#include <stdlib.h>
#include <stdint.h>
#include <string.h>
#include <stdio.h>
#include <assert.h>

/*
This code implements  base ^ exp mod p  with base being a small integer,
g given in g_prime and the r number of bits (4096) being used to pre-calculate
g_r_square = r^2 mod p in order to speed up montgomery reduction

Because our primes conveniently end in 0xf..f for word sizes up to 64, the
multiplicative inverse is ("pre-calculated") g_r_inverse = 1.

Todo: check if little endian internal representation is more efficient, also
write an import/export function for the exponent/result from big endian uint8_t
representation which is the default transport representation.

Currently, with gcc -O3, we measure at around 2,75x the time libgmp needs for
it's powm which is not so bad, considering we don't use any assembler optimization.

Hot spot function is mp_mul_uint_add.

When we're done measuring, we should set PREVENT_TIMING_ATTACKS so that powm()
does not leak information about the number of bits set, but on average doubling the
time needed. Discuss ;)
*/

#if 0
typedef uint32_t leg_t;
typedef uint64_t dleg_t;
#else
typedef uint64_t leg_t;
typedef unsigned int uint128_t __attribute__((mode(TI)));
typedef uint128_t dleg_t;
#endif

//#define WITH_MEASURE_GMP
//#define WITH_PREVENT_TIMING_ATTACKS
//#define WITH_TESTS
//#define WITHOUT_UNROLLING
#define WITH_ROUNDS 128

#ifdef WITH_TESTS
static int run_tests( );
#define KARATSUBA_THRESHOLD 4
#else
#define KARATSUBA_THRESHOLD 16
#endif

#ifdef WITH_MEASURE_GMP
#include <gmp.h>
#endif

#ifdef DEBUG_FUNCTIONS
static void dump_int( int level, char * prefix, leg_t const * p, int l, int nl )
{
  printf( "L%d %s: ", level, prefix );
  while( l-- )
   printf( "%08llX", p[l] );
  if( nl )
    putchar( 10 );
}
#endif

/* Test values for smaller legs sizes */
#if 0
static leg_t g_prime[]     = { 4293918689 };
static leg_t g_r_square[]  = {  333456065 };
static leg_t g_r_inverse   =   4192173023;
#endif

#if 0
static leg_t g_prime[]     = { 0xFFFFFFFF, 0xFFFFFFA1 };
/* g_prime == 18446744073709551521 */
/* g_r     == 18446744073709551616 */
/* g_r^2   == 340282366920938463463374607431768211456 */
static leg_t g_r_square[]  = { 0, 9025 };
/* g_prime^-1 mod g_r = 12815632724892951649 */
static leg_t g_r_inverse   = 3571604383;
#endif

#if 0
static leg_t g_prime[]     = { 0xffffffff, 0xffffffff, 0xffffffff, 0xffffff61 };
/* g_prime == 340282366920938463463374607431768211297 */
/* g_r     == 340282366920938463463374607431768211456 */
/* g_r^2   == 115792089237316195423570985008687907853269984665640564039457584007913129639936 */
static leg_t g_r_square[]  = { 0, 0, 0, 25281 };
/* g_prime^-1 mod g_r = 104866892950477891256008526818595234977 */
static leg_t g_r_inverse   = 783358815;
#endif

#if 0
static leg_t g_prime[]     = { 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0x61 };
/* g_prime == 340282366920938463463374607431768211297 */
/* g_r     == 340282366920938463463374607431768211456 */
/* g_r^2   == 115792089237316195423570985008687907853269984665640564039457584007913129639936 */
static leg_t g_r_square[]  = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0x62, 0xc1 };
/* g_prime^-1 mod g_r = 104866892950477891256008526818595234977 */
static leg_t g_r_inverse   = 0x5f;
#endif

#if 0
/* Standard prime for leg_t defined as uint32_t */
static const leg_t g_prime[] = {
    0xFFFFFFFF, 0xFFFFFFFF, 0xC90FDAA2, 0x2168C234, 0xC4C6628B, 0x80DC1CD1, 0x29024E08, 0x8A67CC74,
    0x020BBEA6, 0x3B139B22, 0x514A0879, 0x8E3404DD, 0xEF9519B3, 0xCD3A431B, 0x302B0A6D, 0xF25F1437,
    0x4FE1356D, 0x6D51C245, 0xE485B576, 0x625E7EC6, 0xF44C42E9, 0xA637ED6B, 0x0BFF5CB6, 0xF406B7ED,
    0xEE386BFB, 0x5A899FA5, 0xAE9F2411, 0x7C4B1FE6, 0x49286651, 0xECE45B3D, 0xC2007CB8, 0xA163BF05,
    0x98DA4836, 0x1C55D39A, 0x69163FA8, 0xFD24CF5F, 0x83655D23, 0xDCA3AD96, 0x1C62F356, 0x208552BB,
    0x9ED52907, 0x7096966D, 0x670C354E, 0x4ABC9804, 0xF1746C08, 0xCA18217C, 0x32905E46, 0x2E36CE3B,
    0xE39E772C, 0x180E8603, 0x9B2783A2, 0xEC07A28F, 0xB5C55DF0, 0x6F4C52C9, 0xDE2BCBF6, 0x95581718,
    0x3995497C, 0xEA956AE5, 0x15D22618, 0x98FA0510, 0x15728E5A, 0x8AAAC42D, 0xAD33170D, 0x04507A33,
    0xA85521AB, 0xDF1CBA64, 0xECFB8504, 0x58DBEF0A, 0x8AEA7157, 0x5D060C7D, 0xB3970F85, 0xA6E1E4C7,
    0xABF5AE8C, 0xDB0933D7, 0x1E8C94E0, 0x4A25619D, 0xCEE3D226, 0x1AD2EE6B, 0xF12FFA06, 0xD98A0864,
    0xD8760273, 0x3EC86A64, 0x521F2B18, 0x177B200C, 0xBBE11757, 0x7A615D6C, 0x770988C0, 0xBAD946E2,
    0x08E24FA0, 0x74E5AB31, 0x43DB5BFC, 0xE0FD108E, 0x4B82D120, 0xA9210801, 0x1A723C12, 0xA787E6D7,
    0x88719A10, 0xBDBA5B26, 0x99C32718, 0x6AF4E23C, 0x1A946834, 0xB6150BDA, 0x2583E9CA, 0x2AD44CE8,
    0xDBBBC2DB, 0x04DE8EF9, 0x2E8EFC14, 0x1FBECAA6, 0x287C5947, 0x4E6BC05D, 0x99B2964F, 0xA090C3A2,
    0x233BA186, 0x515BE7ED, 0x1F612970, 0xCEE2D7AF, 0xB81BDD76, 0x2170481C, 0xD0069127, 0xD5B05AA9,
    0x93B4EA98, 0x8D8FDDC1, 0x86FFB7DC, 0x90A6C08F, 0x4DF435C9, 0x34063199, 0xFFFFFFFF, 0xFFFFFFFF
};

static const leg_t g_r_square[] = {
    0x3DA97659, 0xE280DB0B, 0xE65BCC3A, 0xB78FDAA9, 0xB7B768C8, 0x9931D78D, 0xF8B11725, 0x339EBC93,
    0xAA7FBD95, 0x62059F1F, 0x4C2DE67D, 0xAD47527E, 0x526A653A, 0x7A674BD5, 0x5401EA4F, 0x3ED73A2F,
    0x70B56F52, 0x7F6F604F, 0xF3E56CC2, 0xBD9F048C, 0xFEA80D9A, 0x6EC9FCD3, 0x3AD36FD8, 0x22C39F34,
    0x91F30C52, 0xF798DA6A, 0x18C3DCE2, 0x0188D84C, 0xCB225176, 0x259E080F, 0xA89D1DCD, 0x9D381CC5,
    0xBEACD46F, 0x3CDD1196, 0x91A4F557, 0x2929B90C, 0x67DE8FA0, 0x23864714, 0xC28A61D4, 0x7411402D,
    0x6C09060D, 0x41058639, 0xE4FCCF1D, 0x638F4566, 0x9E10FDE2, 0x8E54806B, 0xF1D27D0B, 0x5C7DC9C2,
    0xB616D6FA, 0x8BE2C91D, 0xE8105464, 0xE9F80A5F, 0x34720869, 0x90DACF1A, 0x7E2C75A5, 0x8E25F142,
    0x8FB4832E, 0xF827DE84, 0xCA06DA91, 0xC2B3E7E2, 0x4F024193, 0x787A8278, 0x8BD70562, 0xDA60E392,
    0x46BDB733, 0x6E8452D9, 0x1D7D37A2, 0x3FB8CF61, 0xB18A9EF1, 0x50C8953A, 0xC93919D1, 0x2A4B1A67,
    0xFDC65A26, 0x9B51C1EF, 0x09954843, 0x20E739F4, 0x6C7951A5, 0x23CEF785, 0xE60C6EFD, 0xFFB7A9A9,
    0x5F666146, 0xCB441F59, 0xAE01E0F3, 0x63A9315D, 0x04BA044A, 0xEB4EEFD4, 0x8563215F, 0x72C8D989,
    0x62D21877, 0x1296EF6A, 0x20BD72B9, 0xB21E6B3D, 0x53C44FAB, 0x734810F7, 0xD203A9E0, 0xD7CE25D0,
    0x2E52989E, 0xCCF85F34, 0x912A0491, 0x3E9EBD87, 0x8267537D, 0x4A612A18, 0x51E75D99, 0x98F001DB,
    0xC9C77F0C, 0x352D408C, 0xA796D182, 0x04A636F7, 0xE4404092, 0x1C1E467C, 0x524E7C7A, 0x7ED36C41,
    0x2A434CEB, 0x230B2DFE, 0x3549C577, 0x7A17FB04, 0xB850DE95, 0xD97AC40A, 0x55EA6F75, 0x41C4F82B,
    0x37BF90FE, 0x52074F19, 0xFA8F75F0, 0x067E82B1, 0x8A1AC024, 0xB30E9B12, 0xC14AB0DD, 0xCC03AA20
};

static const leg_t g_r_inverse = 1;
*/
#endif

#if 1
/* Standard prime for leg_t defined as uint64_t */
static const leg_t g_prime[] = {
    0xFFFFFFFFFFFFFFFF, 0x4DF435C934063199, 0x86FFB7DC90A6C08F, 0x93B4EA988D8FDDC1,
    0xD0069127D5B05AA9, 0xB81BDD762170481C, 0x1F612970CEE2D7AF, 0x233BA186515BE7ED,
    0x99B2964FA090C3A2, 0x287C59474E6BC05D, 0x2E8EFC141FBECAA6, 0xDBBBC2DB04DE8EF9,
    0x2583E9CA2AD44CE8, 0x1A946834B6150BDA, 0x99C327186AF4E23C, 0x88719A10BDBA5B26,
    0x1A723C12A787E6D7, 0x4B82D120A9210801, 0x43DB5BFCE0FD108E, 0x08E24FA074E5AB31,
    0x770988C0BAD946E2, 0xBBE117577A615D6C, 0x521F2B18177B200C, 0xD87602733EC86A64,
    0xF12FFA06D98A0864, 0xCEE3D2261AD2EE6B, 0x1E8C94E04A25619D, 0xABF5AE8CDB0933D7,
    0xB3970F85A6E1E4C7, 0x8AEA71575D060C7D, 0xECFB850458DBEF0A, 0xA85521ABDF1CBA64,
    0xAD33170D04507A33, 0x15728E5A8AAAC42D, 0x15D2261898FA0510, 0x3995497CEA956AE5,
    0xDE2BCBF695581718, 0xB5C55DF06F4C52C9, 0x9B2783A2EC07A28F, 0xE39E772C180E8603,
    0x32905E462E36CE3B, 0xF1746C08CA18217C, 0x670C354E4ABC9804, 0x9ED529077096966D,
    0x1C62F356208552BB, 0x83655D23DCA3AD96, 0x69163FA8FD24CF5F, 0x98DA48361C55D39A,
    0xC2007CB8A163BF05, 0x49286651ECE45B3D, 0xAE9F24117C4B1FE6, 0xEE386BFB5A899FA5,
    0x0BFF5CB6F406B7ED, 0xF44C42E9A637ED6B, 0xE485B576625E7EC6, 0x4FE1356D6D51C245,
    0x302B0A6DF25F1437, 0xEF9519B3CD3A431B, 0x514A08798E3404DD, 0x020BBEA63B139B22,
    0x29024E088A67CC74, 0xC4C6628B80DC1CD1, 0xC90FDAA22168C234, 0xFFFFFFFFFFFFFFFF
  };

static const leg_t g_r_square[] = {
    0xC14AB0DDCC03AA20, 0x8A1AC024B30E9B12, 0xFA8F75F0067E82B1, 0x37BF90FE52074F19,
    0x55EA6F7541C4F82B, 0xB850DE95D97AC40A, 0x3549C5777A17FB04, 0x2A434CEB230B2DFE,
    0x524E7C7A7ED36C41, 0xE44040921C1E467C, 0xA796D18204A636F7, 0xC9C77F0C352D408C,
    0x51E75D9998F001DB, 0x8267537D4A612A18, 0x912A04913E9EBD87, 0x2E52989ECCF85F34,
    0xD203A9E0D7CE25D0, 0x53C44FAB734810F7, 0x20BD72B9B21E6B3D, 0x62D218771296EF6A,
    0x8563215F72C8D989, 0x04BA044AEB4EEFD4, 0xAE01E0F363A9315D, 0x5F666146CB441F59,
    0xE60C6EFDFFB7A9A9, 0x6C7951A523CEF785, 0x0995484320E739F4, 0xFDC65A269B51C1EF,
    0xC93919D12A4B1A67, 0xB18A9EF150C8953A, 0x1D7D37A23FB8CF61, 0x46BDB7336E8452D9,
    0x8BD70562DA60E392, 0x4F024193787A8278, 0xCA06DA91C2B3E7E2, 0x8FB4832EF827DE84,
    0x7E2C75A58E25F142, 0x3472086990DACF1A, 0xE8105464E9F80A5F, 0xB616D6FA8BE2C91D,
    0xF1D27D0B5C7DC9C2, 0x9E10FDE28E54806B, 0xE4FCCF1D638F4566, 0x6C09060D41058639,
    0xC28A61D47411402D, 0x67DE8FA023864714, 0x91A4F5572929B90C, 0xBEACD46F3CDD1196,
    0xA89D1DCD9D381CC5, 0xCB225176259E080F, 0x18C3DCE20188D84C, 0x91F30C52F798DA6A,
    0x3AD36FD822C39F34, 0xFEA80D9A6EC9FCD3, 0xF3E56CC2BD9F048C, 0x70B56F527F6F604F,
    0x5401EA4F3ED73A2F, 0x526A653A7A674BD5, 0x4C2DE67DAD47527E, 0xAA7FBD9562059F1F,
    0xF8B11725339EBC93, 0xB7B768C89931D78D, 0xE65BCC3AB78FDAA9, 0x3DA97659E280DB0B
};

static const leg_t g_r_inverse = 1;
#endif

/* Returns 0 if a and b are equal, -1 if a < b, 1 if a > b */
static int mp_cmp( leg_t const *a, leg_t const *b, int legs )
{
  while( legs-- )
  {
    if( a[legs] < b[legs] ) return -1;
    if( a[legs] > b[legs] ) return  1;
  }
  return 0;
}

/* Subtract b from a, store in a. Expects enough words prepended
   to borrow from */
static void mp_sub( leg_t *a, leg_t const *b, int legs )
{
  int borrow = 0, borrow_temp;
  while( legs-- )
  {
    leg_t temp = *a - *(b++);
    borrow_temp = temp > *a;
    *a = temp - borrow;
    borrow = borrow_temp | ( *(a++) > temp );
  }
  while( borrow )
  {
    leg_t temp = *a - borrow;
    borrow = temp > *a;
    *(a++) = temp;
  }
}

/* Add b to a, store in a. Operates on legs + flegs words, with
   flegs the amount of legs to propagate the carry to */
static void mp_add( leg_t *a, leg_t const *b, int legs, int flegs )
{
  dleg_t acc = 0;
  while( legs-- )
  {
    acc += (dleg_t)*a + (dleg_t)*(b++);
    *(a++) = (leg_t)acc;
    acc >>= 8*sizeof(leg_t);
  }
  while( acc && flegs-- )
  {
    acc += (dleg_t)*a;
    *(a++) = (leg_t)acc;
    acc >>= 8*sizeof(leg_t);
  }
}

/* Subtract b from a and store in result. Expects nothing to borrow.*/
static void mp_sub_mod( leg_t * result, leg_t const *a, leg_t const * b, int legs )
{
  int borrow = 0, borrow_temp;
  while( legs-- )
  {
    leg_t temp = *a - *(b++);
    borrow_temp = temp > *(a++);
    *result = temp - borrow;
    borrow = borrow_temp | (*(result++) > temp );
  }
}

/* Fast negate */
static void mp_negate( leg_t * p, int legs )
{
  /* Only as long as we find 0 along the way, we need to
     propagate the carry of -x == !x+1 */
  int carry = 1;
  while( legs-- )
  {
    leg_t v = carry + ( *p ^ (leg_t)-1 );
    if(v) carry = 0;
    *(p++)=v;
  }
}

/* Multiplies a with fac, adds to result.
   result is guaranteed to be initialized with enough legs prepended to take the carry */
static void mp_mul_uint_add( leg_t *result, leg_t const *a, leg_t fac, int legs )
{
  dleg_t acc8 = 0;
#ifndef WITHOUT_UNROLLING
  dleg_t acc1 = 0, acc2 = 0, acc3 = 0, acc4 = 0;
  dleg_t acc5 = 0, acc6 = 0, acc7 = 0;

  while( legs >= 8 )
  {
    acc1 = ( acc8 >> 8*sizeof(leg_t) ) + (dleg_t)*result + (dleg_t)*(a++) * (dleg_t)fac;
    *(result++) = (leg_t)acc1;

    acc2 = ( acc1 >> 8*sizeof(leg_t) ) + (dleg_t)*result + (dleg_t)*(a++) * (dleg_t)fac;
    *(result++) = (leg_t)acc2;

    acc3 = ( acc2 >> 8*sizeof(leg_t) ) + (dleg_t)*result + (dleg_t)*(a++) * (dleg_t)fac;
    *(result++) = (leg_t)acc3;

    acc4 = ( acc3 >> 8*sizeof(leg_t) ) + (dleg_t)*result + (dleg_t)*(a++) * (dleg_t)fac;
    *(result++) = (leg_t)acc4;

    acc5 = ( acc4 >> 8*sizeof(leg_t) ) + (dleg_t)*result + (dleg_t)*(a++) * (dleg_t)fac;
    *(result++) = (leg_t)acc5;

    acc6 = ( acc5 >> 8*sizeof(leg_t) ) + (dleg_t)*result + (dleg_t)*(a++) * (dleg_t)fac;
    *(result++) = (leg_t)acc6;

    acc7 = ( acc6 >> 8*sizeof(leg_t) ) + (dleg_t)*result + (dleg_t)*(a++) * (dleg_t)fac;
    *(result++) = (leg_t)acc7;

    acc8 = ( acc7 >> 8*sizeof(leg_t) ) + (dleg_t)*result + (dleg_t)*(a++) * (dleg_t)fac;
    *(result++) = (leg_t)acc8;

    legs -= 8;
  }
  acc8 >>= 8*sizeof(leg_t);
#endif
  while( legs-- )
  {
    acc8 += (dleg_t)*result + (dleg_t)*(a++) * (dleg_t)fac;
    *(result++) = (leg_t)acc8;
    acc8 >>= 8*sizeof(leg_t);
  }
  while( acc8 )
  {
    acc8 += (dleg_t)*result;
    *(result++) = (leg_t)acc8;
    acc8 >>= 8*sizeof(leg_t);
  }
}

/* Multiplies a and b, adds to result, base case version. */
static void mp_mul_oper_add( leg_t * result, leg_t const *a, leg_t const *b, int legs )
{
  int leg = legs;
  while( leg-- )
    mp_mul_uint_add( result++, a, *(b++), legs );
}

/* Optimized mp_mul_oper_add for a == b, i.e. squaring */
static void mp_sqr( leg_t *result, leg_t const * a, int legs )
{
  while( legs-- ) {
    leg_t fac = *(a++), *dest = result;
    leg_t const *src = a;
    int leg = legs;

    dleg_t acc = (dleg_t)*dest + (dleg_t)fac * (dleg_t)fac;

    *(dest++) = (leg_t)acc;
    acc >>= 8*sizeof(leg_t);

    while( leg-- )
    {
      dleg_t subresult = (dleg_t)fac * (dleg_t)*(src++);
      int carry = !!( subresult >> (16*sizeof(leg_t)-1));

      acc += 2 * subresult + (dleg_t)*dest;
      *(dest++) = (leg_t)acc;

      acc >>= 8*sizeof(leg_t);
      acc += (dleg_t)carry << 8*sizeof(leg_t);
    }

    while( acc )
    {
      acc += (dleg_t)*dest;
      *(dest++) = (leg_t)acc;
      acc >>= 8*sizeof(leg_t);
    }
    result += 2;
  }
}

/* Optimized karatsuba (toom2.2) for a == b, i.e. squaring */
static void mp_mul_kara_square( leg_t* p, leg_t const *a, int len, leg_t *scratch )
{
  memset( p, 0, 2 * len * sizeof( leg_t ));
  if( len <= KARATSUBA_THRESHOLD )
     mp_sqr( p, a, len );
  else
  {
    int n = len / 2;

    if( mp_cmp( a, a + n, n ) > 0 )
      mp_sub_mod( scratch, a, a + n, n );
    else
      mp_sub_mod( scratch, a + n, a, n );

    mp_mul_kara_square( p + n, scratch, n, scratch + len );
    mp_negate( p + n, len + n );

    mp_mul_kara_square( scratch, a + n, n, scratch + len );
    mp_add( p + len, scratch, len, 0 );
    mp_add( p + n, scratch, len, n );

    mp_mul_kara_square( scratch, a, n, scratch + len );
    mp_add( p + n, scratch, len, n );
    mp_add( p, scratch, len, len );
  }
}

/* karatsuba (toom2.2), generic */
static void mp_mul_kara( leg_t* p, leg_t const *a, leg_t const *b, int len, leg_t *scratch )
{
  memset( p, 0, 2 * len * sizeof( leg_t ));
  if( len <= KARATSUBA_THRESHOLD )
    mp_mul_oper_add( p, a, b, len );
  else
  {
    int sign = 0, n = len / 2;

    if( mp_cmp( a, a + n, n ) > 0 )
      mp_sub_mod( scratch, a, a + n, n );
    else
    {
      mp_sub_mod( scratch, a + n, a, n );
      sign = 1;
    }

    if( mp_cmp( b, b + n, n ) > 0 )
      mp_sub_mod( scratch + n, b, b + n, n );
    else
    {
      mp_sub_mod( scratch + n, b + n, b, n );
      sign ^= 1;
    }

    mp_mul_kara( p + n, scratch, scratch + n, n, scratch + len );
    if( !sign )
      mp_negate( p + n, len + n );

    mp_mul_kara( scratch, a + n, b + n, n, scratch + len );
    mp_add( p + len, scratch, len, 0 );
    mp_add( p + n, scratch, len, n );

    mp_mul_kara( scratch, a, b, n, scratch + len );
    mp_add( p + n, scratch, len, n );
    mp_add( p, scratch, len, len );
  }
}

/* Multiply a and b, store in a, work in montgomery domain
   to achieve multiply, a needs to be reduced by k * g_prime
   until all lower legs are 0, allowing exact division by 2^r

   if !do_mul, we convert from montgomery domain back to
   g_prime domain and thus only multiply by 1 before reducing
 */
static void redc( leg_t *a, leg_t const *b, int legs, int do_mul )
{
  leg_t scratch[ 2 * ( legs - KARATSUBA_THRESHOLD ) ];
  leg_t temp[1 + 2 * legs];
  leg_t leg;

  /* Not necessary for transforming back */
  if( do_mul )
  {
    temp[2*legs] = 0;
    if( a == b )
        mp_mul_kara_square( temp, a, legs, scratch );
    else
        mp_mul_kara( temp, a, b, legs, scratch );
  }
  else
  {
    memset( temp + legs, 0, (legs + 1 ) * sizeof(leg_t));
    memcpy( temp, a, legs * sizeof(leg_t));
  }

  /* m = p * ( m * R_1 ) % R */
  for( leg = 0; leg < legs; ++leg )
  {
    leg_t k = temp[leg] * g_r_inverse;
    mp_mul_uint_add( temp+leg, g_prime, k, legs );
  }

  /* the lower legs of temp are now zero
     if necessary, reduce temp to fit in legs */
  if( temp[2*legs] || mp_cmp( temp + legs, g_prime, legs ) > 0 )
    mp_sub( temp + legs, g_prime, legs );

  memcpy( a, temp + legs, legs * sizeof(leg_t) );
  memset( scratch, 0, sizeof( scratch ) );
  memset( temp, 0, sizeof( temp ) );
}

/* calculate base ^ exponent modulo g_prime */
static void powm( leg_t * result, leg_t const *exponent, leg_t base, int legs )
{
  leg_t acc[legs];
#ifdef WITH_PREVENT_TIMING_ATTACKS
  leg_t dummy[legs];
#endif
  int first = 0, bit;

  memset( acc, 0, sizeof(acc) );
  *acc = base;

  /* Transform base into montgomery domain */
  redc( acc, g_r_square, legs, 1 );

  /* mul in temp and if bit set in exponent, multiply into accumulator */
  for( bit = 0; bit < legs * 8 * sizeof(leg_t); bit++ )
  {
    int this_bit = bit % ( sizeof(leg_t) * 8 );
    if( ( exponent[ bit / ( sizeof(leg_t) * 8 ) ] >> this_bit ) & 1 ) {
      if( first++ )
        redc( result, acc, legs, 1 );
      else
        memcpy( result, acc, sizeof(leg_t) * legs );
    }
#ifdef WITH_PREVENT_TIMING_ATTACKS
    else
      redc( dummy, acc, legs, 1 );
#endif
    redc( acc, acc, legs, 1 );
  }

  /* Transform result in acc back into mod p domain */
  if( first )
    redc( result, 0, legs, 0 );
  else /* base ^ 0 mod p = 1 */
  {
    memset( result+1, 0, legs * sizeof(leg_t) );
    *result = 1;
  }
}

static void mp_import( leg_t *dest, uint8_t const * src, size_t len )
{
  int byte = 0;
  leg_t acc = 0;
  while( len-- )
  {
    acc |= ((leg_t)src[len]) << ( 8 * byte );
    if( ++byte == sizeof(leg_t))
    {
      *(dest++) = acc;
      acc = 0;
      byte = 0;
    }
  }
  if( byte )
    *dest = acc;
}

int main()
{
  int i, legs = 64;
  leg_t input[ /*legs*/ ] = {
      0x281b04b30bb62984, 0xe0cc506399225574, 0xb08ad95f8040183f, 0x56eee1ba0dc1936c,
      0x5ec1d62932d6937a, 0xfa8718685a5b8359, 0xfb4e8dadb3111df4, 0x87e90f5b6397727e,
      0x6b984d2e7b9cc244, 0x9ae478fd69254dc9, 0x21da1ef76099ed94, 0xe407bde8519bc534,
      0x051a88159adc842b, 0x547fd48d3b471c7c, 0xd6b85c29e8e8a528, 0x74fc0b5dcdfaa679,
      0xe0c022a74c073360, 0xc0dc8c759af7492b, 0xa86946df6abbea0f, 0xfcc2a7f6d4b0d4f7,
      0xd2111f6ce69c2bc6, 0x4ab8b5e0adaad6a0, 0xf8df973a9842e7b2, 0x2d44f438b889673c,
      0x82b4e75bf0e60d06, 0x4c3b1333a07fee24, 0x353eb64e067f86cc, 0x518781609b447bf8,
      0xe9228ae741da24cd, 0xd8aec44d430026aa, 0x196bcb0cc0e353f8, 0x942fad7b2cc3dcc5,
      0xa77b229138f2d5d1, 0x5479eaeb9b64ba87, 0x70cc7ff6ea8f1b51, 0x1785bbfb88755c99,
      0xd7ac4896d4e08b35, 0x0e2aab629710919b, 0xa3612061c92319ce, 0xb5444312b3b43541,
      0xa13b801d1ac631d5, 0x87e0db1397d9f2fd, 0x0e84ab94e744b8ba, 0x1f61b1ea0e5aafb4,
      0x70f0c055dde43121, 0x094d72c301b8f25d, 0xd151e2770baec00b, 0x31052a6fc332563d,
      0x2bca50f311573101, 0xb420bd7e538ef88b, 0xac5ecd1fe4997513, 0x3ca892560247ca32,
      0x950e895d7777d182, 0xcb650843a43a4a00, 0xd07dc909e7070b0e, 0x72d720399b4ad2d4,
      0x52292d56ea2960a9, 0x6060e2cd96000219, 0x118727b5eb17dfbd, 0xa6d78571af8d3688,
      0xb563a9cf79c6eaec, 0xe5d85f8fb8fced5c, 0x9ac2961fe160f837, 0x1de7eae6c0d6a0b0,
 };

  leg_t res[legs]; /*
= {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}
*/

#ifdef WITH_TESTS
  run_tests();
#endif

#ifdef WITH_MEASURE_GMP
  mpz_t m, base, exp, result;

  mpz_init( m ); mpz_init( base ); mpz_init( exp ); mpz_init( result );
  mpz_import( m, sizeof(g_prime) / sizeof(uint64_t), -1, sizeof(uint64_t), 0, 0, g_prime );
  mpz_set_ui( base, 2 );
  mpz_import( exp, sizeof(input) / sizeof(uint64_t), -1, sizeof(uint64_t), 0, 0, input );

  for( i=0; i< WITH_ROUNDS; ++i )
    mpz_powm( result, base, exp, m );

  gmp_printf("%ZX\n", result);
#else

  for( i=0; i<WITH_ROUNDS; ++i )
    powm( res, input, 2, legs );

  for( i=legs-1; i>=0; --i )
    printf( "%016llX", res[i] );
  putchar(10); putchar(10);
#endif

  return 0;
}

#ifdef WITH_TESTS
#define TEST_MAX 1024
static void REPORT( char * fail ) { fprintf( stderr, "\nFailed in %s\n", fail ); exit(-1); }
static leg_t test_001[] = { (leg_t)-1, 1 };
static leg_t test_002[] = { (leg_t)-1, 2 };
static leg_t test_003[] = { 0, 1 };
static leg_t test_004[] = { (leg_t)-1, 1, 1 };
static leg_t test_005[] = { 0, (leg_t)-1, 0 };
static leg_t test_006[] = { (leg_t)-2, 4, 0 };
static leg_t test_007[] = { (leg_t)-1, (leg_t)-2 };
static leg_t test_008[] = { (leg_t)-1, (leg_t)-3 };
static leg_t test_009[] = { (leg_t)-2, (leg_t)-4, 0 };
static leg_t test_010[] = { (leg_t)-2, (leg_t)-4, 1 };

static uint8_t test_011[] = {
    0xf2, 0xda, 0xe5, 0x35, 0xa0, 0xbd, 0xb8, 0x4c, 0x88, 0x65, 0x50, 0xca, 0xe8, 0x17, 0xa8, 0xab, 0x34, 0xd6, 0x86, 0xc4, 0x29, 0xa0, 0xc8, 0x51, 0x84, 0x05, 0x3e, 0x9d, 0x4d, 0x23, 0xce, 0x75,
    0x1f, 0xa6, 0xc1, 0x36, 0x86, 0x35, 0xf5, 0x6f, 0xcc, 0x0b, 0x13, 0x40, 0x89, 0xc2, 0x41, 0xb0, 0x0e, 0xa7, 0x68, 0xd4, 0x68, 0x6a, 0xd2, 0x08, 0xf7, 0xf5, 0x2c, 0x05, 0x11, 0x43, 0x00, 0x9b,
    0xf4, 0x92, 0xe1, 0x17, 0xae, 0xcc, 0x05, 0x98, 0x95, 0xc4, 0x33, 0xfd, 0x54, 0x42, 0x06, 0x98, 0x90, 0x62, 0x61, 0x15, 0x34, 0xe8, 0xbd, 0x37, 0xe0, 0x57, 0x27, 0x90, 0x75, 0x08, 0x33, 0x15,
    0xe5, 0x48, 0x19, 0xbb, 0x2c, 0xc1, 0xda, 0xeb, 0x95, 0x72, 0x97, 0x05, 0x60, 0xbf, 0xbf, 0x80, 0xfe, 0x04, 0x7a, 0x4a, 0xff, 0x0e, 0xa0, 0x47, 0x82, 0x0d, 0x90, 0xf0, 0x5b, 0xfc, 0x21, 0x82
};
static uint8_t test_012[] = {
    0x0c, 0xc4, 0xc0, 0xb7, 0x13, 0xee, 0xf6, 0xf3, 0xcd, 0xf7, 0x02, 0x0b, 0x97, 0x70, 0xc0, 0x59, 0x27, 0xd2, 0x74, 0xbf, 0x54, 0xbf, 0x9d, 0x77, 0x59, 0xe0, 0x86, 0x06, 0x18, 0x7a, 0x3c, 0x3e,
    0x12, 0x52, 0xed, 0x77, 0x10, 0xf9, 0x4b, 0xb8, 0xb7, 0x2b, 0xdc, 0x23, 0xe0, 0x45, 0x79, 0xe2, 0xe9, 0xef, 0x12, 0x97, 0x22, 0xf8, 0x72, 0xa2, 0x05, 0x1f, 0x5f, 0x36, 0xae, 0xfb, 0x7a, 0xb2,
    0x7d, 0xd7, 0xa2, 0x15, 0xa4, 0x9f, 0x27, 0x55, 0x65, 0x2a, 0x49, 0x0f, 0xca, 0x9f, 0xf3, 0xf6, 0x26, 0xcb, 0x57, 0x07, 0x82, 0xb7, 0x04, 0x1a, 0x7b, 0x65, 0x40, 0xb8, 0x55, 0x93, 0xee, 0xd4,
    0x09, 0x3e, 0x30, 0xa1, 0x6c, 0x00, 0x90, 0x54, 0x9b, 0x05, 0xb4, 0x08, 0xc4, 0xcc, 0xcd, 0xa2, 0xf6, 0xe6, 0x34, 0xf3, 0x5f, 0xe0, 0xe4, 0xb3, 0xcd, 0xd3, 0x8f, 0x83, 0xb2, 0xe8, 0x08, 0xb6
};
static uint8_t test_013[] = {
    0x0C, 0x1C, 0xE9, 0x27, 0xD9, 0xE2, 0xF8, 0x92, 0x97, 0x0B, 0x8E, 0x79, 0x91, 0x07, 0xE7, 0xAA, 0xF8, 0x5C, 0x11, 0xAB, 0x84, 0xFA, 0x0C, 0x6F, 0xED, 0xA5, 0x0C, 0x1F, 0x4C, 0x55, 0x5F, 0x5B,
    0x70, 0x63, 0x22, 0x85, 0x72, 0x6B, 0x3D, 0x9A, 0x78, 0x6E, 0xBF, 0x2E, 0x29, 0xB8, 0x2E, 0xE4, 0x4C, 0x09, 0x35, 0x45, 0x92, 0x26, 0xCC, 0x29, 0x15, 0xFC, 0x9A, 0x2F, 0x58, 0x4E, 0x70, 0xDD,
    0x00, 0xA9, 0xC7, 0xDE, 0xF5, 0x26, 0x85, 0x23, 0xDF, 0xBF, 0x00, 0x61, 0xC7, 0x47, 0x0E, 0x23, 0x9D, 0xC3, 0xB6, 0x17, 0x8C, 0x36, 0x19, 0x03, 0x70, 0xD1, 0xE1, 0xE7, 0x58, 0xDD, 0x10, 0xD5,
    0x76, 0x3B, 0xA7, 0xAE, 0x73, 0xAC, 0x4E, 0xBF, 0x4B, 0x79, 0xE5, 0xDE, 0x19, 0x68, 0x1D, 0x19, 0xD4, 0xF1, 0xF9, 0x90, 0x0B, 0x55, 0x9A, 0x8A, 0xAE, 0x5C, 0x85, 0x0C, 0x84, 0xB1, 0x33, 0x2F,
    0x0E, 0x99, 0xFC, 0xC6, 0x57, 0xE8, 0x81, 0x53, 0x23, 0x87, 0x5D, 0x41, 0xDF, 0xC9, 0xE0, 0xE5, 0xE1, 0x8C, 0x2C, 0x4E, 0x69, 0xAE, 0x99, 0x54, 0x19, 0x85, 0x7F, 0x7F, 0x19, 0x1E, 0x28, 0xC2,
    0x04, 0xAB, 0x16, 0x1D, 0xE9, 0x22, 0x55, 0xD2, 0x2D, 0x45, 0x44, 0xD2, 0x16, 0x3C, 0x2B, 0xD3, 0xE5, 0x88, 0xF2, 0x5E, 0x97, 0x92, 0x51, 0xF0, 0x64, 0x7D, 0x27, 0x5E, 0x0F, 0x5C, 0x99, 0xD8,
    0x02, 0xA8, 0xD3, 0x87, 0x90, 0x02, 0x9E, 0x4E, 0x9A, 0x9D, 0x72, 0xF3, 0xA8, 0xBC, 0x28, 0x74, 0xBD, 0x60, 0x92, 0xE1, 0x4A, 0x7E, 0x3E, 0xC2, 0xB3, 0x88, 0xBA, 0xAD, 0x8F, 0xFF, 0xCD, 0x57,
    0xE2, 0x7E, 0x43, 0x68, 0x91, 0xC3, 0xBE, 0x50, 0x9E, 0xB8, 0xFA, 0x53, 0x3C, 0xEE, 0x1F, 0x4C, 0x59, 0x5D, 0x4B, 0xE8, 0x42, 0x32, 0xD5, 0xE4, 0xFC, 0xD7, 0xFD, 0x12, 0x08, 0x1B, 0xE2, 0x6C
};

static uint8_t test_014[] = {
    0xE6, 0x62, 0x94, 0x84, 0x90, 0x75, 0x70, 0x5D, 0xF3, 0x6C, 0xC7, 0xFC, 0x45, 0x0C, 0x9F, 0xF7, 0x37, 0xB9, 0x71, 0x65, 0x62, 0x47, 0xC6, 0xD8, 0xFE, 0x93, 0xB0, 0xF1, 0xE2, 0x34, 0x78, 0x99,
    0xE1, 0xED, 0x3C, 0xA7, 0xF0, 0xD3, 0x00, 0x26, 0xA0, 0x18, 0xAA, 0x43, 0xDB, 0xFA, 0xD1, 0xFE, 0x2A, 0x71, 0xC5, 0xD1, 0xA1, 0x1D, 0xC0, 0xD3, 0x26, 0x15, 0x43, 0x6F, 0x06, 0x35, 0x5D, 0x02,
    0x3A, 0xFD, 0x58, 0xAA, 0x91, 0x4A, 0x27, 0xFC, 0xD5, 0x78, 0xBC, 0x65, 0xE6, 0xE6, 0xEC, 0x4E, 0x25, 0xBE, 0xEC, 0xDD, 0x82, 0xDB, 0x6D, 0xDE, 0x42, 0x4D, 0xE6, 0x39, 0x71, 0x80, 0x4B, 0xDA,
    0x2E, 0xD2, 0x25, 0x70, 0x96, 0x9D, 0x7B, 0x1B, 0x61, 0x85, 0xE6, 0x3C, 0x31, 0x4B, 0x60, 0x07, 0xCA, 0x66, 0xEF, 0x98, 0xEA, 0xAE, 0xED, 0xB8, 0x91, 0xC2, 0x8A, 0x70, 0x3B, 0xAC, 0x0B, 0x0A,
    0x3D, 0x80, 0xC9, 0x55, 0x96, 0x2F, 0xF3, 0xED, 0x55, 0xBC, 0x8B, 0x06, 0x62, 0xD9, 0x3F, 0x18, 0x4D, 0x87, 0x51, 0xD2, 0x11, 0x97, 0x7D, 0x16, 0xC5, 0xF0, 0xAB, 0x25, 0xCA, 0xB4, 0xF2, 0x66,
    0xCE, 0xCE, 0x58, 0xBB, 0x23, 0x10, 0x53, 0x6E, 0xA8, 0x96, 0xAD, 0x67, 0x9A, 0xEE, 0x53, 0xF0, 0x37, 0xCF, 0x51, 0x88, 0x51, 0xB8, 0x45, 0x15, 0x19, 0xCD, 0x84, 0xAD, 0xEB, 0xC5, 0xAA, 0x9C,
    0xCC, 0x03, 0x47, 0x2B, 0xDC, 0xCF, 0x86, 0x67, 0x04, 0xA0, 0xBA, 0xF5, 0x1D, 0x6B, 0xB2, 0xAB, 0x2E, 0xAA, 0x83, 0x5F, 0x56, 0x3F, 0x33, 0xB7, 0xA6, 0xA9, 0xF1, 0x0D, 0x95, 0x7B, 0x2F, 0x21,
    0xB6, 0xC0, 0x0E, 0xF2, 0x03, 0x91, 0x56, 0x3C, 0x56, 0x4C, 0x91, 0x87, 0xB3, 0x68, 0x49, 0xC5, 0x8E, 0x12, 0x41, 0xB6, 0xFD, 0xD9, 0xC8, 0xE7, 0xAE, 0xB2, 0x4B, 0xE4, 0x68, 0x52, 0xC6, 0x04
};

static uint8_t test_015[] = {
    0x00, 0xA3, 0x09, 0x48, 0xD3, 0x48, 0x0D, 0xE4, 0xF4, 0x42, 0xA7, 0xA7, 0x7E, 0x47, 0xE0, 0x07, 0x6E, 0x93, 0x73, 0x43, 0xE9, 0x42, 0x6D, 0xCC, 0x87, 0xAE, 0x3B, 0x87, 0x33, 0xAE, 0xDF, 0x4E,
    0x32, 0xE5, 0x78, 0x0C, 0x93, 0xD9, 0x84, 0xCC, 0x31, 0x8B, 0xB4, 0x83, 0xC7, 0x45, 0x71, 0xA9, 0x3C, 0x9F, 0x60, 0x8E, 0xA0, 0xED, 0x80, 0xF5, 0xBD, 0xC1, 0x75, 0x70, 0xB5, 0xFE, 0x39, 0x2A,
    0x90, 0xB9, 0x00, 0xF7, 0x7D, 0x78, 0x91, 0x43, 0xB4, 0xC2, 0x9F, 0xC9, 0x77, 0x8A, 0x26, 0xC9, 0x17, 0xF4, 0x08, 0x0F, 0x5A, 0x5A, 0x64, 0x99, 0x47, 0xF6, 0x6E, 0x41, 0x77, 0x35, 0x20, 0x03,
    0xB6, 0xCC, 0x3C, 0x21, 0x4C, 0x87, 0x18, 0x7E, 0x8A, 0x13, 0xCA, 0x2D, 0x51, 0x53, 0x76, 0xAE, 0x94, 0x37, 0xA4, 0x32, 0x5F, 0x0E, 0x47, 0x50, 0x9A, 0xD6, 0xDB, 0x30, 0xED, 0x43, 0xAC, 0xC3,
    0xD4, 0x1E, 0x53, 0xD2, 0x9C, 0x29, 0x69, 0x98, 0xF6, 0x82, 0xF2, 0x12, 0x50, 0x04, 0x72, 0xC9, 0x57, 0x17, 0xA8, 0xF9, 0x80, 0xED, 0xDF, 0x96, 0x54, 0x5E, 0x1D, 0x21, 0x51, 0xCB, 0xC9, 0xAF,
    0xC5, 0xC2, 0x71, 0xE1, 0xB4, 0xF1, 0xDF, 0xCA, 0xC9, 0xF3, 0x71, 0x4C, 0x91, 0x03, 0xA9, 0x71, 0x3B, 0x08, 0xCA, 0x67, 0x2D, 0x65, 0x35, 0x38, 0x16, 0x02, 0xE5, 0xF9, 0x1C, 0x58, 0x23, 0xE6,
    0x7E, 0x14, 0x97, 0x51, 0x6B, 0xDE, 0xEB, 0x14, 0xB0, 0xE7, 0xB8, 0x59, 0xAD, 0xC4, 0x98, 0xA2, 0xEC, 0x1F, 0x32, 0x35, 0x31, 0x96, 0x64, 0x82, 0x7B, 0x2A, 0x48, 0x59, 0x1B, 0xC8, 0x9C, 0xBB,
    0x55, 0x6D, 0xE7, 0xAD, 0xBC, 0x14, 0x48, 0xF8, 0xF5, 0xBB, 0xD1, 0xF6, 0xC5, 0xD5, 0xDC, 0x62, 0x8F, 0x4D, 0x91, 0x92, 0xA2, 0x95, 0x4C, 0x95, 0xA8, 0x3F, 0xBC, 0xB0, 0xE2, 0x2B, 0xE1, 0x64
};

static int run_tests( )
{
  leg_t val1[TEST_MAX], val2[TEST_MAX], val3[TEST_MAX], scratch[TEST_MAX];
  int i;

  /***********************************************************************/
  printf( "Testing mp_cmp... " );

  printf( "1 " );
  if( mp_cmp( test_001, test_001, sizeof(test_001) / sizeof(leg_t ) ))
    REPORT( "MP_CMP test 1" );

  printf( "2 " );
  if( mp_cmp( test_001, test_002, sizeof(test_001) / sizeof(leg_t ) ) != -1 )
    REPORT( "MP_CMP test 2" );

  printf( "3 " );
  if( mp_cmp( test_002, test_001, sizeof(test_001) / sizeof(leg_t ) ) != 1 )
    REPORT( "MP_CMP test 3" );

  printf( "success.\n" );

  /***********************************************************************/
  printf( "Testing mp_sub... " );
  printf( "1 " );
  memcpy( val1, test_002, sizeof(test_002) );
  mp_sub( val1, test_001, sizeof(test_001) / sizeof(leg_t ) );
  if( memcmp( val1, test_003, sizeof(test_003)) )
    REPORT( "MP_SUB test 1" );

  printf( "2 " );
  memset( val1, 0, sizeof(val1) );
  memcpy( val1, test_004, sizeof(test_004) );
  mp_sub( val1, test_002, sizeof(test_002) / sizeof(leg_t ) );
  if( memcmp( val1, test_005, sizeof(test_005)))
    REPORT( "MP_SUB test 2" );

  printf( "success.\n" );

  /***********************************************************************/
  printf( "Testing mp_add... " );
  printf( "1 " );
  memset( val1, 0, sizeof(val1) );
  memcpy( val1, test_002, sizeof(test_002) );
  mp_add( val1, test_001, sizeof(test_001) / sizeof(leg_t ), 0 );
 if( memcmp( val1, test_006, sizeof(test_006)) )
    REPORT( "MP_ADD test 1" );

  printf( "2 " );
  memset( val1, 0, sizeof(val1) );
  memcpy( val1, test_007, sizeof(test_007) );
  mp_add( val1, test_008, sizeof(test_008) / sizeof(leg_t ), 0 );
  if( memcmp( val1, test_009, sizeof(test_009)) )
    REPORT( "MP_ADD test 2" );

  printf( "3 " );
  memset( val1, 0, sizeof(val1) );
  memcpy( val1, test_007, sizeof(test_007) );
  mp_add( val1, test_008, sizeof(test_008) / sizeof(leg_t ), 1 );
  if( memcmp( val1, test_010, sizeof(test_010)) )
    REPORT( "MP_ADD test 3" );

  printf( "success.\n" );
  /***********************************************************************/

  printf( "Testing mp_mul_oper_add... " );
  printf( "1 " );
  memset( val1, 0, sizeof(val1) );
  mp_import( val2, test_011, sizeof(test_011) );
  mp_import( val3, test_012, sizeof(test_012) );
  mp_mul_oper_add( val1, val2, val3, sizeof(test_011) / sizeof(leg_t ) );
  mp_import( val3, test_013, sizeof(test_013) );

  if( memcmp( val1, val3, sizeof(test_013)) )
    REPORT( "MP_MUL_OPER_ADD test 1" );

  printf( "2 " );
  memset( val1, 0, sizeof(val1) );
  mp_import( val2, test_011, sizeof(test_011) );
  mp_mul_oper_add( val1, val2, val2, sizeof(test_011) / sizeof(leg_t ) );
  mp_import( val3, test_014, sizeof(test_014) );

  if( memcmp( val1, val3, sizeof(test_014)) )
    REPORT( "MP_MUL_OPER_ADD test 2" );

  printf( "success.\n" );
  /***********************************************************************/

  printf( "Testing mp_sqr... " );
  printf( "1 " );
  memset( val1, 0, sizeof(val1) );
  mp_import( val2, test_011, sizeof(test_011) );
  mp_sqr( val1, val2, sizeof(test_011) / sizeof(leg_t ) );
  mp_import( val3, test_014, sizeof(test_014) );

  if( memcmp( val1, val3, sizeof(test_014)) )
    REPORT( "MP_SQR test 1" );

  printf( "2 " );
  memset( val1, 0, sizeof(val1) );
  mp_import( val2, test_012, sizeof(test_012) );
  mp_sqr( val1, val2, sizeof(test_012) / sizeof(leg_t ) );
  mp_import( val3, test_015, sizeof(test_015) );

  if( memcmp( val1, val3, sizeof(test_015)) )
    REPORT( "MP_SQR test 2" );

  printf( "success.\n" );
  /***********************************************************************/

  printf( "Testing mp_mul_kara_square... " );
  printf( "1 " );
  memset( val1, 0, sizeof(val1) );
  mp_import( val2, test_011, sizeof(test_011) );
  mp_mul_kara_square( val1, val2, sizeof(test_011) / sizeof(leg_t ), scratch );
  mp_import( val3, test_014, sizeof(test_014) );

  if( memcmp( val1, val3, sizeof(test_014)) )
    REPORT( "MP_MUL_KARA_SQUARE test 1" );

  printf( "2 " );
  memset( val1, 0, sizeof(val1) );
  mp_import( val2, test_012, sizeof(test_012) );
  mp_mul_kara_square( val1, val2, sizeof(test_012) / sizeof(leg_t ), scratch );
  mp_import( val3, test_015, sizeof(test_015) );

  if( memcmp( val1, val3, sizeof(test_015)) )
    REPORT( "MP_MUL_KARA_SQUARE test 2" );

  printf( "success.\n" );
  /***********************************************************************/

  printf( "Testing mp_mul_kara... " );
  printf( "1 " );
  memset( val1, 0, sizeof(val1) );
  mp_import( val2, test_011, sizeof(test_011) );
  mp_import( val3, test_012, sizeof(test_012) );
  mp_mul_kara( val1, val2, val3, sizeof(test_011) / sizeof(leg_t ), scratch );
  mp_import( val3, test_013, sizeof(test_013) );

  if( memcmp( val1, val3, sizeof(test_013)) )
    REPORT( "MP_MUL_KARA test 1" );

  printf( "2 " );
  memset( val1, 0, sizeof(val1) );
  mp_import( val2, test_011, sizeof(test_011) );
  mp_import( val2, test_011, sizeof(test_011) );
  mp_mul_kara( val1, val2, val2, sizeof(test_011) / sizeof(leg_t ), scratch );
  mp_import( val3, test_014, sizeof(test_014) );

  if( memcmp( val1, val3, sizeof(test_014)) )
    REPORT( "MP_MUL_KARA test 2" );

  printf( "3 " );
  memset( val1, 0, sizeof(val1) );
  mp_import( val2, test_012, sizeof(test_012) );
  mp_import( val2, test_012, sizeof(test_012) );
  mp_mul_kara( val1, val2, val2, sizeof(test_012) / sizeof(leg_t ), scratch );
  mp_import( val3, test_015, sizeof(test_015) );

  if( memcmp( val1, val3, sizeof(test_015)) )
    REPORT( "MP_MUL_KARA test 3" );

  printf( "success.\n" );
  /***********************************************************************/


//  putchar(10);
//  for( i = sizeof(test_013)/sizeof(leg_t )-1; i>=0; --i) printf( "%016llX", val1[i] );
//  putchar(10);
//  for( i = sizeof(test_013)/sizeof(leg_t )-1; i>=0; --i) printf( "%016llX", val3[i] );
//  putchar(10);

}

#endif