1 /* LibTomMath, multiple-precision integer library -- Tom St Denis
\r
3 * LibTomMath is library that provides for multiple-precision
\r
4 * integer arithmetic as well as number theoretic functionality.
\r
6 * The library is designed directly after the MPI library by
\r
7 * Michael Fromberger but has been written from scratch with
\r
8 * additional optimizations in place.
\r
10 * The library is free for all purposes without any express
\r
11 * guarantee it works.
\r
13 * Tom St Denis, tomstdenis@iahu.ca, http://math.libtomcrypt.org
\r
15 #include <tommath.h>
\r
17 /* multiplication using the Toom-Cook 3-way algorithm */
\r
19 mp_toom_mul(mp_int *a, mp_int *b, mp_int *c)
\r
21 mp_int w0, w1, w2, w3, w4, tmp1, tmp2, a0, a1, a2, b0, b1, b2;
\r
25 if ((res = mp_init_multi(&w0, &w1, &w2, &w3, &w4,
\r
26 &a0, &a1, &a2, &b0, &b1,
\r
27 &b2, &tmp1, &tmp2, NULL)) != MP_OKAY) {
\r
32 B = MIN(a->used, b->used) / 3;
\r
34 /* a = a2 * B**2 + a1 * B + a0 */
\r
35 if ((res = mp_mod_2d(a, DIGIT_BIT * B, &a0)) != MP_OKAY) {
\r
39 if ((res = mp_copy(a, &a1)) != MP_OKAY) {
\r
43 mp_mod_2d(&a1, DIGIT_BIT * B, &a1);
\r
45 if ((res = mp_copy(a, &a2)) != MP_OKAY) {
\r
50 /* b = b2 * B**2 + b1 * B + b0 */
\r
51 if ((res = mp_mod_2d(b, DIGIT_BIT * B, &b0)) != MP_OKAY) {
\r
55 if ((res = mp_copy(b, &b1)) != MP_OKAY) {
\r
59 mp_mod_2d(&b1, DIGIT_BIT * B, &b1);
\r
61 if ((res = mp_copy(b, &b2)) != MP_OKAY) {
\r
67 if ((res = mp_mul(&a0, &b0, &w0)) != MP_OKAY) {
\r
72 if ((res = mp_mul(&a2, &b2, &w4)) != MP_OKAY) {
\r
76 /* w1 = (a2 + 2(a1 + 2a0))(b2 + 2(b1 + 2b0)) */
\r
77 if ((res = mp_mul_2(&a0, &tmp1)) != MP_OKAY) {
\r
80 if ((res = mp_add(&tmp1, &a1, &tmp1)) != MP_OKAY) {
\r
83 if ((res = mp_mul_2(&tmp1, &tmp1)) != MP_OKAY) {
\r
86 if ((res = mp_add(&tmp1, &a2, &tmp1)) != MP_OKAY) {
\r
90 if ((res = mp_mul_2(&b0, &tmp2)) != MP_OKAY) {
\r
93 if ((res = mp_add(&tmp2, &b1, &tmp2)) != MP_OKAY) {
\r
96 if ((res = mp_mul_2(&tmp2, &tmp2)) != MP_OKAY) {
\r
99 if ((res = mp_add(&tmp2, &b2, &tmp2)) != MP_OKAY) {
\r
103 if ((res = mp_mul(&tmp1, &tmp2, &w1)) != MP_OKAY) {
\r
107 /* w3 = (a0 + 2(a1 + 2a2))(b0 + 2(b1 + 2b2)) */
\r
108 if ((res = mp_mul_2(&a2, &tmp1)) != MP_OKAY) {
\r
111 if ((res = mp_add(&tmp1, &a1, &tmp1)) != MP_OKAY) {
\r
114 if ((res = mp_mul_2(&tmp1, &tmp1)) != MP_OKAY) {
\r
117 if ((res = mp_add(&tmp1, &a0, &tmp1)) != MP_OKAY) {
\r
121 if ((res = mp_mul_2(&b2, &tmp2)) != MP_OKAY) {
\r
124 if ((res = mp_add(&tmp2, &b1, &tmp2)) != MP_OKAY) {
\r
127 if ((res = mp_mul_2(&tmp2, &tmp2)) != MP_OKAY) {
\r
130 if ((res = mp_add(&tmp2, &b0, &tmp2)) != MP_OKAY) {
\r
134 if ((res = mp_mul(&tmp1, &tmp2, &w3)) != MP_OKAY) {
\r
139 /* w2 = (a2 + a1 + a0)(b2 + b1 + b0) */
\r
140 if ((res = mp_add(&a2, &a1, &tmp1)) != MP_OKAY) {
\r
143 if ((res = mp_add(&tmp1, &a0, &tmp1)) != MP_OKAY) {
\r
146 if ((res = mp_add(&b2, &b1, &tmp2)) != MP_OKAY) {
\r
149 if ((res = mp_add(&tmp2, &b0, &tmp2)) != MP_OKAY) {
\r
152 if ((res = mp_mul(&tmp1, &tmp2, &w2)) != MP_OKAY) {
\r
156 /* now solve the matrix
\r
164 using 12 subtractions, 4 shifts,
\r
165 2 small divisions and 1 small multiplication
\r
169 if ((res = mp_sub(&w1, &w4, &w1)) != MP_OKAY) {
\r
173 if ((res = mp_sub(&w3, &w0, &w3)) != MP_OKAY) {
\r
177 if ((res = mp_div_2(&w1, &w1)) != MP_OKAY) {
\r
181 if ((res = mp_div_2(&w3, &w3)) != MP_OKAY) {
\r
185 if ((res = mp_sub(&w2, &w0, &w2)) != MP_OKAY) {
\r
188 if ((res = mp_sub(&w2, &w4, &w2)) != MP_OKAY) {
\r
192 if ((res = mp_sub(&w1, &w2, &w1)) != MP_OKAY) {
\r
196 if ((res = mp_sub(&w3, &w2, &w3)) != MP_OKAY) {
\r
200 if ((res = mp_mul_2d(&w0, 3, &tmp1)) != MP_OKAY) {
\r
203 if ((res = mp_sub(&w1, &tmp1, &w1)) != MP_OKAY) {
\r
207 if ((res = mp_mul_2d(&w4, 3, &tmp1)) != MP_OKAY) {
\r
210 if ((res = mp_sub(&w3, &tmp1, &w3)) != MP_OKAY) {
\r
213 /* 3r2 - r1 - r3 */
\r
214 if ((res = mp_mul_d(&w2, 3, &w2)) != MP_OKAY) {
\r
217 if ((res = mp_sub(&w2, &w1, &w2)) != MP_OKAY) {
\r
220 if ((res = mp_sub(&w2, &w3, &w2)) != MP_OKAY) {
\r
224 if ((res = mp_sub(&w1, &w2, &w1)) != MP_OKAY) {
\r
228 if ((res = mp_sub(&w3, &w2, &w3)) != MP_OKAY) {
\r
232 if ((res = mp_div_3(&w1, &w1, NULL)) != MP_OKAY) {
\r
236 if ((res = mp_div_3(&w3, &w3, NULL)) != MP_OKAY) {
\r
240 /* at this point shift W[n] by B*n */
\r
241 if ((res = mp_lshd(&w1, 1*B)) != MP_OKAY) {
\r
244 if ((res = mp_lshd(&w2, 2*B)) != MP_OKAY) {
\r
247 if ((res = mp_lshd(&w3, 3*B)) != MP_OKAY) {
\r
250 if ((res = mp_lshd(&w4, 4*B)) != MP_OKAY) {
\r
254 if ((res = mp_add(&w0, &w1, c)) != MP_OKAY) {
\r
257 if ((res = mp_add(&w2, &w3, &tmp1)) != MP_OKAY) {
\r
260 if ((res = mp_add(&w4, &tmp1, &tmp1)) != MP_OKAY) {
\r
263 if ((res = mp_add(&tmp1, c, c)) != MP_OKAY) {
\r
268 mp_clear_multi(&w0, &w1, &w2, &w3, &w4,
\r
269 &a0, &a1, &a2, &b0, &b1,
\r
270 &b2, &tmp1, &tmp2, NULL);
\r