1 // Copyright 2012-2015 The Rust Project Developers. See the COPYRIGHT
2 // file at the top-level directory of this distribution and at
3 // http://rust-lang.org/COPYRIGHT.
5 // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
6 // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license
7 // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your
8 // option. This file may not be copied, modified, or distributed
9 // except according to those terms.
11 // ignore-lexer-test FIXME #15883
13 //! An implementation of SipHash 2-4.
15 //! See: http://131002.net/siphash/
17 //! Consider this as a main "general-purpose" hash for all hashtables: it
18 //! runs at good speed (competitive with spooky and city) and permits
19 //! strong _keyed_ hashing. Key your hashtables from a strong RNG,
20 //! such as `rand::Rng`.
22 //! Although the SipHash algorithm is considered to be cryptographically
23 //! strong, this implementation has not been reviewed for such purposes.
24 //! As such, all cryptographic uses of this implementation are strongly
30 use super::{Hash, Hasher, Writer};
32 /// `SipState` computes a SipHash 2-4 hash over a stream of bytes.
37 length: uint, // how many bytes we've processed
38 v0: u64, // hash state
42 tail: u64, // unprocessed bytes le
43 ntail: uint, // how many bytes in tail are valid
46 // sadly, these macro definitions can't appear later,
47 // because they're needed in the following defs;
48 // this design could be improved.
50 macro_rules! u8to64_le {
51 ($buf:expr, $i:expr) =>
53 ($buf[1+$i] as u64) << 8 |
54 ($buf[2+$i] as u64) << 16 |
55 ($buf[3+$i] as u64) << 24 |
56 ($buf[4+$i] as u64) << 32 |
57 ($buf[5+$i] as u64) << 40 |
58 ($buf[6+$i] as u64) << 48 |
59 ($buf[7+$i] as u64) << 56);
60 ($buf:expr, $i:expr, $len:expr) =>
65 out |= ($buf[t+$i] as u64) << t*8;
74 (($x << $b) | ($x >> (64 - $b)))
77 macro_rules! compress {
78 ($v0:expr, $v1:expr, $v2:expr, $v3:expr) =>
80 $v0 += $v1; $v1 = rotl!($v1, 13); $v1 ^= $v0;
82 $v2 += $v3; $v3 = rotl!($v3, 16); $v3 ^= $v2;
83 $v0 += $v3; $v3 = rotl!($v3, 21); $v3 ^= $v0;
84 $v2 += $v1; $v1 = rotl!($v1, 17); $v1 ^= $v2;
90 /// Creates a `SipState` that is keyed off the provided keys.
92 pub fn new() -> SipState {
93 SipState::new_with_keys(0, 0)
96 /// Creates a `SipState` that is keyed off the provided keys.
98 pub fn new_with_keys(key0: u64, key1: u64) -> SipState {
99 let mut state = SipState {
114 /// Resets the state to its initial state.
116 pub fn reset(&mut self) {
118 self.v0 = self.k0 ^ 0x736f6d6570736575;
119 self.v1 = self.k1 ^ 0x646f72616e646f6d;
120 self.v2 = self.k0 ^ 0x6c7967656e657261;
121 self.v3 = self.k1 ^ 0x7465646279746573;
125 /// Returns the computed hash.
127 pub fn result(&self) -> u64 {
128 let mut v0 = self.v0;
129 let mut v1 = self.v1;
130 let mut v2 = self.v2;
131 let mut v3 = self.v3;
133 let b: u64 = ((self.length as u64 & 0xff) << 56) | self.tail;
136 compress!(v0, v1, v2, v3);
137 compress!(v0, v1, v2, v3);
141 compress!(v0, v1, v2, v3);
142 compress!(v0, v1, v2, v3);
143 compress!(v0, v1, v2, v3);
144 compress!(v0, v1, v2, v3);
150 impl Writer for SipState {
152 fn write(&mut self, msg: &[u8]) {
153 let length = msg.len();
154 self.length += length;
159 needed = 8 - self.ntail;
161 self.tail |= u8to64_le!(msg, 0, length) << 8*self.ntail;
162 self.ntail += length;
166 let m = self.tail | u8to64_le!(msg, 0, needed) << 8*self.ntail;
169 compress!(self.v0, self.v1, self.v2, self.v3);
170 compress!(self.v0, self.v1, self.v2, self.v3);
176 // Buffered tail is now flushed, process new input.
177 let len = length - needed;
178 let end = len & (!0x7);
179 let left = len & 0x7;
183 let mi = u8to64_le!(msg, i);
186 compress!(self.v0, self.v1, self.v2, self.v3);
187 compress!(self.v0, self.v1, self.v2, self.v3);
193 self.tail = u8to64_le!(msg, i, left);
199 impl Clone for SipState {
201 fn clone(&self) -> SipState {
207 impl Default for SipState {
210 fn default() -> SipState {
215 /// `SipHasher` computes the SipHash algorithm from a stream of bytes.
217 #[allow(missing_copy_implementations)]
218 pub struct SipHasher {
226 pub fn new() -> SipHasher {
227 SipHasher::new_with_keys(0, 0)
230 /// Creates a `Sip` that is keyed off the provided keys.
232 pub fn new_with_keys(key0: u64, key1: u64) -> SipHasher {
240 impl Hasher<SipState> for SipHasher {
242 fn hash<Sized? T: Hash<SipState>>(&self, value: &T) -> u64 {
243 let mut state = SipState::new_with_keys(self.k0, self.k1);
244 value.hash(&mut state);
249 impl Default for SipHasher {
251 fn default() -> SipHasher {
256 /// Hashes a value using the SipHash algorithm.
258 pub fn hash<Sized? T: Hash<SipState>>(value: &T) -> u64 {
259 let mut state = SipState::new();
260 value.hash(&mut state);
264 /// Hashes a value with the SipHash algorithm with the provided keys.
266 pub fn hash_with_keys<Sized? T: Hash<SipState>>(k0: u64, k1: u64, value: &T) -> u64 {
267 let mut state = SipState::new_with_keys(k0, k1);
268 value.hash(&mut state);
278 use super::super::{Hash, Writer};
279 use super::{SipState, hash, hash_with_keys};
281 // Hash just the bytes of the slice, without length prefix
282 struct Bytes<'a>(&'a [u8]);
284 impl<'a, S: Writer> Hash<S> for Bytes<'a> {
285 #[allow(unused_must_use)]
286 fn hash(&self, state: &mut S) {
287 let Bytes(v) = *self;
293 #[allow(unused_must_use)]
295 let vecs : [[u8; 8]; 64] = [
296 [ 0x31, 0x0e, 0x0e, 0xdd, 0x47, 0xdb, 0x6f, 0x72, ],
297 [ 0xfd, 0x67, 0xdc, 0x93, 0xc5, 0x39, 0xf8, 0x74, ],
298 [ 0x5a, 0x4f, 0xa9, 0xd9, 0x09, 0x80, 0x6c, 0x0d, ],
299 [ 0x2d, 0x7e, 0xfb, 0xd7, 0x96, 0x66, 0x67, 0x85, ],
300 [ 0xb7, 0x87, 0x71, 0x27, 0xe0, 0x94, 0x27, 0xcf, ],
301 [ 0x8d, 0xa6, 0x99, 0xcd, 0x64, 0x55, 0x76, 0x18, ],
302 [ 0xce, 0xe3, 0xfe, 0x58, 0x6e, 0x46, 0xc9, 0xcb, ],
303 [ 0x37, 0xd1, 0x01, 0x8b, 0xf5, 0x00, 0x02, 0xab, ],
304 [ 0x62, 0x24, 0x93, 0x9a, 0x79, 0xf5, 0xf5, 0x93, ],
305 [ 0xb0, 0xe4, 0xa9, 0x0b, 0xdf, 0x82, 0x00, 0x9e, ],
306 [ 0xf3, 0xb9, 0xdd, 0x94, 0xc5, 0xbb, 0x5d, 0x7a, ],
307 [ 0xa7, 0xad, 0x6b, 0x22, 0x46, 0x2f, 0xb3, 0xf4, ],
308 [ 0xfb, 0xe5, 0x0e, 0x86, 0xbc, 0x8f, 0x1e, 0x75, ],
309 [ 0x90, 0x3d, 0x84, 0xc0, 0x27, 0x56, 0xea, 0x14, ],
310 [ 0xee, 0xf2, 0x7a, 0x8e, 0x90, 0xca, 0x23, 0xf7, ],
311 [ 0xe5, 0x45, 0xbe, 0x49, 0x61, 0xca, 0x29, 0xa1, ],
312 [ 0xdb, 0x9b, 0xc2, 0x57, 0x7f, 0xcc, 0x2a, 0x3f, ],
313 [ 0x94, 0x47, 0xbe, 0x2c, 0xf5, 0xe9, 0x9a, 0x69, ],
314 [ 0x9c, 0xd3, 0x8d, 0x96, 0xf0, 0xb3, 0xc1, 0x4b, ],
315 [ 0xbd, 0x61, 0x79, 0xa7, 0x1d, 0xc9, 0x6d, 0xbb, ],
316 [ 0x98, 0xee, 0xa2, 0x1a, 0xf2, 0x5c, 0xd6, 0xbe, ],
317 [ 0xc7, 0x67, 0x3b, 0x2e, 0xb0, 0xcb, 0xf2, 0xd0, ],
318 [ 0x88, 0x3e, 0xa3, 0xe3, 0x95, 0x67, 0x53, 0x93, ],
319 [ 0xc8, 0xce, 0x5c, 0xcd, 0x8c, 0x03, 0x0c, 0xa8, ],
320 [ 0x94, 0xaf, 0x49, 0xf6, 0xc6, 0x50, 0xad, 0xb8, ],
321 [ 0xea, 0xb8, 0x85, 0x8a, 0xde, 0x92, 0xe1, 0xbc, ],
322 [ 0xf3, 0x15, 0xbb, 0x5b, 0xb8, 0x35, 0xd8, 0x17, ],
323 [ 0xad, 0xcf, 0x6b, 0x07, 0x63, 0x61, 0x2e, 0x2f, ],
324 [ 0xa5, 0xc9, 0x1d, 0xa7, 0xac, 0xaa, 0x4d, 0xde, ],
325 [ 0x71, 0x65, 0x95, 0x87, 0x66, 0x50, 0xa2, 0xa6, ],
326 [ 0x28, 0xef, 0x49, 0x5c, 0x53, 0xa3, 0x87, 0xad, ],
327 [ 0x42, 0xc3, 0x41, 0xd8, 0xfa, 0x92, 0xd8, 0x32, ],
328 [ 0xce, 0x7c, 0xf2, 0x72, 0x2f, 0x51, 0x27, 0x71, ],
329 [ 0xe3, 0x78, 0x59, 0xf9, 0x46, 0x23, 0xf3, 0xa7, ],
330 [ 0x38, 0x12, 0x05, 0xbb, 0x1a, 0xb0, 0xe0, 0x12, ],
331 [ 0xae, 0x97, 0xa1, 0x0f, 0xd4, 0x34, 0xe0, 0x15, ],
332 [ 0xb4, 0xa3, 0x15, 0x08, 0xbe, 0xff, 0x4d, 0x31, ],
333 [ 0x81, 0x39, 0x62, 0x29, 0xf0, 0x90, 0x79, 0x02, ],
334 [ 0x4d, 0x0c, 0xf4, 0x9e, 0xe5, 0xd4, 0xdc, 0xca, ],
335 [ 0x5c, 0x73, 0x33, 0x6a, 0x76, 0xd8, 0xbf, 0x9a, ],
336 [ 0xd0, 0xa7, 0x04, 0x53, 0x6b, 0xa9, 0x3e, 0x0e, ],
337 [ 0x92, 0x59, 0x58, 0xfc, 0xd6, 0x42, 0x0c, 0xad, ],
338 [ 0xa9, 0x15, 0xc2, 0x9b, 0xc8, 0x06, 0x73, 0x18, ],
339 [ 0x95, 0x2b, 0x79, 0xf3, 0xbc, 0x0a, 0xa6, 0xd4, ],
340 [ 0xf2, 0x1d, 0xf2, 0xe4, 0x1d, 0x45, 0x35, 0xf9, ],
341 [ 0x87, 0x57, 0x75, 0x19, 0x04, 0x8f, 0x53, 0xa9, ],
342 [ 0x10, 0xa5, 0x6c, 0xf5, 0xdf, 0xcd, 0x9a, 0xdb, ],
343 [ 0xeb, 0x75, 0x09, 0x5c, 0xcd, 0x98, 0x6c, 0xd0, ],
344 [ 0x51, 0xa9, 0xcb, 0x9e, 0xcb, 0xa3, 0x12, 0xe6, ],
345 [ 0x96, 0xaf, 0xad, 0xfc, 0x2c, 0xe6, 0x66, 0xc7, ],
346 [ 0x72, 0xfe, 0x52, 0x97, 0x5a, 0x43, 0x64, 0xee, ],
347 [ 0x5a, 0x16, 0x45, 0xb2, 0x76, 0xd5, 0x92, 0xa1, ],
348 [ 0xb2, 0x74, 0xcb, 0x8e, 0xbf, 0x87, 0x87, 0x0a, ],
349 [ 0x6f, 0x9b, 0xb4, 0x20, 0x3d, 0xe7, 0xb3, 0x81, ],
350 [ 0xea, 0xec, 0xb2, 0xa3, 0x0b, 0x22, 0xa8, 0x7f, ],
351 [ 0x99, 0x24, 0xa4, 0x3c, 0xc1, 0x31, 0x57, 0x24, ],
352 [ 0xbd, 0x83, 0x8d, 0x3a, 0xaf, 0xbf, 0x8d, 0xb7, ],
353 [ 0x0b, 0x1a, 0x2a, 0x32, 0x65, 0xd5, 0x1a, 0xea, ],
354 [ 0x13, 0x50, 0x79, 0xa3, 0x23, 0x1c, 0xe6, 0x60, ],
355 [ 0x93, 0x2b, 0x28, 0x46, 0xe4, 0xd7, 0x06, 0x66, ],
356 [ 0xe1, 0x91, 0x5f, 0x5c, 0xb1, 0xec, 0xa4, 0x6c, ],
357 [ 0xf3, 0x25, 0x96, 0x5c, 0xa1, 0x6d, 0x62, 0x9f, ],
358 [ 0x57, 0x5f, 0xf2, 0x8e, 0x60, 0x38, 0x1b, 0xe5, ],
359 [ 0x72, 0x45, 0x06, 0xeb, 0x4c, 0x32, 0x8a, 0x95, ]
362 let k0 = 0x_07_06_05_04_03_02_01_00_u64;
363 let k1 = 0x_0f_0e_0d_0c_0b_0a_09_08_u64;
364 let mut buf = Vec::new();
366 let mut state_inc = SipState::new_with_keys(k0, k1);
367 let mut state_full = SipState::new_with_keys(k0, k1);
369 fn to_hex_str(r: &[u8; 8]) -> String {
370 let mut s = String::new();
372 s.push_str(format!("{}", fmt::radix(*b, 16)).as_slice());
377 fn result_bytes(h: u64) -> Vec<u8> {
389 fn result_str(h: u64) -> String {
390 let r = result_bytes(h);
391 let mut s = String::new();
393 s.push_str(format!("{}", fmt::radix(*b, 16)).as_slice());
399 debug!("siphash test {}: {}", t, buf);
400 let vec = u8to64_le!(vecs[t], 0);
401 let out = hash_with_keys(k0, k1, &Bytes(buf.as_slice()));
402 debug!("got {}, expected {}", out, vec);
403 assert_eq!(vec, out);
406 state_full.write(buf.as_slice());
407 let f = result_str(state_full.result());
408 let i = result_str(state_inc.result());
409 let v = to_hex_str(&vecs[t]);
410 debug!("{}: ({}) => inc={} full={}", t, v, i, f);
416 state_inc.write(&[t as u8]);
422 #[test] #[cfg(target_arch = "aarch64")]
423 fn test_hash_uint() {
424 let val = 0xdeadbeef_deadbeef_u64;
425 assert_eq!(hash(&(val as u64)), hash(&(val as uint)));
426 assert!(hash(&(val as u32)) != hash(&(val as uint)));
428 #[test] #[cfg(target_arch = "arm")]
429 fn test_hash_uint() {
430 let val = 0xdeadbeef_deadbeef_u64;
431 assert!(hash(&(val as u64)) != hash(&(val as uint)));
432 assert_eq!(hash(&(val as u32)), hash(&(val as uint)));
434 #[test] #[cfg(target_arch = "x86_64")]
435 fn test_hash_uint() {
436 let val = 0xdeadbeef_deadbeef_u64;
437 assert_eq!(hash(&(val as u64)), hash(&(val as uint)));
438 assert!(hash(&(val as u32)) != hash(&(val as uint)));
440 #[test] #[cfg(target_arch = "x86")]
441 fn test_hash_uint() {
442 let val = 0xdeadbeef_deadbeef_u64;
443 assert!(hash(&(val as u64)) != hash(&(val as uint)));
444 assert_eq!(hash(&(val as u32)), hash(&(val as uint)));
448 fn test_hash_idempotent() {
449 let val64 = 0xdeadbeef_deadbeef_u64;
450 assert_eq!(hash(&val64), hash(&val64));
451 let val32 = 0xdeadbeef_u32;
452 assert_eq!(hash(&val32), hash(&val32));
456 fn test_hash_no_bytes_dropped_64() {
457 let val = 0xdeadbeef_deadbeef_u64;
459 assert!(hash(&val) != hash(&zero_byte(val, 0)));
460 assert!(hash(&val) != hash(&zero_byte(val, 1)));
461 assert!(hash(&val) != hash(&zero_byte(val, 2)));
462 assert!(hash(&val) != hash(&zero_byte(val, 3)));
463 assert!(hash(&val) != hash(&zero_byte(val, 4)));
464 assert!(hash(&val) != hash(&zero_byte(val, 5)));
465 assert!(hash(&val) != hash(&zero_byte(val, 6)));
466 assert!(hash(&val) != hash(&zero_byte(val, 7)));
468 fn zero_byte(val: u64, byte: uint) -> u64 {
470 val & !(0xff << (byte * 8))
475 fn test_hash_no_bytes_dropped_32() {
476 let val = 0xdeadbeef_u32;
478 assert!(hash(&val) != hash(&zero_byte(val, 0)));
479 assert!(hash(&val) != hash(&zero_byte(val, 1)));
480 assert!(hash(&val) != hash(&zero_byte(val, 2)));
481 assert!(hash(&val) != hash(&zero_byte(val, 3)));
483 fn zero_byte(val: u32, byte: uint) -> u32 {
485 val & !(0xff << (byte * 8))
490 fn test_hash_no_concat_alias() {
491 let s = ("aa", "bb");
492 let t = ("aabb", "");
493 let u = ("a", "abb");
495 assert!(s != t && t != u);
496 assert!(hash(&s) != hash(&t) && hash(&s) != hash(&u));
498 let v: (&[u8], &[u8], &[u8]) = (&[1u8], &[0u8, 0], &[0u8]);
499 let w: (&[u8], &[u8], &[u8]) = (&[1u8, 0, 0, 0], &[], &[]);
502 assert!(hash(&v) != hash(&w));
506 fn bench_str_under_8_bytes(b: &mut Bencher) {
509 assert_eq!(hash(&s), 16262950014981195938);
514 fn bench_str_of_8_bytes(b: &mut Bencher) {
517 assert_eq!(hash(&s), 4898293253460910787);
522 fn bench_str_over_8_bytes(b: &mut Bencher) {
523 let s = "foobarbaz0";
525 assert_eq!(hash(&s), 10581415515220175264);
530 fn bench_long_str(b: &mut Bencher) {
531 let s = "Lorem ipsum dolor sit amet, consectetur adipisicing elit, sed do eiusmod tempor \
532 incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud \
533 exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute \
534 irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla \
535 pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui \
536 officia deserunt mollit anim id est laborum.";
538 assert_eq!(hash(&s), 17717065544121360093);
543 fn bench_u64(b: &mut Bencher) {
544 let u = 16262950014981195938u64;
546 assert_eq!(hash(&u), 5254097107239593357);