/* multiplies |a| * |b| and does not compute the lower digs digits * [meant to get the higher part of the product] */ int s_mp_mul_high_digs(mp_int *a, mp_int *b, mp_int *c, int digs) { mp_int t; int res, pa, pb, ix, iy; mp_digit u; mp_word r; mp_digit tmpx, *tmpt, *tmpy; /* can we use the fast multiplier? */ #ifdef BN_FAST_S_MP_MUL_HIGH_DIGS_C if (((a->used + b->used + 1) < MP_WARRAY) && (MIN(a->used, b->used) < (1 << ((CHAR_BIT * sizeof(mp_word)) - (2 * DIGIT_BIT))))) { return fast_s_mp_mul_high_digs(a, b, c, digs); } #endif if ((res = mp_init_size(&t, a->used + b->used + 1)) != MP_OKAY) { return res; } t.used = a->used + b->used + 1; pa = a->used; pb = b->used; for (ix = 0; ix < pa; ix++) { /* clear the carry */ u = 0; /* left hand side of A[ix] * B[iy] */ tmpx = a->dp[ix]; /* alias to the address of where the digits will be stored */ tmpt = &(t.dp[digs]); /* alias for where to read the right hand side from */ tmpy = b->dp + (digs - ix); for (iy = digs - ix; iy < pb; iy++) { /* calculate the double precision result */ r = (mp_word) * tmpt + ((mp_word)tmpx * (mp_word) * tmpy++) + (mp_word)u; /* get the lower part */ *tmpt++ = (mp_digit)(r & ((mp_word)MP_MASK)); /* carry the carry */ u = (mp_digit)(r >> ((mp_word)DIGIT_BIT)); } *tmpt = u; } mp_clamp(&t); mp_exch(&t, c); mp_clear(&t); return MP_OKAY; }
/* reduces x mod m, assumes 0 < x < m**2, mu is * precomputed via mp_reduce_setup. * From HAC pp.604 Algorithm 14.42 */ int mp_reduce (mp_int * x, mp_int * m, mp_int * mu) { mp_int q; int res, um = USED(m); /* q = x */ if ((res = mp_init_copy (&q, x)) != MP_OKAY) { return res; } /* q1 = x / b**(k-1) */ mp_rshd (&q, um - 1); /* according to HAC this optimization is ok */ if (((unsigned long) um) > (((mp_digit)1) << (DIGIT_BIT - 1))) { if ((res = mp_mul (&q, mu, &q)) != MP_OKAY) { goto CLEANUP; } } else { #ifdef BN_S_MP_MUL_HIGH_DIGS_C if ((res = s_mp_mul_high_digs (&q, mu, &q, um)) != MP_OKAY) { goto CLEANUP; } #elif defined(BN_FAST_S_MP_MUL_HIGH_DIGS_C) if ((res = fast_s_mp_mul_high_digs (&q, mu, &q, um)) != MP_OKAY) { goto CLEANUP; } #else { res = MP_VAL; goto CLEANUP; } #endif } /* q3 = q2 / b**(k+1) */ mp_rshd (&q, um + 1); /* x = x mod b**(k+1), quick (no division) */ if ((res = mp_mod_2d (x, DIGIT_BIT * (um + 1), x)) != MP_OKAY) { goto CLEANUP; } /* q = q * m mod b**(k+1), quick (no division) */ if ((res = s_mp_mul_digs (&q, m, &q, um + 1)) != MP_OKAY) { goto CLEANUP; } /* x = x - q */ if ((res = mp_sub (x, &q, x)) != MP_OKAY) { goto CLEANUP; } /* If x < 0, add b**(k+1) to it */ if (mp_cmp_d (x, 0) == MP_LT) { mp_set (&q, 1); if ((res = mp_lshd (&q, um + 1)) != MP_OKAY) goto CLEANUP; if ((res = mp_add (x, &q, x)) != MP_OKAY) goto CLEANUP; } /* Back off if it's too big */ while (mp_cmp (x, m) != MP_LT) { if ((res = s_mp_sub (x, m, x)) != MP_OKAY) { goto CLEANUP; } } CLEANUP: mp_clear (&q); return res; }