cryptoxide 0.4.2

pure implementation of various common modern cryptographic algorithms, WASM compatible
Documentation 
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
227
228
229
230
231
232
233
234
235
236
237
238

#[cfg(target_arch = "x86")]
use core::arch::x86::*;

#[cfg(target_arch = "x86_64")]
use core::arch::x86_64::*;

use super::reference;
use super::sse41;

const K32: [u32; 64] = reference::K32;

// takes 32 bits word from four data 64 bytes block and pack it into one simd u32x4
//
//   | block 3 | block 2 | block 1 | block 0 |
//
// x86 architecture can (currently) do unaligned load (at a cost if unaligned)
unsafe fn gather(block: *const u8) -> __m256i {
    use core::ptr::read;
    let mut temp: __m256i;
    let block = block as *const i32;

    temp = _mm256_castsi128_si256(_mm_cvtsi32_si128(read(block)));
    temp = _mm256_insert_epi32(temp, read(block.add(16)), 1);
    temp = _mm256_insert_epi32(temp, read(block.add(32)), 2);
    temp = _mm256_insert_epi32(temp, read(block.add(48)), 3);
    temp = _mm256_insert_epi32(temp, read(block.add(64)), 4);
    temp = _mm256_insert_epi32(temp, read(block.add(80)), 5);
    temp = _mm256_insert_epi32(temp, read(block.add(96)), 6);
    temp = _mm256_insert_epi32(temp, read(block.add(112)), 7);
    temp
}

unsafe fn sigma0(w: __m256i) -> __m256i {
    _mm256_xor_si256(
        _mm256_xor_si256(
            _mm256_xor_si256(_mm256_srli_epi32(w, 7), _mm256_srli_epi32(w, 18)),
            _mm256_xor_si256(_mm256_srli_epi32(w, 3), _mm256_slli_epi32(w, 25)),
        ),
        _mm256_slli_epi32(w, 14),
    )
}

unsafe fn sigma1(w: __m256i) -> __m256i {
    _mm256_xor_si256(
        _mm256_xor_si256(
            _mm256_xor_si256(_mm256_srli_epi32(w, 17), _mm256_srli_epi32(w, 10)),
            _mm256_xor_si256(_mm256_srli_epi32(w, 19), _mm256_slli_epi32(w, 15)),
        ),
        _mm256_slli_epi32(w, 13),
    )
}

macro_rules! SCHEDULE_ROUND {
    ($schedule: ident, $i:expr, $w1:expr, $w2:expr, $w3:expr, $w4:expr) => {
        let s0 = sigma0($w1);
        let s1 = sigma1($w2);
        $schedule[$i] = _mm256_add_epi32($w3, _mm256_set1_epi32(K32[$i] as i32));
        $w3 = _mm256_add_epi32(_mm256_add_epi32($w3, $w4), _mm256_add_epi32(s0, s1));
    };
}

macro_rules! SCHEDULE_ROUND_INC {
    ($schedule: ident, $i:expr, $w1:expr, $w2:expr, $w3:expr, $w4:expr) => {
        SCHEDULE_ROUND!($schedule, $i, $w1, $w2, $w3, $w4);
        $i += 1;
    };
}

/// compute the message schedule of 8 blocks (512 bytes)
#[inline]
pub unsafe fn message_schedule_8ways(schedule: &mut [__m256i; 64], message: &[u8]) {
    //let bswap_mask: __m128i = _mm_set_epi8(12, 13, 14, 15, 8, 9, 10, 11, 4, 5, 6, 7, 0, 1, 2, 3);
    let bswap_mask: __m256i = _mm256_set_epi8(
        28, 29, 30, 31, 24, 25, 26, 27, 20, 21, 22, 23, 16, 17, 18, 19, 12, 13, 14, 15, 8, 9, 10,
        11, 4, 5, 6, 7, 0, 1, 2, 3,
    );
    let (mut w0, mut w1, mut w2, mut w3, mut w4, mut w5, mut w6, mut w7);
    let (mut w8, mut w9, mut w10, mut w11, mut w12, mut w13, mut w14, mut w15);

    let message = message.as_ptr();
    w0 = gather(message);
    w1 = gather(message.add(4));
    w2 = gather(message.add(8));
    w3 = gather(message.add(12));
    w4 = gather(message.add(16));
    w5 = gather(message.add(20));
    w6 = gather(message.add(24));
    w7 = gather(message.add(28));
    w8 = gather(message.add(32));
    w9 = gather(message.add(36));
    w10 = gather(message.add(40));
    w11 = gather(message.add(44));
    w12 = gather(message.add(48));
    w13 = gather(message.add(52));
    w14 = gather(message.add(56));
    w15 = gather(message.add(60));
    w0 = _mm256_shuffle_epi8(w0, bswap_mask);
    w1 = _mm256_shuffle_epi8(w1, bswap_mask);
    w2 = _mm256_shuffle_epi8(w2, bswap_mask);
    w3 = _mm256_shuffle_epi8(w3, bswap_mask);
    w4 = _mm256_shuffle_epi8(w4, bswap_mask);
    w5 = _mm256_shuffle_epi8(w5, bswap_mask);
    w6 = _mm256_shuffle_epi8(w6, bswap_mask);
    w7 = _mm256_shuffle_epi8(w7, bswap_mask);
    w8 = _mm256_shuffle_epi8(w8, bswap_mask);
    w9 = _mm256_shuffle_epi8(w9, bswap_mask);
    w10 = _mm256_shuffle_epi8(w10, bswap_mask);
    w11 = _mm256_shuffle_epi8(w11, bswap_mask);
    w12 = _mm256_shuffle_epi8(w12, bswap_mask);
    w13 = _mm256_shuffle_epi8(w13, bswap_mask);
    w14 = _mm256_shuffle_epi8(w14, bswap_mask);
    w15 = _mm256_shuffle_epi8(w15, bswap_mask);
    let mut i = 0;
    while i < 32 {
        SCHEDULE_ROUND_INC!(schedule, i, w1, w14, w0, w9);
        SCHEDULE_ROUND_INC!(schedule, i, w2, w15, w1, w10);
        SCHEDULE_ROUND_INC!(schedule, i, w3, w0, w2, w11);
        SCHEDULE_ROUND_INC!(schedule, i, w4, w1, w3, w12);
        SCHEDULE_ROUND_INC!(schedule, i, w5, w2, w4, w13);
        SCHEDULE_ROUND_INC!(schedule, i, w6, w3, w5, w14);
        SCHEDULE_ROUND_INC!(schedule, i, w7, w4, w6, w15);
        SCHEDULE_ROUND_INC!(schedule, i, w8, w5, w7, w0);
        SCHEDULE_ROUND_INC!(schedule, i, w9, w6, w8, w1);
        SCHEDULE_ROUND_INC!(schedule, i, w10, w7, w9, w2);
        SCHEDULE_ROUND_INC!(schedule, i, w11, w8, w10, w3);
        SCHEDULE_ROUND_INC!(schedule, i, w12, w9, w11, w4);
        SCHEDULE_ROUND_INC!(schedule, i, w13, w10, w12, w5);
        SCHEDULE_ROUND_INC!(schedule, i, w14, w11, w13, w6);
        SCHEDULE_ROUND_INC!(schedule, i, w15, w12, w14, w7);
        SCHEDULE_ROUND_INC!(schedule, i, w0, w13, w15, w8);
    }
    SCHEDULE_ROUND_INC!(schedule, i, w1, w14, w0, w9);
    schedule[48] = _mm256_add_epi32(w0, _mm256_set1_epi32(K32[48] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w2, w15, w1, w10);
    schedule[49] = _mm256_add_epi32(w1, _mm256_set1_epi32(K32[49] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w3, w0, w2, w11);
    schedule[50] = _mm256_add_epi32(w2, _mm256_set1_epi32(K32[50] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w4, w1, w3, w12);
    schedule[51] = _mm256_add_epi32(w3, _mm256_set1_epi32(K32[51] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w5, w2, w4, w13);
    schedule[52] = _mm256_add_epi32(w4, _mm256_set1_epi32(K32[52] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w6, w3, w5, w14);
    schedule[53] = _mm256_add_epi32(w5, _mm256_set1_epi32(K32[53] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w7, w4, w6, w15);
    schedule[54] = _mm256_add_epi32(w6, _mm256_set1_epi32(K32[54] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w8, w5, w7, w0);
    schedule[55] = _mm256_add_epi32(w7, _mm256_set1_epi32(K32[55] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w9, w6, w8, w1);
    schedule[56] = _mm256_add_epi32(w8, _mm256_set1_epi32(K32[56] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w10, w7, w9, w2);
    schedule[57] = _mm256_add_epi32(w9, _mm256_set1_epi32(K32[57] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w11, w8, w10, w3);
    schedule[58] = _mm256_add_epi32(w10, _mm256_set1_epi32(K32[58] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w12, w9, w11, w4);
    schedule[59] = _mm256_add_epi32(w11, _mm256_set1_epi32(K32[59] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w13, w10, w12, w5);
    schedule[60] = _mm256_add_epi32(w12, _mm256_set1_epi32(K32[60] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w14, w11, w13, w6);
    schedule[61] = _mm256_add_epi32(w13, _mm256_set1_epi32(K32[61] as i32));
    SCHEDULE_ROUND_INC!(schedule, i, w15, w12, w14, w7);
    schedule[62] = _mm256_add_epi32(w14, _mm256_set1_epi32(K32[62] as i32));
    SCHEDULE_ROUND!(schedule, i, w0, w13, w15, w8);
    schedule[63] = _mm256_add_epi32(w15, _mm256_set1_epi32(K32[63] as i32));
}

unsafe fn compress_8ways(state: &mut [u32; 8], schedule: &[__m256i; 64]) {
    use super::reference::{e0, e1};
    macro_rules! round {
        ($a: ident, $b: ident, $c: ident, $d: ident, $e: ident, $f: ident, $g: ident, $h: ident, $i: expr, $j: expr) => {
            let kwi = _mm256_extract_epi32(*schedule.get_unchecked($i), $j) as u32;
            let t1 = $h
                .wrapping_add(e1($e))
                .wrapping_add($g ^ ($e & ($f ^ $g)))
                .wrapping_add(kwi);
            let t2 = e0($a).wrapping_add(($a & $b) | ($c & ($a | $b)));
            $d = $d.wrapping_add(t1);
            $h = t1.wrapping_add(t2);
        };
    }

    // Compression function main loop:
    macro_rules! compress_once {
        ($j: expr) => {
            let mut a = state[0];
            let mut b = state[1];
            let mut c = state[2];
            let mut d = state[3];
            let mut e = state[4];
            let mut f = state[5];
            let mut g = state[6];
            let mut h = state[7];

            let mut i = 0;
            while i != 64 {
                round!(a, b, c, d, e, f, g, h, i + 0, $j);
                round!(h, a, b, c, d, e, f, g, i + 1, $j);
                round!(g, h, a, b, c, d, e, f, i + 2, $j);
                round!(f, g, h, a, b, c, d, e, i + 3, $j);
                round!(e, f, g, h, a, b, c, d, i + 4, $j);
                round!(d, e, f, g, h, a, b, c, i + 5, $j);
                round!(c, d, e, f, g, h, a, b, i + 6, $j);
                round!(b, c, d, e, f, g, h, a, i + 7, $j);
                i += 8;
            }

            //Add the compressed chunk to the current hash value:
            state[0] = state[0].wrapping_add(a);
            state[1] = state[1].wrapping_add(b);
            state[2] = state[2].wrapping_add(c);
            state[3] = state[3].wrapping_add(d);
            state[4] = state[4].wrapping_add(e);
            state[5] = state[5].wrapping_add(f);
            state[6] = state[6].wrapping_add(g);
            state[7] = state[7].wrapping_add(h);
        };
    }

    compress_once!(0);
    compress_once!(1);
    compress_once!(2);
    compress_once!(3);
    compress_once!(4);
    compress_once!(5);
    compress_once!(6);
    compress_once!(7);
}

pub(crate) fn digest_block(state: &mut [u32; 8], mut block: &[u8]) {
    unsafe {
        let mut schedule = [_mm256_set1_epi32(0); 64];
        while block.len() >= 512 {
            message_schedule_8ways(&mut schedule, &block);
            compress_8ways(state, &schedule);
            block = &block[512..]
        }
    }
    sse41::digest_block(state, block)
}