1 | // Copyright 2016-2023 Brian Smith. |
2 | // |
3 | // Permission to use, copy, modify, and/or distribute this software for any |
4 | // purpose with or without fee is hereby granted, provided that the above |
5 | // copyright notice and this permission notice appear in all copies. |
6 | // |
7 | // THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES |
8 | // WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF |
9 | // MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY |
10 | // SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES |
11 | // WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION |
12 | // OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN |
13 | // CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. |
14 | |
15 | use super::{ |
16 | elem::{binary_op, binary_op_assign}, |
17 | elem_sqr_mul, elem_sqr_mul_acc, PublicModulus, *, |
18 | }; |
19 | |
20 | pub(super) const NUM_LIMBS: usize = 384 / LIMB_BITS; |
21 | |
22 | pub static COMMON_OPS: CommonOps = CommonOps { |
23 | num_limbs: elem::NumLimbs::P384, |
24 | |
25 | q: PublicModulus { |
26 | p: limbs_from_hex("fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffeffffffff0000000000000000ffffffff" ), |
27 | rr: PublicElem::from_hex("10000000200000000fffffffe000000000000000200000000fffffffe00000001" ), |
28 | }, |
29 | n: PublicElem::from_hex("ffffffffffffffffffffffffffffffffffffffffffffffffc7634d81f4372ddf581a0db248b0a77aecec196accc52973" ), |
30 | |
31 | a: PublicElem::from_hex("fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffbfffffffc0000000000000003fffffffc" ), |
32 | b: PublicElem::from_hex("cd08114b604fbff9b62b21f41f022094e3374bee94938ae277f2209b1920022ef729add87a4c32ec081188719d412dcc" ), |
33 | |
34 | elem_mul_mont: p384_elem_mul_mont, |
35 | elem_sqr_mont: p384_elem_sqr_mont, |
36 | }; |
37 | |
38 | pub(super) static GENERATOR: (PublicElem<R>, PublicElem<R>) = ( |
39 | PublicElem::from_hex("4d3aadc2299e1513812ff723614ede2b6454868459a30eff879c3afc541b4d6e20e378e2a0d6ce383dd0756649c0b528" ), |
40 | PublicElem::from_hex("2b78abc25a15c5e9dd8002263969a840c6c3521968f4ffd98bade7562e83b050a1bfa8bf7bb4a9ac23043dad4b03a4fe" ), |
41 | ); |
42 | |
43 | pub static PRIVATE_KEY_OPS: PrivateKeyOps = PrivateKeyOps { |
44 | common: &COMMON_OPS, |
45 | elem_inv_squared: p384_elem_inv_squared, |
46 | point_mul_base_impl: p384_point_mul_base_impl, |
47 | point_mul_impl: p384_point_mul, |
48 | point_add_jacobian_impl: p384_point_add, |
49 | }; |
50 | |
51 | fn p384_elem_inv_squared(q: &Modulus<Q>, a: &Elem<R>) -> Elem<R> { |
52 | // Calculate a**-2 (mod q) == a**(q - 3) (mod q) |
53 | // |
54 | // The exponent (q - 3) is: |
55 | // |
56 | // 0xfffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffe\ |
57 | // ffffffff0000000000000000fffffffc |
58 | |
59 | #[inline ] |
60 | fn sqr_mul(q: &Modulus<Q>, a: &Elem<R>, squarings: LeakyWord, b: &Elem<R>) -> Elem<R> { |
61 | elem_sqr_mul(&COMMON_OPS, a, squarings, b, q.cpu()) |
62 | } |
63 | |
64 | #[inline ] |
65 | fn sqr_mul_acc(q: &Modulus<Q>, a: &mut Elem<R>, squarings: LeakyWord, b: &Elem<R>) { |
66 | elem_sqr_mul_acc(&COMMON_OPS, a, squarings, b, q.cpu()) |
67 | } |
68 | |
69 | let b_1 = &a; |
70 | let b_11 = sqr_mul(q, b_1, 1, b_1); |
71 | let b_111 = sqr_mul(q, &b_11, 1, b_1); |
72 | let f_11 = sqr_mul(q, &b_111, 3, &b_111); |
73 | let fff = sqr_mul(q, &f_11, 6, &f_11); |
74 | let fff_111 = sqr_mul(q, &fff, 3, &b_111); |
75 | let fffffff_11 = sqr_mul(q, &fff_111, 15, &fff_111); |
76 | |
77 | let fffffffffffffff = sqr_mul(q, &fffffff_11, 30, &fffffff_11); |
78 | |
79 | let ffffffffffffffffffffffffffffff = sqr_mul(q, &fffffffffffffff, 60, &fffffffffffffff); |
80 | |
81 | // ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff |
82 | let mut acc = sqr_mul( |
83 | q, |
84 | &ffffffffffffffffffffffffffffff, |
85 | 120, |
86 | &ffffffffffffffffffffffffffffff, |
87 | ); |
88 | |
89 | // fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff_111 |
90 | sqr_mul_acc(q, &mut acc, 15, &fff_111); |
91 | |
92 | // fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffeffffffff |
93 | sqr_mul_acc(q, &mut acc, 1 + 30, &fffffff_11); |
94 | sqr_mul_acc(q, &mut acc, 2, &b_11); |
95 | |
96 | // fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffeffffffff |
97 | // 0000000000000000fffffff_11 |
98 | sqr_mul_acc(q, &mut acc, 64 + 30, &fffffff_11); |
99 | |
100 | // fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffeffffffff |
101 | // 0000000000000000fffffffc |
102 | q.elem_square(&mut acc); |
103 | q.elem_square(&mut acc); |
104 | |
105 | acc |
106 | } |
107 | |
108 | fn p384_point_mul_base_impl(a: &Scalar, cpu: cpu::Features) -> Point { |
109 | // XXX: Not efficient. TODO: Precompute multiples of the generator. |
110 | let generator: (Elem, Elem) = (Elem::from(&GENERATOR.0), Elem::from(&GENERATOR.1)); |
111 | PRIVATE_KEY_OPS.point_mul(p_scalar:a, &generator, cpu) |
112 | } |
113 | |
114 | pub static PUBLIC_KEY_OPS: PublicKeyOps = PublicKeyOps { |
115 | common: &COMMON_OPS, |
116 | }; |
117 | |
118 | pub static SCALAR_OPS: ScalarOps = ScalarOps { |
119 | common: &COMMON_OPS, |
120 | scalar_mul_mont: p384_scalar_mul_mont, |
121 | }; |
122 | |
123 | pub static PUBLIC_SCALAR_OPS: PublicScalarOps = PublicScalarOps { |
124 | scalar_ops: &SCALAR_OPS, |
125 | public_key_ops: &PUBLIC_KEY_OPS, |
126 | twin_mul: |g_scalar: &Elem, p_scalar: &Elem, p_xy: &(Elem, Elem) , cpu: Features| { |
127 | twin_mul_inefficient(&PRIVATE_KEY_OPS, g_scalar, p_scalar, p_xy, cpu) |
128 | }, |
129 | |
130 | q_minus_n: PublicElem::from_hex("389cb27e0bc8d21fa7e5f24cb74f58851313e696333ad68c" ), |
131 | |
132 | // TODO: Use an optimized variable-time implementation. |
133 | scalar_inv_to_mont_vartime: |s: &Elem, cpu: Features| PRIVATE_SCALAR_OPS.scalar_inv_to_mont(a:s, cpu), |
134 | }; |
135 | |
136 | pub static PRIVATE_SCALAR_OPS: PrivateScalarOps = PrivateScalarOps { |
137 | scalar_ops: &SCALAR_OPS, |
138 | |
139 | oneRR_mod_n: PublicScalar::from_hex("c84ee012b39bf213fb05b7a28266895d40d49174aab1cc5bc3e483afcb82947ff3d81e5df1aa4192d319b2419b409a9" ), |
140 | scalar_inv_to_mont: p384_scalar_inv_to_mont, |
141 | }; |
142 | |
143 | fn p384_scalar_inv_to_mont(a: Scalar<R>, _cpu: cpu::Features) -> Scalar<R> { |
144 | // Calculate the modular inverse of scalar |a| using Fermat's Little |
145 | // Theorem: |
146 | // |
147 | // a**-1 (mod n) == a**(n - 2) (mod n) |
148 | // |
149 | // The exponent (n - 2) is: |
150 | // |
151 | // 0xffffffffffffffffffffffffffffffffffffffffffffffffc7634d81f4372ddf\ |
152 | // 581a0db248b0a77aecec196accc52971 |
153 | |
154 | fn mul(a: &Scalar<R>, b: &Scalar<R>) -> Scalar<R> { |
155 | binary_op(p384_scalar_mul_mont, a, b) |
156 | } |
157 | |
158 | fn sqr(a: &Scalar<R>) -> Scalar<R> { |
159 | binary_op(p384_scalar_mul_mont, a, a) |
160 | } |
161 | |
162 | fn sqr_mut(a: &mut Scalar<R>) { |
163 | unary_op_from_binary_op_assign(p384_scalar_mul_mont, a); |
164 | } |
165 | |
166 | // Returns (`a` squared `squarings` times) * `b`. |
167 | fn sqr_mul(a: &Scalar<R>, squarings: LeakyWord, b: &Scalar<R>) -> Scalar<R> { |
168 | debug_assert!(squarings >= 1); |
169 | let mut tmp = sqr(a); |
170 | for _ in 1..squarings { |
171 | sqr_mut(&mut tmp); |
172 | } |
173 | mul(&tmp, b) |
174 | } |
175 | |
176 | // Sets `acc` = (`acc` squared `squarings` times) * `b`. |
177 | fn sqr_mul_acc(acc: &mut Scalar<R>, squarings: LeakyWord, b: &Scalar<R>) { |
178 | debug_assert!(squarings >= 1); |
179 | for _ in 0..squarings { |
180 | sqr_mut(acc); |
181 | } |
182 | binary_op_assign(p384_scalar_mul_mont, acc, b) |
183 | } |
184 | |
185 | // Indexes into `d`. |
186 | const B_1: usize = 0; |
187 | const B_11: usize = 1; |
188 | const B_101: usize = 2; |
189 | const B_111: usize = 3; |
190 | const B_1001: usize = 4; |
191 | const B_1011: usize = 5; |
192 | const B_1101: usize = 6; |
193 | const B_1111: usize = 7; |
194 | const DIGIT_COUNT: usize = 8; |
195 | |
196 | let mut d = [Scalar::zero(); DIGIT_COUNT]; |
197 | d[B_1] = a; |
198 | let b_10 = sqr(&d[B_1]); |
199 | for i in B_11..DIGIT_COUNT { |
200 | d[i] = mul(&d[i - 1], &b_10); |
201 | } |
202 | |
203 | let ff = sqr_mul(&d[B_1111], 0 + 4, &d[B_1111]); |
204 | let ffff = sqr_mul(&ff, 0 + 8, &ff); |
205 | let ffffffff = sqr_mul(&ffff, 0 + 16, &ffff); |
206 | |
207 | let ffffffffffffffff = sqr_mul(&ffffffff, 0 + 32, &ffffffff); |
208 | |
209 | let ffffffffffffffffffffffff = sqr_mul(&ffffffffffffffff, 0 + 32, &ffffffff); |
210 | |
211 | // ffffffffffffffffffffffffffffffffffffffffffffffff |
212 | let mut acc = sqr_mul(&ffffffffffffffffffffffff, 0 + 96, &ffffffffffffffffffffffff); |
213 | |
214 | // The rest of the exponent, in binary, is: |
215 | // |
216 | // 1100011101100011010011011000000111110100001101110010110111011111 |
217 | // 0101100000011010000011011011001001001000101100001010011101111010 |
218 | // 1110110011101100000110010110101011001100110001010010100101110001 |
219 | |
220 | #[allow (clippy::cast_possible_truncation)] |
221 | static REMAINING_WINDOWS: [(u8, u8); 39] = [ |
222 | (2, B_11 as u8), |
223 | (3 + 3, B_111 as u8), |
224 | (1 + 2, B_11 as u8), |
225 | (3 + 2, B_11 as u8), |
226 | (1 + 4, B_1001 as u8), |
227 | (4, B_1011 as u8), |
228 | (6 + 4, B_1111 as u8), |
229 | (3, B_101 as u8), |
230 | (4 + 1, B_1 as u8), |
231 | (4, B_1011 as u8), |
232 | (4, B_1001 as u8), |
233 | (1 + 4, B_1101 as u8), |
234 | (4, B_1101 as u8), |
235 | (4, B_1111 as u8), |
236 | (1 + 4, B_1011 as u8), |
237 | (6 + 4, B_1101 as u8), |
238 | (5 + 4, B_1101 as u8), |
239 | (4, B_1011 as u8), |
240 | (2 + 4, B_1001 as u8), |
241 | (2 + 1, B_1 as u8), |
242 | (3 + 4, B_1011 as u8), |
243 | (4 + 3, B_101 as u8), |
244 | (2 + 3, B_111 as u8), |
245 | (1 + 4, B_1111 as u8), |
246 | (1 + 4, B_1011 as u8), |
247 | (4, B_1011 as u8), |
248 | (2 + 3, B_111 as u8), |
249 | (1 + 2, B_11 as u8), |
250 | (5 + 2, B_11 as u8), |
251 | (2 + 4, B_1011 as u8), |
252 | (1 + 3, B_101 as u8), |
253 | (1 + 2, B_11 as u8), |
254 | (2 + 2, B_11 as u8), |
255 | (2 + 2, B_11 as u8), |
256 | (3 + 3, B_101 as u8), |
257 | (2 + 3, B_101 as u8), |
258 | (2 + 3, B_101 as u8), |
259 | (2, B_11 as u8), |
260 | (3 + 1, B_1 as u8), |
261 | ]; |
262 | |
263 | for &(squarings, digit) in &REMAINING_WINDOWS[..] { |
264 | sqr_mul_acc(&mut acc, LeakyWord::from(squarings), &d[usize::from(digit)]); |
265 | } |
266 | |
267 | acc |
268 | } |
269 | |
270 | unsafe extern "C" fn p384_elem_sqr_mont( |
271 | r: *mut Limb, // [COMMON_OPS.num_limbs] |
272 | a: *const Limb, // [COMMON_OPS.num_limbs] |
273 | ) { |
274 | // XXX: Inefficient. TODO: Make a dedicated squaring routine. |
275 | unsafe { |
276 | p384_elem_mul_mont(r, a, b:a); |
277 | } |
278 | } |
279 | |
280 | prefixed_extern! { |
281 | fn p384_elem_mul_mont( |
282 | r: *mut Limb, // [COMMON_OPS.num_limbs] |
283 | a: *const Limb, // [COMMON_OPS.num_limbs] |
284 | b: *const Limb, // [COMMON_OPS.num_limbs] |
285 | ); |
286 | |
287 | fn p384_point_add( |
288 | r: *mut Limb, // [3][COMMON_OPS.num_limbs] |
289 | a: *const Limb, // [3][COMMON_OPS.num_limbs] |
290 | b: *const Limb, // [3][COMMON_OPS.num_limbs] |
291 | ); |
292 | fn p384_point_mul( |
293 | r: *mut Limb, // [3][COMMON_OPS.num_limbs] |
294 | p_scalar: *const Limb, // [COMMON_OPS.num_limbs] |
295 | p_x: *const Limb, // [COMMON_OPS.num_limbs] |
296 | p_y: *const Limb, // [COMMON_OPS.num_limbs] |
297 | ); |
298 | |
299 | fn p384_scalar_mul_mont( |
300 | r: *mut Limb, // [COMMON_OPS.num_limbs] |
301 | a: *const Limb, // [COMMON_OPS.num_limbs] |
302 | b: *const Limb, // [COMMON_OPS.num_limbs] |
303 | ); |
304 | } |
305 | |