Repository navigation
Expand file tree
/
Copy pathhash.h
More file actions
313 lines (300 loc) · 9.98 KB
/
Copy pathhash.h
File metadata and controls
313 lines (300 loc) · 9.98 KB
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
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
/* CC0 (Public domain) - see LICENSE file for details */
#ifndef CCAN_HASH_H
#define CCAN_HASH_H
#include "config.h"
#include <stdint.h>
#include <stdlib.h>
#include <ccan/build_assert/build_assert.h>
/* Stolen mostly from: lookup3.c, by Bob Jenkins, May 2006, Public Domain.
*
* http://burtleburtle.net/bob/c/lookup3.c
*/
/**
* hash - fast hash of an array for internal use
* @p: the array or pointer to first element
* @num: the number of elements to hash
* @base: the base number to roll into the hash (usually 0)
*
* The memory region pointed to by p is combined with the base to form
* a 32-bit hash.
*
* This hash will have different results on different machines, so is
* only useful for internal hashes (ie. not hashes sent across the
* network or saved to disk).
*
* It may also change with future versions: it could even detect at runtime
* what the fastest hash to use is.
*
* See also: hash64, hash_stable.
*
* Example:
* #include <ccan/hash/hash.h>
* #include <err.h>
* #include <stdio.h>
* #include <string.h>
*
* // Simple demonstration: idential strings will have the same hash, but
* // two different strings will probably not.
* int main(int argc, char *argv[])
* {
* uint32_t hash1, hash2;
*
* if (argc != 3)
* err(1, "Usage: %s <string1> <string2>", argv[0]);
*
* hash1 = hash(argv[1], strlen(argv[1]), 0);
* hash2 = hash(argv[2], strlen(argv[2]), 0);
* printf("Hash is %s\n", hash1 == hash2 ? "same" : "different");
* return 0;
* }
*/
#define hash(p, num, base) hash_any((p), (num)*sizeof(*(p)), (base))
/**
* hash_stable - hash of an array for external use
* @p: the array or pointer to first element
* @num: the number of elements to hash
* @base: the base number to roll into the hash (usually 0)
*
* The array of simple integer types pointed to by p is combined with
* the base to form a 32-bit hash.
*
* This hash will have the same results on different machines, so can
* be used for external hashes (ie. hashes sent across the network or
* saved to disk). The results will not change in future versions of
* this module.
*
* Note that it is only legal to hand an array of simple integer types
* to this hash (ie. char, uint16_t, int64_t, etc). In these cases,
* the same values will have the same hash result, even though the
* memory representations of integers depend on the machine
* endianness.
*
* See also:
* hash64_stable
*
* Example:
* #include <ccan/hash/hash.h>
* #include <err.h>
* #include <stdio.h>
* #include <string.h>
*
* int main(int argc, char *argv[])
* {
* if (argc != 2)
* err(1, "Usage: %s <string-to-hash>", argv[0]);
*
* printf("Hash stable result is %u\n",
* hash_stable(argv[1], strlen(argv[1]), 0));
* return 0;
* }
*/
#define hash_stable(p, num, base) \
(BUILD_ASSERT_OR_ZERO(sizeof(*(p)) == 8 || sizeof(*(p)) == 4 \
|| sizeof(*(p)) == 2 || sizeof(*(p)) == 1) + \
sizeof(*(p)) == 8 ? hash_stable_64((p), (num), (base)) \
: sizeof(*(p)) == 4 ? hash_stable_32((p), (num), (base)) \
: sizeof(*(p)) == 2 ? hash_stable_16((p), (num), (base)) \
: hash_stable_8((p), (num), (base)))
/**
* hash_u32 - fast hash an array of 32-bit values for internal use
* @key: the array of uint32_t
* @num: the number of elements to hash
* @base: the base number to roll into the hash (usually 0)
*
* The array of uint32_t pointed to by @key is combined with the base
* to form a 32-bit hash. This is 2-3 times faster than hash() on small
* arrays, but the advantage vanishes over large hashes.
*
* This hash will have different results on different machines, so is
* only useful for internal hashes (ie. not hashes sent across the
* network or saved to disk).
*/
uint32_t hash_u32(const uint32_t *key, size_t num, uint32_t base);
/**
* hash_string - very fast hash of an ascii string
* @str: the nul-terminated string
*
* The string is hashed, using a hash function optimized for ASCII and
* similar strings. It's weaker than the other hash functions.
*
* This hash may have different results on different machines, so is
* only useful for internal hashes (ie. not hashes sent across the
* network or saved to disk). The results will be different from the
* other hash functions in this module, too.
*/
static inline uint32_t hash_string(const char *string)
{
/* This is Karl Nelson <kenelson@ece.ucdavis.edu>'s X31 hash.
* It's a little faster than the (much better) lookup3 hash(): 56ns vs
* 84ns on my 2GHz Intel Core Duo 2 laptop for a 10 char string. */
uint32_t ret;
for (ret = 0; *string; string++)
ret = (ret << 5) - ret + *string;
return ret;
}
/**
* hash64 - fast 64-bit hash of an array for internal use
* @p: the array or pointer to first element
* @num: the number of elements to hash
* @base: the 64-bit base number to roll into the hash (usually 0)
*
* The memory region pointed to by p is combined with the base to form
* a 64-bit hash.
*
* This hash will have different results on different machines, so is
* only useful for internal hashes (ie. not hashes sent across the
* network or saved to disk).
*
* It may also change with future versions: it could even detect at runtime
* what the fastest hash to use is.
*
* See also: hash.
*
* Example:
* #include <ccan/hash/hash.h>
* #include <err.h>
* #include <stdio.h>
* #include <string.h>
*
* // Simple demonstration: idential strings will have the same hash, but
* // two different strings will probably not.
* int main(int argc, char *argv[])
* {
* uint64_t hash1, hash2;
*
* if (argc != 3)
* err(1, "Usage: %s <string1> <string2>", argv[0]);
*
* hash1 = hash64(argv[1], strlen(argv[1]), 0);
* hash2 = hash64(argv[2], strlen(argv[2]), 0);
* printf("Hash is %s\n", hash1 == hash2 ? "same" : "different");
* return 0;
* }
*/
#define hash64(p, num, base) hash64_any((p), (num)*sizeof(*(p)), (base))
/**
* hash64_stable - 64 bit hash of an array for external use
* @p: the array or pointer to first element
* @num: the number of elements to hash
* @base: the base number to roll into the hash (usually 0)
*
* The array of simple integer types pointed to by p is combined with
* the base to form a 64-bit hash.
*
* This hash will have the same results on different machines, so can
* be used for external hashes (ie. hashes sent across the network or
* saved to disk). The results will not change in future versions of
* this module.
*
* Note that it is only legal to hand an array of simple integer types
* to this hash (ie. char, uint16_t, int64_t, etc). In these cases,
* the same values will have the same hash result, even though the
* memory representations of integers depend on the machine
* endianness.
*
* See also:
* hash_stable
*
* Example:
* #include <ccan/hash/hash.h>
* #include <err.h>
* #include <stdio.h>
* #include <string.h>
*
* int main(int argc, char *argv[])
* {
* if (argc != 2)
* err(1, "Usage: %s <string-to-hash>", argv[0]);
*
* printf("Hash stable result is %llu\n",
* (long long)hash64_stable(argv[1], strlen(argv[1]), 0));
* return 0;
* }
*/
#define hash64_stable(p, num, base) \
(BUILD_ASSERT_OR_ZERO(sizeof(*(p)) == 8 || sizeof(*(p)) == 4 \
|| sizeof(*(p)) == 2 || sizeof(*(p)) == 1) + \
sizeof(*(p)) == 8 ? hash64_stable_64((p), (num), (base)) \
: sizeof(*(p)) == 4 ? hash64_stable_32((p), (num), (base)) \
: sizeof(*(p)) == 2 ? hash64_stable_16((p), (num), (base)) \
: hash64_stable_8((p), (num), (base)))
/**
* hashl - fast 32/64-bit hash of an array for internal use
* @p: the array or pointer to first element
* @num: the number of elements to hash
* @base: the base number to roll into the hash (usually 0)
*
* This is either hash() or hash64(), on 32/64 bit long machines.
*/
#define hashl(p, num, base) \
(BUILD_ASSERT_OR_ZERO(sizeof(long) == sizeof(uint32_t) \
|| sizeof(long) == sizeof(uint64_t)) + \
(sizeof(long) == sizeof(uint64_t) \
? hash64((p), (num), (base)) : hash((p), (num), (base))))
/* Our underlying operations. */
uint32_t hash_any(const void *key, size_t length, uint32_t base);
uint32_t hash_stable_64(const void *key, size_t n, uint32_t base);
uint32_t hash_stable_32(const void *key, size_t n, uint32_t base);
uint32_t hash_stable_16(const void *key, size_t n, uint32_t base);
uint32_t hash_stable_8(const void *key, size_t n, uint32_t base);
uint64_t hash64_any(const void *key, size_t length, uint64_t base);
uint64_t hash64_stable_64(const void *key, size_t n, uint64_t base);
uint64_t hash64_stable_32(const void *key, size_t n, uint64_t base);
uint64_t hash64_stable_16(const void *key, size_t n, uint64_t base);
uint64_t hash64_stable_8(const void *key, size_t n, uint64_t base);
/**
* hash_pointer - hash a pointer for internal use
* @p: the pointer value to hash
* @base: the base number to roll into the hash (usually 0)
*
* The pointer p (not what p points to!) is combined with the base to form
* a 32-bit hash.
*
* This hash will have different results on different machines, so is
* only useful for internal hashes (ie. not hashes sent across the
* network or saved to disk).
*
* Example:
* #include <ccan/hash/hash.h>
*
* // Code to keep track of memory regions.
* struct region {
* struct region *chain;
* void *start;
* unsigned int size;
* };
* // We keep a simple hash table.
* static struct region *region_hash[128];
*
* static void add_region(struct region *r)
* {
* unsigned int h = hash_pointer(r->start, 0);
*
* r->chain = region_hash[h];
* region_hash[h] = r->chain;
* }
*
* static struct region *find_region(const void *start)
* {
* struct region *r;
*
* for (r = region_hash[hash_pointer(start, 0)]; r; r = r->chain)
* if (r->start == start)
* return r;
* return NULL;
* }
*/
static inline uint32_t hash_pointer(const void *p, uint32_t base)
{
if (sizeof(p) % sizeof(uint32_t) == 0) {
/* This convoluted union is the right way of aliasing. */
union {
uint32_t a[sizeof(p) / sizeof(uint32_t)];
const void *p;
} u;
u.p = p;
return hash_u32(u.a, sizeof(p) / sizeof(uint32_t), base);
} else
return hash(&p, 1, base);
}
#endif /* HASH_H */