1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
use alloc::vec::Vec;
use core::mem;
use core::ops::Shl;
use num_traits::One;

use crate::big_digit::{self, BigDigit, DoubleBigDigit};
use crate::biguint::BigUint;

struct MontyReducer {
    n0inv: BigDigit,
}

// k0 = -m**-1 mod 2**BITS. Algorithm from: Dumas, J.G. "On Newton–Raphson
// Iteration for Multiplicative Inverses Modulo Prime Powers".
fn inv_mod_alt(b: BigDigit) -> BigDigit {
    assert_ne!(b & 1, 0);

    let mut k0 = BigDigit::wrapping_sub(2, b);
    let mut t = b - 1;
    let mut i = 1;
    while i < big_digit::BITS {
        t = t.wrapping_mul(t);
        k0 = k0.wrapping_mul(t + 1);

        i <<= 1;
    }
    debug_assert_eq!(k0.wrapping_mul(b), 1);
    k0.wrapping_neg()
}

impl MontyReducer {
    fn new(n: &BigUint) -> Self {
        let n0inv = inv_mod_alt(n.data[0]);
        MontyReducer { n0inv }
    }
}

/// Computes z mod m = x * y * 2 ** (-n*_W) mod m
/// assuming k = -1/m mod 2**_W
/// See Gueron, "Efficient Software Implementations of Modular Exponentiation".
/// <https://eprint.iacr.org/2011/239.pdf>
/// In the terminology of that paper, this is an "Almost Montgomery Multiplication":
/// x and y are required to satisfy 0 <= z < 2**(n*_W) and then the result
/// z is guaranteed to satisfy 0 <= z < 2**(n*_W), but it may not be < m.
#[allow(clippy::many_single_char_names)]
fn montgomery(x: &BigUint, y: &BigUint, m: &BigUint, k: BigDigit, n: usize) -> BigUint {
    // This code assumes x, y, m are all the same length, n.
    // (required by addMulVVW and the for loop).
    // It also assumes that x, y are already reduced mod m,
    // or else the result will not be properly reduced.
    assert!(
        x.data.len() == n && y.data.len() == n && m.data.len() == n,
        "{:?} {:?} {:?} {}",
        x,
        y,
        m,
        n
    );

    let mut z = BigUint::ZERO;
    z.data.resize(n * 2, 0);

    let mut c: BigDigit = 0;
    for i in 0..n {
        let c2 = add_mul_vvw(&mut z.data[i..n + i], &x.data, y.data[i]);
        let t = z.data[i].wrapping_mul(k);
        let c3 = add_mul_vvw(&mut z.data[i..n + i], &m.data, t);
        let cx = c.wrapping_add(c2);
        let cy = cx.wrapping_add(c3);
        z.data[n + i] = cy;
        if cx < c2 || cy < c3 {
            c = 1;
        } else {
            c = 0;
        }
    }

    if c == 0 {
        z.data = z.data[n..].to_vec();
    } else {
        {
            let (first, second) = z.data.split_at_mut(n);
            sub_vv(first, second, &m.data);
        }
        z.data = z.data[..n].to_vec();
    }

    z
}

#[inline(always)]
fn add_mul_vvw(z: &mut [BigDigit], x: &[BigDigit], y: BigDigit) -> BigDigit {
    let mut c = 0;
    for (zi, xi) in z.iter_mut().zip(x.iter()) {
        let (z1, z0) = mul_add_www(*xi, y, *zi);
        let (c_, zi_) = add_ww(z0, c, 0);
        *zi = zi_;
        c = c_ + z1;
    }

    c
}

/// The resulting carry c is either 0 or 1.
#[inline(always)]
fn sub_vv(z: &mut [BigDigit], x: &[BigDigit], y: &[BigDigit]) -> BigDigit {
    let mut c = 0;
    for (i, (xi, yi)) in x.iter().zip(y.iter()).enumerate().take(z.len()) {
        let zi = xi.wrapping_sub(*yi).wrapping_sub(c);
        z[i] = zi;
        // see "Hacker's Delight", section 2-12 (overflow detection)
        c = ((yi & !xi) | ((yi | !xi) & zi)) >> (big_digit::BITS - 1)
    }

    c
}

/// z1<<_W + z0 = x+y+c, with c == 0 or 1
#[inline(always)]
fn add_ww(x: BigDigit, y: BigDigit, c: BigDigit) -> (BigDigit, BigDigit) {
    let yc = y.wrapping_add(c);
    let z0 = x.wrapping_add(yc);
    let z1 = if z0 < x || yc < y { 1 } else { 0 };

    (z1, z0)
}

/// z1 << _W + z0 = x * y + c
#[inline(always)]
fn mul_add_www(x: BigDigit, y: BigDigit, c: BigDigit) -> (BigDigit, BigDigit) {
    let z = x as DoubleBigDigit * y as DoubleBigDigit + c as DoubleBigDigit;
    ((z >> big_digit::BITS) as BigDigit, z as BigDigit)
}

/// Calculates x ** y mod m using a fixed, 4-bit window.
#[allow(clippy::many_single_char_names)]
pub(super) fn monty_modpow(x: &BigUint, y: &BigUint, m: &BigUint) -> BigUint {
    assert!(m.data[0] & 1 == 1);
    let mr = MontyReducer::new(m);
    let num_words = m.data.len();

    let mut x = x.clone();

    // We want the lengths of x and m to be equal.
    // It is OK if x >= m as long as len(x) == len(m).
    if x.data.len() > num_words {
        x %= m;
        // Note: now len(x) <= numWords, not guaranteed ==.
    }
    if x.data.len() < num_words {
        x.data.resize(num_words, 0);
    }

    // rr = 2**(2*_W*len(m)) mod m
    let mut rr = BigUint::one();
    rr = (rr.shl(2 * num_words as u64 * u64::from(big_digit::BITS))) % m;
    if rr.data.len() < num_words {
        rr.data.resize(num_words, 0);
    }
    // one = 1, with equal length to that of m
    let mut one = BigUint::one();
    one.data.resize(num_words, 0);

    let n = 4;
    // powers[i] contains x^i
    let mut powers = Vec::with_capacity(1 << n);
    powers.push(montgomery(&one, &rr, m, mr.n0inv, num_words));
    powers.push(montgomery(&x, &rr, m, mr.n0inv, num_words));
    for i in 2..1 << n {
        let r = montgomery(&powers[i - 1], &powers[1], m, mr.n0inv, num_words);
        powers.push(r);
    }

    // initialize z = 1 (Montgomery 1)
    let mut z = powers[0].clone();
    z.data.resize(num_words, 0);
    let mut zz = BigUint::ZERO;
    zz.data.resize(num_words, 0);

    // same windowed exponent, but with Montgomery multiplications
    for i in (0..y.data.len()).rev() {
        let mut yi = y.data[i];
        let mut j = 0;
        while j < big_digit::BITS {
            if i != y.data.len() - 1 || j != 0 {
                zz = montgomery(&z, &z, m, mr.n0inv, num_words);
                z = montgomery(&zz, &zz, m, mr.n0inv, num_words);
                zz = montgomery(&z, &z, m, mr.n0inv, num_words);
                z = montgomery(&zz, &zz, m, mr.n0inv, num_words);
            }
            zz = montgomery(
                &z,
                &powers[(yi >> (big_digit::BITS - n)) as usize],
                m,
                mr.n0inv,
                num_words,
            );
            mem::swap(&mut z, &mut zz);
            yi <<= n;
            j += n;
        }
    }

    // convert to regular number
    zz = montgomery(&z, &one, m, mr.n0inv, num_words);

    zz.normalize();
    // One last reduction, just in case.
    // See golang.org/issue/13907.
    if zz >= *m {
        // Common case is m has high bit set; in that case,
        // since zz is the same length as m, there can be just
        // one multiple of m to remove. Just subtract.
        // We think that the subtract should be sufficient in general,
        // so do that unconditionally, but double-check,
        // in case our beliefs are wrong.
        // The div is not expected to be reached.
        zz -= m;
        if zz >= *m {
            zz %= m;
        }
    }

    zz.normalize();
    zz
}