void fp3_sqrn_low(dv3_t c, fp3_t a) {
align dig_t t0[2 * FP_DIGS], t1[2 * FP_DIGS], t2[2 * FP_DIGS];
align dig_t t3[2 * FP_DIGS], t4[2 * FP_DIGS], t5[2 * FP_DIGS];
/* t0 = a_0^2. */
fp_sqrn_low(t0, a[0]);
/* t1 = 2 * a_1 * a_2. */
#ifdef FP_SPACE
fp_dbln_low(t2, a[1]);
#else
fp_dblm_low(t2, a[1]);
#endif
fp_muln_low(t1, t2, a[2]);
/* t2 = a_2^2. */
fp_sqrn_low(t2, a[2]);
/* t3 = (a_0 + a_2 + a_1)^2, t4 = (a_0 + a_2 - a_1)^2. */
#ifdef FP_SPACE
fp_addn_low(t3, a[0], a[2]);
fp_addn_low(t4, t3, a[1]);
#else
fp_addm_low(t3, a[0], a[2]);
fp_addm_low(t4, t3, a[1]);
#endif
fp_subm_low(t5, t3, a[1]);
fp_sqrn_low(t3, t4);
fp_sqrn_low(t4, t5);
/* t4 = (t4 + t3)/2. */
fp_addd_low(t4, t4, t3);
fp_hlvd_low(t4, t4);
/* t3 = t3 - t4 - t1. */
fp_addc_low(t5, t1, t4);
fp_subc_low(t3, t3, t5);
/* c_2 = t4 - t0 - t2. */
fp_addc_low(t5, t0, t2);
fp_subc_low(c[2], t4, t5);
/* c_0 = t0 + t1 * B. */
fp_subc_low(c[0], t0, t1);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_subc_low(c[0], c[0], t1);
}
/* c_1 = t3 + t2 * B. */
fp_subc_low(c[1], t3, t2);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_subc_low(c[1], c[1], t2);
}
}
void fp3_muln_low(dv3_t c, fp3_t a, fp3_t b) {
align dig_t t0[2 * FP_DIGS], t1[2 * FP_DIGS], t2[2 * FP_DIGS], t3[2 * FP_DIGS];
align dig_t t4[2 * FP_DIGS], t5[2 * FP_DIGS], t6[2 * FP_DIGS];
/* Karatsuba algorithm. */
/* t0 = a_0 * b_0, t1 = a_1 * b_1, t2 = a_2 * b_2. */
fp_muln_low(t0, a[0], b[0]);
fp_muln_low(t1, a[1], b[1]);
fp_muln_low(t2, a[2], b[2]);
/* t3 = (a_1 + a_2) * (b_1 + b_2). */
#ifdef FP_SPACE
fp_addn_low(t3, a[1], a[2]);
fp_addn_low(t4, b[1], b[2]);
#else
fp_addm_low(t3, a[1], a[2]);
fp_addm_low(t4, b[1], b[2]);
#endif
fp_muln_low(t5, t3, t4);
fp_addd_low(t6, t1, t2);
fp_subc_low(t4, t5, t6);
fp_subc_low(c[0], t0, t4);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_subc_low(c[0], c[0], t4);
}
#ifdef FP_SPACE
fp_addn_low(t4, a[0], a[1]);
fp_addn_low(t5, b[0], b[1]);
#else
fp_addm_low(t4, a[0], a[1]);
fp_addm_low(t5, b[0], b[1]);
#endif
fp_muln_low(t6, t4, t5);
fp_addd_low(t4, t0, t1);
fp_subc_low(t4, t6, t4);
fp_subc_low(c[1], t4, t2);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_subc_low(c[1], c[1], t2);
}
#ifdef FP_SPACE
fp_addn_low(t5, a[0], a[2]);
fp_addn_low(t6, b[0], b[2]);
#else
fp_addm_low(t5, a[0], a[2]);
fp_addm_low(t6, b[0], b[2]);
#endif
fp_muln_low(t4, t5, t6);
fp_addd_low(t6, t0, t2);
fp_subc_low(t5, t4, t6);
fp_addc_low(c[2], t5, t1);
}
void fp_prime_calc() {
#ifdef WITH_FPX
if (fp_prime_get_qnr() != 0) {
fp2_calc();
}
if (fp_prime_get_cnr() != 0) {
fp3_calc();
}
#endif
}
void fp3_sqr_basic(fp3_t c, fp3_t a) {
dv_t t0, t1, t2, t3, t4, t5;
dv_null(t0);
dv_null(t1);
dv_null(t2);
dv_null(t3);
dv_null(t4);
dv_null(t5);
TRY {
dv_new(t0);
dv_new(t1);
dv_new(t2);
dv_new(t3);
dv_new(t4);
dv_new(t5);
/* t0 = a_0^2. */
fp_sqrn_low(t0, a[0]);
/* t1 = 2 * a_1 * a_2. */
fp_dbl(t2, a[1]);
fp_muln_low(t1, t2, a[2]);
/* t2 = a_2^2. */
fp_sqrn_low(t2, a[2]);
/* t3 = (a_0 + a_2 + a_1)^2, t4 = (a_0 + a_2 - a_1)^2. */
fp_add(t3, a[0], a[2]);
fp_add(t4, t3, a[1]);
fp_sub(t5, t3, a[1]);
fp_sqrn_low(t3, t4);
fp_sqrn_low(t4, t5);
/* t4 = (t4 + t3)/2. */
fp_addd_low(t4, t4, t3);
fp_hlvd_low(t4, t4);
/* t3 = t3 - t4 - t1. */
fp_addc_low(t5, t1, t4);
fp_subc_low(t3, t3, t5);
/* c_2 = t4 - t0 - t2. */
fp_addc_low(t5, t0, t2);
fp_subc_low(t4, t4, t5);
fp_rdc(c[2], t4);
/* c_0 = t0 + t1 * B. */
fp_subc_low(t0, t0, t1);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_subc_low(t0, t0, t1);
}
fp_rdc(c[0], t0);
/* c_1 = t3 + t2 * B. */
fp_subc_low(t3, t3, t2);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_subc_low(t3, t3, t2);
}
fp_rdc(c[1], t3);
}
CATCH_ANY {
THROW(ERR_CAUGHT);
}
FINALLY {
dv_free(t0);
dv_free(t1);
dv_free(t2);
dv_free(t3);
dv_free(t4);
dv_free(t5);
}
}
/**
* Computes the constants required for evaluating Frobenius maps.
*/
static void fp3_calc() {
bn_t e;
fp3_t t0, t1, t2;
ctx_t *ctx = core_get();
bn_null(e);
fp3_null(t0);
fp3_null(t1);
fp3_null(t2);
TRY {
bn_new(e);
fp3_new(t0);
fp3_new(t1);
fp3_new(t2);
fp_set_dig(ctx->fp3_base[0], -fp_prime_get_cnr());
fp_neg(ctx->fp3_base[0], ctx->fp3_base[0]);
e->used = FP_DIGS;
dv_copy(e->dp, fp_prime_get(), FP_DIGS);
bn_sub_dig(e, e, 1);
bn_div_dig(e, e, 3);
fp_exp(ctx->fp3_base[0], ctx->fp3_base[0], e);
fp_sqr(ctx->fp3_base[1], ctx->fp3_base[0]);
fp3_zero(t0);
fp_set_dig(t0[1], 1);
dv_copy(e->dp, fp_prime_get(), FP_DIGS);
bn_sub_dig(e, e, 1);
bn_div_dig(e, e, 6);
/* t0 = u^((p-1)/6). */
fp3_exp(t0, t0, e);
fp_copy(ctx->fp3_p[0], t0[2]);
fp3_sqr(t1, t0);
fp_copy(ctx->fp3_p[1], t1[1]);
fp3_mul(t2, t1, t0);
fp_copy(ctx->fp3_p[2], t2[0]);
fp3_sqr(t2, t1);
fp_copy(ctx->fp3_p[3], t2[2]);
fp3_mul(t2, t2, t0);
fp_copy(ctx->fp3_p[4], t2[1]);
fp_mul(ctx->fp3_p2[0], ctx->fp3_p[0], ctx->fp3_base[1]);
fp_mul(t0[0], ctx->fp3_p2[0], ctx->fp3_p[0]);
fp_neg(ctx->fp3_p2[0], t0[0]);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_sub(ctx->fp3_p2[0], ctx->fp3_p2[0], t0[0]);
}
fp_mul(ctx->fp3_p2[1], ctx->fp3_p[1], ctx->fp3_base[0]);
fp_mul(ctx->fp3_p2[1], ctx->fp3_p2[1], ctx->fp3_p[1]);
fp_sqr(ctx->fp3_p2[2], ctx->fp3_p[2]);
fp_mul(ctx->fp3_p2[3], ctx->fp3_p[3], ctx->fp3_base[1]);
fp_mul(t0[0], ctx->fp3_p2[3], ctx->fp3_p[3]);
fp_neg(ctx->fp3_p2[3], t0[0]);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_sub(ctx->fp3_p2[3], ctx->fp3_p2[3], t0[0]);
}
fp_mul(ctx->fp3_p2[4], ctx->fp3_p[4], ctx->fp3_base[0]);
fp_mul(ctx->fp3_p2[4], ctx->fp3_p2[4], ctx->fp3_p[4]);
fp_mul(ctx->fp3_p3[0], ctx->fp3_p[0], ctx->fp3_base[0]);
fp_mul(t0[0], ctx->fp3_p3[0], ctx->fp3_p2[0]);
fp_neg(ctx->fp3_p3[0], t0[0]);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_sub(ctx->fp3_p3[0], ctx->fp3_p3[0], t0[0]);
}
fp_mul(ctx->fp3_p3[1], ctx->fp3_p[1], ctx->fp3_base[1]);
fp_mul(t0[0], ctx->fp3_p3[1], ctx->fp3_p2[1]);
fp_neg(ctx->fp3_p3[1], t0[0]);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_sub(ctx->fp3_p3[1], ctx->fp3_p3[1], t0[0]);
}
fp_mul(ctx->fp3_p3[2], ctx->fp3_p[2], ctx->fp3_p2[2]);
fp_mul(ctx->fp3_p3[3], ctx->fp3_p[3], ctx->fp3_base[0]);
fp_mul(t0[0], ctx->fp3_p3[3], ctx->fp3_p2[3]);
fp_neg(ctx->fp3_p3[3], t0[0]);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_sub(ctx->fp3_p3[3], ctx->fp3_p3[3], t0[0]);
}
fp_mul(ctx->fp3_p3[4], ctx->fp3_p[4], ctx->fp3_base[1]);
fp_mul(t0[0], ctx->fp3_p3[4], ctx->fp3_p2[4]);
fp_neg(ctx->fp3_p3[4], t0[0]);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_sub(ctx->fp3_p3[4], ctx->fp3_p3[4], t0[0]);
}
for (int i = 0; i < 5; i++) {
fp_mul(ctx->fp3_p4[i], ctx->fp3_p[i], ctx->fp3_p3[i]);
fp_mul(ctx->fp3_p5[i], ctx->fp3_p2[i], ctx->fp3_p3[i]);
}
} CATCH_ANY {
THROW(ERR_CAUGHT);
} FINALLY {
bn_free(e);
fp3_free(t0);
fp3_free(t1);
fp3_free(t2);
}
}
void fp3_mul_basic(fp3_t c, fp3_t a, fp3_t b) {
dv_t t, t0, t1, t2, t3, t4, t5, t6;
dv_null(t);
dv_null(t0);
dv_null(t1);
dv_null(t2);
dv_null(t3);
dv_null(t4);
dv_null(t5);
dv_null(t6);
TRY {
dv_new(t);
dv_new(t0);
dv_new(t1);
dv_new(t2);
dv_new(t3);
dv_new(t4);
dv_new(t5);
dv_new(t6);
/* Karatsuba algorithm. */
/* t0 = a_0 * b_0, t1 = a_1 * b_1, t2 = a_2 * b_2. */
fp_muln_low(t0, a[0], b[0]);
fp_muln_low(t1, a[1], b[1]);
fp_muln_low(t2, a[2], b[2]);
/* t3 = (a_1 + a_2) * (b_1 + b_2). */
fp_add(t3, a[1], a[2]);
fp_add(t4, b[1], b[2]);
fp_muln_low(t, t3, t4);
fp_addd_low(t6, t1, t2);
fp_subc_low(t4, t, t6);
fp_subc_low(t3, t0, t4);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_subc_low(t3, t3, t4);
}
fp_add(t4, a[0], a[1]);
fp_add(t5, b[0], b[1]);
fp_muln_low(t, t4, t5);
fp_addd_low(t4, t0, t1);
fp_subc_low(t4, t, t4);
fp_subc_low(t4, t4, t2);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_subc_low(t4, t4, t2);
}
fp_add(t5, a[0], a[2]);
fp_add(t6, b[0], b[2]);
fp_muln_low(t, t5, t6);
fp_addd_low(t6, t0, t2);
fp_subc_low(t5, t, t6);
fp_addc_low(t5, t5, t1);
/* c_0 = t3 mod p. */
fp_rdc(c[0], t3);
/* c_1 = t4 mod p. */
fp_rdc(c[1], t4);
/* c_2 = t5 mod p. */
fp_rdc(c[2], t5);
}
CATCH_ANY {
THROW(ERR_CAUGHT);
}
FINALLY {
dv_free(t);
dv_free(t0);
dv_free(t1);
dv_free(t2);
dv_free(t3);
dv_free(t4);
dv_free(t5);
dv_free(t6);
}
}
void fp3_inv(fp3_t c, fp3_t a) {
fp_t v0;
fp_t v1;
fp_t v2;
fp_t t0;
fp_null(v0);
fp_null(v1);
fp_null(v2);
fp_null(t0);
TRY {
fp_new(v0);
fp_new(v1);
fp_new(v2);
fp_new(t0);
/* v0 = a_0^2 - B * a_1 * a_2. */
fp_sqr(t0, a[0]);
fp_mul(v0, a[1], a[2]);
fp_neg(v2, v0);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_sub(v2, v2, v0);
}
fp_sub(v0, t0, v2);
/* v1 = B * a_2^2 - a_0 * a_1. */
fp_sqr(t0, a[2]);
fp_neg(v2, t0);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_sub(v2, v2, t0);
}
fp_mul(v1, a[0], a[1]);
fp_sub(v1, v2, v1);
/* v2 = a_1^2 - a_0 * a_2. */
fp_sqr(t0, a[1]);
fp_mul(v2, a[0], a[2]);
fp_sub(v2, t0, v2);
fp_mul(t0, a[1], v2);
fp_neg(c[1], t0);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_sub(c[1], c[1], t0);
}
fp_mul(c[0], a[0], v0);
fp_mul(t0, a[2], v1);
fp_neg(c[2], t0);
for (int i = -1; i > fp_prime_get_cnr(); i--) {
fp_sub(c[2], c[2], t0);
}
fp_add(t0, c[0], c[1]);
fp_add(t0, t0, c[2]);
fp_inv(t0, t0);
fp_mul(c[0], v0, t0);
fp_mul(c[1], v1, t0);
fp_mul(c[2], v2, t0);
} CATCH_ANY {
THROW(ERR_CAUGHT);
} FINALLY {
fp_free(v0);
fp_free(v1);
fp_free(v2);
fp_free(t0);
}
}
