#include #include #include // for intptr_t #include //for rand, pbc_malloc, pbc_free #include //for strcmp #include #include "pbc_utils.h" #include "pbc_field.h" #include "pbc_fp.h" #include "pbc_fieldquadratic.h" #include "pbc_param.h" #include "pbc_pairing.h" #include "pbc_curve.h" #include "pbc_random.h" #include "pbc_memory.h" #include "ecc/param.h" #include "pbc_a_param.h" #include "pbc_a1_param.h" typedef struct { int exp2; int exp1; int sign1; int sign0; mpz_t r; // r = 2^exp2 + sign1 * 2^exp1 + sign0 * 1 mpz_t q; // we work in E(F_q) (and E(F_q^2)) mpz_t h; // r * h = q + 1 } *a_param_ptr; typedef struct { field_t Fq, Fq2, Eq; int exp2, exp1; int sign1; } *a_pairing_data_ptr; static void a_out_str(FILE *stream, void *data) { a_param_ptr p = data; param_out_type(stream, "a"); param_out_mpz(stream, "q", p->q); param_out_mpz(stream, "h", p->h); param_out_mpz(stream, "r", p->r); param_out_int(stream, "exp2", p->exp2); param_out_int(stream, "exp1", p->exp1); param_out_int(stream, "sign1", p->sign1); param_out_int(stream, "sign0", p->sign0); } static void a_clear(void *data) { a_param_ptr sp = data; mpz_clear(sp->r); mpz_clear(sp->q); mpz_clear(sp->h); pbc_free(data); } static void phi_identity(element_ptr out, element_ptr in, pairing_ptr pairing) { UNUSED_VAR(pairing); element_set(out, in); } static void compute_abc_tangent(element_ptr a, element_ptr b, element_ptr c, element_ptr Vx, element_ptr Vy, element_ptr e0) { //a = -slope_tangent(V.x, V.y); //b = 1; //c = -(V.y + aV.x); //but we multiply by -2*V.y to avoid division so: //a = -(3 Vx^2 + cc->a) //b = 2 * Vy //c = -(2 Vy^2 + a Vx); element_square(a, Vx); //element_mul_si(a, a, 3); element_add(e0, a, a); element_add(a, e0, a); element_set1(b); element_add(a, a, b); element_neg(a, a); element_double(b, Vy); element_mul(e0, b, Vy); element_mul(c, a, Vx); element_add(c, c, e0); element_neg(c, c); } static void compute_abc_tangent_proj(element_ptr a, element_ptr b, element_ptr c, element_ptr Vx, element_ptr Vy, element_ptr z, element_ptr z2, element_ptr e0) { //a = -(3x^2 + cca z^4) //for this case cca = 1 //b = 2 y z^3 //c = -(2 y^2 + x a) //a = z^2 a element_square(a, z2); element_square(b, Vx); ////element_mul_si(b, b, 3); element_double(e0, b); element_add(b, e0, b); element_add(a, a, b); element_neg(a, a); ////element_mul_si(e0, Vy, 2); element_double(e0, Vy); element_mul(b, e0, z2); element_mul(b, b, z); element_mul(c, Vx, a); element_mul(a, a, z2); element_mul(e0, e0, Vy); element_add(c, c, e0); element_neg(c, c); } static void compute_abc_line(element_ptr a, element_ptr b, element_ptr c, element_ptr Vx, element_ptr Vy, element_ptr V1x, element_ptr V1y, element_ptr e0) { //a = -(B.y - A.y) / (B.x - A.x); //b = 1; //c = -(A.y + a * A.x); //but we'll multiply by B.x - A.x to avoid division, so //a = -(By - Ay) //b = Bx - Ax //c = -(Ay b + a Ax); element_sub(a, Vy, V1y); element_sub(b, V1x, Vx); element_mul(c, Vx, V1y); element_mul(e0, Vy, V1x); element_sub(c, c, e0); } struct pp_coeff_s { element_t a; element_t b; element_t c; }; typedef struct pp_coeff_s pp_coeff_t[1]; typedef struct pp_coeff_s *pp_coeff_ptr; static void pp_coeff_set(pp_coeff_ptr p, element_t a, element_t b, element_t c) { element_init(p->a, a->field); element_init(p->b, b->field); element_init(p->c, c->field); element_set(p->a, a); element_set(p->b, b); element_set(p->c, c); } static void a_pairing_pp_init(pairing_pp_t p, element_ptr in1, pairing_t pairing) { int i, n; a_pairing_data_ptr ainfo = pairing->data; p->data = pbc_malloc(sizeof(pp_coeff_t) * (ainfo->exp2 + 1)); pp_coeff_t *coeff = (pp_coeff_t *) p->data; element_t V, V1; element_t a, b, c; element_t e0; element_ptr Vx, Vy; element_ptr V1x, V1y; #define do_tangent() \ compute_abc_tangent(a, b, c, Vx, Vy, e0); \ pp_coeff_set(coeff[i], a, b, c); #define do_line() \ compute_abc_line(a, b, c, Vx, Vy, V1x, V1y, e0); \ pp_coeff_set(coeff[i], a, b, c); element_init(V, ainfo->Eq); element_init(V1, ainfo->Eq); element_set(V, in1); Vx = curve_x_coord(V); Vy = curve_y_coord(V); V1x = curve_x_coord(V1); V1y = curve_y_coord(V1); element_init(e0, ainfo->Fq); element_init(a, ainfo->Fq); element_init(b, ainfo->Fq); element_init(c, ainfo->Fq); n = ainfo->exp1; for (i=0; isign1 < 0) { element_neg(V1, V); } else { element_set(V1, V); } n = ainfo->exp2; for (; ipairing->data; pp_coeff_t *coeff = (pp_coeff_t *) p->data; int i, n = ainfo->exp2 + 1; for (i=0; ia); element_clear(pp->b); element_clear(pp->c); } pbc_free(p->data); } // Requires cofactor to be odd. // Overwrites in and temp, out != in. // Luckily this touchy routine is only used internally. // TODO: rewrite to allow (out == in)? would simplify a_finalpow() static void lucas_odd(element_ptr out, element_ptr in, element_ptr temp, mpz_t cofactor) { element_ptr in0 = element_x(in); element_ptr in1 = element_y(in); element_ptr v0 = element_x(out); element_ptr v1 = element_y(out); element_ptr t0 = element_x(temp); element_ptr t1 = element_y(temp); int j; element_set_si(t0, 2); element_double(t1, in0); element_set(v0, t0); element_set(v1, t1); j = mpz_sizeinbase(cofactor, 2) - 1; for (;;) { if (!j) { element_mul(v1, v0, v1); element_sub(v1, v1, t1); element_square(v0, v0); element_sub(v0, v0, t0); break; } if (mpz_tstbit(cofactor, j)) { element_mul(v0, v0, v1); element_sub(v0, v0, t1); element_square(v1, v1); element_sub(v1, v1, t0); } else { element_mul(v1, v0, v1); element_sub(v1, v1, t1); element_square(v0, v0); element_sub(v0, v0, t0); } j--; } //assume cofactor = (q + 1) / r is even //(r should be odd and q + 1 is always even) //thus v0 = V_k, v1 = V_{k+1} //and V_{k-1} = P v0 - v1 //so U_k = (P V_k - 2 V_{k-1}) / (P^2 - 4) // = (2 v1 - P v0) / (P^2 - 4) element_mul(in0, v0, t1); element_double(v1, v1); element_sub(v1, v1, in0); element_square(t1, t1); element_sub(t1, t1, t0); element_sub(t1, t1, t0); element_div(v1, v1, t1); element_halve(v0, v0); element_mul(v1, v1, in1); } static inline void a_tateexp(element_ptr out, element_ptr in, element_ptr temp, mpz_t cofactor) { element_ptr in1 = element_y(in); //simpler but slower: //element_pow_mpz(out, f, tateexp); //1. Exponentiate by q-1 //which is equivalent to the following element_invert(temp, in); element_neg(in1, in1); element_mul(in, in, temp); //2. Exponentiate by (q+1)/r //Instead of: // element_pow_mpz(out, in, cofactor); //we use Lucas sequences (see "Compressed Pairings", Scott and Barreto) lucas_odd(out, in, temp, cofactor); } //computes a Qx + b Qy + c for type A pairing static inline void a_miller_evalfn(element_ptr out, element_ptr a, element_ptr b, element_ptr c, element_ptr Qx, element_ptr Qy) { //we'll map Q via (x,y) --> (-x, iy) //hence Re(a Qx + b Qy + c) = -a Q'x + c and //Im(a Qx + b Qy + c) = b Q'y element_mul(element_y(out), a, Qx); element_sub(element_x(out), c, element_y(out)); element_mul(element_y(out), b, Qy); } static void a_pairing_pp_apply(element_ptr out, element_ptr in2, pairing_pp_t p) { //TODO: use proj coords here too to shave off a little time element_ptr Qx = curve_x_coord(in2); element_ptr Qy = curve_y_coord(in2); element_t f, f0; int i, n; a_pairing_data_ptr ainfo = p->pairing->data; pp_coeff_t *coeff = p->data; element_init(f, ainfo->Fq2); element_init(f0, ainfo->Fq2); element_set1(f); n = ainfo->exp1; for (i=0; ia, pp->b, pp->c, Qx, Qy); element_mul(f, f, f0); } if (ainfo->sign1 < 0) { element_invert(out, f); } else { element_set(out, f); } n = ainfo->exp2; for (; ia, pp->b, pp->c, Qx, Qy); element_mul(f, f, f0); } element_mul(f, f, out); { pp_coeff_ptr pp = coeff[i]; a_miller_evalfn(f0, pp->a, pp->b, pp->c, Qx, Qy); element_mul(f, f, f0); } a_tateexp(out, f, f0, p->pairing->phikonr); element_clear(f); element_clear(f0); } // in1, in2 are from E(F_q), out from F_q^2. // Pairing via elliptic nets (see Stange). static void a_pairing_ellnet(element_ptr out, element_ptr in1, element_ptr in2, pairing_t pairing) { element_ptr x = curve_x_coord(in1); element_ptr y = curve_y_coord(in1); element_ptr x2 = curve_x_coord(in2); element_ptr y2 = curve_y_coord(in2); //we map (x2,y2) to (-x2, i y2) before pairing //notation: cmi means c_{k-i}, ci means c_{k+i} element_t cm3, cm2, cm1, c0, c1, c2, c3, c4; element_t dm1, d0, d1; element_t A, B, C; element_init_same_as(cm3, x); element_init_same_as(cm2, x); element_init_same_as(cm1, x); element_init_same_as(c0, x); element_init_same_as(c1, x); element_init_same_as(c2, x); element_init_same_as(c3, x); element_init_same_as(c4, x); element_init_same_as(C, x); element_init_same_as(dm1, out); element_init_same_as(d0, out); element_init_same_as(d1, out); element_init_same_as(A, x); element_init_same_as(B, out); // c1 = 2y // c0 = 1 // cm2 = -1 // cm3 = -2y element_double(c1, y); element_set1(c0); element_neg(cm3, c1); element_neg(cm2, c0); // a = 1, b = 0 for Y^2 = X^3 + X //hence c3 = c_{k+3} = c_4 = 4y(x^6 + 5(x^4 - x^2) - 1) //use cm1, C, c2 as temp variables for now element_square(cm1, x); element_square(C, cm1); element_sub(c2, C, cm1); element_double(c3, c2); element_double(c3, c3); element_add(c3, c3, c2); element_mul(c2, C, cm1); element_add(c3, c3, c2); element_add(c3, c3, cm2); element_mul(c3, c3, c1); element_double(c3, c3); // c2 = c_3 = 3x^4 + 6x^2 - 1 element_double(cm1, cm1); element_add(cm1, cm1, C); element_double(C, cm1); element_add(C, C, cm1); element_add(c2, C, cm2); // c4 = c_5 = c_2^3 c_4 - c_3^3 = c1^3 c3 - c2^3 element_square(C, c1); element_mul(c4, C, c1); element_mul(c4, c4, c3); element_square(C, c2); element_mul(C, C, c2); element_sub(c4, c4, C); //compute A, B, d1 (which is d_2 since k = 1) //(recall phi takes x2 to -x2, y2 to i y2) element_add(A, x, x2); element_double(C, x); element_sub(C, C, x2); element_square(cm1, A); element_mul(cm1, C, cm1); element_set(element_x(d1), y); element_set(element_y(d1), y2); element_square(d1, d1); element_sub(element_x(d1), element_x(d1), cm1); element_neg(B, d1); element_invert(B, B); element_invert(A, A); element_mul(element_x(d1), y, A); element_neg(element_x(d1), element_x(d1)); element_mul(element_y(d1), y2, A); element_square(d1, d1); element_sub(element_x(d1), C, element_x(d1)); element_neg(element_y(d1), element_y(d1)); // cm1 = 0 // C = (2y)^-1 element_set0(cm1); element_invert(C, c1); element_set1(dm1); element_set1(d0); element_t sm2, sm1; element_t s0, s1, s2, s3; element_t tm2, tm1; element_t t0, t1, t2, t3; element_t e0, e1; element_t u, v; element_init_same_as(sm2, x); element_init_same_as(sm1, x); element_init_same_as(s0, x); element_init_same_as(s1, x); element_init_same_as(s2, x); element_init_same_as(s3, x); element_init_same_as(tm2, x); element_init_same_as(tm1, x); element_init_same_as(t0, x); element_init_same_as(t1, x); element_init_same_as(t2, x); element_init_same_as(t3, x); element_init_same_as(e0, x); element_init_same_as(e1, x); element_init_same_as(u, d0); element_init_same_as(v, d0); int m = mpz_sizeinbase(pairing->r, 2) - 2; for (;;) { element_square(sm2, cm2); element_square(sm1, cm1); element_square(s0, c0); element_square(s1, c1); element_square(s2, c2); element_square(s3, c3); element_mul(tm2, cm3, cm1); element_mul(tm1, cm2, c0); element_mul(t0, cm1, c1); element_mul(t1, c0, c2); element_mul(t2, c1, c3); element_mul(t3, c2, c4); element_square(u, d0); element_mul(v, dm1, d1); if (mpz_tstbit(pairing->r, m)) { //double-and-add element_mul(e0, t0, sm2); element_mul(e1, tm2, s0); element_sub(cm3, e0, e1); element_mul(cm3, cm3, C); element_mul(e0, t0, sm1); element_mul(e1, tm1, s0); element_sub(cm2, e0, e1); element_mul(e0, t1, sm1); element_mul(e1, tm1, s1); element_sub(cm1, e0, e1); element_mul(cm1, cm1, C); element_mul(e0, t1, s0); element_mul(e1, t0, s1); element_sub(c0, e0, e1); element_mul(e0, t2, s0); element_mul(e1, t0, s2); element_sub(c1, e0, e1); element_mul(c1, c1, C); element_mul(e0, t2, s1); element_mul(e1, t1, s2); element_sub(c2, e0, e1); element_mul(e0, t3, s1); element_mul(e1, t1, s3); element_sub(c3, e0, e1); element_mul(c3, c3, C); element_mul(e0, t3, s2); element_mul(e1, t2, s3); element_sub(c4, e0, e1); element_mul(element_x(out), element_x(u), t0); element_mul(element_y(out), element_y(u), t0); element_mul(element_x(dm1), element_x(v), s0); element_mul(element_y(dm1), element_y(v), s0); element_sub(dm1, dm1, out); element_mul(element_x(out), element_x(u), t1); element_mul(element_y(out), element_y(u), t1); element_mul(element_x(d0), element_x(v), s1); element_mul(element_y(d0), element_y(v), s1); element_sub(d0, d0, out); element_mul(element_x(d0), element_x(d0), A); element_mul(element_y(d0), element_y(d0), A); element_mul(element_x(out), element_x(u), t2); element_mul(element_y(out), element_y(u), t2); element_mul(element_x(d1), element_x(v), s2); element_mul(element_y(d1), element_y(v), s2); element_sub(d1, d1, out); element_mul(d1, d1, B); } else { //double element_mul(e0, tm1, sm2); element_mul(e1, tm2, sm1); element_sub(cm3, e0, e1); element_mul(e0, t0, sm2); element_mul(e1, tm2, s0); element_sub(cm2, e0, e1); element_mul(cm2, cm2, C); element_mul(e0, t0, sm1); element_mul(e1, tm1, s0); element_sub(cm1, e0, e1); element_mul(e0, t1, sm1); element_mul(e1, tm1, s1); element_sub(c0, e0, e1); element_mul(c0, c0, C); element_mul(e0, t1, s0); element_mul(e1, t0, s1); element_sub(c1, e0, e1); element_mul(e0, t2, s0); element_mul(e1, t0, s2); element_sub(c2, e0, e1); element_mul(c2, c2, C); element_mul(e0, t2, s1); element_mul(e1, t1, s2); element_sub(c3, e0, e1); element_mul(e0, t3, s1); element_mul(e1, t1, s3); element_sub(c4, e0, e1); element_mul(c4, c4, C); element_mul(element_x(out), element_x(u), tm1); element_mul(element_y(out), element_y(u), tm1); element_mul(element_x(dm1), element_x(v), sm1); element_mul(element_y(dm1), element_y(v), sm1); element_sub(dm1, dm1, out); element_mul(element_x(out), element_x(u), t0); element_mul(element_y(out), element_y(u), t0); element_mul(element_x(d0), element_x(v), s0); element_mul(element_y(d0), element_y(v), s0); element_sub(d0, d0, out); element_mul(element_x(out), element_x(u), t1); element_mul(element_y(out), element_y(u), t1); element_mul(element_x(d1), element_x(v), s1); element_mul(element_y(d1), element_y(v), s1); element_sub(d1, d1, out); element_mul(element_x(d1), element_x(d1), A); element_mul(element_y(d1), element_y(d1), A); } if (!m) break; m--; } // since c_k lies base field // it gets killed by the final powering //element_invert(c1, c1); //element_mul(element_x(d1), element_x(d1), c1); //element_mul(element_y(d1), element_y(d1), c1); a_tateexp(out, d1, d0, pairing->phikonr); element_clear(dm1); element_clear(d0); element_clear(d1); element_clear(cm3); element_clear(cm2); element_clear(cm1); element_clear(c0); element_clear(c1); element_clear(c2); element_clear(c3); element_clear(c4); element_clear(sm2); element_clear(sm1); element_clear(s0); element_clear(s1); element_clear(s2); element_clear(s3); element_clear(tm2); element_clear(tm1); element_clear(t0); element_clear(t1); element_clear(t2); element_clear(t3); element_clear(e0); element_clear(e1); element_clear(A); element_clear(B); element_clear(C); element_clear(u); element_clear(v); } struct ellnet_pp_st_s { element_t sm1, s0, s1, s2; element_t tm1, t0, t1, t2; }; typedef struct ellnet_pp_st_s ellnet_pp_st_t[1]; typedef struct ellnet_pp_st_s *ellnet_pp_st_ptr; struct ellnet_pp_s { element_t x; element_t y; ellnet_pp_st_t *seq; }; typedef struct ellnet_pp_s ellnet_pp_t[1]; typedef struct ellnet_pp_s *ellnet_pp_ptr; static void a_pairing_ellnet_pp_init(pairing_pp_t p, element_ptr in1, pairing_t pairing) { element_ptr x = curve_x_coord(in1); element_ptr y = curve_y_coord(in1); int i, rbits = mpz_sizeinbase(pairing->r, 2); ellnet_pp_ptr pp = p->data = pbc_malloc(sizeof(ellnet_pp_t)); pp->seq = pbc_malloc(sizeof(ellnet_pp_st_t) * rbits); element_init_same_as(pp->x, x); element_init_same_as(pp->y, y); element_set(pp->x, x); element_set(pp->y, y); for (i=0; iseq[i]; element_init_same_as(seq->sm1, x); element_init_same_as(seq->s0, x); element_init_same_as(seq->s1, x); element_init_same_as(seq->s2, x); element_init_same_as(seq->tm1, x); element_init_same_as(seq->t0, x); element_init_same_as(seq->t1, x); element_init_same_as(seq->t2, x); } //we map (x2,y2) to (-x2, i y2) before pairing //notation: cmi means c_{k-i}, ci means c_{k+i} element_t cm3, cm2, cm1, c0, c1, c2, c3, c4; element_t C; element_init_same_as(cm3, x); element_init_same_as(cm2, x); element_init_same_as(cm1, x); element_init_same_as(c0, x); element_init_same_as(c1, x); element_init_same_as(c2, x); element_init_same_as(c3, x); element_init_same_as(c4, x); element_init_same_as(C, x); // c1 = 2y // c0 = 1 // cm2 = -1 // cm3 = -2y element_double(c1, y); element_set1(c0); element_neg(cm3, c1); element_neg(cm2, c0); // a = 1, b = 0 for Y^2 = X^3 + X //hence c3 = c_{k+3} = c_4 = 4y(x^6 + 5(x^4 - x^2) - 1) //use cm1, C, c2 as temp variables for now element_square(cm1, x); element_square(C, cm1); element_sub(c2, C, cm1); element_double(c3, c2); element_double(c3, c3); element_add(c3, c3, c2); element_mul(c2, C, cm1); element_add(c3, c3, c2); element_add(c3, c3, cm2); element_mul(c3, c3, c1); element_double(c3, c3); // c2 = c_3 = 3x^4 + 6x^2 - 1 element_double(cm1, cm1); element_add(cm1, cm1, C); element_double(C, cm1); element_add(C, C, cm1); element_add(c2, C, cm2); // c4 = c_5 = c_2^3 c_4 - c_3^3 = c1^3 c3 - c2^3 element_square(C, c1); element_mul(c4, C, c1); element_mul(c4, c4, c3); element_square(C, c2); element_mul(C, C, c2); element_sub(c4, c4, C); // cm1 = 0 // C = (2y)^-1 element_set0(cm1); element_invert(C, c1); int k = 0; element_t sm2, s3; element_t tm2, t3; element_ptr sm1, s0, s1, s2; element_ptr tm1, t0, t1, t2; element_t e0, e1; element_init_same_as(sm2, x); element_init_same_as(s3, x); element_init_same_as(tm2, x); element_init_same_as(t3, x); element_init_same_as(e0, x); element_init_same_as(e1, x); int m = rbits - 2; for (;;) { ellnet_pp_st_ptr seq = pp->seq[k]; sm1 = seq->sm1; s0 = seq->s0; s1 = seq->s1; s2 = seq->s2; tm1 = seq->tm1; t0 = seq->t0; t1 = seq->t1; t2 = seq->t2; element_square(sm2, cm2); element_square(sm1, cm1); element_square(s0, c0); element_square(s1, c1); element_square(s2, c2); element_square(s3, c3); element_mul(tm2, cm3, cm1); element_mul(tm1, cm2, c0); element_mul(t0, cm1, c1); element_mul(t1, c0, c2); element_mul(t2, c1, c3); element_mul(t3, c2, c4); if (!m) break; k++; if (mpz_tstbit(pairing->r, m)) { //double-and-add element_mul(e0, t0, sm2); element_mul(e1, tm2, s0); element_sub(cm3, e0, e1); element_mul(cm3, cm3, C); element_mul(e0, t0, sm1); element_mul(e1, tm1, s0); element_sub(cm2, e0, e1); element_mul(e0, t1, sm1); element_mul(e1, tm1, s1); element_sub(cm1, e0, e1); element_mul(cm1, cm1, C); element_mul(e0, t1, s0); element_mul(e1, t0, s1); element_sub(c0, e0, e1); element_mul(e0, t2, s0); element_mul(e1, t0, s2); element_sub(c1, e0, e1); element_mul(c1, c1, C); element_mul(e0, t2, s1); element_mul(e1, t1, s2); element_sub(c2, e0, e1); element_mul(e0, t3, s1); element_mul(e1, t1, s3); element_sub(c3, e0, e1); element_mul(c3, c3, C); element_mul(e0, t3, s2); element_mul(e1, t2, s3); element_sub(c4, e0, e1); } else { //double element_mul(e0, tm1, sm2); element_mul(e1, tm2, sm1); element_sub(cm3, e0, e1); element_mul(e0, t0, sm2); element_mul(e1, tm2, s0); element_sub(cm2, e0, e1); element_mul(cm2, cm2, C); element_mul(e0, t0, sm1); element_mul(e1, tm1, s0); element_sub(cm1, e0, e1); element_mul(e0, t1, sm1); element_mul(e1, tm1, s1); element_sub(c0, e0, e1); element_mul(c0, c0, C); element_mul(e0, t1, s0); element_mul(e1, t0, s1); element_sub(c1, e0, e1); element_mul(e0, t2, s0); element_mul(e1, t0, s2); element_sub(c2, e0, e1); element_mul(c2, c2, C); element_mul(e0, t2, s1); element_mul(e1, t1, s2); element_sub(c3, e0, e1); element_mul(e0, t3, s1); element_mul(e1, t1, s3); element_sub(c4, e0, e1); element_mul(c4, c4, C); } m--; } element_clear(cm3); element_clear(cm2); element_clear(cm1); element_clear(c0); element_clear(c1); element_clear(c2); element_clear(c3); element_clear(c4); element_clear(sm2); element_clear(s3); element_clear(tm2); element_clear(t3); element_clear(e0); element_clear(e1); element_clear(C); } static void a_pairing_ellnet_pp_clear(pairing_pp_t p) { ellnet_pp_ptr pp = p->data; int i, rbits = mpz_sizeinbase(p->pairing->r, 2); for (i=0; iseq[i]; element_clear(seq->sm1); element_clear(seq->s0); element_clear(seq->s1); element_clear(seq->s2); element_clear(seq->tm1); element_clear(seq->t0); element_clear(seq->t1); element_clear(seq->t2); } element_clear(pp->x); element_clear(pp->y); pbc_free(pp->seq); pbc_free(p->data); } static void a_pairing_ellnet_pp_apply(element_ptr out, element_ptr in2, pairing_pp_t p) { element_ptr x2 = curve_x_coord(in2); element_ptr y2 = curve_y_coord(in2); ellnet_pp_ptr pp = p->data; int rbits = mpz_sizeinbase(p->pairing->r, 2); int k = 0; int m = rbits - 2; element_t A, B; element_t e0, e1; element_t dm1, d0, d1; element_t u, v; element_init_same_as(A, x2); element_init_same_as(B, out); element_init_same_as(e0, x2); element_init_same_as(e1, x2); element_init_same_as(dm1, out); element_init_same_as(d0, out); element_init_same_as(d1, out); element_init_same_as(u, out); element_init_same_as(v, out); element_add(A, pp->x, x2); element_double(e0, pp->x); element_sub(e0, e0, x2); element_square(e1, A); element_mul(e1, e0, e1); element_set(element_x(d1), pp->y); element_set(element_y(d1), y2); element_square(d1, d1); element_sub(element_x(d1), element_x(d1), e1); element_neg(B, d1); element_invert(B, B); element_invert(A, A); element_mul(element_x(d1), pp->y, A); element_neg(element_x(d1), element_x(d1)); element_mul(element_y(d1), y2, A); element_square(d1, d1); element_sub(element_x(d1), e0, element_x(d1)); element_neg(element_y(d1), element_y(d1)); element_set1(dm1); element_set1(d0); for (;;) { element_ptr sm1, s0, s1, s2; element_ptr tm1, t0, t1, t2; ellnet_pp_st_ptr seq = pp->seq[k]; sm1 = seq->sm1; s0 = seq->s0; s1 = seq->s1; s2 = seq->s2; tm1 = seq->tm1; t0 = seq->t0; t1 = seq->t1; t2 = seq->t2; k++; element_square(u, d0); element_mul(v, dm1, d1); if (mpz_tstbit(p->pairing->r, m)) { //double-and-add element_mul(element_x(out), element_x(u), t0); element_mul(element_y(out), element_y(u), t0); element_mul(element_x(dm1), element_x(v), s0); element_mul(element_y(dm1), element_y(v), s0); element_sub(dm1, dm1, out); element_mul(element_x(out), element_x(u), t1); element_mul(element_y(out), element_y(u), t1); element_mul(element_x(d0), element_x(v), s1); element_mul(element_y(d0), element_y(v), s1); element_sub(d0, d0, out); element_mul(element_x(d0), element_x(d0), A); element_mul(element_y(d0), element_y(d0), A); element_mul(element_x(out), element_x(u), t2); element_mul(element_y(out), element_y(u), t2); element_mul(element_x(d1), element_x(v), s2); element_mul(element_y(d1), element_y(v), s2); element_sub(d1, d1, out); element_mul(d1, d1, B); } else { //double element_mul(element_x(out), element_x(u), tm1); element_mul(element_y(out), element_y(u), tm1); element_mul(element_x(dm1), element_x(v), sm1); element_mul(element_y(dm1), element_y(v), sm1); element_sub(dm1, dm1, out); element_mul(element_x(out), element_x(u), t0); element_mul(element_y(out), element_y(u), t0); element_mul(element_x(d0), element_x(v), s0); element_mul(element_y(d0), element_y(v), s0); element_sub(d0, d0, out); element_mul(element_x(out), element_x(u), t1); element_mul(element_y(out), element_y(u), t1); element_mul(element_x(d1), element_x(v), s1); element_mul(element_y(d1), element_y(v), s1); element_sub(d1, d1, out); element_mul(element_x(d1), element_x(d1), A); element_mul(element_y(d1), element_y(d1), A); } if (!m) break; m--; } a_tateexp(out, d1, d0, p->pairing->phikonr); element_clear(A); element_clear(B); element_clear(e0); element_clear(e1); element_clear(dm1); element_clear(d0); element_clear(d1); element_clear(u); element_clear(v); } //in1, in2 are from E(F_q), out from F_q^2 static void a_pairing_proj(element_ptr out, element_ptr in1, element_ptr in2, pairing_t pairing) { a_pairing_data_ptr p = pairing->data; element_t V, V1; element_t z, z2; element_t f, f0, f1; element_t a, b, c; element_t e0; const element_ptr e1 = a, e2 = b, e3 = c; int i, n; element_ptr Vx, Vy; element_ptr V1x, V1y; element_ptr Qx = curve_x_coord(in2); element_ptr Qy = curve_y_coord(in2); //could save a couple of inversions by avoiding //this function and rewriting do_line() to handle projective coords //convert V from weighted projective (Jacobian) to affine //i.e. (X, Y, Z) --> (X/Z^2, Y/Z^3) //also sets z to 1 #define point_to_affine() \ element_invert(z, z); \ element_square(e0, z); \ element_mul(Vx, Vx, e0); \ element_mul(e0, e0, z); \ element_mul(Vy, Vy, e0); \ element_set1(z); \ element_set1(z2); #define proj_double() { \ /* e0 = 3x^2 + (cc->a) z^4 */ \ /* for this case a = 1 */ \ element_square(e0, Vx); \ /*element_mul_si(e0, e0, 3);*/ \ element_double(e1, e0); \ element_add(e0, e1, e0); \ element_square(e1, z2); \ element_add(e0, e0, e1); \ \ /* z_out = 2 y z */ \ element_mul(z, Vy, z); \ /*element_mul_si(z, z, 2);*/ \ element_double(z, z); \ element_square(z2, z); \ \ /* e1 = 4 x y^2 */ \ element_square(e2, Vy); \ element_mul(e1, Vx, e2); \ /*element_mul_si(e1, e1, 4);*/ \ element_double(e1, e1); \ element_double(e1, e1); \ \ /* x_out = e0^2 - 2 e1 */ \ element_double(e3, e1); \ element_square(Vx, e0); \ element_sub(Vx, Vx, e3); \ \ /* e2 = 8y^4 */ \ element_square(e2, e2); \ /*element_mul_si(e2, e2, 8);*/ \ element_double(e2, e2); \ element_double(e2, e2); \ element_double(e2, e2); \ \ /*y_out = e0(e1 - x_out) - e2*/\ element_sub(e1, e1, Vx); \ element_mul(e0, e0, e1); \ element_sub(Vy, e0, e2); \ } #define do_tangent() \ compute_abc_tangent_proj(a, b, c, Vx, Vy, z, z2, e0); \ a_miller_evalfn(f0, a, b, c, Qx, Qy); \ element_mul(f, f, f0); #define do_line() \ compute_abc_line(a, b, c, Vx, Vy, V1x, V1y, e0); \ a_miller_evalfn(f0, a, b, c, Qx, Qy); \ element_mul(f, f, f0); element_init(V, p->Eq); element_init(V1, p->Eq); element_set(V, in1); Vx = curve_x_coord(V); Vy = curve_y_coord(V); V1x = curve_x_coord(V1); V1y = curve_y_coord(V1); element_init(f, p->Fq2); element_init(f0, p->Fq2); element_init(f1, p->Fq2); element_set1(f); element_init(a, p->Fq); element_init(b, p->Fq); element_init(c, p->Fq); element_init(e0, p->Fq); element_init(z, p->Fq); element_init(z2, p->Fq); element_set1(z); element_set1(z2); n = p->exp1; for (i=0; isign1 < 0) { element_neg(V1, V); element_invert(f1, f); } else { element_set(V1, V); element_set(f1, f); } n = p->exp2; for (; iphikonr); element_clear(f); element_clear(f0); element_clear(f1); element_clear(z); element_clear(z2); element_clear(V); element_clear(V1); element_clear(a); element_clear(b); element_clear(c); element_clear(e0); #undef point_to_affine #undef proj_double #undef do_tangent #undef do_line } //in1, in2 are from E(F_q), out from F_q^2 static void a_pairing_affine(element_ptr out, element_ptr in1, element_ptr in2, pairing_t pairing) { a_pairing_data_ptr p = pairing->data; element_t V, V1; element_t f, f0, f1; element_t a, b, c; element_t e0; int i, n; element_ptr Qx = curve_x_coord(in2); element_ptr Qy = curve_y_coord(in2); element_ptr Vx, Vy; element_ptr V1x, V1y; #define do_tangent() \ compute_abc_tangent(a, b, c, Vx, Vy, e0); \ a_miller_evalfn(f0, a, b, c, Qx, Qy); \ element_mul(f, f, f0); #define do_line() \ compute_abc_line(a, b, c, Vx, Vy, V1x, V1y, e0); \ a_miller_evalfn(f0, a, b, c, Qx, Qy); \ element_mul(f, f, f0); element_init(V, p->Eq); element_init(V1, p->Eq); Vx = curve_x_coord(V); Vy = curve_y_coord(V); V1x = curve_x_coord(V1); V1y = curve_y_coord(V1); element_set(V, in1); element_init(f, p->Fq2); element_init(f0, p->Fq2); element_init(f1, p->Fq2); element_set1(f); element_init(a, p->Fq); element_init(b, p->Fq); element_init(c, p->Fq); element_init(e0, p->Fq); n = p->exp1; for (i=0; isign1 < 0) { element_neg(V1, V); element_invert(f1, f); } else { element_set(V1, V); element_set(f1, f); } n = p->exp2; for (; iphikonr); element_clear(f); element_clear(f0); element_clear(f1); element_clear(V); element_clear(V1); element_clear(a); element_clear(b); element_clear(c); element_clear(e0); #undef do_tangent #undef do_line } // On Computing Products of Pairing //in1, in2 are from E(F_q), out from F_q^2 void a_pairings_affine(element_ptr out, element_t in1[], element_t in2[], int n_prod, pairing_t pairing) { a_pairing_data_ptr p = pairing->data; element_t* V = pbc_malloc(sizeof(element_t)*n_prod); element_t* V1 = pbc_malloc(sizeof(element_t)*n_prod); element_t f, f0, f1; element_t a, b, c; element_t e0; int i, j, n; element_ptr Qx, Qy; element_ptr Vx, Vy; element_ptr V1x, V1y; #define do_tangents() \ for(j=0; jEq); element_init(V1[i],p->Eq); element_set(V[i],in1[i]); } element_init(f, p->Fq2); element_init(f0, p->Fq2); element_init(f1, p->Fq2); element_set1(f); element_init(a, p->Fq); element_init(b, p->Fq); element_init(c, p->Fq); element_init(e0, p->Fq); n = p->exp1; for (i=0; isign1 < 0) { for(j=0; jexp2; for (; iphikonr); element_clear(f); element_clear(f0); element_clear(f1); for(j=0;jGT); a_pairing_data_ptr p = pairing->data; field_clear(p->Eq); field_clear(p->Fq); field_clear(p->Fq2); pbc_free(p); mpz_clear(pairing->r); mpz_clear(pairing->phikonr); field_clear(pairing->Zr); } static void a_pairing_option_set(pairing_t pairing, char *key, char *value) { if (!strcmp(key, "method")) { if (!strcmp(value, "miller")) { pairing->map = a_pairing_proj; pairing->pp_init = a_pairing_pp_init; pairing->pp_clear = a_pairing_pp_clear; pairing->pp_apply = a_pairing_pp_apply; } else if (!strcmp(value, "miller-affine")) { pairing->map = a_pairing_affine; pairing->pp_init = a_pairing_pp_init; pairing->pp_clear = a_pairing_pp_clear; pairing->pp_apply = a_pairing_pp_apply; } else if (!strcmp(value, "shipsey-stange")) { pairing->map = a_pairing_ellnet; pairing->pp_init = a_pairing_ellnet_pp_init; pairing->pp_clear = a_pairing_ellnet_pp_clear; pairing->pp_apply = a_pairing_ellnet_pp_apply; } } } static void a_finalpow(element_t e) { pairing_ptr pairing = e->field->pairing; element_t t0, t1; element_init_same_as(t0, e->data); element_init_same_as(t1, e->data); a_tateexp(t0, e->data, t1, pairing->phikonr); element_set(e->data, t0); element_clear(t0); element_clear(t1); } static void a_init_pairing(pairing_ptr pairing, void *data) { a_param_ptr param = data; element_t a, b; a_pairing_data_ptr p; p = pairing->data = pbc_malloc(sizeof(*p)); p->exp2 = param->exp2; p->exp1 = param->exp1; p->sign1 = param->sign1; mpz_init(pairing->r); mpz_set(pairing->r, param->r); field_init_fp(pairing->Zr, pairing->r); pairing->map = a_pairing_proj; pairing->prod_pairings = a_pairings_affine; field_init_fp(p->Fq, param->q); element_init(a, p->Fq); element_init(b, p->Fq); element_set1(a); element_set0(b); field_init_curve_ab(p->Eq, a, b, pairing->r, param->h); element_clear(a); element_clear(b); field_init_fi(p->Fq2, p->Fq); //k=2, hence phi_k(q) = q + 1, phikonr = (q+1)/r mpz_init(pairing->phikonr); mpz_set(pairing->phikonr, param->h); pairing->G1 = p->Eq; pairing->G2 = pairing->G1; pairing->phi = phi_identity; pairing_GT_init(pairing, p->Fq2); pairing->finalpow = a_finalpow; pairing->clear_func = a_pairing_clear; pairing->option_set = a_pairing_option_set; pairing->pp_init = a_pairing_pp_init; pairing->pp_clear = a_pairing_pp_clear; pairing->pp_apply = a_pairing_pp_apply; } static void a_param_init(pbc_param_ptr par) { static pbc_param_interface_t interface = {{ a_clear, a_init_pairing, a_out_str, }}; par->api = interface; a_param_ptr p = par->data = pbc_malloc(sizeof(*p)); mpz_init(p->r); mpz_init(p->q); mpz_init(p->h); } // Public interface for type A pairings: int pbc_param_init_a(pbc_param_ptr par, struct symtab_s *tab) { a_param_init(par); a_param_ptr p = par->data; int err = 0; err += lookup_mpz(p->q, tab, "q"); err += lookup_mpz(p->r, tab, "r"); err += lookup_mpz(p->h, tab, "h"); err += lookup_int(&p->exp2, tab, "exp2"); err += lookup_int(&p->exp1, tab, "exp1"); err += lookup_int(&p->sign1, tab, "sign1"); err += lookup_int(&p->sign0, tab, "sign0"); return err; } void pbc_param_init_a_gen(pbc_param_ptr par, int rbits, int qbits) { a_param_init(par); a_param_ptr sp = par->data; int found = 0; mpz_ptr q = sp->q; mpz_ptr r = sp->r; mpz_ptr h = sp->h; do { int i; mpz_set_ui(r, 0); if (rand() % 2) { sp->exp2 = rbits - 1; sp->sign1 = 1; } else { sp->exp2 = rbits; sp->sign1 = -1; } mpz_setbit(r, sp->exp2); //use q as a temp variable mpz_set_ui(q, 0); sp->exp1 = (rand() % (sp->exp2 - 1)) + 1; mpz_setbit(q, sp->exp1); if (sp->sign1 > 0) { mpz_add(r, r, q); } else { mpz_sub(r, r, q); } if (rand() % 2) { sp->sign0 = 1; mpz_add_ui(r, r, 1); } else { sp->sign0 = -1; mpz_sub_ui(r, r, 1); } if (!mpz_probab_prime_p(r, 10)) continue; for (i=0; i<10; i++) { int bit; //use q as a temp variable mpz_set_ui(q, 0); bit = qbits - rbits - 4 + 1; if (bit < 3) bit = 3; mpz_setbit(q, bit); pbc_mpz_random(h, q); mpz_mul_ui(h, h, 12); //finally q takes the value it should mpz_mul(q, h, r); mpz_sub_ui(q, q, 1); if (mpz_probab_prime_p(q, 10)) { found = 1; break; } } } while (!found); } // Type A1 pairings: struct a1_param_s { mpz_t p; mpz_t n; int l; }; typedef struct a1_param_s a1_param_t[1]; typedef struct a1_param_s *a1_param_ptr; struct a1_pairing_data_s { field_t Fp, Fp2, Ep; }; typedef struct a1_pairing_data_s a1_pairing_data_t[1]; typedef struct a1_pairing_data_s *a1_pairing_data_ptr; static void a1_clear(void *data) { a1_param_ptr param = data; mpz_clear(param->p); mpz_clear(param->n); pbc_free(data); } static void a1_out_str(FILE *stream, void *data) { a1_param_ptr p = data; param_out_type(stream, "a1"); param_out_mpz(stream, "p", p->p); param_out_mpz(stream, "n", p->n); param_out_int(stream, "l", p->l); } struct pp2_coeff_s { element_t cx2; element_t cy2; element_t cxy; element_t cx; element_t cy; element_t c; }; typedef struct pp2_coeff_s pp2_coeff_t[1]; typedef struct pp2_coeff_s *pp2_coeff_ptr; static void pp2_coeff_set(pp2_coeff_ptr p, element_t cx2, element_t cy2, element_t cxy, element_t cx, element_t cy, element_t c) { element_init(p->cx2, cx2->field); element_init(p->cy2, cy2->field); element_init(p->cxy, cxy->field); element_init(p->cx, cx->field); element_init(p->cy, cy->field); element_init(p->c, c->field); element_set(p->cx2, cx2); element_set(p->cy2, cy2); element_set(p->cxy, cxy); element_set(p->cx, cx); element_set(p->cy, cy); element_set(p->c, c); } static void a1_pairing_pp_clear(pairing_pp_t p) { void **pp = p->data; while (*pp) { pbc_free(*pp); pp++; } pbc_free(p->data); } static void a1_pairing_pp_init(pairing_pp_t p, element_ptr in1, pairing_t pairing) { int m; element_ptr Px = curve_x_coord(in1); element_ptr Py = curve_y_coord(in1); a1_pairing_data_ptr a1info = pairing->data; p->data = pbc_malloc(sizeof(void *) * mpz_sizeinbase(pairing->r, 2)); void **pp = p->data; element_t V; element_t a, b, c; element_t a2, b2, c2; element_t e0, e1, e2; element_ptr Vx, Vy; #define do_tangent() compute_abc_tangent(a, b, c, Vx, Vy, e0); #define do_line() compute_abc_line(a2, b2, c2, Vx, Vy, Px, Py, e0); element_init(V, a1info->Ep); element_set(V, in1); Vx = curve_x_coord(V); Vy = curve_y_coord(V); element_init(a, a1info->Fp); element_init(b, a1info->Fp); element_init(c, a1info->Fp); element_init(e0, a1info->Fp); element_init(e1, a1info->Fp); element_init(e2, a1info->Fp); element_init(a2, a1info->Fp); element_init(b2, a1info->Fp); element_init(c2, a1info->Fp); m = mpz_sizeinbase(pairing->r, 2) - 2; for(;;) { do_tangent(); if (!m) break; element_double(V, V); if (mpz_tstbit(pairing->r, m)) { do_line(); element_add(V, V, in1); //preprocess two at once //e0 = coeff of x element_mul(e0, a, c2); element_mul(e1, a2, c); element_add(e0, e0, e1); //e1 = coeff of y element_mul(e1, b2, c); element_mul(e2, b, c2); element_add(e1, e1, e2); //c = constant term element_mul(c, c, c2); //c2 = coeff of xy element_mul(c2, a, b2); element_mul(e2, a2, b); element_add(c2, c2, e2); //a = coeff of x^2 element_mul(a, a, a2); //b = coeff of y^2 element_mul(b, b, b2); *pp = pbc_malloc(sizeof(pp2_coeff_t)); pp2_coeff_set(*pp, a, b, c2, e0, e1, c); } else { *pp = pbc_malloc(sizeof(pp_coeff_t)); pp_coeff_set(*pp, a, b, c); } pp++; m--; } *pp = pbc_malloc(sizeof(pp_coeff_t)); pp_coeff_set(*pp, a, b, c); pp++; *pp = NULL; element_clear(a2); element_clear(b2); element_clear(c2); element_clear(e2); element_clear(e1); element_clear(e0); element_clear(a); element_clear(b); element_clear(c); element_clear(V); #undef do_tangent #undef do_line } static void a1_pairing_pp_apply(element_ptr out, element_ptr in2, pairing_pp_t p) { void **pp = p->data; a1_pairing_data_ptr a1info = p->pairing->data; element_t f, f0; element_t e0, e1; int m; element_ptr Qx = curve_x_coord(in2); element_ptr Qy = curve_y_coord(in2); element_t Qx2, Qy2, Qxy; #define do_tangent() \ pp_coeff_ptr ppp = *pp; \ a_miller_evalfn(f0, ppp->a, ppp->b, ppp->c, Qx, Qy); #define do_line() { \ pp2_coeff_ptr ppp = *pp; \ /*we'll map Q via (x,y) --> (-x, iy) */ \ /*hence Qx^2 = x^2, Qy^2 = -y^2, Qx Qy = -ixy */\ /*where x = Q'x, y = Q'y */ \ \ /* Re = cx2 x^2 - cy2 y^2 - cx x + c */ \ /* Im = -cxy xy + cy y */ \ element_mul(e0, ppp->cx2, Qx2); \ element_mul(e1, ppp->cy2, Qy2); \ element_sub(e0, e0, e1); \ element_mul(e1, ppp->cx, Qx); \ element_sub(e0, e0, e1); \ element_add(element_x(f0), e0, ppp->c); \ \ element_mul(e0, ppp->cy, Qy); \ element_mul(e1, ppp->cxy, Qxy); \ element_sub(element_y(f0), e0, e1); \ } element_init(f, out->field); element_init(f0, out->field); element_set1(f); element_init(e0, a1info->Fp); element_init(e1, a1info->Fp); element_init(Qx2, a1info->Fp); element_init(Qy2, a1info->Fp); element_init(Qxy, a1info->Fp); element_square(Qx2, Qx); element_square(Qy2, Qy); element_mul(Qxy, Qx, Qy); m = mpz_sizeinbase(p->pairing->r, 2) - 2; while (m > 0) { if (mpz_tstbit(p->pairing->r, m)) { do_line(); } else { do_tangent(); } element_mul(f, f, f0); pp++; m--; element_square(f, f); } do_tangent(); element_mul(f, f, f0); //Tate exponentiation //simpler but slower: //element_pow_mpz(out, f, p->tateexp); //use this trick instead: element_invert(f0, f); element_neg(element_y(f), element_y(f)); element_mul(f, f, f0); element_pow_mpz(out, f, p->pairing->phikonr); /* We could use this instead but p->h is small so this does not help much a_tateexp(out, f, f0, p->h); */ element_clear(Qx2); element_clear(Qy2); element_clear(Qxy); element_clear(f); element_clear(f0); element_clear(e1); element_clear(e0); #undef do_tangent #undef do_line } // e0 is a temp var. // Mixed coordinates. static void compute_abc_line_proj(element_ptr a, element_ptr b, element_ptr c, element_ptr Vx, element_ptr Vy, element_ptr z, element_ptr z2, element_ptr V1x, element_ptr V1y, element_ptr e0) { //temporally used to store Z1^3 element_mul(c,z,z2); //a = Y1-Y2*Z1^3 element_mul(e0,V1y,c); element_sub(a,Vy,e0); //b = -(X1*Z1-X2*Z1^3) element_mul(b,c,V1x); element_mul(e0,Vx,z); element_sub(b,b,e0); //c = -(Y2*b+X2*a) element_mul(c,b,V1y); element_mul(e0,a,V1x); element_add(c,c,e0); element_neg(c,c); } // in1, in2 are from E(F_q), out from F_q^2 static void a1_pairing_proj(element_ptr out, element_ptr in1, element_ptr in2, pairing_t pairing) { a1_pairing_data_ptr p = pairing->data; element_t V; element_t z, z2; element_t f, f0; element_t a, b, c; element_t e0; const element_ptr e1 = a, e2 = b, e3 = c; // used in point_to_affine() etc. int m; element_ptr Px = curve_x_coord(in1); element_ptr Py = curve_y_coord(in1); element_ptr Qx = curve_x_coord(in2); element_ptr Qy = curve_y_coord(in2); element_ptr Vx; element_ptr Vy; #define point_to_affine() \ element_invert(z, z); \ element_square(e0, z); \ element_mul(Vx, Vx, e0); \ element_mul(e0, e0, z); \ element_mul(Vy, Vy, e0); \ element_set1(z); \ element_set1(z2); //TODO: do I need to check if V=-in1? //Where V=(Vx,Vy,z) and in1=(Px,Py,1), a mixed coordinates. #define proj_add() { \ /* H=X2*Z1^2-X1 */ \ element_mul(e0,Px,z2); \ element_sub(e0,e0,Vx); \ /* H^2 */ \ element_square(e1,e0); \ /* r=Y2*Z1^3-Y1 */ \ element_mul(e2,z,z2); \ element_mul(e2,e2,Py); \ element_sub(e2,e2,Vy); \ \ /* X3=r^2-H^3-2X1*H^2 */ \ element_set(z2,Vx); /* use z2 to store X1 and update Vx=X3 */ \ element_square(Vx,e2); \ element_mul(e3,e0,e1); /* e3=H^3 */ \ element_sub(Vx,Vx,e3); \ element_double(e3,z2); \ element_mul(e3,e3,e1); /* 2X1*H^2 */ \ element_sub(Vx,Vx,e3); \ /* Y3=r(X1*H^2-X3)-Y1*H^3 */ \ element_mul(e3,z2,e1); \ element_sub(e3,e3,Vx); \ element_mul(e3,e3,e2); \ element_mul(e2,e0,e1); /* e2 no longer used. */ \ element_mul(e2,e2,Vy); \ element_sub(Vy,e3,e2); \ /* Z3=Z1*H */ \ element_mul(z,z,e0); \ element_square(z2,z); \ } #define proj_double() { \ /* e0 = 3x^2 + (cc->a) z^4 */ \ /* for this case a = 1 */ \ element_square(e0, Vx); \ /* element_mul_si(e0, e0, 3); */ \ element_double(e1, e0); \ element_add(e0, e1, e0); \ element_square(e1, z2); \ element_add(e0, e0, e1); \ \ /* z_out = 2 y z */ \ element_mul(z, Vy, z); \ /* element_mul_si(z, z, 2); */ \ element_double(z, z); \ element_square(z2, z); \ \ /* e1 = 4 x y^2 */ \ element_square(e2, Vy); \ element_mul(e1, Vx, e2); \ /* element_mul_si(e1, e1, 4); */ \ element_double(e1, e1); \ element_double(e1, e1); \ \ /* x_out = e0^2 - 2 e1 */ \ element_double(e3, e1); \ element_square(Vx, e0); \ element_sub(Vx, Vx, e3); \ \ /* e2 = 8y^4 */ \ element_square(e2, e2); \ /* element_mul_si(e2, e2, 8); */ \ element_double(e2, e2); \ element_double(e2, e2); \ element_double(e2, e2); \ \ /* y_out = e0(e1 - x_out) - e2 */ \ element_sub(e1, e1, Vx); \ element_mul(e0, e0, e1); \ element_sub(Vy, e0, e2); \ } #define do_tangent() { \ compute_abc_tangent_proj(a, b, c, Vx, Vy, z, z2, e0); \ a_miller_evalfn(f0, a, b, c, Qx, Qy); \ element_mul(f, f, f0); \ } #define do_line() { \ compute_abc_line_proj(a, b, c, Vx, Vy, z, z2, Px, Py, e0); \ a_miller_evalfn(f0, a, b, c, Qx, Qy); \ element_mul(f, f, f0); \ } element_init(V, p->Ep); element_set(V, in1); Vx = curve_x_coord(V); Vy = curve_y_coord(V); element_init(f, p->Fp2); element_init(f0, p->Fp2); element_set1(f); element_init(a, p->Fp); element_init(b, p->Fp); element_init(c, p->Fp); element_init(e0, p->Fp); element_init(z, p->Fp); element_init(z2, p->Fp); element_set1(z); element_set1(z2); m = mpz_sizeinbase(pairing->r, 2) - 2; //TODO: sliding NAF for(;;) { do_tangent(); if (!m) break; proj_double(); //V=2V if (mpz_tstbit(pairing->r, m)) { // point_to_affine(); do_line(); proj_add(); //V=V+in1 } m--; element_square(f, f); } // Tate exponentiation. // Simpler but slower: // element_pow_mpz(out, f, p->tateexp); // Use this trick instead: element_invert(f0, f); element_neg(element_y(f), element_y(f)); element_mul(f, f, f0); element_pow_mpz(out, f, pairing->phikonr); /* We could use this instead but p->h is small so this does not help much a_tateexp(out, f, f0, p->h); */ element_clear(f); element_clear(f0); element_clear(z); element_clear(z2); element_clear(V); element_clear(a); element_clear(b); element_clear(c); element_clear(e0); #undef point_to_affine #undef proj_add #undef proj_double #undef do_tangent #undef do_line } //in1, in2 are from E(F_q), out from F_q^2 static void a1_pairing(element_ptr out, element_ptr in1, element_ptr in2, pairing_t pairing) { a1_pairing_data_ptr p = pairing->data; element_t V; element_t f, f0; element_t a, b, c; element_t e0; int m; element_ptr Px = curve_x_coord(in1); element_ptr Py = curve_y_coord(in1); element_ptr Qx = curve_x_coord(in2); element_ptr Qy = curve_y_coord(in2); element_ptr Vx; element_ptr Vy; #define do_tangent() { \ compute_abc_tangent(a, b, c, Vx, Vy, e0); \ a_miller_evalfn(f0, a, b, c, Qx, Qy); \ element_mul(f, f, f0); \ } #define do_line() { \ compute_abc_line(a, b, c, Vx, Vy, Px, Py, e0); \ a_miller_evalfn(f0, a, b, c, Qx, Qy); \ element_mul(f, f, f0); \ } element_init(V, p->Ep); element_set(V, in1); Vx = curve_x_coord(V); Vy = curve_y_coord(V); element_init(f, p->Fp2); element_init(f0, p->Fp2); element_set1(f); element_init(a, p->Fp); element_init(b, p->Fp); element_init(c, p->Fp); element_init(e0, p->Fp); m = mpz_sizeinbase(pairing->r, 2) - 2; //TODO: sliding NAF for(;;) { do_tangent(); if (!m) break; element_double(V, V); if (mpz_tstbit(pairing->r, m)) { do_line(); element_add(V, V, in1); } m--; element_square(f, f); } // Tate exponentiation. // Simpler but slower: // element_pow_mpz(out, f, p->tateexp); // Use this trick instead: element_invert(f0, f); element_neg(element_y(f), element_y(f)); element_mul(f, f, f0); element_pow_mpz(out, f, pairing->phikonr); /* We could use this instead but p->h is small so this does not help much a_tateexp(out, f, f0, p->h); */ element_clear(f); element_clear(f0); element_clear(V); element_clear(a); element_clear(b); element_clear(c); element_clear(e0); #undef do_tangent #undef do_line } //in1, in2 are from E(F_q), out from F_q^2 void a1_pairings_affine(element_ptr out, element_t in1[], element_t in2[], int n_prod, pairing_t pairing) { a1_pairing_data_ptr p = pairing->data; element_t* V = pbc_malloc(sizeof(element_t)*n_prod); element_t f, f0; element_t a, b, c; element_t e0; int m, i; element_ptr Px, Py; element_ptr Qx, Qy; element_ptr Vx, Vy; #define do_tangents() { \ for(i=0; iEp); element_set(V[i], in1[i]); } element_init(f, p->Fp2); element_init(f0, p->Fp2); element_set1(f); element_init(a, p->Fp); element_init(b, p->Fp); element_init(c, p->Fp); element_init(e0, p->Fp); m = mpz_sizeinbase(pairing->r, 2) - 2; //TODO: sliding NAF for(;;) { do_tangents(); if (!m) break; element_multi_double(V, V, n_prod); if (mpz_tstbit(pairing->r, m)) { do_lines(); element_multi_add(V, V, in1, n_prod); } m--; element_square(f, f); } // Tate exponentiation. // Simpler but slower: // element_pow_mpz(out, f, p->tateexp); // Use this trick instead: element_invert(f0, f); element_neg(element_y(f), element_y(f)); element_mul(f, f, f0); element_pow_mpz(out, f, pairing->phikonr); /* We could use this instead but p->h is small so this does not help much a_tateexp(out, f, f0, p->h); */ element_clear(f); element_clear(f0); for(i=0; iGT); a1_pairing_data_ptr p = pairing->data; field_clear(p->Ep); field_clear(p->Fp2); field_clear(p->Fp); pbc_free(p); mpz_clear(pairing->phikonr); mpz_clear(pairing->r); field_clear(pairing->Zr); } static void a1_pairing_option_set(pairing_t pairing, char *key, char *value) { if (!strcmp(key, "method")) { if (!strcmp(value, "miller")) { pairing->map = a1_pairing_proj; pairing->pp_init = a1_pairing_pp_init; pairing->pp_clear = a1_pairing_pp_clear; pairing->pp_apply = a1_pairing_pp_apply; } else if (!strcmp(value, "miller-affine")){ pairing->map = a1_pairing; pairing->pp_init = a1_pairing_pp_init; pairing->pp_clear = a1_pairing_pp_clear; pairing->pp_apply = a1_pairing_pp_apply; } else if (!strcmp(value, "shipsey-stange")) { pairing->map = a_pairing_ellnet; pairing->pp_init = a_pairing_ellnet_pp_init; pairing->pp_clear = a_pairing_ellnet_pp_clear; pairing->pp_apply = a_pairing_ellnet_pp_apply; } } } static void a1_init_pairing(pairing_t pairing, void *data) { a1_param_ptr param = data; element_t a, b; mpz_init(pairing->r); mpz_set(pairing->r, param->n); field_init_fp(pairing->Zr, pairing->r); a1_pairing_data_ptr p; p = pairing->data = pbc_malloc(sizeof(a1_pairing_data_t)); //k=2, hence phi_k(q) = q + 1, phikonr = (q+1)/r mpz_init(pairing->phikonr); mpz_set_ui(pairing->phikonr, param->l); field_init_fp(p->Fp, param->p); element_init(a, p->Fp); element_init(b, p->Fp); element_set1(a); element_set0(b); field_init_curve_ab(p->Ep, a, b, pairing->r, pairing->phikonr); // Turns out to be faster. field_curve_use_random_solvefory(p->Ep); element_clear(a); element_clear(b); field_init_fi(p->Fp2, p->Fp); pairing->finalpow = a_finalpow; pairing->G1 = pbc_malloc(sizeof(field_t)); pairing->G2 = pairing->G1 = p->Ep; pairing_GT_init(pairing, p->Fp2); pairing->map = a1_pairing_proj; //default uses projective coordinates. pairing->phi = phi_identity; pairing->prod_pairings = a1_pairings_affine; pairing->clear_func = a1_pairing_clear; pairing->pp_init = a1_pairing_pp_init; pairing->pp_clear = a1_pairing_pp_clear; pairing->pp_apply = a1_pairing_pp_apply; pairing->option_set = a1_pairing_option_set; } static void a1_init(pbc_param_t p) { static pbc_param_interface_t interface = {{ a1_clear, a1_init_pairing, a1_out_str, }}; p->api = interface; a1_param_ptr param = p->data = pbc_malloc(sizeof(*param)); mpz_init(param->p); mpz_init(param->n); } // Public interface: int pbc_param_init_a1(pbc_param_ptr par, struct symtab_s *tab) { a1_init(par); a1_param_ptr p = par->data; int err = 0; err += lookup_mpz(p->p, tab, "p"); err += lookup_mpz(p->n, tab, "n"); err += lookup_int(&p->l, tab, "l"); return err; } void pbc_param_init_a1_gen(pbc_param_ptr par, mpz_t order) { a1_init(par); a1_param_ptr param = par->data; // If order is even, ideally check all even l, not just multiples of 4 // but I don't see a good reason for having an even order. unsigned int l = 4; mpz_t n; mpz_ptr p = param->p; mpz_init(n); mpz_mul_ui(n, order, 4); mpz_sub_ui(p, n, 1); for (;;) { if (mpz_probab_prime_p(p, 20)) { break; } mpz_add(p, p, n); l += 4; } param->l = l; mpz_set(param->n, order); mpz_clear(n); }