Line data Source code
1 : /* Slightly modified by Lennart Poettering, to avoid name clashes, and
2 : * unexport a few functions. */
3 :
4 : #include "lookup3.h"
5 :
6 : /*
7 : -------------------------------------------------------------------------------
8 : lookup3.c, by Bob Jenkins, May 2006, Public Domain.
9 :
10 : These are functions for producing 32-bit hashes for hash table lookup.
11 : hashword(), hashlittle(), hashlittle2(), hashbig(), mix(), and final()
12 : are externally useful functions. Routines to test the hash are included
13 : if SELF_TEST is defined. You can use this free for any purpose. It's in
14 : the public domain. It has no warranty.
15 :
16 : You probably want to use hashlittle(). hashlittle() and hashbig()
17 : hash byte arrays. hashlittle() is faster than hashbig() on
18 : little-endian machines. Intel and AMD are little-endian machines.
19 : On second thought, you probably want hashlittle2(), which is identical to
20 : hashlittle() except it returns two 32-bit hashes for the price of one.
21 : You could implement hashbig2() if you wanted but I haven't bothered here.
22 :
23 : If you want to find a hash of, say, exactly 7 integers, do
24 : a = i1; b = i2; c = i3;
25 : mix(a,b,c);
26 : a += i4; b += i5; c += i6;
27 : mix(a,b,c);
28 : a += i7;
29 : final(a,b,c);
30 : then use c as the hash value. If you have a variable length array of
31 : 4-byte integers to hash, use hashword(). If you have a byte array (like
32 : a character string), use hashlittle(). If you have several byte arrays, or
33 : a mix of things, see the comments above hashlittle().
34 :
35 : Why is this so big? I read 12 bytes at a time into 3 4-byte integers,
36 : then mix those integers. This is fast (you can do a lot more thorough
37 : mixing with 12*3 instructions on 3 integers than you can with 3 instructions
38 : on 1 byte), but shoehorning those bytes into integers efficiently is messy.
39 : -------------------------------------------------------------------------------
40 : */
41 : /* #define SELF_TEST 1 */
42 :
43 : #include <stdint.h> /* defines uint32_t etc */
44 : #include <stdio.h> /* defines printf for tests */
45 : #include <sys/param.h> /* attempt to define endianness */
46 : #include <time.h> /* defines time_t for timings in the test */
47 : #ifdef linux
48 : # include <endian.h> /* attempt to define endianness */
49 : #endif
50 :
51 : #if __GNUC__ >= 7
52 : _Pragma("GCC diagnostic ignored \"-Wimplicit-fallthrough\"")
53 : #endif
54 :
55 : /*
56 : * My best guess at if you are big-endian or little-endian. This may
57 : * need adjustment.
58 : */
59 : #if (defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && \
60 : __BYTE_ORDER == __LITTLE_ENDIAN) || \
61 : (defined(i386) || defined(__i386__) || defined(__i486__) || \
62 : defined(__i586__) || defined(__i686__) || defined(vax) || defined(MIPSEL))
63 : # define HASH_LITTLE_ENDIAN 1
64 : # define HASH_BIG_ENDIAN 0
65 : #elif (defined(__BYTE_ORDER) && defined(__BIG_ENDIAN) && \
66 : __BYTE_ORDER == __BIG_ENDIAN) || \
67 : (defined(sparc) || defined(POWERPC) || defined(mc68000) || defined(sel))
68 : # define HASH_LITTLE_ENDIAN 0
69 : # define HASH_BIG_ENDIAN 1
70 : #else
71 : # define HASH_LITTLE_ENDIAN 0
72 : # define HASH_BIG_ENDIAN 0
73 : #endif
74 :
75 : #define hashsize(n) ((uint32_t)1<<(n))
76 : #define hashmask(n) (hashsize(n)-1)
77 : #define rot(x,k) (((x)<<(k)) | ((x)>>(32-(k))))
78 :
79 : /*
80 : -------------------------------------------------------------------------------
81 : mix -- mix 3 32-bit values reversibly.
82 :
83 : This is reversible, so any information in (a,b,c) before mix() is
84 : still in (a,b,c) after mix().
85 :
86 : If four pairs of (a,b,c) inputs are run through mix(), or through
87 : mix() in reverse, there are at least 32 bits of the output that
88 : are sometimes the same for one pair and different for another pair.
89 : This was tested for:
90 : * pairs that differed by one bit, by two bits, in any combination
91 : of top bits of (a,b,c), or in any combination of bottom bits of
92 : (a,b,c).
93 : * "differ" is defined as +, -, ^, or ~^. For + and -, I transformed
94 : the output delta to a Gray code (a^(a>>1)) so a string of 1's (as
95 : is commonly produced by subtraction) look like a single 1-bit
96 : difference.
97 : * the base values were pseudorandom, all zero but one bit set, or
98 : all zero plus a counter that starts at zero.
99 :
100 : Some k values for my "a-=c; a^=rot(c,k); c+=b;" arrangement that
101 : satisfy this are
102 : 4 6 8 16 19 4
103 : 9 15 3 18 27 15
104 : 14 9 3 7 17 3
105 : Well, "9 15 3 18 27 15" didn't quite get 32 bits diffing
106 : for "differ" defined as + with a one-bit base and a two-bit delta. I
107 : used http://burtleburtle.net/bob/hash/avalanche.html to choose
108 : the operations, constants, and arrangements of the variables.
109 :
110 : This does not achieve avalanche. There are input bits of (a,b,c)
111 : that fail to affect some output bits of (a,b,c), especially of a. The
112 : most thoroughly mixed value is c, but it doesn't really even achieve
113 : avalanche in c.
114 :
115 : This allows some parallelism. Read-after-writes are good at doubling
116 : the number of bits affected, so the goal of mixing pulls in the opposite
117 : direction as the goal of parallelism. I did what I could. Rotates
118 : seem to cost as much as shifts on every machine I could lay my hands
119 : on, and rotates are much kinder to the top and bottom bits, so I used
120 : rotates.
121 : -------------------------------------------------------------------------------
122 : */
123 : #define mix(a,b,c) \
124 : { \
125 : a -= c; a ^= rot(c, 4); c += b; \
126 : b -= a; b ^= rot(a, 6); a += c; \
127 : c -= b; c ^= rot(b, 8); b += a; \
128 : a -= c; a ^= rot(c,16); c += b; \
129 : b -= a; b ^= rot(a,19); a += c; \
130 : c -= b; c ^= rot(b, 4); b += a; \
131 : }
132 :
133 : /*
134 : -------------------------------------------------------------------------------
135 : final -- final mixing of 3 32-bit values (a,b,c) into c
136 :
137 : Pairs of (a,b,c) values differing in only a few bits will usually
138 : produce values of c that look totally different. This was tested for
139 : * pairs that differed by one bit, by two bits, in any combination
140 : of top bits of (a,b,c), or in any combination of bottom bits of
141 : (a,b,c).
142 : * "differ" is defined as +, -, ^, or ~^. For + and -, I transformed
143 : the output delta to a Gray code (a^(a>>1)) so a string of 1's (as
144 : is commonly produced by subtraction) look like a single 1-bit
145 : difference.
146 : * the base values were pseudorandom, all zero but one bit set, or
147 : all zero plus a counter that starts at zero.
148 :
149 : These constants passed:
150 : 14 11 25 16 4 14 24
151 : 12 14 25 16 4 14 24
152 : and these came close:
153 : 4 8 15 26 3 22 24
154 : 10 8 15 26 3 22 24
155 : 11 8 15 26 3 22 24
156 : -------------------------------------------------------------------------------
157 : */
158 : #define final(a,b,c) \
159 : { \
160 : c ^= b; c -= rot(b,14); \
161 : a ^= c; a -= rot(c,11); \
162 : b ^= a; b -= rot(a,25); \
163 : c ^= b; c -= rot(b,16); \
164 : a ^= c; a -= rot(c,4); \
165 : b ^= a; b -= rot(a,14); \
166 : c ^= b; c -= rot(b,24); \
167 : }
168 :
169 : /*
170 : --------------------------------------------------------------------
171 : This works on all machines. To be useful, it requires
172 : -- that the key be an array of uint32_t's, and
173 : -- that the length be the number of uint32_t's in the key
174 :
175 : The function hashword() is identical to hashlittle() on little-endian
176 : machines, and identical to hashbig() on big-endian machines,
177 : except that the length has to be measured in uint32_ts rather than in
178 : bytes. hashlittle() is more complicated than hashword() only because
179 : hashlittle() has to dance around fitting the key bytes into registers.
180 : --------------------------------------------------------------------
181 : */
182 0 : uint32_t jenkins_hashword(
183 : const uint32_t *k, /* the key, an array of uint32_t values */
184 : size_t length, /* the length of the key, in uint32_ts */
185 : uint32_t initval) /* the previous hash, or an arbitrary value */
186 : {
187 : uint32_t a,b,c;
188 :
189 : /* Set up the internal state */
190 0 : a = b = c = 0xdeadbeef + (((uint32_t)length)<<2) + initval;
191 :
192 : /*------------------------------------------------- handle most of the key */
193 0 : while (length > 3)
194 : {
195 0 : a += k[0];
196 0 : b += k[1];
197 0 : c += k[2];
198 0 : mix(a,b,c);
199 0 : length -= 3;
200 0 : k += 3;
201 : }
202 :
203 : /*------------------------------------------- handle the last 3 uint32_t's */
204 0 : switch(length) /* all the case statements fall through */
205 : {
206 0 : case 3 : c+=k[2];
207 0 : case 2 : b+=k[1];
208 0 : case 1 : a+=k[0];
209 0 : final(a,b,c);
210 0 : case 0: /* case 0: nothing left to add */
211 0 : break;
212 : }
213 : /*------------------------------------------------------ report the result */
214 0 : return c;
215 : }
216 :
217 : /*
218 : --------------------------------------------------------------------
219 : hashword2() -- same as hashword(), but take two seeds and return two
220 : 32-bit values. pc and pb must both be nonnull, and *pc and *pb must
221 : both be initialized with seeds. If you pass in (*pb)==0, the output
222 : (*pc) will be the same as the return value from hashword().
223 : --------------------------------------------------------------------
224 : */
225 0 : void jenkins_hashword2 (
226 : const uint32_t *k, /* the key, an array of uint32_t values */
227 : size_t length, /* the length of the key, in uint32_ts */
228 : uint32_t *pc, /* IN: seed OUT: primary hash value */
229 : uint32_t *pb) /* IN: more seed OUT: secondary hash value */
230 : {
231 : uint32_t a,b,c;
232 :
233 : /* Set up the internal state */
234 0 : a = b = c = 0xdeadbeef + ((uint32_t)(length<<2)) + *pc;
235 0 : c += *pb;
236 :
237 : /*------------------------------------------------- handle most of the key */
238 0 : while (length > 3)
239 : {
240 0 : a += k[0];
241 0 : b += k[1];
242 0 : c += k[2];
243 0 : mix(a,b,c);
244 0 : length -= 3;
245 0 : k += 3;
246 : }
247 :
248 : /*------------------------------------------- handle the last 3 uint32_t's */
249 0 : switch(length) /* all the case statements fall through */
250 : {
251 0 : case 3 : c+=k[2];
252 0 : case 2 : b+=k[1];
253 0 : case 1 : a+=k[0];
254 0 : final(a,b,c);
255 0 : case 0: /* case 0: nothing left to add */
256 0 : break;
257 : }
258 : /*------------------------------------------------------ report the result */
259 0 : *pc=c; *pb=b;
260 0 : }
261 :
262 : /*
263 : -------------------------------------------------------------------------------
264 : hashlittle() -- hash a variable-length key into a 32-bit value
265 : k : the key (the unaligned variable-length array of bytes)
266 : length : the length of the key, counting by bytes
267 : initval : can be any 4-byte value
268 : Returns a 32-bit value. Every bit of the key affects every bit of
269 : the return value. Two keys differing by one or two bits will have
270 : totally different hash values.
271 :
272 : The best hash table sizes are powers of 2. There is no need to do
273 : mod a prime (mod is sooo slow!). If you need less than 32 bits,
274 : use a bitmask. For example, if you need only 10 bits, do
275 : h = (h & hashmask(10));
276 : In which case, the hash table should have hashsize(10) elements.
277 :
278 : If you are hashing n strings (uint8_t **)k, do it like this:
279 : for (i=0, h=0; i<n; ++i) h = hashlittle( k[i], len[i], h);
280 :
281 : By Bob Jenkins, 2006. bob_jenkins@burtleburtle.net. You may use this
282 : code any way you wish, private, educational, or commercial. It's free.
283 :
284 : Use for hash table lookup, or anything where one collision in 2^^32 is
285 : acceptable. Do NOT use for cryptographic purposes.
286 : -------------------------------------------------------------------------------
287 : */
288 :
289 0 : uint32_t jenkins_hashlittle( const void *key, size_t length, uint32_t initval)
290 : {
291 : uint32_t a,b,c; /* internal state */
292 : union { const void *ptr; size_t i; } u; /* needed for Mac Powerbook G4 */
293 :
294 : /* Set up the internal state */
295 0 : a = b = c = 0xdeadbeef + ((uint32_t)length) + initval;
296 :
297 0 : u.ptr = key;
298 0 : if (HASH_LITTLE_ENDIAN && ((u.i & 0x3) == 0)) {
299 0 : const uint32_t *k = (const uint32_t *)key; /* read 32-bit chunks */
300 :
301 : /*------ all but last block: aligned reads and affect 32 bits of (a,b,c) */
302 0 : while (length > 12)
303 : {
304 0 : a += k[0];
305 0 : b += k[1];
306 0 : c += k[2];
307 0 : mix(a,b,c);
308 0 : length -= 12;
309 0 : k += 3;
310 : }
311 :
312 : /*----------------------------- handle the last (probably partial) block */
313 : /*
314 : * "k[2]&0xffffff" actually reads beyond the end of the string, but
315 : * then masks off the part it's not allowed to read. Because the
316 : * string is aligned, the masked-off tail is in the same word as the
317 : * rest of the string. Every machine with memory protection I've seen
318 : * does it on word boundaries, so is OK with this. But valgrind will
319 : * still catch it and complain. The masking trick does make the hash
320 : * noticeably faster for short strings (like English words).
321 : */
322 : #if !VALGRIND && !HAS_FEATURE_ADDRESS_SANITIZER && !HAS_FEATURE_MEMORY_SANITIZER
323 :
324 0 : switch(length)
325 : {
326 0 : case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
327 0 : case 11: c+=k[2]&0xffffff; b+=k[1]; a+=k[0]; break;
328 0 : case 10: c+=k[2]&0xffff; b+=k[1]; a+=k[0]; break;
329 0 : case 9 : c+=k[2]&0xff; b+=k[1]; a+=k[0]; break;
330 0 : case 8 : b+=k[1]; a+=k[0]; break;
331 0 : case 7 : b+=k[1]&0xffffff; a+=k[0]; break;
332 0 : case 6 : b+=k[1]&0xffff; a+=k[0]; break;
333 0 : case 5 : b+=k[1]&0xff; a+=k[0]; break;
334 0 : case 4 : a+=k[0]; break;
335 0 : case 3 : a+=k[0]&0xffffff; break;
336 0 : case 2 : a+=k[0]&0xffff; break;
337 0 : case 1 : a+=k[0]&0xff; break;
338 0 : case 0 : return c; /* zero length strings require no mixing */
339 : }
340 :
341 : #else /* make valgrind happy */
342 : {
343 : const uint8_t *k8 = (const uint8_t *) k;
344 :
345 : switch(length)
346 : {
347 : case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
348 : case 11: c+=((uint32_t)k8[10])<<16; /* fall through */
349 : case 10: c+=((uint32_t)k8[9])<<8; /* fall through */
350 : case 9 : c+=k8[8]; /* fall through */
351 : case 8 : b+=k[1]; a+=k[0]; break;
352 : case 7 : b+=((uint32_t)k8[6])<<16; /* fall through */
353 : case 6 : b+=((uint32_t)k8[5])<<8; /* fall through */
354 : case 5 : b+=k8[4]; /* fall through */
355 : case 4 : a+=k[0]; break;
356 : case 3 : a+=((uint32_t)k8[2])<<16; /* fall through */
357 : case 2 : a+=((uint32_t)k8[1])<<8; /* fall through */
358 : case 1 : a+=k8[0]; break;
359 : case 0 : return c;
360 : }
361 : }
362 :
363 : #endif /* !valgrind */
364 :
365 0 : } else if (HASH_LITTLE_ENDIAN && ((u.i & 0x1) == 0)) {
366 0 : const uint16_t *k = (const uint16_t *)key; /* read 16-bit chunks */
367 : const uint8_t *k8;
368 :
369 : /*--------------- all but last block: aligned reads and different mixing */
370 0 : while (length > 12)
371 : {
372 0 : a += k[0] + (((uint32_t)k[1])<<16);
373 0 : b += k[2] + (((uint32_t)k[3])<<16);
374 0 : c += k[4] + (((uint32_t)k[5])<<16);
375 0 : mix(a,b,c);
376 0 : length -= 12;
377 0 : k += 6;
378 : }
379 :
380 : /*----------------------------- handle the last (probably partial) block */
381 0 : k8 = (const uint8_t *)k;
382 0 : switch(length)
383 : {
384 0 : case 12: c+=k[4]+(((uint32_t)k[5])<<16);
385 0 : b+=k[2]+(((uint32_t)k[3])<<16);
386 0 : a+=k[0]+(((uint32_t)k[1])<<16);
387 0 : break;
388 0 : case 11: c+=((uint32_t)k8[10])<<16; /* fall through */
389 0 : case 10: c+=k[4];
390 0 : b+=k[2]+(((uint32_t)k[3])<<16);
391 0 : a+=k[0]+(((uint32_t)k[1])<<16);
392 0 : break;
393 0 : case 9 : c+=k8[8]; /* fall through */
394 0 : case 8 : b+=k[2]+(((uint32_t)k[3])<<16);
395 0 : a+=k[0]+(((uint32_t)k[1])<<16);
396 0 : break;
397 0 : case 7 : b+=((uint32_t)k8[6])<<16; /* fall through */
398 0 : case 6 : b+=k[2];
399 0 : a+=k[0]+(((uint32_t)k[1])<<16);
400 0 : break;
401 0 : case 5 : b+=k8[4]; /* fall through */
402 0 : case 4 : a+=k[0]+(((uint32_t)k[1])<<16);
403 0 : break;
404 0 : case 3 : a+=((uint32_t)k8[2])<<16; /* fall through */
405 0 : case 2 : a+=k[0];
406 0 : break;
407 0 : case 1 : a+=k8[0];
408 0 : break;
409 0 : case 0 : return c; /* zero length requires no mixing */
410 : }
411 :
412 0 : } else { /* need to read the key one byte at a time */
413 0 : const uint8_t *k = (const uint8_t *)key;
414 :
415 : /*--------------- all but the last block: affect some 32 bits of (a,b,c) */
416 0 : while (length > 12)
417 : {
418 0 : a += k[0];
419 0 : a += ((uint32_t)k[1])<<8;
420 0 : a += ((uint32_t)k[2])<<16;
421 0 : a += ((uint32_t)k[3])<<24;
422 0 : b += k[4];
423 0 : b += ((uint32_t)k[5])<<8;
424 0 : b += ((uint32_t)k[6])<<16;
425 0 : b += ((uint32_t)k[7])<<24;
426 0 : c += k[8];
427 0 : c += ((uint32_t)k[9])<<8;
428 0 : c += ((uint32_t)k[10])<<16;
429 0 : c += ((uint32_t)k[11])<<24;
430 0 : mix(a,b,c);
431 0 : length -= 12;
432 0 : k += 12;
433 : }
434 :
435 : /*-------------------------------- last block: affect all 32 bits of (c) */
436 0 : switch(length) /* all the case statements fall through */
437 : {
438 0 : case 12: c+=((uint32_t)k[11])<<24;
439 0 : case 11: c+=((uint32_t)k[10])<<16;
440 0 : case 10: c+=((uint32_t)k[9])<<8;
441 0 : case 9 : c+=k[8];
442 0 : case 8 : b+=((uint32_t)k[7])<<24;
443 0 : case 7 : b+=((uint32_t)k[6])<<16;
444 0 : case 6 : b+=((uint32_t)k[5])<<8;
445 0 : case 5 : b+=k[4];
446 0 : case 4 : a+=((uint32_t)k[3])<<24;
447 0 : case 3 : a+=((uint32_t)k[2])<<16;
448 0 : case 2 : a+=((uint32_t)k[1])<<8;
449 0 : case 1 : a+=k[0];
450 0 : break;
451 0 : case 0 : return c;
452 : }
453 0 : }
454 :
455 0 : final(a,b,c);
456 0 : return c;
457 : }
458 :
459 : /*
460 : * hashlittle2: return 2 32-bit hash values
461 : *
462 : * This is identical to hashlittle(), except it returns two 32-bit hash
463 : * values instead of just one. This is good enough for hash table
464 : * lookup with 2^^64 buckets, or if you want a second hash if you're not
465 : * happy with the first, or if you want a probably-unique 64-bit ID for
466 : * the key. *pc is better mixed than *pb, so use *pc first. If you want
467 : * a 64-bit value do something like "*pc + (((uint64_t)*pb)<<32)".
468 : */
469 260407 : void jenkins_hashlittle2(
470 : const void *key, /* the key to hash */
471 : size_t length, /* length of the key */
472 : uint32_t *pc, /* IN: primary initval, OUT: primary hash */
473 : uint32_t *pb) /* IN: secondary initval, OUT: secondary hash */
474 : {
475 : uint32_t a,b,c; /* internal state */
476 : union { const void *ptr; size_t i; } u; /* needed for Mac Powerbook G4 */
477 :
478 : /* Set up the internal state */
479 260407 : a = b = c = 0xdeadbeef + ((uint32_t)length) + *pc;
480 260407 : c += *pb;
481 :
482 260407 : u.ptr = key;
483 260407 : if (HASH_LITTLE_ENDIAN && ((u.i & 0x3) == 0)) {
484 260384 : const uint32_t *k = (const uint32_t *)key; /* read 32-bit chunks */
485 :
486 : /*------ all but last block: aligned reads and affect 32 bits of (a,b,c) */
487 801676 : while (length > 12)
488 : {
489 541292 : a += k[0];
490 541292 : b += k[1];
491 541292 : c += k[2];
492 541292 : mix(a,b,c);
493 541292 : length -= 12;
494 541292 : k += 3;
495 : }
496 :
497 : /*----------------------------- handle the last (probably partial) block */
498 : /*
499 : * "k[2]&0xffffff" actually reads beyond the end of the string, but
500 : * then masks off the part it's not allowed to read. Because the
501 : * string is aligned, the masked-off tail is in the same word as the
502 : * rest of the string. Every machine with memory protection I've seen
503 : * does it on word boundaries, so is OK with this. But valgrind will
504 : * still catch it and complain. The masking trick does make the hash
505 : * noticeably faster for short strings (like English words).
506 : */
507 : #if !VALGRIND && !HAS_FEATURE_ADDRESS_SANITIZER && !HAS_FEATURE_MEMORY_SANITIZER
508 :
509 260384 : switch(length)
510 : {
511 8966 : case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
512 16155 : case 11: c+=k[2]&0xffffff; b+=k[1]; a+=k[0]; break;
513 21314 : case 10: c+=k[2]&0xffff; b+=k[1]; a+=k[0]; break;
514 24840 : case 9 : c+=k[2]&0xff; b+=k[1]; a+=k[0]; break;
515 23378 : case 8 : b+=k[1]; a+=k[0]; break;
516 41242 : case 7 : b+=k[1]&0xffffff; a+=k[0]; break;
517 27559 : case 6 : b+=k[1]&0xffff; a+=k[0]; break;
518 27838 : case 5 : b+=k[1]&0xff; a+=k[0]; break;
519 19338 : case 4 : a+=k[0]; break;
520 19671 : case 3 : a+=k[0]&0xffffff; break;
521 15652 : case 2 : a+=k[0]&0xffff; break;
522 14431 : case 1 : a+=k[0]&0xff; break;
523 0 : case 0 : *pc=c; *pb=b; return; /* zero length strings require no mixing */
524 : }
525 :
526 : #else /* make valgrind happy */
527 :
528 : {
529 : const uint8_t *k8 = (const uint8_t *)k;
530 : switch(length)
531 : {
532 : case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
533 : case 11: c+=((uint32_t)k8[10])<<16; /* fall through */
534 : case 10: c+=((uint32_t)k8[9])<<8; /* fall through */
535 : case 9 : c+=k8[8]; /* fall through */
536 : case 8 : b+=k[1]; a+=k[0]; break;
537 : case 7 : b+=((uint32_t)k8[6])<<16; /* fall through */
538 : case 6 : b+=((uint32_t)k8[5])<<8; /* fall through */
539 : case 5 : b+=k8[4]; /* fall through */
540 : case 4 : a+=k[0]; break;
541 : case 3 : a+=((uint32_t)k8[2])<<16; /* fall through */
542 : case 2 : a+=((uint32_t)k8[1])<<8; /* fall through */
543 : case 1 : a+=k8[0]; break;
544 : case 0 : *pc=c; *pb=b; return; /* zero length strings require no mixing */
545 : }
546 : }
547 :
548 : #endif /* !valgrind */
549 :
550 260407 : } else if (HASH_LITTLE_ENDIAN && ((u.i & 0x1) == 0)) {
551 5 : const uint16_t *k = (const uint16_t *)key; /* read 16-bit chunks */
552 : const uint8_t *k8;
553 :
554 : /*--------------- all but last block: aligned reads and different mixing */
555 5 : while (length > 12)
556 : {
557 0 : a += k[0] + (((uint32_t)k[1])<<16);
558 0 : b += k[2] + (((uint32_t)k[3])<<16);
559 0 : c += k[4] + (((uint32_t)k[5])<<16);
560 0 : mix(a,b,c);
561 0 : length -= 12;
562 0 : k += 6;
563 : }
564 :
565 : /*----------------------------- handle the last (probably partial) block */
566 5 : k8 = (const uint8_t *)k;
567 5 : switch(length)
568 : {
569 0 : case 12: c+=k[4]+(((uint32_t)k[5])<<16);
570 0 : b+=k[2]+(((uint32_t)k[3])<<16);
571 0 : a+=k[0]+(((uint32_t)k[1])<<16);
572 0 : break;
573 0 : case 11: c+=((uint32_t)k8[10])<<16; /* fall through */
574 1 : case 10: c+=k[4];
575 1 : b+=k[2]+(((uint32_t)k[3])<<16);
576 1 : a+=k[0]+(((uint32_t)k[1])<<16);
577 1 : break;
578 1 : case 9 : c+=k8[8]; /* fall through */
579 1 : case 8 : b+=k[2]+(((uint32_t)k[3])<<16);
580 1 : a+=k[0]+(((uint32_t)k[1])<<16);
581 1 : break;
582 0 : case 7 : b+=((uint32_t)k8[6])<<16; /* fall through */
583 1 : case 6 : b+=k[2];
584 1 : a+=k[0]+(((uint32_t)k[1])<<16);
585 1 : break;
586 1 : case 5 : b+=k8[4]; /* fall through */
587 2 : case 4 : a+=k[0]+(((uint32_t)k[1])<<16);
588 2 : break;
589 0 : case 3 : a+=((uint32_t)k8[2])<<16; /* fall through */
590 0 : case 2 : a+=k[0];
591 0 : break;
592 0 : case 1 : a+=k8[0];
593 0 : break;
594 0 : case 0 : *pc=c; *pb=b; return; /* zero length strings require no mixing */
595 : }
596 :
597 5 : } else { /* need to read the key one byte at a time */
598 18 : const uint8_t *k = (const uint8_t *)key;
599 :
600 : /*--------------- all but the last block: affect some 32 bits of (a,b,c) */
601 18 : while (length > 12)
602 : {
603 0 : a += k[0];
604 0 : a += ((uint32_t)k[1])<<8;
605 0 : a += ((uint32_t)k[2])<<16;
606 0 : a += ((uint32_t)k[3])<<24;
607 0 : b += k[4];
608 0 : b += ((uint32_t)k[5])<<8;
609 0 : b += ((uint32_t)k[6])<<16;
610 0 : b += ((uint32_t)k[7])<<24;
611 0 : c += k[8];
612 0 : c += ((uint32_t)k[9])<<8;
613 0 : c += ((uint32_t)k[10])<<16;
614 0 : c += ((uint32_t)k[11])<<24;
615 0 : mix(a,b,c);
616 0 : length -= 12;
617 0 : k += 12;
618 : }
619 :
620 : /*-------------------------------- last block: affect all 32 bits of (c) */
621 18 : switch(length) /* all the case statements fall through */
622 : {
623 0 : case 12: c+=((uint32_t)k[11])<<24;
624 1 : case 11: c+=((uint32_t)k[10])<<16;
625 5 : case 10: c+=((uint32_t)k[9])<<8;
626 9 : case 9 : c+=k[8];
627 11 : case 8 : b+=((uint32_t)k[7])<<24;
628 14 : case 7 : b+=((uint32_t)k[6])<<16;
629 16 : case 6 : b+=((uint32_t)k[5])<<8;
630 17 : case 5 : b+=k[4];
631 18 : case 4 : a+=((uint32_t)k[3])<<24;
632 18 : case 3 : a+=((uint32_t)k[2])<<16;
633 18 : case 2 : a+=((uint32_t)k[1])<<8;
634 18 : case 1 : a+=k[0];
635 18 : break;
636 0 : case 0 : *pc=c; *pb=b; return; /* zero length strings require no mixing */
637 : }
638 260407 : }
639 :
640 260407 : final(a,b,c);
641 260407 : *pc=c; *pb=b;
642 : }
643 :
644 : /*
645 : * hashbig():
646 : * This is the same as hashword() on big-endian machines. It is different
647 : * from hashlittle() on all machines. hashbig() takes advantage of
648 : * big-endian byte ordering.
649 : */
650 0 : uint32_t jenkins_hashbig( const void *key, size_t length, uint32_t initval)
651 : {
652 : uint32_t a,b,c;
653 : union { const void *ptr; size_t i; } u; /* to cast key to (size_t) happily */
654 :
655 : /* Set up the internal state */
656 0 : a = b = c = 0xdeadbeef + ((uint32_t)length) + initval;
657 :
658 0 : u.ptr = key;
659 : if (HASH_BIG_ENDIAN && ((u.i & 0x3) == 0)) {
660 : const uint32_t *k = (const uint32_t *)key; /* read 32-bit chunks */
661 :
662 : /*------ all but last block: aligned reads and affect 32 bits of (a,b,c) */
663 : while (length > 12)
664 : {
665 : a += k[0];
666 : b += k[1];
667 : c += k[2];
668 : mix(a,b,c);
669 : length -= 12;
670 : k += 3;
671 : }
672 :
673 : /*----------------------------- handle the last (probably partial) block */
674 : /*
675 : * "k[2]<<8" actually reads beyond the end of the string, but
676 : * then shifts out the part it's not allowed to read. Because the
677 : * string is aligned, the illegal read is in the same word as the
678 : * rest of the string. Every machine with memory protection I've seen
679 : * does it on word boundaries, so is OK with this. But valgrind will
680 : * still catch it and complain. The masking trick does make the hash
681 : * noticeably faster for short strings (like English words).
682 : */
683 : #if !VALGRIND && !HAS_FEATURE_ADDRESS_SANITIZER && !HAS_FEATURE_MEMORY_SANITIZER
684 :
685 : switch(length)
686 : {
687 : case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
688 : case 11: c+=k[2]&0xffffff00; b+=k[1]; a+=k[0]; break;
689 : case 10: c+=k[2]&0xffff0000; b+=k[1]; a+=k[0]; break;
690 : case 9 : c+=k[2]&0xff000000; b+=k[1]; a+=k[0]; break;
691 : case 8 : b+=k[1]; a+=k[0]; break;
692 : case 7 : b+=k[1]&0xffffff00; a+=k[0]; break;
693 : case 6 : b+=k[1]&0xffff0000; a+=k[0]; break;
694 : case 5 : b+=k[1]&0xff000000; a+=k[0]; break;
695 : case 4 : a+=k[0]; break;
696 : case 3 : a+=k[0]&0xffffff00; break;
697 : case 2 : a+=k[0]&0xffff0000; break;
698 : case 1 : a+=k[0]&0xff000000; break;
699 : case 0 : return c; /* zero length strings require no mixing */
700 : }
701 :
702 : #else /* make valgrind happy */
703 :
704 : {
705 : const uint8_t *k8 = (const uint8_t *)k;
706 : switch(length) /* all the case statements fall through */
707 : {
708 : case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
709 : case 11: c+=((uint32_t)k8[10])<<8; /* fall through */
710 : case 10: c+=((uint32_t)k8[9])<<16; /* fall through */
711 : case 9 : c+=((uint32_t)k8[8])<<24; /* fall through */
712 : case 8 : b+=k[1]; a+=k[0]; break;
713 : case 7 : b+=((uint32_t)k8[6])<<8; /* fall through */
714 : case 6 : b+=((uint32_t)k8[5])<<16; /* fall through */
715 : case 5 : b+=((uint32_t)k8[4])<<24; /* fall through */
716 : case 4 : a+=k[0]; break;
717 : case 3 : a+=((uint32_t)k8[2])<<8; /* fall through */
718 : case 2 : a+=((uint32_t)k8[1])<<16; /* fall through */
719 : case 1 : a+=((uint32_t)k8[0])<<24; break;
720 : case 0 : return c;
721 : }
722 : }
723 :
724 : #endif /* !VALGRIND */
725 :
726 : } else { /* need to read the key one byte at a time */
727 0 : const uint8_t *k = (const uint8_t *)key;
728 :
729 : /*--------------- all but the last block: affect some 32 bits of (a,b,c) */
730 0 : while (length > 12)
731 : {
732 0 : a += ((uint32_t)k[0])<<24;
733 0 : a += ((uint32_t)k[1])<<16;
734 0 : a += ((uint32_t)k[2])<<8;
735 0 : a += ((uint32_t)k[3]);
736 0 : b += ((uint32_t)k[4])<<24;
737 0 : b += ((uint32_t)k[5])<<16;
738 0 : b += ((uint32_t)k[6])<<8;
739 0 : b += ((uint32_t)k[7]);
740 0 : c += ((uint32_t)k[8])<<24;
741 0 : c += ((uint32_t)k[9])<<16;
742 0 : c += ((uint32_t)k[10])<<8;
743 0 : c += ((uint32_t)k[11]);
744 0 : mix(a,b,c);
745 0 : length -= 12;
746 0 : k += 12;
747 : }
748 :
749 : /*-------------------------------- last block: affect all 32 bits of (c) */
750 0 : switch(length) /* all the case statements fall through */
751 : {
752 0 : case 12: c+=k[11];
753 0 : case 11: c+=((uint32_t)k[10])<<8;
754 0 : case 10: c+=((uint32_t)k[9])<<16;
755 0 : case 9 : c+=((uint32_t)k[8])<<24;
756 0 : case 8 : b+=k[7];
757 0 : case 7 : b+=((uint32_t)k[6])<<8;
758 0 : case 6 : b+=((uint32_t)k[5])<<16;
759 0 : case 5 : b+=((uint32_t)k[4])<<24;
760 0 : case 4 : a+=k[3];
761 0 : case 3 : a+=((uint32_t)k[2])<<8;
762 0 : case 2 : a+=((uint32_t)k[1])<<16;
763 0 : case 1 : a+=((uint32_t)k[0])<<24;
764 0 : break;
765 0 : case 0 : return c;
766 : }
767 0 : }
768 :
769 0 : final(a,b,c);
770 0 : return c;
771 : }
772 :
773 : #ifdef SELF_TEST
774 :
775 : /* used for timings */
776 : void driver1()
777 : {
778 : uint8_t buf[256];
779 : uint32_t i;
780 : uint32_t h=0;
781 : time_t a,z;
782 :
783 : time(&a);
784 : for (i=0; i<256; ++i) buf[i] = 'x';
785 : for (i=0; i<1; ++i)
786 : {
787 : h = hashlittle(&buf[0],1,h);
788 : }
789 : time(&z);
790 : if (z-a > 0) printf("time %d %.8x\n", z-a, h);
791 : }
792 :
793 : /* check that every input bit changes every output bit half the time */
794 : #define HASHSTATE 1
795 : #define HASHLEN 1
796 : #define MAXPAIR 60
797 : #define MAXLEN 70
798 : void driver2()
799 : {
800 : uint8_t qa[MAXLEN+1], qb[MAXLEN+2], *a = &qa[0], *b = &qb[1];
801 : uint32_t c[HASHSTATE], d[HASHSTATE], i=0, j=0, k, l, m=0, z;
802 : uint32_t e[HASHSTATE],f[HASHSTATE],g[HASHSTATE],h[HASHSTATE];
803 : uint32_t x[HASHSTATE],y[HASHSTATE];
804 : uint32_t hlen;
805 :
806 : printf("No more than %d trials should ever be needed \n",MAXPAIR/2);
807 : for (hlen=0; hlen < MAXLEN; ++hlen)
808 : {
809 : z=0;
810 : for (i=0; i<hlen; ++i) /*----------------------- for each input byte, */
811 : {
812 : for (j=0; j<8; ++j) /*------------------------ for each input bit, */
813 : {
814 : for (m=1; m<8; ++m) /*------------- for several possible initvals, */
815 : {
816 : for (l=0; l<HASHSTATE; ++l)
817 : e[l]=f[l]=g[l]=h[l]=x[l]=y[l]=~((uint32_t)0);
818 :
819 : /*---- check that every output bit is affected by that input bit */
820 : for (k=0; k<MAXPAIR; k+=2)
821 : {
822 : uint32_t finished=1;
823 : /* keys have one bit different */
824 : for (l=0; l<hlen+1; ++l) {a[l] = b[l] = (uint8_t)0;}
825 : /* have a and b be two keys differing in only one bit */
826 : a[i] ^= (k<<j);
827 : a[i] ^= (k>>(8-j));
828 : c[0] = hashlittle(a, hlen, m);
829 : b[i] ^= ((k+1)<<j);
830 : b[i] ^= ((k+1)>>(8-j));
831 : d[0] = hashlittle(b, hlen, m);
832 : /* check every bit is 1, 0, set, and not set at least once */
833 : for (l=0; l<HASHSTATE; ++l)
834 : {
835 : e[l] &= (c[l]^d[l]);
836 : f[l] &= ~(c[l]^d[l]);
837 : g[l] &= c[l];
838 : h[l] &= ~c[l];
839 : x[l] &= d[l];
840 : y[l] &= ~d[l];
841 : if (e[l]|f[l]|g[l]|h[l]|x[l]|y[l]) finished=0;
842 : }
843 : if (finished) break;
844 : }
845 : if (k>z) z=k;
846 : if (k==MAXPAIR)
847 : {
848 : printf("Some bit didn't change: ");
849 : printf("%.8x %.8x %.8x %.8x %.8x %.8x ",
850 : e[0],f[0],g[0],h[0],x[0],y[0]);
851 : printf("i %d j %d m %d len %d\n", i, j, m, hlen);
852 : }
853 : if (z==MAXPAIR) goto done;
854 : }
855 : }
856 : }
857 : done:
858 : if (z < MAXPAIR)
859 : {
860 : printf("Mix success %2d bytes %2d initvals ",i,m);
861 : printf("required %d trials\n", z/2);
862 : }
863 : }
864 : printf("\n");
865 : }
866 :
867 : /* Check for reading beyond the end of the buffer and alignment problems */
868 : void driver3()
869 : {
870 : uint8_t buf[MAXLEN+20], *b;
871 : uint32_t len;
872 : uint8_t q[] = "This is the time for all good men to come to the aid of their country...";
873 : uint32_t h;
874 : uint8_t qq[] = "xThis is the time for all good men to come to the aid of their country...";
875 : uint32_t i;
876 : uint8_t qqq[] = "xxThis is the time for all good men to come to the aid of their country...";
877 : uint32_t j;
878 : uint8_t qqqq[] = "xxxThis is the time for all good men to come to the aid of their country...";
879 : uint32_t ref,x,y;
880 : uint8_t *p;
881 :
882 : printf("Endianness. These lines should all be the same (for values filled in):\n");
883 : printf("%.8x %.8x %.8x\n",
884 : hashword((const uint32_t *)q, (sizeof(q)-1)/4, 13),
885 : hashword((const uint32_t *)q, (sizeof(q)-5)/4, 13),
886 : hashword((const uint32_t *)q, (sizeof(q)-9)/4, 13));
887 : p = q;
888 : printf("%.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x\n",
889 : hashlittle(p, sizeof(q)-1, 13), hashlittle(p, sizeof(q)-2, 13),
890 : hashlittle(p, sizeof(q)-3, 13), hashlittle(p, sizeof(q)-4, 13),
891 : hashlittle(p, sizeof(q)-5, 13), hashlittle(p, sizeof(q)-6, 13),
892 : hashlittle(p, sizeof(q)-7, 13), hashlittle(p, sizeof(q)-8, 13),
893 : hashlittle(p, sizeof(q)-9, 13), hashlittle(p, sizeof(q)-10, 13),
894 : hashlittle(p, sizeof(q)-11, 13), hashlittle(p, sizeof(q)-12, 13));
895 : p = &qq[1];
896 : printf("%.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x\n",
897 : hashlittle(p, sizeof(q)-1, 13), hashlittle(p, sizeof(q)-2, 13),
898 : hashlittle(p, sizeof(q)-3, 13), hashlittle(p, sizeof(q)-4, 13),
899 : hashlittle(p, sizeof(q)-5, 13), hashlittle(p, sizeof(q)-6, 13),
900 : hashlittle(p, sizeof(q)-7, 13), hashlittle(p, sizeof(q)-8, 13),
901 : hashlittle(p, sizeof(q)-9, 13), hashlittle(p, sizeof(q)-10, 13),
902 : hashlittle(p, sizeof(q)-11, 13), hashlittle(p, sizeof(q)-12, 13));
903 : p = &qqq[2];
904 : printf("%.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x\n",
905 : hashlittle(p, sizeof(q)-1, 13), hashlittle(p, sizeof(q)-2, 13),
906 : hashlittle(p, sizeof(q)-3, 13), hashlittle(p, sizeof(q)-4, 13),
907 : hashlittle(p, sizeof(q)-5, 13), hashlittle(p, sizeof(q)-6, 13),
908 : hashlittle(p, sizeof(q)-7, 13), hashlittle(p, sizeof(q)-8, 13),
909 : hashlittle(p, sizeof(q)-9, 13), hashlittle(p, sizeof(q)-10, 13),
910 : hashlittle(p, sizeof(q)-11, 13), hashlittle(p, sizeof(q)-12, 13));
911 : p = &qqqq[3];
912 : printf("%.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x\n",
913 : hashlittle(p, sizeof(q)-1, 13), hashlittle(p, sizeof(q)-2, 13),
914 : hashlittle(p, sizeof(q)-3, 13), hashlittle(p, sizeof(q)-4, 13),
915 : hashlittle(p, sizeof(q)-5, 13), hashlittle(p, sizeof(q)-6, 13),
916 : hashlittle(p, sizeof(q)-7, 13), hashlittle(p, sizeof(q)-8, 13),
917 : hashlittle(p, sizeof(q)-9, 13), hashlittle(p, sizeof(q)-10, 13),
918 : hashlittle(p, sizeof(q)-11, 13), hashlittle(p, sizeof(q)-12, 13));
919 : printf("\n");
920 :
921 : /* check that hashlittle2 and hashlittle produce the same results */
922 : i=47; j=0;
923 : hashlittle2(q, sizeof(q), &i, &j);
924 : if (hashlittle(q, sizeof(q), 47) != i)
925 : printf("hashlittle2 and hashlittle mismatch\n");
926 :
927 : /* check that hashword2 and hashword produce the same results */
928 : len = 0xdeadbeef;
929 : i=47, j=0;
930 : hashword2(&len, 1, &i, &j);
931 : if (hashword(&len, 1, 47) != i)
932 : printf("hashword2 and hashword mismatch %x %x\n",
933 : i, hashword(&len, 1, 47));
934 :
935 : /* check hashlittle doesn't read before or after the ends of the string */
936 : for (h=0, b=buf+1; h<8; ++h, ++b)
937 : {
938 : for (i=0; i<MAXLEN; ++i)
939 : {
940 : len = i;
941 : for (j=0; j<i; ++j) *(b+j)=0;
942 :
943 : /* these should all be equal */
944 : ref = hashlittle(b, len, (uint32_t)1);
945 : *(b+i)=(uint8_t)~0;
946 : *(b-1)=(uint8_t)~0;
947 : x = hashlittle(b, len, (uint32_t)1);
948 : y = hashlittle(b, len, (uint32_t)1);
949 : if ((ref != x) || (ref != y))
950 : {
951 : printf("alignment error: %.8x %.8x %.8x %d %d\n",ref,x,y,
952 : h, i);
953 : }
954 : }
955 : }
956 : }
957 :
958 : /* check for problems with nulls */
959 : void driver4()
960 : {
961 : uint8_t buf[1];
962 : uint32_t h,i,state[HASHSTATE];
963 :
964 : buf[0] = ~0;
965 : for (i=0; i<HASHSTATE; ++i) state[i] = 1;
966 : printf("These should all be different\n");
967 : for (i=0, h=0; i<8; ++i)
968 : {
969 : h = hashlittle(buf, 0, h);
970 : printf("%2ld 0-byte strings, hash is %.8x\n", i, h);
971 : }
972 : }
973 :
974 : void driver5()
975 : {
976 : uint32_t b,c;
977 : b=0, c=0, hashlittle2("", 0, &c, &b);
978 : printf("hash is %.8lx %.8lx\n", c, b); /* deadbeef deadbeef */
979 : b=0xdeadbeef, c=0, hashlittle2("", 0, &c, &b);
980 : printf("hash is %.8lx %.8lx\n", c, b); /* bd5b7dde deadbeef */
981 : b=0xdeadbeef, c=0xdeadbeef, hashlittle2("", 0, &c, &b);
982 : printf("hash is %.8lx %.8lx\n", c, b); /* 9c093ccd bd5b7dde */
983 : b=0, c=0, hashlittle2("Four score and seven years ago", 30, &c, &b);
984 : printf("hash is %.8lx %.8lx\n", c, b); /* 17770551 ce7226e6 */
985 : b=1, c=0, hashlittle2("Four score and seven years ago", 30, &c, &b);
986 : printf("hash is %.8lx %.8lx\n", c, b); /* e3607cae bd371de4 */
987 : b=0, c=1, hashlittle2("Four score and seven years ago", 30, &c, &b);
988 : printf("hash is %.8lx %.8lx\n", c, b); /* cd628161 6cbea4b3 */
989 : c = hashlittle("Four score and seven years ago", 30, 0);
990 : printf("hash is %.8lx\n", c); /* 17770551 */
991 : c = hashlittle("Four score and seven years ago", 30, 1);
992 : printf("hash is %.8lx\n", c); /* cd628161 */
993 : }
994 :
995 : int main()
996 : {
997 : driver1(); /* test that the key is hashed: used for timings */
998 : driver2(); /* test that whole key is hashed thoroughly */
999 : driver3(); /* test that nothing but the key is hashed */
1000 : driver4(); /* test hashing multiple buffers (all buffers are null) */
1001 : driver5(); /* test the hash against known vectors */
1002 : return 1;
1003 : }
1004 :
1005 : #endif /* SELF_TEST */
|
