rscrypto 0.1.0

Rust crypto with zero default deps: BLAKE3, Ed25519/X25519, hashes, MACs, KDFs, AEADs, and checksums with SIMD/ASM acceleration.
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

use core::arch::x86_64::{
  __m128i, _mm_add_epi32, _mm_loadu_si128, _mm_or_si128, _mm_set_epi8, _mm_set1_epi32, _mm_setr_epi32,
  _mm_shuffle_epi8, _mm_slli_epi32, _mm_srli_epi32, _mm_storeu_si128, _mm_unpackhi_epi32, _mm_unpackhi_epi64,
  _mm_unpacklo_epi32, _mm_unpacklo_epi64, _mm_xor_si128,
};

use super::{BLOCK_SIZE, KEY_SIZE, NONCE_SIZE, load_u32_le};

pub(super) const BLOCKS_PER_BATCH: usize = 4;

#[inline]
pub(super) unsafe fn xor_blocks(
  key: &[u8; KEY_SIZE],
  initial_counter: u32,
  nonce: &[u8; NONCE_SIZE],
  chunk: &mut [u8],
) {
  debug_assert_eq!(chunk.len(), BLOCK_SIZE * BLOCKS_PER_BATCH);
  // SAFETY: callers only reach this helper from x86 SIMD backends whose
  // dispatch gates guarantee AVX/SSSE3-capable hardware.
  unsafe { xor_blocks_impl(key, initial_counter, nonce, chunk) }
}

#[target_feature(enable = "ssse3,avx")]
unsafe fn xor_blocks_impl(key: &[u8; KEY_SIZE], initial_counter: u32, nonce: &[u8; NONCE_SIZE], chunk: &mut [u8]) {
  let rot16 = _mm_set_epi8(13, 12, 15, 14, 9, 8, 11, 10, 5, 4, 7, 6, 1, 0, 3, 2);
  let rot8 = _mm_set_epi8(14, 13, 12, 15, 10, 9, 8, 11, 6, 5, 4, 7, 2, 1, 0, 3);

  let mut x0 = _mm_set1_epi32(0x6170_7865u32 as i32);
  let mut x1 = _mm_set1_epi32(0x3320_646eu32 as i32);
  let mut x2 = _mm_set1_epi32(0x7962_2d32u32 as i32);
  let mut x3 = _mm_set1_epi32(0x6b20_6574u32 as i32);
  let mut x4 = _mm_set1_epi32(load_u32_le(&key[0..4]) as i32);
  let mut x5 = _mm_set1_epi32(load_u32_le(&key[4..8]) as i32);
  let mut x6 = _mm_set1_epi32(load_u32_le(&key[8..12]) as i32);
  let mut x7 = _mm_set1_epi32(load_u32_le(&key[12..16]) as i32);
  let mut x8 = _mm_set1_epi32(load_u32_le(&key[16..20]) as i32);
  let mut x9 = _mm_set1_epi32(load_u32_le(&key[20..24]) as i32);
  let mut x10 = _mm_set1_epi32(load_u32_le(&key[24..28]) as i32);
  let mut x11 = _mm_set1_epi32(load_u32_le(&key[28..32]) as i32);
  let mut x12 = _mm_setr_epi32(
    initial_counter as i32,
    initial_counter.wrapping_add(1) as i32,
    initial_counter.wrapping_add(2) as i32,
    initial_counter.wrapping_add(3) as i32,
  );
  let mut x13 = _mm_set1_epi32(load_u32_le(&nonce[0..4]) as i32);
  let mut x14 = _mm_set1_epi32(load_u32_le(&nonce[4..8]) as i32);
  let mut x15 = _mm_set1_epi32(load_u32_le(&nonce[8..12]) as i32);

  let o0 = x0;
  let o1 = x1;
  let o2 = x2;
  let o3 = x3;
  let o4 = x4;
  let o5 = x5;
  let o6 = x6;
  let o7 = x7;
  let o8 = x8;
  let o9 = x9;
  let o10 = x10;
  let o11 = x11;
  let o12 = x12;
  let o13 = x13;
  let o14 = x14;
  let o15 = x15;

  let mut round = 0usize;
  while round < 10 {
    quarter_round(&mut x0, &mut x4, &mut x8, &mut x12, rot16, rot8);
    quarter_round(&mut x1, &mut x5, &mut x9, &mut x13, rot16, rot8);
    quarter_round(&mut x2, &mut x6, &mut x10, &mut x14, rot16, rot8);
    quarter_round(&mut x3, &mut x7, &mut x11, &mut x15, rot16, rot8);

    quarter_round(&mut x0, &mut x5, &mut x10, &mut x15, rot16, rot8);
    quarter_round(&mut x1, &mut x6, &mut x11, &mut x12, rot16, rot8);
    quarter_round(&mut x2, &mut x7, &mut x8, &mut x13, rot16, rot8);
    quarter_round(&mut x3, &mut x4, &mut x9, &mut x14, rot16, rot8);

    round = round.strict_add(1);
  }

  x0 = _mm_add_epi32(x0, o0);
  x1 = _mm_add_epi32(x1, o1);
  x2 = _mm_add_epi32(x2, o2);
  x3 = _mm_add_epi32(x3, o3);
  x4 = _mm_add_epi32(x4, o4);
  x5 = _mm_add_epi32(x5, o5);
  x6 = _mm_add_epi32(x6, o6);
  x7 = _mm_add_epi32(x7, o7);
  x8 = _mm_add_epi32(x8, o8);
  x9 = _mm_add_epi32(x9, o9);
  x10 = _mm_add_epi32(x10, o10);
  x11 = _mm_add_epi32(x11, o11);
  x12 = _mm_add_epi32(x12, o12);
  x13 = _mm_add_epi32(x13, o13);
  x14 = _mm_add_epi32(x14, o14);
  x15 = _mm_add_epi32(x15, o15);

  let b03 = transpose_words(x0, x1, x2, x3);
  let b47 = transpose_words(x4, x5, x6, x7);
  let b811 = transpose_words(x8, x9, x10, x11);
  let b1215 = transpose_words(x12, x13, x14, x15);

  let ptr = chunk.as_mut_ptr();
  // SAFETY: `chunk` is exactly four 64-byte blocks.
  unsafe {
    xor_block(ptr, 0, b03[0], b47[0], b811[0], b1215[0]);
    xor_block(ptr, 1, b03[1], b47[1], b811[1], b1215[1]);
    xor_block(ptr, 2, b03[2], b47[2], b811[2], b1215[2]);
    xor_block(ptr, 3, b03[3], b47[3], b811[3], b1215[3]);
  }
}

#[inline(always)]
fn transpose_words(w0: __m128i, w1: __m128i, w2: __m128i, w3: __m128i) -> [__m128i; 4] {
  // SAFETY: all operations are lane-local integer shuffles.
  unsafe {
    let s0 = _mm_unpacklo_epi32(w0, w1);
    let s1 = _mm_unpackhi_epi32(w0, w1);
    let s2 = _mm_unpacklo_epi32(w2, w3);
    let s3 = _mm_unpackhi_epi32(w2, w3);
    [
      _mm_unpacklo_epi64(s0, s2),
      _mm_unpackhi_epi64(s0, s2),
      _mm_unpacklo_epi64(s1, s3),
      _mm_unpackhi_epi64(s1, s3),
    ]
  }
}

#[inline(always)]
unsafe fn xor_block(buf: *mut u8, idx: usize, w03: __m128i, w47: __m128i, w811: __m128i, w1215: __m128i) {
  // SAFETY: caller guarantees `buf` points to four full ChaCha20 blocks and
  // `idx < 4`, so all four 16-byte load/store pairs are in bounds.
  unsafe {
    let p = buf.add(idx.strict_mul(BLOCK_SIZE));
    xor_store(p, w03);
    xor_store(p.add(16), w47);
    xor_store(p.add(32), w811);
    xor_store(p.add(48), w1215);
  }
}

#[inline(always)]
unsafe fn xor_store(ptr: *mut u8, keystream: __m128i) {
  // SAFETY: caller guarantees `ptr..ptr+16` is in bounds for unaligned access.
  unsafe {
    let plaintext = _mm_loadu_si128(ptr.cast());
    _mm_storeu_si128(ptr.cast(), _mm_xor_si128(plaintext, keystream));
  }
}

#[inline(always)]
fn quarter_round(a: &mut __m128i, b: &mut __m128i, c: &mut __m128i, d: &mut __m128i, rot16: __m128i, rot8: __m128i) {
  // SAFETY: this helper is only reached from the SSSE3-enabled backend.
  unsafe {
    *a = _mm_add_epi32(*a, *b);
    *d = _mm_shuffle_epi8(_mm_xor_si128(*d, *a), rot16);
    *c = _mm_add_epi32(*c, *d);
    *b = rotl::<12, 20>(_mm_xor_si128(*b, *c));
    *a = _mm_add_epi32(*a, *b);
    *d = _mm_shuffle_epi8(_mm_xor_si128(*d, *a), rot8);
    *c = _mm_add_epi32(*c, *d);
    *b = rotl::<7, 25>(_mm_xor_si128(*b, *c));
  }
}

#[inline(always)]
fn rotl<const LEFT: i32, const RIGHT: i32>(value: __m128i) -> __m128i {
  const { assert!(LEFT + RIGHT == 32, "rotation amounts must sum to 32") };
  // SAFETY: only reached from the SSSE3-enabled backend.
  unsafe { _mm_or_si128(_mm_slli_epi32(value, LEFT), _mm_srli_epi32(value, RIGHT)) }
}