void fp_rdcn_low(dig_t *c, dig_t *a) {
int i;
dig_t r, c0, c1, u, *tmp;
const dig_t *m;
u = *(fp_prime_get_rdc());
m = fp_prime_get();
tmp = a;
c1 = 0;
for (i = 0; i < FP_DIGS; i++, tmp++) {
r = (dig_t)(*tmp * u);
c0 = mpn_addmul_1(tmp, m, FP_DIGS, r);
c1 += mpn_add_1(tmp + FP_DIGS, tmp + FP_DIGS, FP_DIGS - i, c0);
}
for (i = 0; i < FP_DIGS; i++, tmp++) {
c[i] = *tmp;
}
for (i = 0; i < c1; i++) {
fp_subn_low(c, c, m);
}
if (fp_cmp(c, m) != CMP_LT) {
fp_subn_low(c, c, m);
}
}
void fp_rdc_monty_basic(fp_t c, dv_t a) {
int i;
dig_t r, c0, c1, *tmp, u0;
const dig_t *p = NULL;
tmp = a;
u0 = *(fp_prime_get_rdc());
p = fp_prime_get();
c1 = 0;
for (i = 0; i < FP_DIGS; i++, tmp++) {
r = (dig_t)(*tmp * u0);
c0 = fp_mula_low(tmp, fp_prime_get(), r);
/* We must use this because the size (FP_DIGS - i) is variable. */
c1 += bn_add1_low(tmp + FP_DIGS, tmp + FP_DIGS, c0, FP_DIGS - i);
}
fp_copy(c, a + FP_DIGS);
for (i = 0; i < c1; i++) {
fp_subn_low(c, c, p);
}
if (fp_cmpn_low(c, p) != CMP_LT) {
fp_subn_low(c, c, p);
}
}
void fp_rdcn_low2(dig_t *c, dig_t *a) {
int i, j;
dig_t r0, r1, r2, u;
dig_t *m, *tmp, *tmpm, *tmpc;
u = *(fp_prime_get_rdc());
m = fp_prime_get();
tmpc = c;
r0 = r1 = r2 = 0;
for (i = 0; i < FP_DIGS; i++, tmpc++, a++) {
tmp = c;
tmpm = m + i;
for (j = 0; j < i; j++, tmp++, tmpm--) {
COMBA_STEP(r2, r1, r0, *tmp, *tmpm);
}
COMBA_ADD(r2, r1, r0, *a);
*tmpc = (dig_t)(r0 * u);
COMBA_STEP(r2, r1, r0, *tmpc, *m);
r0 = r1;
r1 = r2;
r2 = 0;
}
for (i = FP_DIGS; i < 2 * FP_DIGS - 1; i++, a++) {
tmp = c + (i - FP_DIGS + 1);
tmpm = m + FP_DIGS - 1;
for (j = i - FP_DIGS + 1; j < FP_DIGS; j++, tmp++, tmpm--) {
COMBA_STEP(r2, r1, r0, *tmp, *tmpm);
}
COMBA_ADD(r2, r1, r0, *a);
c[i - FP_DIGS] = r0;
r0 = r1;
r1 = r2;
r2 = 0;
}
COMBA_ADD(r2, r1, r0, *a);
c[FP_DIGS - 1] = r0;
if (r1 || fp_cmpn_low(c, m) != CMP_LT) {
fp_subn_low(c, c, m);
}
}
