1 /* ecc-mul-m.c
2 
3    Point multiplication using Montgomery curve representation.
4 
5    Copyright (C) 2014 Niels Möller
6 
7    This file is part of GNU Nettle.
8 
9    GNU Nettle is free software: you can redistribute it and/or
10    modify it under the terms of either:
11 
12      * the GNU Lesser General Public License as published by the Free
13        Software Foundation; either version 3 of the License, or (at your
14        option) any later version.
15 
16    or
17 
18      * the GNU General Public License as published by the Free
19        Software Foundation; either version 2 of the License, or (at your
20        option) any later version.
21 
22    or both in parallel, as here.
23 
24    GNU Nettle is distributed in the hope that it will be useful,
25    but WITHOUT ANY WARRANTY; without even the implied warranty of
26    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
27    General Public License for more details.
28 
29    You should have received copies of the GNU General Public License and
30    the GNU Lesser General Public License along with this program.  If
31    not, see http://www.gnu.org/licenses/.
32 */
33 
34 #if HAVE_CONFIG_H
35 # include "config.h"
36 #endif
37 
38 #include <assert.h>
39 
40 #include "ecc.h"
41 #include "ecc-internal.h"
42 
43 void
ecc_mul_m(const struct ecc_modulo * m,mp_limb_t a24,unsigned bit_low,unsigned bit_high,mp_limb_t * qx,const uint8_t * n,const mp_limb_t * px,mp_limb_t * scratch)44 ecc_mul_m (const struct ecc_modulo *m,
45 	   mp_limb_t a24,
46 	   unsigned bit_low, unsigned bit_high,
47 	   mp_limb_t *qx, const uint8_t *n, const mp_limb_t *px,
48 	   mp_limb_t *scratch)
49 {
50   unsigned i;
51   mp_limb_t swap;
52 
53 #define x2 (scratch)
54 #define z2 (scratch + m->size)
55 #define x3 (scratch + 2*m->size)
56 #define z3 (scratch + 3*m->size)
57 
58   /* Formulas from RFC 7748:
59 
60        A = x_2 + z_2
61        AA = A^2
62        B = x_2 - z_2
63        BB = B^2
64        E = AA - BB
65        C = x_3 + z_3
66        D = x_3 - z_3
67        DA = D * A
68        CB = C * B
69        x_3 = (DA + CB)^2
70        z_3 = x_1 * (DA - CB)^2
71        x_2 = AA * BB
72        z_2 = E * (AA + a24 * E)
73 
74      For pure doubling, we use:
75 
76        A = x_2 + z_2
77        AA = A^2
78        B = x_2 - z_2
79        BB = B^2
80        E = AA - BB
81        x3 = AA * BB
82        z3 =  E * (AA + a24 * E)
83   */
84 
85 #define A (scratch + 4*m->size)
86 #define AA A
87 #define D (scratch + 5*m->size)
88 #define DA D
89 
90 #define tp (scratch + 6*m->size)
91 
92   /* For the doubling formulas. */
93 #define B D
94 #define BB D
95 #define E D
96 
97   /* Initialize, x2 = px, z2 = 1 */
98   mpn_copyi (x2, px, m->size);
99   z2[0] = 1;
100   mpn_zero (z2+1, m->size - 1);
101 
102   /* Get x3, z3 from doubling. Since most significant bit is forced to 1. */
103   ecc_mod_add (m, A, x2, z2);
104   ecc_mod_sub (m, B, x2, z2);
105   ecc_mod_sqr (m, AA, A, tp);
106   ecc_mod_sqr (m, BB, B, tp);
107   ecc_mod_mul (m, x3, AA, BB, tp);
108   ecc_mod_sub (m, E, AA, BB);
109   ecc_mod_addmul_1 (m, AA, E, a24);
110   ecc_mod_mul (m, z3, E, AA, tp);
111 
112   for (i = bit_high, swap = 0; i >= bit_low; i--)
113     {
114       mp_limb_t bit = (n[i/8] >> (i & 7)) & 1;
115 
116       mpn_cnd_swap (swap ^ bit, x2, x3, 2*m->size);
117       swap = bit;
118 
119       ecc_mod_add (m, A, x2, z2);
120       ecc_mod_sub (m, D, x3, z3);
121       ecc_mod_mul (m, DA, D, A, tp);
122       ecc_mod_sqr (m, AA, A, tp);
123 
124       /* Store B, BB and E at z2 */
125       ecc_mod_sub (m, z2, x2, z2);	/* B */
126       /* Store C and CB at z3 */
127       ecc_mod_add (m, z3, x3, z3);	/* C */
128       ecc_mod_mul (m, z3, z3, z2, tp);	/* CB */
129       ecc_mod_sqr (m, z2, z2, tp);	/* BB */
130 
131       /* Finish x2 */
132       ecc_mod_mul (m, x2, AA, z2, tp);
133 
134       ecc_mod_sub (m, z2, AA, z2);	/* E */
135 
136       /* Finish z2 */
137       ecc_mod_addmul_1 (m, AA, z2, a24);
138       ecc_mod_mul (m, z2, z2, AA, tp);
139 
140       /* Finish x3 */
141       ecc_mod_add (m, x3, DA, z3);
142       ecc_mod_sqr (m, x3, x3, tp);
143 
144       /* Finish z3 */
145       ecc_mod_sub (m, z3, DA, z3);	/* DA - CB */
146       ecc_mod_sqr (m, z3, z3, tp);
147       ecc_mod_mul (m, z3, z3, px, tp);
148     }
149   mpn_cnd_swap (swap, x2, x3, 2*m->size);
150 
151   /* Do the low zero bits, just duplicating x2 */
152   for (i = 0; i < bit_low; i++)
153     {
154       ecc_mod_add (m, A, x2, z2);
155       ecc_mod_sub (m, B, x2, z2);
156       ecc_mod_sqr (m, AA, A, tp);
157       ecc_mod_sqr (m, BB, B, tp);
158       ecc_mod_mul (m, x2, AA, BB, tp);
159       ecc_mod_sub (m, E, AA, BB);
160       ecc_mod_addmul_1 (m, AA, E, a24);
161       ecc_mod_mul (m, z2, E, AA, tp);
162     }
163   assert (m->invert_itch <= 7 * m->size);
164   m->invert (m, x3, z2, z3 + m->size);
165   ecc_mod_mul_canonical (m, qx, x2, x3, z3);
166 }
167