpa, *tmpb, *tmpc;
    register int i;

    /* alias for digit pointers */

    /* first input */
    tmpa = a->dp;

    /* second input */
    tmpb = b->dp;

    /* destination */
    tmpc = c->dp;

    /* zero the carry */
    u = 0;
    for (i = 0; i < min; i++) {
      /* Compute the sum at one digit, T[i] = A[i] + B[i] + U */
      *tmpc = *tmpa++ + *tmpb++ + u;

      /* U = carry bit of T[i] */
      u = *tmpc >> ((mp_digit)DIGIT_BIT);

      /* take away carry bit from T[i] */
      *tmpc++ &= MP_MASK;
    }

    /* now copy higher words if any, that is in A+B 
     * if A or B has more digits add those in 
     */
    if (min != max) {
      for (; i < max; i++) {
        /* T[i] = X[i] + U */
        *tmpc = x->dp[i] + u;

        /* U = carry bit of T[i] */
        u = *tmpc >> ((mp_digit)DIGIT_BIT);

        /* take away carry bit from T[i] */
        *tmpc++ &= MP_MASK;
      }
    }

    /* add carry */
    *tmpc++ = u;

    /* clear digits above oldused */
    for (i = c->used; i < olduse; i++) {
      *tmpc++ = 0;
    }
  }

  mp_clamp (c);
  return MP_OKAY;
}
예제 #14
0
/* low level subtraction (assumes |a| > |b|), HAC pp.595 Algorithm 14.9 */
int
s_mp_sub (mp_int * a, mp_int * b, mp_int * c)
{
  int     olduse, res, min, max;

  /* find sizes */
  min = b->used;
  max = a->used;

  /* init result */
  if (c->alloc < max) {
    if ((res = mp_grow (c, max)) != MP_OKAY) {
      return res;
    }
  }
  olduse = c->used;
  c->used = max;

  {
    register mp_digit u, *tmpa, *tmpb, *tmpc;
    register int i;

    /* alias for digit pointers */
    tmpa = a->dp;
    tmpb = b->dp;
    tmpc = c->dp;

    /* set carry to zero */
    u = 0;
    for (i = 0; i < min; i++) {
      /* T[i] = A[i] - B[i] - U */
      *tmpc = *tmpa++ - *tmpb++ - u;

      /* U = carry bit of T[i]
       * Note this saves performing an AND operation since
       * if a carry does occur it will propagate all the way to the
       * MSB.  As a result a single shift is enough to get the carry
       */
      u = *tmpc >> ((mp_digit)(CHAR_BIT * sizeof (mp_digit) - 1));

      /* Clear carry from T[i] */
      *tmpc++ &= MP_MASK;
    }

    /* now copy higher words if any, e.g. if A has more digits than B  */
    for (; i < max; i++) {
      /* T[i] = A[i] - U */
      *tmpc = *tmpa++ - u;

      /* U = carry bit of T[i] */
      u = *tmpc >> ((mp_digit)(CHAR_BIT * sizeof (mp_digit) - 1));

      /* Clear carry from T[i] */
      *tmpc++ &= MP_MASK;
    }

    /* clear digits above used (since we may not have grown result above) */
    for (i = c->used; i < olduse; i++) {
      *tmpc++ = 0;
    }
  }

  mp_clamp (c);
  return MP_OKAY;
}
예제 #15
0
int fast_s_mp_sqr (mp_int * a, mp_int * b)
{
    int       olduse, res, pa, ix, iz;
    mp_digit   W[MP_WARRAY], *tmpx;
    mp_word   W1;

    /* grow the destination as required */
    pa = a->used + a->used;
    if (b->alloc < pa) {
        if ((res = mp_grow (b, pa)) != MP_OKAY) {
            return res;
        }
    }

    /* number of output digits to produce */
    W1 = 0;
    for (ix = 0; ix < pa; ix++) {
        int      tx, ty, iy;
        mp_word  _W;
        mp_digit *tmpy;

        /* clear counter */
        _W = 0;

        /* get offsets into the two bignums */
        ty = MIN(a->used-1, ix);
        tx = ix - ty;

        /* setup temp aliases */
        tmpx = a->dp + tx;
        tmpy = a->dp + ty;

        /* this is the number of times the loop will iterrate, essentially its
           while (tx++ < a->used && ty-- >= 0) { ... }
         */
        iy = MIN(a->used-tx, ty+1);

        /* now for squaring tx can never equal ty
         * we halve the distance since they approach at a rate of 2x
         * and we have to round because odd cases need to be executed
         */
        iy = MIN(iy, (ty-tx+1)>>1);

        /* execute loop */
        for (iz = 0; iz < iy; iz++) {
            _W += ((mp_word)*tmpx++)*((mp_word)*tmpy--);
        }

        /* double the inner product and add carry */
        _W = _W + _W + W1;

        /* even columns have the square term in them */
        if ((ix&1) == 0) {
            _W += ((mp_word)a->dp[ix>>1])*((mp_word)a->dp[ix>>1]);
        }

        /* store it */
        W[ix] = _W;

        /* make next carry */
        W1 = _W >> ((mp_word)DIGIT_BIT);
    }
예제 #16
0
/* this is a modified version of fast_s_mul_digs that only produces
 * output digits *above* digs.  See the comments for fast_s_mul_digs
 * to see how it works.
 *
 * This is used in the Barrett reduction since for one of the multiplications
 * only the higher digits were needed.  This essentially halves the work.
 *
 * Based on Algorithm 14.12 on pp.595 of HAC.
 */
int fast_s_mp_mul_high_digs (mp_int * a, mp_int * b, mp_int * c, int digs)
{
  int     olduse, res, pa, ix, iz;
  mp_digit W[MP_WARRAY];
  mp_word  _W;

  /* grow the destination as required */
  pa = a->used + b->used;
  if (c->alloc < pa) {
    if ((res = mp_grow (c, pa)) != MP_OKAY) {
      return res;
    }
  }

  /* number of output digits to produce */
  pa = a->used + b->used;
  _W = 0;
  for (ix = digs; ix < pa; ix++) { 
      int      tx, ty, iy;
      mp_digit *tmpx, *tmpy;

      /* get offsets into the two bignums */
      ty = MIN(b->used-1, ix);
      tx = ix - ty;

      /* setup temp aliases */
      tmpx = a->dp + tx;
      tmpy = b->dp + ty;

      /* this is the number of times the loop will iterrate, essentially its 
         while (tx++ < a->used && ty-- >= 0) { ... }
       */
      iy = MIN(a->used-tx, ty+1);

      /* execute loop */
      for (iz = 0; iz < iy; iz++) {
         _W += ((mp_word)*tmpx++)*((mp_word)*tmpy--);
      }

      /* store term */
      W[ix] = ((mp_digit)_W) & MP_MASK;

      /* make next carry */
      _W = _W >> ((mp_word)DIGIT_BIT);
  }
  
  /* setup dest */
  olduse  = c->used;
  c->used = pa;

  {
    mp_digit *tmpc;

    tmpc = c->dp + digs;
    for (ix = digs; ix < pa; ix++) {
      /* now extract the previous digit [below the carry] */
      *tmpc++ = W[ix];
    }

    /* clear unused digits [that existed in the old copy of c] */
    for (; ix < olduse; ix++) {
      *tmpc++ = 0;
    }
  }
  mp_clamp (c);
  return MP_OKAY;
}
/* computes xR**-1 == x (mod N) via Montgomery Reduction */
int
mp_montgomery_reduce (mp_int * x, mp_int * n, mp_digit rho)
{
  int     ix, res, digs;
  mp_digit mu;

  /* can the fast reduction [comba] method be used?
   *
   * Note that unlike in mul you're safely allowed *less*
   * than the available columns [255 per default] since carries
   * are fixed up in the inner loop.
   */
  digs = n->used * 2 + 1;
  if ((digs < MP_WARRAY) &&
      n->used <
      (1 << ((CHAR_BIT * sizeof (mp_word)) - (2 * DIGIT_BIT)))) {
    return fast_mp_montgomery_reduce (x, n, rho);
  }

  /* grow the input as required */
  if (x->alloc < digs) {
    if ((res = mp_grow (x, digs)) != MP_OKAY) {
      return res;
    }
  }
  x->used = digs;

  for (ix = 0; ix < n->used; ix++) {
    /* mu = ai * rho mod b
     *
     * The value of rho must be precalculated via
     * montgomery_setup() such that
     * it equals -1/n0 mod b this allows the
     * following inner loop to reduce the
     * input one digit at a time
     */
    mu = (mp_digit) (((mp_word)x->dp[ix]) * ((mp_word)rho) & MP_MASK);

    /* a = a + mu * m * b**i */
    {
      register int iy;
      register mp_digit *tmpn, *tmpx, u;
      register mp_word r;

      /* alias for digits of the modulus */
      tmpn = n->dp;

      /* alias for the digits of x [the input] */
      tmpx = x->dp + ix;

      /* set the carry to zero */
      u = 0;

      /* Multiply and add in place */
      for (iy = 0; iy < n->used; iy++) {
        /* compute product and sum */
        r       = ((mp_word)mu) * ((mp_word)*tmpn++) +
                  ((mp_word) u) + ((mp_word) * tmpx);

        /* get carry */
        u       = (mp_digit)(r >> ((mp_word) DIGIT_BIT));

        /* fix digit */
        *tmpx++ = (mp_digit)(r & ((mp_word) MP_MASK));
      }
      /* At this point the ix'th digit of x should be zero */


      /* propagate carries upwards as required*/
      while (u) {
        *tmpx   += u;
        u        = *tmpx >> DIGIT_BIT;
        *tmpx++ &= MP_MASK;
      }
    }
  }

  /* at this point the n.used'th least
   * significant digits of x are all zero
   * which means we can shift x to the
   * right by n.used digits and the
   * residue is unchanged.
   */

  /* x = x/b**n.used */
  mp_clamp(x);
  mp_rshd (x, n->used);

  /* if x >= n then x = x - n */
  if (mp_cmp_mag (x, n) != MP_LT) {
    return s_mp_sub (x, n, x);
  }

  return MP_OKAY;
}
예제 #18
0
파일: demo.c 프로젝트: DavidMulder/heimdal
int main(void)
{
   mp_int a, b, c, d, e, f;
   unsigned long expt_n, add_n, sub_n, mul_n, div_n, sqr_n, mul2d_n, div2d_n,
      gcd_n, lcm_n, inv_n, div2_n, mul2_n, add_d_n, sub_d_n, t;
   unsigned rr;
   int i, n, err, cnt, ix, old_kara_m, old_kara_s;
   mp_digit mp;


   mp_init(&a);
   mp_init(&b);
   mp_init(&c);
   mp_init(&d);
   mp_init(&e);
   mp_init(&f);

   srand(time(NULL));

#if 0
   // test montgomery
   printf("Testing montgomery...\n");
   for (i = 1; i < 10; i++) {
      printf("Testing digit size: %d\n", i);
      for (n = 0; n < 1000; n++) {
         mp_rand(&a, i);
         a.dp[0] |= 1;

         // let's see if R is right
         mp_montgomery_calc_normalization(&b, &a);
         mp_montgomery_setup(&a, &mp);

         // now test a random reduction
         for (ix = 0; ix < 100; ix++) {
             mp_rand(&c, 1 + abs(rand()) % (2*i));
             mp_copy(&c, &d);
             mp_copy(&c, &e);

             mp_mod(&d, &a, &d);
             mp_montgomery_reduce(&c, &a, mp);
             mp_mulmod(&c, &b, &a, &c);

             if (mp_cmp(&c, &d) != MP_EQ) {
printf("d = e mod a, c = e MOD a\n");
mp_todecimal(&a, buf); printf("a = %s\n", buf);
mp_todecimal(&e, buf); printf("e = %s\n", buf);
mp_todecimal(&d, buf); printf("d = %s\n", buf);
mp_todecimal(&c, buf); printf("c = %s\n", buf);
printf("compare no compare!\n"); exit(EXIT_FAILURE); }
         }
      }
   }
   printf("done\n");

   // test mp_get_int
   printf("Testing: mp_get_int\n");
   for (i = 0; i < 1000; ++i) {
      t = ((unsigned long) rand() * rand() + 1) & 0xFFFFFFFF;
      mp_set_int(&a, t);
      if (t != mp_get_int(&a)) {
	 printf("mp_get_int() bad result!\n");
	 return 1;
      }
   }
   mp_set_int(&a, 0);
   if (mp_get_int(&a) != 0) {
      printf("mp_get_int() bad result!\n");
      return 1;
   }
   mp_set_int(&a, 0xffffffff);
   if (mp_get_int(&a) != 0xffffffff) {
      printf("mp_get_int() bad result!\n");
      return 1;
   }
   // test mp_sqrt
   printf("Testing: mp_sqrt\n");
   for (i = 0; i < 1000; ++i) {
      printf("%6d\r", i);
      fflush(stdout);
      n = (rand() & 15) + 1;
      mp_rand(&a, n);
      if (mp_sqrt(&a, &b) != MP_OKAY) {
	 printf("mp_sqrt() error!\n");
	 return 1;
      }
      mp_n_root(&a, 2, &a);
      if (mp_cmp_mag(&b, &a) != MP_EQ) {
	 printf("mp_sqrt() bad result!\n");
	 return 1;
      }
   }

   printf("\nTesting: mp_is_square\n");
   for (i = 0; i < 1000; ++i) {
      printf("%6d\r", i);
      fflush(stdout);

      /* test mp_is_square false negatives */
      n = (rand() & 7) + 1;
      mp_rand(&a, n);
      mp_sqr(&a, &a);
      if (mp_is_square(&a, &n) != MP_OKAY) {
	 printf("fn:mp_is_square() error!\n");
	 return 1;
      }
      if (n == 0) {
	 printf("fn:mp_is_square() bad result!\n");
	 return 1;
      }

      /* test for false positives */
      mp_add_d(&a, 1, &a);
      if (mp_is_square(&a, &n) != MP_OKAY) {
	 printf("fp:mp_is_square() error!\n");
	 return 1;
      }
      if (n == 1) {
	 printf("fp:mp_is_square() bad result!\n");
	 return 1;
      }

   }
   printf("\n\n");

   /* test for size */
   for (ix = 10; ix < 128; ix++) {
      printf("Testing (not safe-prime): %9d bits    \r", ix);
      fflush(stdout);
      err =
	 mp_prime_random_ex(&a, 8, ix,
			    (rand() & 1) ? LTM_PRIME_2MSB_OFF :
			    LTM_PRIME_2MSB_ON, myrng, NULL);
      if (err != MP_OKAY) {
	 printf("failed with err code %d\n", err);
	 return EXIT_FAILURE;
      }
      if (mp_count_bits(&a) != ix) {
	 printf("Prime is %d not %d bits!!!\n", mp_count_bits(&a), ix);
	 return EXIT_FAILURE;
      }
   }

   for (ix = 16; ix < 128; ix++) {
      printf("Testing (   safe-prime): %9d bits    \r", ix);
      fflush(stdout);
      err =
	 mp_prime_random_ex(&a, 8, ix,
			    ((rand() & 1) ? LTM_PRIME_2MSB_OFF :
			     LTM_PRIME_2MSB_ON) | LTM_PRIME_SAFE, myrng,
			    NULL);
      if (err != MP_OKAY) {
	 printf("failed with err code %d\n", err);
	 return EXIT_FAILURE;
      }
      if (mp_count_bits(&a) != ix) {
	 printf("Prime is %d not %d bits!!!\n", mp_count_bits(&a), ix);
	 return EXIT_FAILURE;
      }
      /* let's see if it's really a safe prime */
      mp_sub_d(&a, 1, &a);
      mp_div_2(&a, &a);
      mp_prime_is_prime(&a, 8, &cnt);
      if (cnt != MP_YES) {
	 printf("sub is not prime!\n");
	 return EXIT_FAILURE;
      }
   }

   printf("\n\n");

   mp_read_radix(&a, "123456", 10);
   mp_toradix_n(&a, buf, 10, 3);
   printf("a == %s\n", buf);
   mp_toradix_n(&a, buf, 10, 4);
   printf("a == %s\n", buf);
   mp_toradix_n(&a, buf, 10, 30);
   printf("a == %s\n", buf);


#if 0
   for (;;) {
      fgets(buf, sizeof(buf), stdin);
      mp_read_radix(&a, buf, 10);
      mp_prime_next_prime(&a, 5, 1);
      mp_toradix(&a, buf, 10);
      printf("%s, %lu\n", buf, a.dp[0] & 3);
   }
#endif

   /* test mp_cnt_lsb */
   printf("testing mp_cnt_lsb...\n");
   mp_set(&a, 1);
   for (ix = 0; ix < 1024; ix++) {
      if (mp_cnt_lsb(&a) != ix) {
	 printf("Failed at %d, %d\n", ix, mp_cnt_lsb(&a));
	 return 0;
      }
      mp_mul_2(&a, &a);
   }

/* test mp_reduce_2k */
   printf("Testing mp_reduce_2k...\n");
   for (cnt = 3; cnt <= 128; ++cnt) {
      mp_digit tmp;

      mp_2expt(&a, cnt);
      mp_sub_d(&a, 2, &a);	/* a = 2**cnt - 2 */


      printf("\nTesting %4d bits", cnt);
      printf("(%d)", mp_reduce_is_2k(&a));
      mp_reduce_2k_setup(&a, &tmp);
      printf("(%d)", tmp);
      for (ix = 0; ix < 1000; ix++) {
	 if (!(ix & 127)) {
	    printf(".");
	    fflush(stdout);
	 }
	 mp_rand(&b, (cnt / DIGIT_BIT + 1) * 2);
	 mp_copy(&c, &b);
	 mp_mod(&c, &a, &c);
	 mp_reduce_2k(&b, &a, 2);
	 if (mp_cmp(&c, &b)) {
	    printf("FAILED\n");
	    exit(0);
	 }
      }
   }

/* test mp_div_3  */
   printf("Testing mp_div_3...\n");
   mp_set(&d, 3);
   for (cnt = 0; cnt < 10000;) {
      mp_digit r1, r2;

      if (!(++cnt & 127))
	 printf("%9d\r", cnt);
      mp_rand(&a, abs(rand()) % 128 + 1);
      mp_div(&a, &d, &b, &e);
      mp_div_3(&a, &c, &r2);

      if (mp_cmp(&b, &c) || mp_cmp_d(&e, r2)) {
	 printf("\n\nmp_div_3 => Failure\n");
      }
   }
   printf("\n\nPassed div_3 testing\n");

/* test the DR reduction */
   printf("testing mp_dr_reduce...\n");
   for (cnt = 2; cnt < 32; cnt++) {
      printf("%d digit modulus\n", cnt);
      mp_grow(&a, cnt);
      mp_zero(&a);
      for (ix = 1; ix < cnt; ix++) {
	 a.dp[ix] = MP_MASK;
      }
      a.used = cnt;
      a.dp[0] = 3;

      mp_rand(&b, cnt - 1);
      mp_copy(&b, &c);

      rr = 0;
      do {
	 if (!(rr & 127)) {
	    printf("%9lu\r", rr);
	    fflush(stdout);
	 }
	 mp_sqr(&b, &b);
	 mp_add_d(&b, 1, &b);
	 mp_copy(&b, &c);

	 mp_mod(&b, &a, &b);
	 mp_dr_reduce(&c, &a, (((mp_digit) 1) << DIGIT_BIT) - a.dp[0]);

	 if (mp_cmp(&b, &c) != MP_EQ) {
	    printf("Failed on trial %lu\n", rr);
	    exit(-1);

	 }
      } while (++rr < 500);
      printf("Passed DR test for %d digits\n", cnt);
   }

#endif

/* test the mp_reduce_2k_l code */
#if 0
#if 0
/* first load P with 2^1024 - 0x2A434 B9FDEC95 D8F9D550 FFFFFFFF FFFFFFFF */
   mp_2expt(&a, 1024);
   mp_read_radix(&b, "2A434B9FDEC95D8F9D550FFFFFFFFFFFFFFFF", 16);
   mp_sub(&a, &b, &a);
#elif 1
/*  p = 2^2048 - 0x1 00000000 00000000 00000000 00000000 4945DDBF 8EA2A91D 5776399B B83E188F  */
   mp_2expt(&a, 2048);
   mp_read_radix(&b,
		 "1000000000000000000000000000000004945DDBF8EA2A91D5776399BB83E188F",
		 16);
   mp_sub(&a, &b, &a);
#endif

   mp_todecimal(&a, buf);
   printf("p==%s\n", buf);
/* now mp_reduce_is_2k_l() should return */
   if (mp_reduce_is_2k_l(&a) != 1) {
      printf("mp_reduce_is_2k_l() return 0, should be 1\n");
      return EXIT_FAILURE;
   }
   mp_reduce_2k_setup_l(&a, &d);
   /* now do a million square+1 to see if it varies */
   mp_rand(&b, 64);
   mp_mod(&b, &a, &b);
   mp_copy(&b, &c);
   printf("testing mp_reduce_2k_l...");
   fflush(stdout);
   for (cnt = 0; cnt < (1UL << 20); cnt++) {
      mp_sqr(&b, &b);
      mp_add_d(&b, 1, &b);
      mp_reduce_2k_l(&b, &a, &d);
      mp_sqr(&c, &c);
      mp_add_d(&c, 1, &c);
      mp_mod(&c, &a, &c);
      if (mp_cmp(&b, &c) != MP_EQ) {
	 printf("mp_reduce_2k_l() failed at step %lu\n", cnt);
	 mp_tohex(&b, buf);
	 printf("b == %s\n", buf);
	 mp_tohex(&c, buf);
	 printf("c == %s\n", buf);
	 return EXIT_FAILURE;
      }
   }
   printf("...Passed\n");
#endif

   div2_n = mul2_n = inv_n = expt_n = lcm_n = gcd_n = add_n =
      sub_n = mul_n = div_n = sqr_n = mul2d_n = div2d_n = cnt = add_d_n =
      sub_d_n = 0;

   /* force KARA and TOOM to enable despite cutoffs */
   KARATSUBA_SQR_CUTOFF = KARATSUBA_MUL_CUTOFF = 8;
   TOOM_SQR_CUTOFF = TOOM_MUL_CUTOFF = 16;

   for (;;) {
      /* randomly clear and re-init one variable, this has the affect of triming the alloc space */
      switch (abs(rand()) % 7) {
      case 0:
	 mp_clear(&a);
	 mp_init(&a);
	 break;
      case 1:
	 mp_clear(&b);
	 mp_init(&b);
	 break;
      case 2:
	 mp_clear(&c);
	 mp_init(&c);
	 break;
      case 3:
	 mp_clear(&d);
	 mp_init(&d);
	 break;
      case 4:
	 mp_clear(&e);
	 mp_init(&e);
	 break;
      case 5:
	 mp_clear(&f);
	 mp_init(&f);
	 break;
      case 6:
	 break;			/* don't clear any */
      }


      printf
	 ("%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu ",
	  add_n, sub_n, mul_n, div_n, sqr_n, mul2d_n, div2d_n, gcd_n, lcm_n,
	  expt_n, inv_n, div2_n, mul2_n, add_d_n, sub_d_n);
      fgets(cmd, 4095, stdin);
      cmd[strlen(cmd) - 1] = 0;
      printf("%s  ]\r", cmd);
      fflush(stdout);
      if (!strcmp(cmd, "mul2d")) {
	 ++mul2d_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 sscanf(buf, "%d", &rr);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);

	 mp_mul_2d(&a, rr, &a);
	 a.sign = b.sign;
	 if (mp_cmp(&a, &b) != MP_EQ) {
	    printf("mul2d failed, rr == %d\n", rr);
	    draw(&a);
	    draw(&b);
	    return 0;
	 }
      } else if (!strcmp(cmd, "div2d")) {
	 ++div2d_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 sscanf(buf, "%d", &rr);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);

	 mp_div_2d(&a, rr, &a, &e);
	 a.sign = b.sign;
	 if (a.used == b.used && a.used == 0) {
	    a.sign = b.sign = MP_ZPOS;
	 }
	 if (mp_cmp(&a, &b) != MP_EQ) {
	    printf("div2d failed, rr == %d\n", rr);
	    draw(&a);
	    draw(&b);
	    return 0;
	 }
      } else if (!strcmp(cmd, "add")) {
	 ++add_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&c, buf, 64);
	 mp_copy(&a, &d);
	 mp_add(&d, &b, &d);
	 if (mp_cmp(&c, &d) != MP_EQ) {
	    printf("add %lu failure!\n", add_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    return 0;
	 }

	 /* test the sign/unsigned storage functions */

	 rr = mp_signed_bin_size(&c);
	 mp_to_signed_bin(&c, (unsigned char *) cmd);
	 memset(cmd + rr, rand() & 255, sizeof(cmd) - rr);
	 mp_read_signed_bin(&d, (unsigned char *) cmd, rr);
	 if (mp_cmp(&c, &d) != MP_EQ) {
	    printf("mp_signed_bin failure!\n");
	    draw(&c);
	    draw(&d);
	    return 0;
	 }


	 rr = mp_unsigned_bin_size(&c);
	 mp_to_unsigned_bin(&c, (unsigned char *) cmd);
	 memset(cmd + rr, rand() & 255, sizeof(cmd) - rr);
	 mp_read_unsigned_bin(&d, (unsigned char *) cmd, rr);
	 if (mp_cmp_mag(&c, &d) != MP_EQ) {
	    printf("mp_unsigned_bin failure!\n");
	    draw(&c);
	    draw(&d);
	    return 0;
	 }

      } else if (!strcmp(cmd, "sub")) {
	 ++sub_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&c, buf, 64);
	 mp_copy(&a, &d);
	 mp_sub(&d, &b, &d);
	 if (mp_cmp(&c, &d) != MP_EQ) {
	    printf("sub %lu failure!\n", sub_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    return 0;
	 }
      } else if (!strcmp(cmd, "mul")) {
	 ++mul_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&c, buf, 64);
	 mp_copy(&a, &d);
	 mp_mul(&d, &b, &d);
	 if (mp_cmp(&c, &d) != MP_EQ) {
	    printf("mul %lu failure!\n", mul_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    return 0;
	 }
      } else if (!strcmp(cmd, "div")) {
	 ++div_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&c, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&d, buf, 64);

	 mp_div(&a, &b, &e, &f);
	 if (mp_cmp(&c, &e) != MP_EQ || mp_cmp(&d, &f) != MP_EQ) {
	    printf("div %lu %d, %d, failure!\n", div_n, mp_cmp(&c, &e),
		   mp_cmp(&d, &f));
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    draw(&e);
	    draw(&f);
	    return 0;
	 }

      } else if (!strcmp(cmd, "sqr")) {
	 ++sqr_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 mp_copy(&a, &c);
	 mp_sqr(&c, &c);
	 if (mp_cmp(&b, &c) != MP_EQ) {
	    printf("sqr %lu failure!\n", sqr_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    return 0;
	 }
      } else if (!strcmp(cmd, "gcd")) {
	 ++gcd_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&c, buf, 64);
	 mp_copy(&a, &d);
	 mp_gcd(&d, &b, &d);
	 d.sign = c.sign;
	 if (mp_cmp(&c, &d) != MP_EQ) {
	    printf("gcd %lu failure!\n", gcd_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    return 0;
	 }
      } else if (!strcmp(cmd, "lcm")) {
	 ++lcm_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&c, buf, 64);
	 mp_copy(&a, &d);
	 mp_lcm(&d, &b, &d);
	 d.sign = c.sign;
	 if (mp_cmp(&c, &d) != MP_EQ) {
	    printf("lcm %lu failure!\n", lcm_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    return 0;
	 }
      } else if (!strcmp(cmd, "expt")) {
	 ++expt_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&c, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&d, buf, 64);
	 mp_copy(&a, &e);
	 mp_exptmod(&e, &b, &c, &e);
	 if (mp_cmp(&d, &e) != MP_EQ) {
	    printf("expt %lu failure!\n", expt_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    draw(&e);
	    return 0;
	 }
      } else if (!strcmp(cmd, "invmod")) {
	 ++inv_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&c, buf, 64);
	 mp_invmod(&a, &b, &d);
	 mp_mulmod(&d, &a, &b, &e);
	 if (mp_cmp_d(&e, 1) != MP_EQ) {
	    printf("inv [wrong value from MPI?!] failure\n");
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    mp_gcd(&a, &b, &e);
	    draw(&e);
	    return 0;
	 }

      } else if (!strcmp(cmd, "div2")) {
	 ++div2_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 mp_div_2(&a, &c);
	 if (mp_cmp(&c, &b) != MP_EQ) {
	    printf("div_2 %lu failure\n", div2_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    return 0;
	 }
      } else if (!strcmp(cmd, "mul2")) {
	 ++mul2_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 mp_mul_2(&a, &c);
	 if (mp_cmp(&c, &b) != MP_EQ) {
	    printf("mul_2 %lu failure\n", mul2_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    return 0;
	 }
      } else if (!strcmp(cmd, "add_d")) {
	 ++add_d_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 sscanf(buf, "%d", &ix);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 mp_add_d(&a, ix, &c);
	 if (mp_cmp(&b, &c) != MP_EQ) {
	    printf("add_d %lu failure\n", add_d_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    printf("d == %d\n", ix);
	    return 0;
	 }
      } else if (!strcmp(cmd, "sub_d")) {
	 ++sub_d_n;
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&a, buf, 64);
	 fgets(buf, 4095, stdin);
	 sscanf(buf, "%d", &ix);
	 fgets(buf, 4095, stdin);
	 mp_read_radix(&b, buf, 64);
	 mp_sub_d(&a, ix, &c);
	 if (mp_cmp(&b, &c) != MP_EQ) {
	    printf("sub_d %lu failure\n", sub_d_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    printf("d == %d\n", ix);
	    return 0;
	 }
      }
   }
   return 0;
}
예제 #19
0
/* single digit subtraction */
int
mp_sub_d (mp_int * a, mp_digit b, mp_int * c)
{
  mp_digit *tmpa, *tmpc, mu;
  int       res, ix, oldused;

  /* grow c as required */
  if (c->alloc < a->used + 1) {
     if ((res = mp_grow(c, a->used + 1)) != MP_OKAY) {
        return res;
     }
  }

  /* if a is negative just do an unsigned
   * addition [with fudged signs]
   */
  if (a->sign == MP_NEG) {
     a->sign = MP_ZPOS;
     res     = mp_add_d(a, b, c);
     a->sign = c->sign = MP_NEG;

     /* clamp */
     mp_clamp(c);

     return res;
  }

  /* setup regs */
  oldused = c->used;
  tmpa    = a->dp;
  tmpc    = c->dp;

  /* if a <= b simply fix the single digit */
  if ((a->used == 1 && a->dp[0] <= b) || a->used == 0) {
     if (a->used == 1) {
        *tmpc++ = b - *tmpa;
     } else {
        *tmpc++ = b;
     }
     ix      = 1;

     /* negative/1digit */
     c->sign = MP_NEG;
     c->used = 1;
  } else {
     /* positive/size */
     c->sign = MP_ZPOS;
     c->used = a->used;

     /* subtract first digit */
     *tmpc    = *tmpa++ - b;
     mu       = *tmpc >> (sizeof(mp_digit) * CHAR_BIT - 1);
     *tmpc++ &= MP_MASK;

     /* handle rest of the digits */
     for (ix = 1; ix < a->used; ix++) {
        *tmpc    = *tmpa++ - mu;
        mu       = *tmpc >> (sizeof(mp_digit) * CHAR_BIT - 1);
        *tmpc++ &= MP_MASK;
     }
  }

  /* zero excess digits */
  while (ix++ < oldused) {
     *tmpc++ = 0;
  }
  mp_clamp(c);
  return MP_OKAY;
}
/* computes xR**-1 == x (mod N) via Montgomery Reduction
 *
 * This is an optimized implementation of montgomery_reduce
 * which uses the comba method to quickly calculate the columns of the
 * reduction.
 *
 * Based on Algorithm 14.32 on pp.601 of HAC.
*/
int fast_mp_montgomery_reduce (mp_int * x, mp_int * n, mp_digit rho)
{
  int     ix, res, olduse;
  mp_word W[MP_WARRAY] = { 0 };

  /* get old used count */
  olduse = x->used;

  /* grow a as required */
  if (x->alloc < n->used + 1) {
    if ((res = mp_grow (x, n->used + 1)) != MP_OKAY) {
      return res;
    }
  }

  /* first we have to get the digits of the input into
   * an array of double precision words W[...]
   */
  {
    register mp_word *_W;
    register mp_digit *tmpx;

    /* alias for the W[] array */
    _W   = W;

    /* alias for the digits of  x*/
    tmpx = x->dp;

    /* copy the digits of a into W[0..a->used-1] */
    for (ix = 0; ix < x->used; ix++) {
      *_W++ = *tmpx++;
    }

    /* zero the high words of W[a->used..m->used*2] */
    for (; ix < n->used * 2 + 1; ix++) {
      *_W++ = 0;
    }
  }

  /* now we proceed to zero successive digits
   * from the least significant upwards
   */
  for (ix = 0; ix < n->used; ix++) {
    /* mu = ai * m' mod b
     *
     * We avoid a double precision multiplication (which isn't required)
     * by casting the value down to a mp_digit.  Note this requires
     * that W[ix-1] have  the carry cleared (see after the inner loop)
     */
    register mp_digit mu;
    mu = (mp_digit) (((W[ix] & MP_MASK) * rho) & MP_MASK);

    /* a = a + mu * m * b**i
     *
     * This is computed in place and on the fly.  The multiplication
     * by b**i is handled by offseting which columns the results
     * are added to.
     *
     * Note the comba method normally doesn't handle carries in the
     * inner loop In this case we fix the carry from the previous
     * column since the Montgomery reduction requires digits of the
     * result (so far) [see above] to work.  This is
     * handled by fixing up one carry after the inner loop.  The
     * carry fixups are done in order so after these loops the
     * first m->used words of W[] have the carries fixed
     */
    {
      register int iy;
      register mp_digit *tmpn;
      register mp_word *_W;

      /* alias for the digits of the modulus */
      tmpn = n->dp;

      /* Alias for the columns set by an offset of ix */
      _W = W + ix;

      /* inner loop */
      for (iy = 0; iy < n->used; iy++) {
          *_W++ += ((mp_word)mu) * ((mp_word)*tmpn++);
      }
    }

    /* now fix carry for next digit, W[ix+1] */
    W[ix + 1] += W[ix] >> ((mp_word) DIGIT_BIT);
  }

  /* now we have to propagate the carries and
   * shift the words downward [all those least
   * significant digits we zeroed].
   */
  {
    register mp_digit *tmpx;
    register mp_word *_W, *_W1;

    /* nox fix rest of carries */

    /* alias for current word */
    _W1 = W + ix;

    /* alias for next word, where the carry goes */
    _W = W + ++ix;

    for (; ix <= n->used * 2 + 1; ix++) {
      *_W++ += *_W1++ >> ((mp_word) DIGIT_BIT);
    }

    /* copy out, A = A/b**n
     *
     * The result is A/b**n but instead of converting from an
     * array of mp_word to mp_digit than calling mp_rshd
     * we just copy them in the right order
     */

    /* alias for destination word */
    tmpx = x->dp;

    /* alias for shifted double precision result */
    _W = W + n->used;

    for (ix = 0; ix < n->used + 1; ix++) {
      *tmpx++ = (mp_digit)(*_W++ & ((mp_word) MP_MASK));
    }

    /* zero oldused digits, if the input a was larger than
     * m->used+1 we'll have to clear the digits
     */
    for (; ix < olduse; ix++) {
      *tmpx++ = 0;
    }
  }

  /* set the max used and clamp */
  x->used = n->used + 1;
  mp_clamp (x);

  /* if A >= m then A = A - m */
  if (mp_cmp_mag (x, n) != MP_LT) {
    return s_mp_sub (x, n, x);
  }
  return MP_OKAY;
}
예제 #21
0
/* single digit addition */
int
mp_add_d (mp_int * a, mp_digit b, mp_int * c)
{
  int     res, ix, oldused;
  mp_digit *tmpa, *tmpc, mu;

  /* grow c as required */
  if (c->alloc < a->used + 1) {
     if ((res = mp_grow(c, a->used + 1)) != MP_OKAY) {
        return res;
     }
  }

  /* if a is negative and |a| >= b, call c = |a| - b */
  if (a->sign == MP_NEG && (a->used > 1 || a->dp[0] >= b)) {
     /* temporarily fix sign of a */
     a->sign = MP_ZPOS;

     /* c = |a| - b */
     res = mp_sub_d(a, b, c);

     /* fix sign  */
     a->sign = c->sign = MP_NEG;

     /* clamp */
     mp_clamp(c);

     return res;
  }

  /* old number of used digits in c */
  oldused = c->used;

  /* sign always positive */
  c->sign = MP_ZPOS;

  /* source alias */
  tmpa    = a->dp;

  /* destination alias */
  tmpc    = c->dp;

  /* if a is positive */
  if (a->sign == MP_ZPOS) {
     /* add digit, after this we're propagating
      * the carry.
      */
     *tmpc   = *tmpa++ + b;
     mu      = *tmpc >> DIGIT_BIT;
     *tmpc++ &= MP_MASK;

     /* now handle rest of the digits */
     for (ix = 1; ix < a->used; ix++) {
        *tmpc   = *tmpa++ + mu;
        mu      = *tmpc >> DIGIT_BIT;
        *tmpc++ &= MP_MASK;
     }
     /* set final carry */
     ix++;
     *tmpc++  = mu;

     /* setup size */
     c->used = a->used + 1;
  } else {
예제 #22
0
/* Fast (comba) multiplier
 *
 * This is the fast column-array [comba] multiplier.  It is 
 * designed to compute the columns of the product first 
 * then handle the carries afterwards.  This has the effect 
 * of making the nested loops that compute the columns very
 * simple and schedulable on super-scalar processors.
 *
 * This has been modified to produce a variable number of 
 * digits of output so if say only a half-product is required 
 * you don't have to compute the upper half (a feature 
 * required for fast Barrett reduction).
 *
 * Based on Algorithm 14.12 on pp.595 of HAC.
 *
 */
int fast_s_mp_mul_digs(mp_int * a, mp_int * b, mp_int * c, int digs)
{
	int olduse, res, pa, ix;
	extern mp_word *W;

	/* grow the destination as required */
	if (c->alloc < digs) {
		if ((res = mp_grow(c, digs)) != MP_OKAY) {
			return res;
		}
	}

	/* clear temp buf (the columns) */
	memset(W, 0, sizeof(mp_word) * digs);

	/* calculate the columns */
	pa = a->used;
	for (ix = 0; ix < pa; ix++) {
		/* this multiplier has been modified to allow you to 
		 * control how many digits of output are produced.  
		 * So at most we want to make upto "digs" digits of output.
		 *
		 * this adds products to distinct columns (at ix+iy) of W
		 * note that each step through the loop is not dependent on
		 * the previous which means the compiler can easily unroll
		 * the loop without scheduling problems
		 */
		{
			register mp_digit tmpx, *tmpy;
			register mp_word *_W;
			register int iy, pb;

			/* alias for the the word on the left e.g. A[ix] * A[iy] */
			tmpx = a->dp[ix];

			/* alias for the right side */
			tmpy = b->dp;

			/* alias for the columns, each step through the loop adds a new
			   term to each column
			 */
			_W = W + ix;

			/* the number of digits is limited by their placement.  E.g.
			   we avoid multiplying digits that will end up above the # of
			   digits of precision requested
			 */
			pb = MIN(b->used, digs - ix);

			for (iy = 0; iy < pb; iy++) {
				*_W++ +=
				    ((mp_word) tmpx) * ((mp_word) *
							tmpy++);
			}
		}

	}

	/* setup dest */
	olduse = c->used;
	c->used = digs;

	{
		register mp_digit *tmpc;

		/* At this point W[] contains the sums of each column.  To get the
		 * correct result we must take the extra bits from each column and
		 * carry them down
		 *
		 * Note that while this adds extra code to the multiplier it 
		 * saves time since the carry propagation is removed from the 
		 * above nested loop.This has the effect of reducing the work 
		 * from N*(N+N*c)==N**2 + c*N**2 to N**2 + N*c where c is the 
		 * cost of the shifting.  On very small numbers this is slower 
		 * but on most cryptographic size numbers it is faster.
		 *
		 * In this particular implementation we feed the carries from
		 * behind which means when the loop terminates we still have one
		 * last digit to copy
		 */
		tmpc = c->dp;
		for (ix = 1; ix < digs; ix++) {
			/* forward the carry from the previous temp */
			W[ix] += (W[ix - 1] >> ((mp_word) DIGIT_BIT));

			/* now extract the previous digit [below the carry] */
			*tmpc++ =
			    (mp_digit) (W[ix - 1] & ((mp_word) MP_MASK));
		}
		/* fetch the last digit */
		*tmpc++ = (mp_digit) (W[digs - 1] & ((mp_word) MP_MASK));

		/* clear unused digits [that existed in the old copy of c] */
		for (; ix < olduse; ix++) {
			*tmpc++ = 0;
		}
	}
	mp_clamp(c);
	return MP_OKAY;
}
예제 #23
0
파일: mpint.hpp 프로젝트: Fadis/hermit
 void reserve( int size_ ) {
   MPINT_SAFE_CALL( mp_grow( &value, size_ ) );
 }
예제 #24
0
파일: demo.c 프로젝트: mkj/dropbear
int main(void)
{
   unsigned rr;
   int cnt, ix;
#if LTM_DEMO_TEST_VS_MTEST
   unsigned long expt_n, add_n, sub_n, mul_n, div_n, sqr_n, mul2d_n, div2d_n,
      gcd_n, lcm_n, inv_n, div2_n, mul2_n, add_d_n, sub_d_n;
   char* ret;
#else
   unsigned long s, t;
   unsigned long long q, r;
   mp_digit mp;
   int i, n, err, should;
#endif

   if (mp_init_multi(&a, &b, &c, &d, &e, &f, NULL)!= MP_OKAY)
     return EXIT_FAILURE;

   atexit(_cleanup);

#if defined(LTM_DEMO_REAL_RAND)
   if (!fd_urandom) {
      fd_urandom = fopen("/dev/urandom", "r");
      if (!fd_urandom) {
#if !defined(_WIN32)
         fprintf(stderr, "\ncould not open /dev/urandom\n");
#endif
      }
   }
#endif
   srand(LTM_DEMO_RAND_SEED);

#ifdef MP_8BIT
   printf("Digit size 8 Bit \n");
#endif
#ifdef MP_16BIT
   printf("Digit size 16 Bit \n");
#endif
#ifdef MP_32BIT
   printf("Digit size 32 Bit \n");
#endif
#ifdef MP_64BIT
   printf("Digit size 64 Bit \n");
#endif
   printf("Size of mp_digit: %u\n", (unsigned int)sizeof(mp_digit));
   printf("Size of mp_word: %u\n", (unsigned int)sizeof(mp_word));
   printf("DIGIT_BIT: %d\n", DIGIT_BIT);
   printf("MP_PREC: %d\n", MP_PREC);

#if LTM_DEMO_TEST_VS_MTEST == 0
   // trivial stuff
   // a: 0->5
   mp_set_int(&a, 5);
   // a: 5-> b: -5
   mp_neg(&a, &b);
   if (mp_cmp(&a, &b) != MP_GT) {
      return EXIT_FAILURE;
   }
   if (mp_cmp(&b, &a) != MP_LT) {
      return EXIT_FAILURE;
   }
   // a: 5-> a: -5
   mp_neg(&a, &a);
   if (mp_cmp(&b, &a) != MP_EQ) {
      return EXIT_FAILURE;
   }
   // a: -5-> b: 5
   mp_abs(&a, &b);
   if (mp_isneg(&b) != MP_NO) {
      return EXIT_FAILURE;
   }
   // a: -5-> b: -4
   mp_add_d(&a, 1, &b);
   if (mp_isneg(&b) != MP_YES) {
      return EXIT_FAILURE;
   }
   if (mp_get_int(&b) != 4) {
      return EXIT_FAILURE;
   }
   // a: -5-> b: 1
   mp_add_d(&a, 6, &b);
   if (mp_get_int(&b) != 1) {
      return EXIT_FAILURE;
   }
   // a: -5-> a: 1
   mp_add_d(&a, 6, &a);
   if (mp_get_int(&a) != 1) {
      return EXIT_FAILURE;
   }
   mp_zero(&a);
   // a: 0-> a: 6
   mp_add_d(&a, 6, &a);
   if (mp_get_int(&a) != 6) {
      return EXIT_FAILURE;
   }


   mp_set_int(&a, 0);
   mp_set_int(&b, 1);
   if ((err = mp_jacobi(&a, &b, &i)) != MP_OKAY) {
      printf("Failed executing mp_jacobi(0 | 1) %s.\n", mp_error_to_string(err));
      return EXIT_FAILURE;
   }
   if (i != 1) {
      printf("Failed trivial mp_jacobi(0 | 1) %d != 1\n", i);
      return EXIT_FAILURE;
   }
   for (cnt = 0; cnt < (int)(sizeof(jacobi)/sizeof(jacobi[0])); ++cnt) {
      mp_set_int(&b, jacobi[cnt].n);
      /* only test positive values of a */
      for (n = -5; n <= 10; ++n) {
         mp_set_int(&a, abs(n));
         should = MP_OKAY;
         if (n < 0) {
            mp_neg(&a, &a);
            /* Until #44 is fixed the negative a's must fail */
            should = MP_VAL;
         }
         if ((err = mp_jacobi(&a, &b, &i)) != should) {
            printf("Failed executing mp_jacobi(%d | %lu) %s.\n", n, jacobi[cnt].n, mp_error_to_string(err));
            return EXIT_FAILURE;
         }
         if (err == MP_OKAY && i != jacobi[cnt].c[n + 5]) {
            printf("Failed trivial mp_jacobi(%d | %lu) %d != %d\n", n, jacobi[cnt].n, i, jacobi[cnt].c[n + 5]);
            return EXIT_FAILURE;
         }
      }
   }

   // test mp_get_int
   printf("\n\nTesting: mp_get_int");
   for (i = 0; i < 1000; ++i) {
      t = ((unsigned long) rand () * rand () + 1) & 0xFFFFFFFF;
      mp_set_int (&a, t);
      if (t != mp_get_int (&a)) {
         printf ("\nmp_get_int() bad result!");
         return EXIT_FAILURE;
      }
   }
   mp_set_int(&a, 0);
   if (mp_get_int(&a) != 0) {
      printf("\nmp_get_int() bad result!");
      return EXIT_FAILURE;
   }
   mp_set_int(&a, 0xffffffff);
   if (mp_get_int(&a) != 0xffffffff) {
      printf("\nmp_get_int() bad result!");
      return EXIT_FAILURE;
   }

   printf("\n\nTesting: mp_get_long\n");
   for (i = 0; i < (int)(sizeof(unsigned long)*CHAR_BIT) - 1; ++i) {
      t = (1ULL << (i+1)) - 1;
      if (!t)
         t = -1;
      printf(" t = 0x%lx i = %d\r", t, i);
      do {
         if (mp_set_long(&a, t) != MP_OKAY) {
            printf("\nmp_set_long() error!");
            return EXIT_FAILURE;
         }
         s = mp_get_long(&a);
         if (s != t) {
            printf("\nmp_get_long() bad result! 0x%lx != 0x%lx", s, t);
            return EXIT_FAILURE;
         }
         t <<= 1;
      } while(t);
   }

   printf("\n\nTesting: mp_get_long_long\n");
   for (i = 0; i < (int)(sizeof(unsigned long long)*CHAR_BIT) - 1; ++i) {
      r = (1ULL << (i+1)) - 1;
      if (!r)
         r = -1;
      printf(" r = 0x%llx i = %d\r", r, i);
      do {
         if (mp_set_long_long(&a, r) != MP_OKAY) {
            printf("\nmp_set_long_long() error!");
            return EXIT_FAILURE;
         }
         q = mp_get_long_long(&a);
         if (q != r) {
            printf("\nmp_get_long_long() bad result! 0x%llx != 0x%llx", q, r);
            return EXIT_FAILURE;
         }
         r <<= 1;
      } while(r);
   }

   // test mp_sqrt
   printf("\n\nTesting: mp_sqrt\n");
   for (i = 0; i < 1000; ++i) {
      printf ("%6d\r", i);
      fflush (stdout);
      n = (rand () & 15) + 1;
      mp_rand (&a, n);
      if (mp_sqrt (&a, &b) != MP_OKAY) {
         printf ("\nmp_sqrt() error!");
         return EXIT_FAILURE;
      }
      mp_n_root_ex (&a, 2, &c, 0);
      mp_n_root_ex (&a, 2, &d, 1);
      if (mp_cmp_mag (&c, &d) != MP_EQ) {
         printf ("\nmp_n_root_ex() bad result!");
         return EXIT_FAILURE;
      }
      if (mp_cmp_mag (&b, &c) != MP_EQ) {
         printf ("mp_sqrt() bad result!\n");
         return EXIT_FAILURE;
      }
   }

   printf("\n\nTesting: mp_is_square\n");
   for (i = 0; i < 1000; ++i) {
      printf ("%6d\r", i);
      fflush (stdout);

      /* test mp_is_square false negatives */
      n = (rand () & 7) + 1;
      mp_rand (&a, n);
      mp_sqr (&a, &a);
      if (mp_is_square (&a, &n) != MP_OKAY) {
         printf ("\nfn:mp_is_square() error!");
         return EXIT_FAILURE;
      }
      if (n == 0) {
         printf ("\nfn:mp_is_square() bad result!");
         return EXIT_FAILURE;
      }

      /* test for false positives */
      mp_add_d (&a, 1, &a);
      if (mp_is_square (&a, &n) != MP_OKAY) {
         printf ("\nfp:mp_is_square() error!");
         return EXIT_FAILURE;
      }
      if (n == 1) {
         printf ("\nfp:mp_is_square() bad result!");
         return EXIT_FAILURE;
      }

   }
   printf("\n\n");

   // r^2 = n (mod p)
   for (i = 0; i < (int)(sizeof(sqrtmod_prime)/sizeof(sqrtmod_prime[0])); ++i) {
      mp_set_int(&a, sqrtmod_prime[i].p);
      mp_set_int(&b, sqrtmod_prime[i].n);
      if (mp_sqrtmod_prime(&b, &a, &c) != MP_OKAY) {
         printf("Failed executing %d. mp_sqrtmod_prime\n", (i+1));
         return EXIT_FAILURE;
      }
      if (mp_cmp_d(&c, sqrtmod_prime[i].r) != MP_EQ) {
         printf("Failed %d. trivial mp_sqrtmod_prime\n", (i+1));
         ndraw(&c, "r");
         return EXIT_FAILURE;
      }
   }

   /* test for size */
   for (ix = 10; ix < 128; ix++) {
      printf ("Testing (not safe-prime): %9d bits    \r", ix);
      fflush (stdout);
      err = mp_prime_random_ex (&a, 8, ix,
                                (rand () & 1) ? 0 : LTM_PRIME_2MSB_ON, myrng,
                                NULL);
      if (err != MP_OKAY) {
         printf ("failed with err code %d\n", err);
         return EXIT_FAILURE;
      }
      if (mp_count_bits (&a) != ix) {
         printf ("Prime is %d not %d bits!!!\n", mp_count_bits (&a), ix);
         return EXIT_FAILURE;
      }
   }
   printf("\n");

   for (ix = 16; ix < 128; ix++) {
      printf ("Testing (    safe-prime): %9d bits    \r", ix);
      fflush (stdout);
      err = mp_prime_random_ex (
            &a, 8, ix, ((rand () & 1) ? 0 : LTM_PRIME_2MSB_ON) | LTM_PRIME_SAFE,
            myrng, NULL);
      if (err != MP_OKAY) {
         printf ("failed with err code %d\n", err);
         return EXIT_FAILURE;
      }
      if (mp_count_bits (&a) != ix) {
         printf ("Prime is %d not %d bits!!!\n", mp_count_bits (&a), ix);
         return EXIT_FAILURE;
      }
      /* let's see if it's really a safe prime */
      mp_sub_d (&a, 1, &a);
      mp_div_2 (&a, &a);
      mp_prime_is_prime (&a, 8, &cnt);
      if (cnt != MP_YES) {
         printf ("sub is not prime!\n");
         return EXIT_FAILURE;
      }
   }

   printf("\n\n");

   // test montgomery
   printf("Testing: montgomery...\n");
   for (i = 1; i <= 10; i++) {
      if (i == 10)
         i = 1000;
      printf(" digit size: %2d\r", i);
      fflush(stdout);
      for (n = 0; n < 1000; n++) {
         mp_rand(&a, i);
         a.dp[0] |= 1;

         // let's see if R is right
         mp_montgomery_calc_normalization(&b, &a);
         mp_montgomery_setup(&a, &mp);

         // now test a random reduction
         for (ix = 0; ix < 100; ix++) {
             mp_rand(&c, 1 + abs(rand()) % (2*i));
             mp_copy(&c, &d);
             mp_copy(&c, &e);

             mp_mod(&d, &a, &d);
             mp_montgomery_reduce(&c, &a, mp);
             mp_mulmod(&c, &b, &a, &c);

             if (mp_cmp(&c, &d) != MP_EQ) {
printf("d = e mod a, c = e MOD a\n");
mp_todecimal(&a, buf); printf("a = %s\n", buf);
mp_todecimal(&e, buf); printf("e = %s\n", buf);
mp_todecimal(&d, buf); printf("d = %s\n", buf);
mp_todecimal(&c, buf); printf("c = %s\n", buf);
printf("compare no compare!\n"); return EXIT_FAILURE; }
             /* only one big montgomery reduction */
             if (i > 10)
             {
                n = 1000;
                ix = 100;
             }
         }
      }
   }

   printf("\n\n");

   mp_read_radix(&a, "123456", 10);
   mp_toradix_n(&a, buf, 10, 3);
   printf("a == %s\n", buf);
   mp_toradix_n(&a, buf, 10, 4);
   printf("a == %s\n", buf);
   mp_toradix_n(&a, buf, 10, 30);
   printf("a == %s\n", buf);


#if 0
   for (;;) {
      fgets(buf, sizeof(buf), stdin);
      mp_read_radix(&a, buf, 10);
      mp_prime_next_prime(&a, 5, 1);
      mp_toradix(&a, buf, 10);
      printf("%s, %lu\n", buf, a.dp[0] & 3);
   }
#endif

   /* test mp_cnt_lsb */
   printf("\n\nTesting: mp_cnt_lsb");
   mp_set(&a, 1);
   for (ix = 0; ix < 1024; ix++) {
      if (mp_cnt_lsb (&a) != ix) {
         printf ("Failed at %d, %d\n", ix, mp_cnt_lsb (&a));
         return EXIT_FAILURE;
      }
      mp_mul_2 (&a, &a);
   }

/* test mp_reduce_2k */
   printf("\n\nTesting: mp_reduce_2k\n");
   for (cnt = 3; cnt <= 128; ++cnt) {
      mp_digit tmp;

      mp_2expt (&a, cnt);
      mp_sub_d (&a, 2, &a); /* a = 2**cnt - 2 */

      printf ("\r %4d bits", cnt);
      printf ("(%d)", mp_reduce_is_2k (&a));
      mp_reduce_2k_setup (&a, &tmp);
      printf ("(%lu)", (unsigned long) tmp);
      for (ix = 0; ix < 1000; ix++) {
         if (!(ix & 127)) {
            printf (".");
            fflush (stdout);
         }
         mp_rand (&b, (cnt / DIGIT_BIT + 1) * 2);
         mp_copy (&c, &b);
         mp_mod (&c, &a, &c);
         mp_reduce_2k (&b, &a, 2);
         if (mp_cmp (&c, &b)) {
            printf ("FAILED\n");
            return EXIT_FAILURE;
         }
      }
   }

/* test mp_div_3  */
   printf("\n\nTesting: mp_div_3...\n");
   mp_set(&d, 3);
   for (cnt = 0; cnt < 10000;) {
      mp_digit r2;

      if (!(++cnt & 127))
      {
        printf("%9d\r", cnt);
        fflush(stdout);
      }
      mp_rand(&a, abs(rand()) % 128 + 1);
      mp_div(&a, &d, &b, &e);
      mp_div_3(&a, &c, &r2);

      if (mp_cmp(&b, &c) || mp_cmp_d(&e, r2)) {
	 printf("\nmp_div_3 => Failure\n");
      }
   }
   printf("\nPassed div_3 testing");

/* test the DR reduction */
   printf("\n\nTesting: mp_dr_reduce...\n");
   for (cnt = 2; cnt < 32; cnt++) {
      printf ("\r%d digit modulus", cnt);
      mp_grow (&a, cnt);
      mp_zero (&a);
      for (ix = 1; ix < cnt; ix++) {
         a.dp[ix] = MP_MASK;
      }
      a.used = cnt;
      a.dp[0] = 3;

      mp_rand (&b, cnt - 1);
      mp_copy (&b, &c);

      rr = 0;
      do {
         if (!(rr & 127)) {
            printf (".");
            fflush (stdout);
         }
         mp_sqr (&b, &b);
         mp_add_d (&b, 1, &b);
         mp_copy (&b, &c);

         mp_mod (&b, &a, &b);
         mp_dr_setup(&a, &mp),
         mp_dr_reduce (&c, &a, mp);

         if (mp_cmp (&b, &c) != MP_EQ) {
            printf ("Failed on trial %u\n", rr);
            return EXIT_FAILURE;
         }
      } while (++rr < 500);
      printf (" passed");
      fflush (stdout);
   }

#if LTM_DEMO_TEST_REDUCE_2K_L
/* test the mp_reduce_2k_l code */
#if LTM_DEMO_TEST_REDUCE_2K_L == 1
/* first load P with 2^1024 - 0x2A434 B9FDEC95 D8F9D550 FFFFFFFF FFFFFFFF */
   mp_2expt(&a, 1024);
   mp_read_radix(&b, "2A434B9FDEC95D8F9D550FFFFFFFFFFFFFFFF", 16);
   mp_sub(&a, &b, &a);
#elif LTM_DEMO_TEST_REDUCE_2K_L == 2
/*  p = 2^2048 - 0x1 00000000 00000000 00000000 00000000 4945DDBF 8EA2A91D 5776399B B83E188F  */
   mp_2expt(&a, 2048);
   mp_read_radix(&b,
		 "1000000000000000000000000000000004945DDBF8EA2A91D5776399BB83E188F",
		 16);
   mp_sub(&a, &b, &a);
#else
#error oops
#endif

   mp_todecimal(&a, buf);
   printf("\n\np==%s\n", buf);
/* now mp_reduce_is_2k_l() should return */
   if (mp_reduce_is_2k_l(&a) != 1) {
      printf("mp_reduce_is_2k_l() return 0, should be 1\n");
      return EXIT_FAILURE;
   }
   mp_reduce_2k_setup_l(&a, &d);
   /* now do a million square+1 to see if it varies */
   mp_rand(&b, 64);
   mp_mod(&b, &a, &b);
   mp_copy(&b, &c);
   printf("Testing: mp_reduce_2k_l...");
   fflush(stdout);
   for (cnt = 0; cnt < (int)(1UL << 20); cnt++) {
      mp_sqr(&b, &b);
      mp_add_d(&b, 1, &b);
      mp_reduce_2k_l(&b, &a, &d);
      mp_sqr(&c, &c);
      mp_add_d(&c, 1, &c);
      mp_mod(&c, &a, &c);
      if (mp_cmp(&b, &c) != MP_EQ) {
	 printf("mp_reduce_2k_l() failed at step %d\n", cnt);
	 mp_tohex(&b, buf);
	 printf("b == %s\n", buf);
	 mp_tohex(&c, buf);
	 printf("c == %s\n", buf);
	 return EXIT_FAILURE;
      }
   }
   printf("...Passed\n");
#endif /* LTM_DEMO_TEST_REDUCE_2K_L */

#else

   div2_n = mul2_n = inv_n = expt_n = lcm_n = gcd_n = add_n =
      sub_n = mul_n = div_n = sqr_n = mul2d_n = div2d_n = cnt = add_d_n =
      sub_d_n = 0;

   /* force KARA and TOOM to enable despite cutoffs */
   KARATSUBA_SQR_CUTOFF = KARATSUBA_MUL_CUTOFF = 8;
   TOOM_SQR_CUTOFF = TOOM_MUL_CUTOFF = 16;

   for (;;) {
      /* randomly clear and re-init one variable, this has the affect of triming the alloc space */
      switch (abs(rand()) % 7) {
      case 0:
	 mp_clear(&a);
	 mp_init(&a);
	 break;
      case 1:
	 mp_clear(&b);
	 mp_init(&b);
	 break;
      case 2:
	 mp_clear(&c);
	 mp_init(&c);
	 break;
      case 3:
	 mp_clear(&d);
	 mp_init(&d);
	 break;
      case 4:
	 mp_clear(&e);
	 mp_init(&e);
	 break;
      case 5:
	 mp_clear(&f);
	 mp_init(&f);
	 break;
      case 6:
	 break;			/* don't clear any */
      }


      printf
	 ("%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu/%4lu ",
	  add_n, sub_n, mul_n, div_n, sqr_n, mul2d_n, div2d_n, gcd_n, lcm_n,
	  expt_n, inv_n, div2_n, mul2_n, add_d_n, sub_d_n);
      ret=fgets(cmd, 4095, stdin); if(!ret){_panic(__LINE__);}
      cmd[strlen(cmd) - 1] = 0;
      printf("%-6s ]\r", cmd);
      fflush(stdout);
      if (!strcmp(cmd, "mul2d")) {
	 ++mul2d_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 sscanf(buf, "%d", &rr);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);

	 mp_mul_2d(&a, rr, &a);
	 a.sign = b.sign;
	 if (mp_cmp(&a, &b) != MP_EQ) {
	    printf("mul2d failed, rr == %d\n", rr);
	    draw(&a);
	    draw(&b);
	    return EXIT_FAILURE;
	 }
      } else if (!strcmp(cmd, "div2d")) {
	 ++div2d_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 sscanf(buf, "%d", &rr);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);

	 mp_div_2d(&a, rr, &a, &e);
	 a.sign = b.sign;
	 if (a.used == b.used && a.used == 0) {
	    a.sign = b.sign = MP_ZPOS;
	 }
	 if (mp_cmp(&a, &b) != MP_EQ) {
	    printf("div2d failed, rr == %d\n", rr);
	    draw(&a);
	    draw(&b);
	    return EXIT_FAILURE;
	 }
      } else if (!strcmp(cmd, "add")) {
	 ++add_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&c, buf, 64);
	 mp_copy(&a, &d);
	 mp_add(&d, &b, &d);
	 if (mp_cmp(&c, &d) != MP_EQ) {
	    printf("add %lu failure!\n", add_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    return EXIT_FAILURE;
	 }

	 /* test the sign/unsigned storage functions */

	 rr = mp_signed_bin_size(&c);
	 mp_to_signed_bin(&c, (unsigned char *) cmd);
	 memset(cmd + rr, rand() & 255, sizeof(cmd) - rr);
	 mp_read_signed_bin(&d, (unsigned char *) cmd, rr);
	 if (mp_cmp(&c, &d) != MP_EQ) {
	    printf("mp_signed_bin failure!\n");
	    draw(&c);
	    draw(&d);
	    return EXIT_FAILURE;
	 }


	 rr = mp_unsigned_bin_size(&c);
	 mp_to_unsigned_bin(&c, (unsigned char *) cmd);
	 memset(cmd + rr, rand() & 255, sizeof(cmd) - rr);
	 mp_read_unsigned_bin(&d, (unsigned char *) cmd, rr);
	 if (mp_cmp_mag(&c, &d) != MP_EQ) {
	    printf("mp_unsigned_bin failure!\n");
	    draw(&c);
	    draw(&d);
	    return EXIT_FAILURE;
	 }

      } else if (!strcmp(cmd, "sub")) {
	 ++sub_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&c, buf, 64);
	 mp_copy(&a, &d);
	 mp_sub(&d, &b, &d);
	 if (mp_cmp(&c, &d) != MP_EQ) {
	    printf("sub %lu failure!\n", sub_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    return EXIT_FAILURE;
	 }
      } else if (!strcmp(cmd, "mul")) {
	 ++mul_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&c, buf, 64);
	 mp_copy(&a, &d);
	 mp_mul(&d, &b, &d);
	 if (mp_cmp(&c, &d) != MP_EQ) {
	    printf("mul %lu failure!\n", mul_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    return EXIT_FAILURE;
	 }
      } else if (!strcmp(cmd, "div")) {
	 ++div_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&c, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&d, buf, 64);

	 mp_div(&a, &b, &e, &f);
	 if (mp_cmp(&c, &e) != MP_EQ || mp_cmp(&d, &f) != MP_EQ) {
	    printf("div %lu %d, %d, failure!\n", div_n, mp_cmp(&c, &e),
		   mp_cmp(&d, &f));
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    draw(&e);
	    draw(&f);
	    return EXIT_FAILURE;
	 }

      } else if (!strcmp(cmd, "sqr")) {
	 ++sqr_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 mp_copy(&a, &c);
	 mp_sqr(&c, &c);
	 if (mp_cmp(&b, &c) != MP_EQ) {
	    printf("sqr %lu failure!\n", sqr_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    return EXIT_FAILURE;
	 }
      } else if (!strcmp(cmd, "gcd")) {
	 ++gcd_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&c, buf, 64);
	 mp_copy(&a, &d);
	 mp_gcd(&d, &b, &d);
	 d.sign = c.sign;
	 if (mp_cmp(&c, &d) != MP_EQ) {
	    printf("gcd %lu failure!\n", gcd_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    return EXIT_FAILURE;
	 }
      } else if (!strcmp(cmd, "lcm")) {
	 ++lcm_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&c, buf, 64);
	 mp_copy(&a, &d);
	 mp_lcm(&d, &b, &d);
	 d.sign = c.sign;
	 if (mp_cmp(&c, &d) != MP_EQ) {
	    printf("lcm %lu failure!\n", lcm_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    return EXIT_FAILURE;
	 }
      } else if (!strcmp(cmd, "expt")) {
	 ++expt_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&c, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&d, buf, 64);
	 mp_copy(&a, &e);
	 mp_exptmod(&e, &b, &c, &e);
	 if (mp_cmp(&d, &e) != MP_EQ) {
	    printf("expt %lu failure!\n", expt_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    draw(&e);
	    return EXIT_FAILURE;
	 }
      } else if (!strcmp(cmd, "invmod")) {
	 ++inv_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&c, buf, 64);
	 mp_invmod(&a, &b, &d);
	 mp_mulmod(&d, &a, &b, &e);
	 if (mp_cmp_d(&e, 1) != MP_EQ) {
	    printf("inv [wrong value from MPI?!] failure\n");
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    draw(&d);
	    draw(&e);
	    mp_gcd(&a, &b, &e);
	    draw(&e);
	    return EXIT_FAILURE;
	 }

      } else if (!strcmp(cmd, "div2")) {
	 ++div2_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 mp_div_2(&a, &c);
	 if (mp_cmp(&c, &b) != MP_EQ) {
	    printf("div_2 %lu failure\n", div2_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    return EXIT_FAILURE;
	 }
      } else if (!strcmp(cmd, "mul2")) {
	 ++mul2_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 mp_mul_2(&a, &c);
	 if (mp_cmp(&c, &b) != MP_EQ) {
	    printf("mul_2 %lu failure\n", mul2_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    return EXIT_FAILURE;
	 }
      } else if (!strcmp(cmd, "add_d")) {
	 ++add_d_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 sscanf(buf, "%d", &ix);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 mp_add_d(&a, ix, &c);
	 if (mp_cmp(&b, &c) != MP_EQ) {
	    printf("add_d %lu failure\n", add_d_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    printf("d == %d\n", ix);
	    return EXIT_FAILURE;
	 }
      } else if (!strcmp(cmd, "sub_d")) {
	 ++sub_d_n;
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&a, buf, 64);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 sscanf(buf, "%d", &ix);
	 ret=fgets(buf, 4095, stdin); if(!ret){_panic(__LINE__);}
	 mp_read_radix(&b, buf, 64);
	 mp_sub_d(&a, ix, &c);
	 if (mp_cmp(&b, &c) != MP_EQ) {
	    printf("sub_d %lu failure\n", sub_d_n);
	    draw(&a);
	    draw(&b);
	    draw(&c);
	    printf("d == %d\n", ix);
	    return EXIT_FAILURE;
	 }
      } else if (!strcmp(cmd, "exit")) {
         printf("\nokay, exiting now\n");
         break;
      }
   }
#endif
   return 0;
}