/rust/registry/src/index.crates.io-1949cf8c6b5b557f/rand_chacha-0.3.1/src/guts.rs
Line | Count | Source |
1 | | // Copyright 2019 The CryptoCorrosion Contributors |
2 | | // Copyright 2020 Developers of the Rand project. |
3 | | // |
4 | | // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or |
5 | | // https://www.apache.org/licenses/LICENSE-2.0> or the MIT license |
6 | | // <LICENSE-MIT or https://opensource.org/licenses/MIT>, at your |
7 | | // option. This file may not be copied, modified, or distributed |
8 | | // except according to those terms. |
9 | | |
10 | | //! The ChaCha random number generator. |
11 | | |
12 | | use ppv_lite86::{dispatch, dispatch_light128}; |
13 | | |
14 | | pub use ppv_lite86::Machine; |
15 | | use ppv_lite86::{vec128_storage, ArithOps, BitOps32, LaneWords4, MultiLane, StoreBytes, Vec4}; |
16 | | |
17 | | pub(crate) const BLOCK: usize = 64; |
18 | | pub(crate) const BLOCK64: u64 = BLOCK as u64; |
19 | | const LOG2_BUFBLOCKS: u64 = 2; |
20 | | const BUFBLOCKS: u64 = 1 << LOG2_BUFBLOCKS; |
21 | | pub(crate) const BUFSZ64: u64 = BLOCK64 * BUFBLOCKS; |
22 | | pub(crate) const BUFSZ: usize = BUFSZ64 as usize; |
23 | | |
24 | | const STREAM_PARAM_NONCE: u32 = 1; |
25 | | const STREAM_PARAM_BLOCK: u32 = 0; |
26 | | |
27 | | #[derive(Clone, PartialEq, Eq)] |
28 | | pub struct ChaCha { |
29 | | pub(crate) b: vec128_storage, |
30 | | pub(crate) c: vec128_storage, |
31 | | pub(crate) d: vec128_storage, |
32 | | } |
33 | | |
34 | | #[derive(Clone)] |
35 | | pub struct State<V> { |
36 | | pub(crate) a: V, |
37 | | pub(crate) b: V, |
38 | | pub(crate) c: V, |
39 | | pub(crate) d: V, |
40 | | } |
41 | | |
42 | | #[inline(always)] |
43 | 0 | pub(crate) fn round<V: ArithOps + BitOps32>(mut x: State<V>) -> State<V> { |
44 | 0 | x.a += x.b; |
45 | 0 | x.d = (x.d ^ x.a).rotate_each_word_right16(); |
46 | 0 | x.c += x.d; |
47 | 0 | x.b = (x.b ^ x.c).rotate_each_word_right20(); |
48 | 0 | x.a += x.b; |
49 | 0 | x.d = (x.d ^ x.a).rotate_each_word_right24(); |
50 | 0 | x.c += x.d; |
51 | 0 | x.b = (x.b ^ x.c).rotate_each_word_right25(); |
52 | 0 | x |
53 | 0 | } Unexecuted instantiation: rand_chacha::guts::round::<ppv_lite86::soft::x2<ppv_lite86::x86_64::sse2::avx2::u32x4x2_avx2<ppv_lite86::x86_64::NoNI>, ppv_lite86::x86_64::sse2::G0>> Unexecuted instantiation: rand_chacha::guts::round::<ppv_lite86::soft::x4<ppv_lite86::x86_64::sse2::u32x4_sse2<ppv_lite86::x86_64::NoS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>>> Unexecuted instantiation: rand_chacha::guts::round::<ppv_lite86::soft::x4<ppv_lite86::x86_64::sse2::u32x4_sse2<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>>> Unexecuted instantiation: rand_chacha::guts::round::<ppv_lite86::soft::x4<ppv_lite86::x86_64::sse2::u32x4_sse2<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::YesS4, ppv_lite86::x86_64::NoNI>>> |
54 | | |
55 | | #[inline(always)] |
56 | 0 | pub(crate) fn diagonalize<V: LaneWords4>(mut x: State<V>) -> State<V> { |
57 | 0 | x.b = x.b.shuffle_lane_words3012(); |
58 | 0 | x.c = x.c.shuffle_lane_words2301(); |
59 | 0 | x.d = x.d.shuffle_lane_words1230(); |
60 | 0 | x |
61 | 0 | } Unexecuted instantiation: rand_chacha::guts::diagonalize::<ppv_lite86::soft::x2<ppv_lite86::x86_64::sse2::avx2::u32x4x2_avx2<ppv_lite86::x86_64::NoNI>, ppv_lite86::x86_64::sse2::G0>> Unexecuted instantiation: rand_chacha::guts::diagonalize::<ppv_lite86::soft::x4<ppv_lite86::x86_64::sse2::u32x4_sse2<ppv_lite86::x86_64::NoS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>>> Unexecuted instantiation: rand_chacha::guts::diagonalize::<ppv_lite86::soft::x4<ppv_lite86::x86_64::sse2::u32x4_sse2<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>>> Unexecuted instantiation: rand_chacha::guts::diagonalize::<ppv_lite86::soft::x4<ppv_lite86::x86_64::sse2::u32x4_sse2<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::YesS4, ppv_lite86::x86_64::NoNI>>> |
62 | | #[inline(always)] |
63 | 0 | pub(crate) fn undiagonalize<V: LaneWords4>(mut x: State<V>) -> State<V> { |
64 | 0 | x.b = x.b.shuffle_lane_words1230(); |
65 | 0 | x.c = x.c.shuffle_lane_words2301(); |
66 | 0 | x.d = x.d.shuffle_lane_words3012(); |
67 | 0 | x |
68 | 0 | } Unexecuted instantiation: rand_chacha::guts::undiagonalize::<ppv_lite86::soft::x2<ppv_lite86::x86_64::sse2::avx2::u32x4x2_avx2<ppv_lite86::x86_64::NoNI>, ppv_lite86::x86_64::sse2::G0>> Unexecuted instantiation: rand_chacha::guts::undiagonalize::<ppv_lite86::soft::x4<ppv_lite86::x86_64::sse2::u32x4_sse2<ppv_lite86::x86_64::NoS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>>> Unexecuted instantiation: rand_chacha::guts::undiagonalize::<ppv_lite86::soft::x4<ppv_lite86::x86_64::sse2::u32x4_sse2<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>>> Unexecuted instantiation: rand_chacha::guts::undiagonalize::<ppv_lite86::soft::x4<ppv_lite86::x86_64::sse2::u32x4_sse2<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::YesS4, ppv_lite86::x86_64::NoNI>>> |
69 | | |
70 | | impl ChaCha { |
71 | | #[inline(always)] |
72 | 0 | pub fn new(key: &[u8; 32], nonce: &[u8]) -> Self { |
73 | 0 | init_chacha(key, nonce) |
74 | 0 | } |
75 | | |
76 | | #[inline(always)] |
77 | 0 | fn pos64<M: Machine>(&self, m: M) -> u64 { |
78 | 0 | let d: M::u32x4 = m.unpack(self.d); |
79 | 0 | ((d.extract(1) as u64) << 32) | d.extract(0) as u64 |
80 | 0 | } Unexecuted instantiation: <rand_chacha::guts::ChaCha>::pos64::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::NoS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: <rand_chacha::guts::ChaCha>::pos64::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: <rand_chacha::guts::ChaCha>::pos64::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::YesS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: <rand_chacha::guts::ChaCha>::pos64::<ppv_lite86::x86_64::Avx2Machine<ppv_lite86::x86_64::NoNI>> |
81 | | |
82 | | /// Produce 4 blocks of output, advancing the state |
83 | | #[inline(always)] |
84 | 0 | pub fn refill4(&mut self, drounds: u32, out: &mut [u8; BUFSZ]) { |
85 | 0 | refill_wide(self, drounds, out) |
86 | 0 | } |
87 | | |
88 | | #[inline(always)] |
89 | 0 | pub fn set_block_pos(&mut self, value: u64) { |
90 | 0 | set_stream_param(self, STREAM_PARAM_BLOCK, value) |
91 | 0 | } |
92 | | |
93 | | #[inline(always)] |
94 | 0 | pub fn get_block_pos(&self) -> u64 { |
95 | 0 | get_stream_param(self, STREAM_PARAM_BLOCK) |
96 | 0 | } |
97 | | |
98 | | #[inline(always)] |
99 | 0 | pub fn set_nonce(&mut self, value: u64) { |
100 | 0 | set_stream_param(self, STREAM_PARAM_NONCE, value) |
101 | 0 | } |
102 | | |
103 | | #[inline(always)] |
104 | 0 | pub fn get_nonce(&self) -> u64 { |
105 | 0 | get_stream_param(self, STREAM_PARAM_NONCE) |
106 | 0 | } |
107 | | |
108 | | #[inline(always)] |
109 | 0 | pub fn get_seed(&self) -> [u8; 32] { |
110 | 0 | get_seed(self) |
111 | 0 | } |
112 | | } |
113 | | |
114 | | #[allow(clippy::many_single_char_names)] |
115 | | #[inline(always)] |
116 | 0 | fn refill_wide_impl<Mach: Machine>( |
117 | 0 | m: Mach, state: &mut ChaCha, drounds: u32, out: &mut [u8; BUFSZ], |
118 | 0 | ) { |
119 | 0 | let k = m.vec([0x6170_7865, 0x3320_646e, 0x7962_2d32, 0x6b20_6574]); |
120 | 0 | let mut pos = state.pos64(m); |
121 | 0 | let d0: Mach::u32x4 = m.unpack(state.d); |
122 | 0 | pos = pos.wrapping_add(1); |
123 | 0 | let d1 = d0.insert((pos >> 32) as u32, 1).insert(pos as u32, 0); |
124 | 0 | pos = pos.wrapping_add(1); |
125 | 0 | let d2 = d0.insert((pos >> 32) as u32, 1).insert(pos as u32, 0); |
126 | 0 | pos = pos.wrapping_add(1); |
127 | 0 | let d3 = d0.insert((pos >> 32) as u32, 1).insert(pos as u32, 0); |
128 | | |
129 | 0 | let b = m.unpack(state.b); |
130 | 0 | let c = m.unpack(state.c); |
131 | 0 | let mut x = State { |
132 | 0 | a: Mach::u32x4x4::from_lanes([k, k, k, k]), |
133 | 0 | b: Mach::u32x4x4::from_lanes([b, b, b, b]), |
134 | 0 | c: Mach::u32x4x4::from_lanes([c, c, c, c]), |
135 | 0 | d: m.unpack(Mach::u32x4x4::from_lanes([d0, d1, d2, d3]).into()), |
136 | 0 | }; |
137 | 0 | for _ in 0..drounds { |
138 | 0 | x = round(x); |
139 | 0 | x = undiagonalize(round(diagonalize(x))); |
140 | 0 | } |
141 | 0 | let mut pos = state.pos64(m); |
142 | 0 | let d0: Mach::u32x4 = m.unpack(state.d); |
143 | 0 | pos = pos.wrapping_add(1); |
144 | 0 | let d1 = d0.insert((pos >> 32) as u32, 1).insert(pos as u32, 0); |
145 | 0 | pos = pos.wrapping_add(1); |
146 | 0 | let d2 = d0.insert((pos >> 32) as u32, 1).insert(pos as u32, 0); |
147 | 0 | pos = pos.wrapping_add(1); |
148 | 0 | let d3 = d0.insert((pos >> 32) as u32, 1).insert(pos as u32, 0); |
149 | 0 | pos = pos.wrapping_add(1); |
150 | 0 | let d4 = d0.insert((pos >> 32) as u32, 1).insert(pos as u32, 0); |
151 | | |
152 | 0 | let (a, b, c, d) = ( |
153 | 0 | x.a.to_lanes(), |
154 | 0 | x.b.to_lanes(), |
155 | 0 | x.c.to_lanes(), |
156 | 0 | x.d.to_lanes(), |
157 | 0 | ); |
158 | 0 | let sb = m.unpack(state.b); |
159 | 0 | let sc = m.unpack(state.c); |
160 | 0 | let sd = [m.unpack(state.d), d1, d2, d3]; |
161 | 0 | state.d = d4.into(); |
162 | 0 | let mut words = out.chunks_exact_mut(16); |
163 | 0 | for ((((&a, &b), &c), &d), &sd) in a.iter().zip(&b).zip(&c).zip(&d).zip(&sd) { |
164 | 0 | (a + k).write_le(words.next().unwrap()); |
165 | 0 | (b + sb).write_le(words.next().unwrap()); |
166 | 0 | (c + sc).write_le(words.next().unwrap()); |
167 | 0 | (d + sd).write_le(words.next().unwrap()); |
168 | 0 | } |
169 | 0 | } Unexecuted instantiation: rand_chacha::guts::refill_wide_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::NoS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: rand_chacha::guts::refill_wide_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: rand_chacha::guts::refill_wide_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::YesS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: rand_chacha::guts::refill_wide_impl::<ppv_lite86::x86_64::Avx2Machine<ppv_lite86::x86_64::NoNI>> |
170 | | |
171 | | dispatch!(m, Mach, { |
172 | 0 | fn refill_wide(state: &mut ChaCha, drounds: u32, out: &mut [u8; BUFSZ]) { |
173 | 0 | refill_wide_impl(m, state, drounds, out); Unexecuted instantiation: rand_chacha::guts::refill_wide::fn_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::NoS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: rand_chacha::guts::refill_wide::fn_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: rand_chacha::guts::refill_wide::fn_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::YesS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: rand_chacha::guts::refill_wide::fn_impl::<ppv_lite86::x86_64::Avx2Machine<ppv_lite86::x86_64::NoNI>> |
174 | | } |
175 | | }); |
176 | | |
177 | | // Single-block, rounds-only; shared by try_apply_keystream for tails shorter than BUFSZ |
178 | | // and XChaCha's setup step. |
179 | | dispatch!(m, Mach, { |
180 | 0 | fn refill_narrow_rounds(state: &mut ChaCha, drounds: u32) -> State<vec128_storage> { |
181 | 0 | let k: Mach::u32x4 = m.vec([0x6170_7865, 0x3320_646e, 0x7962_2d32, 0x6b20_6574]); |
182 | 0 | let mut x = State { |
183 | 0 | a: k, |
184 | 0 | b: m.unpack(state.b), |
185 | 0 | c: m.unpack(state.c), |
186 | 0 | d: m.unpack(state.d), |
187 | 0 | }; |
188 | 0 | for _ in 0..drounds { |
189 | 0 | x = round(x); |
190 | 0 | x = undiagonalize(round(diagonalize(x))); |
191 | 0 | } |
192 | 0 | State { |
193 | 0 | a: x.a.into(), |
194 | 0 | b: x.b.into(), |
195 | 0 | c: x.c.into(), |
196 | 0 | d: x.d.into(), |
197 | 0 | } |
198 | | } |
199 | | }); |
200 | | |
201 | | dispatch_light128!(m, Mach, { |
202 | 0 | fn set_stream_param(state: &mut ChaCha, param: u32, value: u64) { |
203 | 0 | let d: Mach::u32x4 = m.unpack(state.d); |
204 | 0 | state.d = d |
205 | 0 | .insert((value >> 32) as u32, (param << 1) | 1) |
206 | 0 | .insert(value as u32, param << 1) |
207 | 0 | .into(); Unexecuted instantiation: rand_chacha::guts::set_stream_param::fn_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::NoS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: rand_chacha::guts::set_stream_param::fn_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::YesS4, ppv_lite86::x86_64::NoNI>> |
208 | | } |
209 | | }); |
210 | | |
211 | | dispatch_light128!(m, Mach, { |
212 | 0 | fn get_stream_param(state: &ChaCha, param: u32) -> u64 { |
213 | 0 | let d: Mach::u32x4 = m.unpack(state.d); |
214 | 0 | ((d.extract((param << 1) | 1) as u64) << 32) | d.extract(param << 1) as u64 Unexecuted instantiation: rand_chacha::guts::get_stream_param::fn_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::NoS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: rand_chacha::guts::get_stream_param::fn_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::YesS4, ppv_lite86::x86_64::NoNI>> |
215 | | } |
216 | | }); |
217 | | |
218 | | dispatch_light128!(m, Mach, { |
219 | 0 | fn get_seed(state: &ChaCha) -> [u8; 32] { |
220 | 0 | let b: Mach::u32x4 = m.unpack(state.b); |
221 | 0 | let c: Mach::u32x4 = m.unpack(state.c); |
222 | 0 | let mut key = [0u8; 32]; |
223 | 0 | b.write_le(&mut key[..16]); |
224 | 0 | c.write_le(&mut key[16..]); |
225 | 0 | key Unexecuted instantiation: rand_chacha::guts::get_seed::fn_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::NoS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: rand_chacha::guts::get_seed::fn_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::YesS4, ppv_lite86::x86_64::NoNI>> |
226 | | } |
227 | | }); |
228 | | |
229 | 0 | fn read_u32le(xs: &[u8]) -> u32 { |
230 | 0 | assert_eq!(xs.len(), 4); |
231 | 0 | u32::from(xs[0]) | (u32::from(xs[1]) << 8) | (u32::from(xs[2]) << 16) | (u32::from(xs[3]) << 24) |
232 | 0 | } |
233 | | |
234 | | dispatch_light128!(m, Mach, { |
235 | 0 | fn init_chacha(key: &[u8; 32], nonce: &[u8]) -> ChaCha { |
236 | 0 | let ctr_nonce = [ |
237 | | 0, |
238 | 0 | if nonce.len() == 12 { |
239 | 0 | read_u32le(&nonce[0..4]) |
240 | | } else { |
241 | 0 | 0 |
242 | | }, |
243 | 0 | read_u32le(&nonce[nonce.len() - 8..nonce.len() - 4]), |
244 | 0 | read_u32le(&nonce[nonce.len() - 4..]), |
245 | | ]; |
246 | 0 | let key0: Mach::u32x4 = m.read_le(&key[..16]); |
247 | 0 | let key1: Mach::u32x4 = m.read_le(&key[16..]); |
248 | 0 | ChaCha { |
249 | 0 | b: key0.into(), |
250 | 0 | c: key1.into(), |
251 | 0 | d: ctr_nonce.into(), |
252 | 0 | } Unexecuted instantiation: rand_chacha::guts::init_chacha::fn_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::NoS3, ppv_lite86::x86_64::NoS4, ppv_lite86::x86_64::NoNI>> Unexecuted instantiation: rand_chacha::guts::init_chacha::fn_impl::<ppv_lite86::x86_64::SseMachine<ppv_lite86::x86_64::YesS3, ppv_lite86::x86_64::YesS4, ppv_lite86::x86_64::NoNI>> |
253 | | } |
254 | | }); |
255 | | |
256 | | dispatch_light128!(m, Mach, { |
257 | 0 | fn init_chacha_x(key: &[u8; 32], nonce: &[u8; 24], rounds: u32) -> ChaCha { |
258 | 0 | let key0: Mach::u32x4 = m.read_le(&key[..16]); |
259 | 0 | let key1: Mach::u32x4 = m.read_le(&key[16..]); |
260 | 0 | let nonce0: Mach::u32x4 = m.read_le(&nonce[..16]); |
261 | 0 | let mut state = ChaCha { |
262 | 0 | b: key0.into(), |
263 | 0 | c: key1.into(), |
264 | 0 | d: nonce0.into(), |
265 | 0 | }; |
266 | 0 | let x = refill_narrow_rounds(&mut state, rounds); |
267 | 0 | let ctr_nonce1 = [0, 0, read_u32le(&nonce[16..20]), read_u32le(&nonce[20..24])]; |
268 | 0 | state.b = x.a; |
269 | 0 | state.c = x.d; |
270 | 0 | state.d = ctr_nonce1.into(); |
271 | 0 | state |
272 | | } |
273 | | }); |