2 -------------------------------------------------------------------------------
3 lookup3.c, by Bob Jenkins, May 2006, Public Domain.
5 These are functions for producing 32-bit hashes for hash table lookup.
6 hashword(), hashlittle(), hashlittle2(), hashbig(), mix(), and final()
7 are externally useful functions. Routines to test the hash are included
8 if SELF_TEST is defined. You can use this free for any purpose. It's in
9 the public domain. It has no warranty.
11 You probably want to use hashlittle(). hashlittle() and hashbig()
12 hash byte arrays. hashlittle() is is faster than hashbig() on
13 little-endian machines. Intel and AMD are little-endian machines.
14 On second thought, you probably want hashlittle2(), which is identical to
15 hashlittle() except it returns two 32-bit hashes for the price of one.
16 You could implement hashbig2() if you wanted but I haven't bothered here.
18 If you want to find a hash of, say, exactly 7 integers, do
19 a = i1; b = i2; c = i3;
21 a += i4; b += i5; c += i6;
25 then use c as the hash value. If you have a variable length array of
26 4-byte integers to hash, use hashword(). If you have a byte array (like
27 a character string), use hashlittle(). If you have several byte arrays, or
28 a mix of things, see the comments above hashlittle().
30 Why is this so big? I read 12 bytes at a time into 3 4-byte integers,
31 then mix those integers. This is fast (you can do a lot more thorough
32 mixing with 12*3 instructions on 3 integers than you can with 3 instructions
33 on 1 byte), but shoehorning those bytes into integers efficiently is messy.
34 -------------------------------------------------------------------------------
38 #include <stdio.h> /* defines printf for tests */
39 #include <time.h> /* defines time_t for timings in the test */
40 #include <stdint.h> /* defines uint32_t etc */
41 #include <sys/param.h> /* attempt to define endianness */
43 # include <endian.h> /* attempt to define endianness */
47 * My best guess at if you are big-endian or little-endian. This may
50 #if (defined(__BYTE_ORDER) && defined(__LITTLE_ENDIAN) && \
51 __BYTE_ORDER == __LITTLE_ENDIAN) || \
52 (defined(i386) || defined(__i386__) || defined(__i486__) || \
53 defined(__i586__) || defined(__i686__) || defined(vax) || defined(MIPSEL))
54 # define HASH_LITTLE_ENDIAN 1
55 # define HASH_BIG_ENDIAN 0
56 #elif (defined(__BYTE_ORDER) && defined(__BIG_ENDIAN) && \
57 __BYTE_ORDER == __BIG_ENDIAN) || \
58 (defined(sparc) || defined(POWERPC) || defined(mc68000) || defined(sel))
59 # define HASH_LITTLE_ENDIAN 0
60 # define HASH_BIG_ENDIAN 1
62 # define HASH_LITTLE_ENDIAN 0
63 # define HASH_BIG_ENDIAN 0
66 #define hashsize(n) ((uint32_t)1<<(n))
67 #define hashmask(n) (hashsize(n)-1)
68 #define rot(x,k) (((x)<<(k)) | ((x)>>(32-(k))))
71 -------------------------------------------------------------------------------
72 mix -- mix 3 32-bit values reversibly.
74 This is reversible, so any information in (a,b,c) before mix() is
75 still in (a,b,c) after mix().
77 If four pairs of (a,b,c) inputs are run through mix(), or through
78 mix() in reverse, there are at least 32 bits of the output that
79 are sometimes the same for one pair and different for another pair.
81 * pairs that differed by one bit, by two bits, in any combination
82 of top bits of (a,b,c), or in any combination of bottom bits of
84 * "differ" is defined as +, -, ^, or ~^. For + and -, I transformed
85 the output delta to a Gray code (a^(a>>1)) so a string of 1's (as
86 is commonly produced by subtraction) look like a single 1-bit
88 * the base values were pseudorandom, all zero but one bit set, or
89 all zero plus a counter that starts at zero.
91 Some k values for my "a-=c; a^=rot(c,k); c+=b;" arrangement that
96 Well, "9 15 3 18 27 15" didn't quite get 32 bits diffing
97 for "differ" defined as + with a one-bit base and a two-bit delta. I
98 used http://burtleburtle.net/bob/hash/avalanche.html to choose
99 the operations, constants, and arrangements of the variables.
101 This does not achieve avalanche. There are input bits of (a,b,c)
102 that fail to affect some output bits of (a,b,c), especially of a. The
103 most thoroughly mixed value is c, but it doesn't really even achieve
106 This allows some parallelism. Read-after-writes are good at doubling
107 the number of bits affected, so the goal of mixing pulls in the opposite
108 direction as the goal of parallelism. I did what I could. Rotates
109 seem to cost as much as shifts on every machine I could lay my hands
110 on, and rotates are much kinder to the top and bottom bits, so I used
112 -------------------------------------------------------------------------------
116 a -= c; a ^= rot(c, 4); c += b; \
117 b -= a; b ^= rot(a, 6); a += c; \
118 c -= b; c ^= rot(b, 8); b += a; \
119 a -= c; a ^= rot(c,16); c += b; \
120 b -= a; b ^= rot(a,19); a += c; \
121 c -= b; c ^= rot(b, 4); b += a; \
125 -------------------------------------------------------------------------------
126 final -- final mixing of 3 32-bit values (a,b,c) into c
128 Pairs of (a,b,c) values differing in only a few bits will usually
129 produce values of c that look totally different. This was tested for
130 * pairs that differed by one bit, by two bits, in any combination
131 of top bits of (a,b,c), or in any combination of bottom bits of
133 * "differ" is defined as +, -, ^, or ~^. For + and -, I transformed
134 the output delta to a Gray code (a^(a>>1)) so a string of 1's (as
135 is commonly produced by subtraction) look like a single 1-bit
137 * the base values were pseudorandom, all zero but one bit set, or
138 all zero plus a counter that starts at zero.
140 These constants passed:
143 and these came close:
147 -------------------------------------------------------------------------------
149 #define final(a,b,c) \
151 c ^= b; c -= rot(b,14); \
152 a ^= c; a -= rot(c,11); \
153 b ^= a; b -= rot(a,25); \
154 c ^= b; c -= rot(b,16); \
155 a ^= c; a -= rot(c,4); \
156 b ^= a; b -= rot(a,14); \
157 c ^= b; c -= rot(b,24); \
161 --------------------------------------------------------------------
162 This works on all machines. To be useful, it requires
163 -- that the key be an array of uint32_t's, and
164 -- that the length be the number of uint32_t's in the key
166 The function hashword() is identical to hashlittle() on little-endian
167 machines, and identical to hashbig() on big-endian machines,
168 except that the length has to be measured in uint32_ts rather than in
169 bytes. hashlittle() is more complicated than hashword() only because
170 hashlittle() has to dance around fitting the key bytes into registers.
171 --------------------------------------------------------------------
173 #if 0 // libcitadel doesn't use this.
175 const uint32_t *k, /* the key, an array of uint32_t values */
176 size_t length, /* the length of the key, in uint32_ts */
177 uint32_t initval) /* the previous hash, or an arbitrary value */
181 /* Set up the internal state */
182 a = b = c = 0xdeadbeef + (((uint32_t)length)<<2) + initval;
184 /*------------------------------------------------- handle most of the key */
195 /*------------------------------------------- handle the last 3 uint32_t's */
196 switch(length) /* all the case statements fall through */
202 case 0: /* case 0: nothing left to add */
205 /*------------------------------------------------------ report the result */
211 --------------------------------------------------------------------
212 hashword2() -- same as hashword(), but take two seeds and return two
213 32-bit values. pc and pb must both be nonnull, and *pc and *pb must
214 both be initialized with seeds. If you pass in (*pb)==0, the output
215 (*pc) will be the same as the return value from hashword().
216 --------------------------------------------------------------------
218 #if 0 // libcitadel doesn't use this.
220 const uint32_t *k, /* the key, an array of uint32_t values */
221 size_t length, /* the length of the key, in uint32_ts */
222 uint32_t *pc, /* IN: seed OUT: primary hash value */
223 uint32_t *pb) /* IN: more seed OUT: secondary hash value */
227 /* Set up the internal state */
228 a = b = c = 0xdeadbeef + ((uint32_t)(length<<2)) + *pc;
231 /*------------------------------------------------- handle most of the key */
242 /*------------------------------------------- handle the last 3 uint32_t's */
243 switch(length) /* all the case statements fall through */
249 case 0: /* case 0: nothing left to add */
252 /*------------------------------------------------------ report the result */
258 -------------------------------------------------------------------------------
259 hashlittle() -- hash a variable-length key into a 32-bit value
260 k : the key (the unaligned variable-length array of bytes)
261 length : the length of the key, counting by bytes
262 initval : can be any 4-byte value
263 Returns a 32-bit value. Every bit of the key affects every bit of
264 the return value. Two keys differing by one or two bits will have
265 totally different hash values.
267 The best hash table sizes are powers of 2. There is no need to do
268 mod a prime (mod is sooo slow!). If you need less than 32 bits,
269 use a bitmask. For example, if you need only 10 bits, do
270 h = (h & hashmask(10));
271 In which case, the hash table should have hashsize(10) elements.
273 If you are hashing n strings (uint8_t **)k, do it like this:
274 for (i=0, h=0; i<n; ++i) h = hashlittle( k[i], len[i], h);
276 By Bob Jenkins, 2006. bob_jenkins@burtleburtle.net. You may use this
277 code any way you wish, private, educational, or commercial. It's free.
279 Use for hash table lookup, or anything where one collision in 2^^32 is
280 acceptable. Do NOT use for cryptographic purposes.
281 -------------------------------------------------------------------------------
284 uint32_t hashlittle( const void *key, size_t length, uint32_t initval)
286 uint32_t a,b,c; /* internal state */
287 union { const void *ptr; size_t i; } u; /* needed for Mac Powerbook G4 */
289 /* Set up the internal state */
290 a = b = c = 0xdeadbeef + ((uint32_t)length) + initval;
293 if (HASH_LITTLE_ENDIAN && ((u.i & 0x3) == 0)) {
294 const uint32_t *k = (const uint32_t *)key; /* read 32-bit chunks */
298 /*------ all but last block: aligned reads and affect 32 bits of (a,b,c) */
309 /*----------------------------- handle the last (probably partial) block */
311 * "k[2]&0xffffff" actually reads beyond the end of the string, but
312 * then masks off the part it's not allowed to read. Because the
313 * string is aligned, the masked-off tail is in the same word as the
314 * rest of the string. Every machine with memory protection I've seen
315 * does it on word boundaries, so is OK with this. But VALGRIND will
316 * still catch it and complain. The masking trick does make the hash
317 * noticably faster for short strings (like English words).
323 case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
324 case 11: c+=k[2]&0xffffff; b+=k[1]; a+=k[0]; break;
325 case 10: c+=k[2]&0xffff; b+=k[1]; a+=k[0]; break;
326 case 9 : c+=k[2]&0xff; b+=k[1]; a+=k[0]; break;
327 case 8 : b+=k[1]; a+=k[0]; break;
328 case 7 : b+=k[1]&0xffffff; a+=k[0]; break;
329 case 6 : b+=k[1]&0xffff; a+=k[0]; break;
330 case 5 : b+=k[1]&0xff; a+=k[0]; break;
331 case 4 : a+=k[0]; break;
332 case 3 : a+=k[0]&0xffffff; break;
333 case 2 : a+=k[0]&0xffff; break;
334 case 1 : a+=k[0]&0xff; break;
335 case 0 : return c; /* zero length strings require no mixing */
338 #else /* make valgrind happy */
340 k8 = (const uint8_t *)k;
343 case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
344 case 11: c+=((uint32_t)k8[10])<<16; /* fall through */
345 case 10: c+=((uint32_t)k8[9])<<8; /* fall through */
346 case 9 : c+=k8[8]; /* fall through */
347 case 8 : b+=k[1]; a+=k[0]; break;
348 case 7 : b+=((uint32_t)k8[6])<<16; /* fall through */
349 case 6 : b+=((uint32_t)k8[5])<<8; /* fall through */
350 case 5 : b+=k8[4]; /* fall through */
351 case 4 : a+=k[0]; break;
352 case 3 : a+=((uint32_t)k8[2])<<16; /* fall through */
353 case 2 : a+=((uint32_t)k8[1])<<8; /* fall through */
354 case 1 : a+=k8[0]; break;
358 #endif /* !valgrind */
360 } else if (HASH_LITTLE_ENDIAN && ((u.i & 0x1) == 0)) {
361 const uint16_t *k = (const uint16_t *)key; /* read 16-bit chunks */
364 /*--------------- all but last block: aligned reads and different mixing */
367 a += k[0] + (((uint32_t)k[1])<<16);
368 b += k[2] + (((uint32_t)k[3])<<16);
369 c += k[4] + (((uint32_t)k[5])<<16);
375 /*----------------------------- handle the last (probably partial) block */
376 k8 = (const uint8_t *)k;
379 case 12: c+=k[4]+(((uint32_t)k[5])<<16);
380 b+=k[2]+(((uint32_t)k[3])<<16);
381 a+=k[0]+(((uint32_t)k[1])<<16);
383 case 11: c+=((uint32_t)k8[10])<<16; /* fall through */
385 b+=k[2]+(((uint32_t)k[3])<<16);
386 a+=k[0]+(((uint32_t)k[1])<<16);
388 case 9 : c+=k8[8]; /* fall through */
389 case 8 : b+=k[2]+(((uint32_t)k[3])<<16);
390 a+=k[0]+(((uint32_t)k[1])<<16);
392 case 7 : b+=((uint32_t)k8[6])<<16; /* fall through */
394 a+=k[0]+(((uint32_t)k[1])<<16);
396 case 5 : b+=k8[4]; /* fall through */
397 case 4 : a+=k[0]+(((uint32_t)k[1])<<16);
399 case 3 : a+=((uint32_t)k8[2])<<16; /* fall through */
404 case 0 : return c; /* zero length requires no mixing */
407 } else { /* need to read the key one byte at a time */
408 const uint8_t *k = (const uint8_t *)key;
410 /*--------------- all but the last block: affect some 32 bits of (a,b,c) */
414 a += ((uint32_t)k[1])<<8;
415 a += ((uint32_t)k[2])<<16;
416 a += ((uint32_t)k[3])<<24;
418 b += ((uint32_t)k[5])<<8;
419 b += ((uint32_t)k[6])<<16;
420 b += ((uint32_t)k[7])<<24;
422 c += ((uint32_t)k[9])<<8;
423 c += ((uint32_t)k[10])<<16;
424 c += ((uint32_t)k[11])<<24;
430 /*-------------------------------- last block: affect all 32 bits of (c) */
431 switch(length) /* all the case statements fall through */
433 case 12: c+=((uint32_t)k[11])<<24;
434 case 11: c+=((uint32_t)k[10])<<16;
435 case 10: c+=((uint32_t)k[9])<<8;
437 case 8 : b+=((uint32_t)k[7])<<24;
438 case 7 : b+=((uint32_t)k[6])<<16;
439 case 6 : b+=((uint32_t)k[5])<<8;
441 case 4 : a+=((uint32_t)k[3])<<24;
442 case 3 : a+=((uint32_t)k[2])<<16;
443 case 2 : a+=((uint32_t)k[1])<<8;
456 * hashlittle2: return 2 32-bit hash values
458 * This is identical to hashlittle(), except it returns two 32-bit hash
459 * values instead of just one. This is good enough for hash table
460 * lookup with 2^^64 buckets, or if you want a second hash if you're not
461 * happy with the first, or if you want a probably-unique 64-bit ID for
462 * the key. *pc is better mixed than *pb, so use *pc first. If you want
463 * a 64-bit value do something like "*pc + (((uint64_t)*pb)<<32)".
465 #if 0 // libcitadel doesn't use this.
467 const void *key, /* the key to hash */
468 size_t length, /* length of the key */
469 uint32_t *pc, /* IN: primary initval, OUT: primary hash */
470 uint32_t *pb) /* IN: secondary initval, OUT: secondary hash */
472 uint32_t a,b,c; /* internal state */
473 union { const void *ptr; size_t i; } u; /* needed for Mac Powerbook G4 */
475 /* Set up the internal state */
476 a = b = c = 0xdeadbeef + ((uint32_t)length) + *pc;
480 if (HASH_LITTLE_ENDIAN && ((u.i & 0x3) == 0)) {
481 const uint32_t *k = (const uint32_t *)key; /* read 32-bit chunks */
486 /*------ all but last block: aligned reads and affect 32 bits of (a,b,c) */
497 /*----------------------------- handle the last (probably partial) block */
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 * noticably faster for short strings (like English words).
511 case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
512 case 11: c+=k[2]&0xffffff; b+=k[1]; a+=k[0]; break;
513 case 10: c+=k[2]&0xffff; b+=k[1]; a+=k[0]; break;
514 case 9 : c+=k[2]&0xff; b+=k[1]; a+=k[0]; break;
515 case 8 : b+=k[1]; a+=k[0]; break;
516 case 7 : b+=k[1]&0xffffff; a+=k[0]; break;
517 case 6 : b+=k[1]&0xffff; a+=k[0]; break;
518 case 5 : b+=k[1]&0xff; a+=k[0]; break;
519 case 4 : a+=k[0]; break;
520 case 3 : a+=k[0]&0xffffff; break;
521 case 2 : a+=k[0]&0xffff; break;
522 case 1 : a+=k[0]&0xff; break;
523 case 0 : *pc=c; *pb=b; return; /* zero length strings require no mixing */
526 #else /* make valgrind happy */
528 k8 = (const uint8_t *)k;
531 case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
532 case 11: c+=((uint32_t)k8[10])<<16; /* fall through */
533 case 10: c+=((uint32_t)k8[9])<<8; /* fall through */
534 case 9 : c+=k8[8]; /* fall through */
535 case 8 : b+=k[1]; a+=k[0]; break;
536 case 7 : b+=((uint32_t)k8[6])<<16; /* fall through */
537 case 6 : b+=((uint32_t)k8[5])<<8; /* fall through */
538 case 5 : b+=k8[4]; /* fall through */
539 case 4 : a+=k[0]; break;
540 case 3 : a+=((uint32_t)k8[2])<<16; /* fall through */
541 case 2 : a+=((uint32_t)k8[1])<<8; /* fall through */
542 case 1 : a+=k8[0]; break;
543 case 0 : *pc=c; *pb=b; return; /* zero length strings require no mixing */
546 #endif /* !valgrind */
548 } else if (HASH_LITTLE_ENDIAN && ((u.i & 0x1) == 0)) {
549 const uint16_t *k = (const uint16_t *)key; /* read 16-bit chunks */
552 /*--------------- all but last block: aligned reads and different mixing */
555 a += k[0] + (((uint32_t)k[1])<<16);
556 b += k[2] + (((uint32_t)k[3])<<16);
557 c += k[4] + (((uint32_t)k[5])<<16);
563 /*----------------------------- handle the last (probably partial) block */
564 k8 = (const uint8_t *)k;
567 case 12: c+=k[4]+(((uint32_t)k[5])<<16);
568 b+=k[2]+(((uint32_t)k[3])<<16);
569 a+=k[0]+(((uint32_t)k[1])<<16);
571 case 11: c+=((uint32_t)k8[10])<<16; /* fall through */
573 b+=k[2]+(((uint32_t)k[3])<<16);
574 a+=k[0]+(((uint32_t)k[1])<<16);
576 case 9 : c+=k8[8]; /* fall through */
577 case 8 : b+=k[2]+(((uint32_t)k[3])<<16);
578 a+=k[0]+(((uint32_t)k[1])<<16);
580 case 7 : b+=((uint32_t)k8[6])<<16; /* fall through */
582 a+=k[0]+(((uint32_t)k[1])<<16);
584 case 5 : b+=k8[4]; /* fall through */
585 case 4 : a+=k[0]+(((uint32_t)k[1])<<16);
587 case 3 : a+=((uint32_t)k8[2])<<16; /* fall through */
592 case 0 : *pc=c; *pb=b; return; /* zero length strings require no mixing */
595 } else { /* need to read the key one byte at a time */
596 const uint8_t *k = (const uint8_t *)key;
598 /*--------------- all but the last block: affect some 32 bits of (a,b,c) */
602 a += ((uint32_t)k[1])<<8;
603 a += ((uint32_t)k[2])<<16;
604 a += ((uint32_t)k[3])<<24;
606 b += ((uint32_t)k[5])<<8;
607 b += ((uint32_t)k[6])<<16;
608 b += ((uint32_t)k[7])<<24;
610 c += ((uint32_t)k[9])<<8;
611 c += ((uint32_t)k[10])<<16;
612 c += ((uint32_t)k[11])<<24;
618 /*-------------------------------- last block: affect all 32 bits of (c) */
619 switch(length) /* all the case statements fall through */
621 case 12: c+=((uint32_t)k[11])<<24;
622 case 11: c+=((uint32_t)k[10])<<16;
623 case 10: c+=((uint32_t)k[9])<<8;
625 case 8 : b+=((uint32_t)k[7])<<24;
626 case 7 : b+=((uint32_t)k[6])<<16;
627 case 6 : b+=((uint32_t)k[5])<<8;
629 case 4 : a+=((uint32_t)k[3])<<24;
630 case 3 : a+=((uint32_t)k[2])<<16;
631 case 2 : a+=((uint32_t)k[1])<<8;
634 case 0 : *pc=c; *pb=b; return; /* zero length strings require no mixing */
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.
650 #if 0 // libcitadel doesn't use this.
651 uint32_t hashbig( const void *key, size_t length, uint32_t initval)
654 union { const void *ptr; size_t i; } u; /* to cast key to (size_t) happily */
656 /* Set up the internal state */
657 a = b = c = 0xdeadbeef + ((uint32_t)length) + initval;
660 if (HASH_BIG_ENDIAN && ((u.i & 0x3) == 0)) {
661 const uint32_t *k = (const uint32_t *)key; /* read 32-bit chunks */
665 /*------ all but last block: aligned reads and affect 32 bits of (a,b,c) */
676 /*----------------------------- handle the last (probably partial) block */
678 * "k[2]<<8" actually reads beyond the end of the string, but
679 * then shifts out the part it's not allowed to read. Because the
680 * string is aligned, the illegal read is in the same word as the
681 * rest of the string. Every machine with memory protection I've seen
682 * does it on word boundaries, so is OK with this. But VALGRIND will
683 * still catch it and complain. The masking trick does make the hash
684 * noticably faster for short strings (like English words).
690 case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
691 case 11: c+=k[2]&0xffffff00; b+=k[1]; a+=k[0]; break;
692 case 10: c+=k[2]&0xffff0000; b+=k[1]; a+=k[0]; break;
693 case 9 : c+=k[2]&0xff000000; b+=k[1]; a+=k[0]; break;
694 case 8 : b+=k[1]; a+=k[0]; break;
695 case 7 : b+=k[1]&0xffffff00; a+=k[0]; break;
696 case 6 : b+=k[1]&0xffff0000; a+=k[0]; break;
697 case 5 : b+=k[1]&0xff000000; a+=k[0]; break;
698 case 4 : a+=k[0]; break;
699 case 3 : a+=k[0]&0xffffff00; break;
700 case 2 : a+=k[0]&0xffff0000; break;
701 case 1 : a+=k[0]&0xff000000; break;
702 case 0 : return c; /* zero length strings require no mixing */
705 #else /* make valgrind happy */
707 k8 = (const uint8_t *)k;
708 switch(length) /* all the case statements fall through */
710 case 12: c+=k[2]; b+=k[1]; a+=k[0]; break;
711 case 11: c+=((uint32_t)k8[10])<<8; /* fall through */
712 case 10: c+=((uint32_t)k8[9])<<16; /* fall through */
713 case 9 : c+=((uint32_t)k8[8])<<24; /* fall through */
714 case 8 : b+=k[1]; a+=k[0]; break;
715 case 7 : b+=((uint32_t)k8[6])<<8; /* fall through */
716 case 6 : b+=((uint32_t)k8[5])<<16; /* fall through */
717 case 5 : b+=((uint32_t)k8[4])<<24; /* fall through */
718 case 4 : a+=k[0]; break;
719 case 3 : a+=((uint32_t)k8[2])<<8; /* fall through */
720 case 2 : a+=((uint32_t)k8[1])<<16; /* fall through */
721 case 1 : a+=((uint32_t)k8[0])<<24; break;
725 #endif /* !VALGRIND */
727 } else { /* need to read the key one byte at a time */
728 const uint8_t *k = (const uint8_t *)key;
730 /*--------------- all but the last block: affect some 32 bits of (a,b,c) */
733 a += ((uint32_t)k[0])<<24;
734 a += ((uint32_t)k[1])<<16;
735 a += ((uint32_t)k[2])<<8;
736 a += ((uint32_t)k[3]);
737 b += ((uint32_t)k[4])<<24;
738 b += ((uint32_t)k[5])<<16;
739 b += ((uint32_t)k[6])<<8;
740 b += ((uint32_t)k[7]);
741 c += ((uint32_t)k[8])<<24;
742 c += ((uint32_t)k[9])<<16;
743 c += ((uint32_t)k[10])<<8;
744 c += ((uint32_t)k[11]);
750 /*-------------------------------- last block: affect all 32 bits of (c) */
751 switch(length) /* all the case statements fall through */
754 case 11: c+=((uint32_t)k[10])<<8;
755 case 10: c+=((uint32_t)k[9])<<16;
756 case 9 : c+=((uint32_t)k[8])<<24;
758 case 7 : b+=((uint32_t)k[6])<<8;
759 case 6 : b+=((uint32_t)k[5])<<16;
760 case 5 : b+=((uint32_t)k[4])<<24;
762 case 3 : a+=((uint32_t)k[2])<<8;
763 case 2 : a+=((uint32_t)k[1])<<16;
764 case 1 : a+=((uint32_t)k[0])<<24;
777 /* used for timings */
786 for (i=0; i<256; ++i) buf[i] = 'x';
789 h = hashlittle(&buf[0],1,h);
792 if (z-a > 0) printf("time %d %.8x\n", z-a, h);
795 /* check that every input bit changes every output bit half the time */
802 uint8_t qa[MAXLEN+1], qb[MAXLEN+2], *a = &qa[0], *b = &qb[1];
803 uint32_t c[HASHSTATE], d[HASHSTATE], i=0, j=0, k, l, m=0, z;
804 uint32_t e[HASHSTATE],f[HASHSTATE],g[HASHSTATE],h[HASHSTATE];
805 uint32_t x[HASHSTATE],y[HASHSTATE];
808 printf("No more than %d trials should ever be needed \n",MAXPAIR/2);
809 for (hlen=0; hlen < MAXLEN; ++hlen)
812 for (i=0; i<hlen; ++i) /*----------------------- for each input byte, */
814 for (j=0; j<8; ++j) /*------------------------ for each input bit, */
816 for (m=1; m<8; ++m) /*------------ for serveral possible initvals, */
818 for (l=0; l<HASHSTATE; ++l)
819 e[l]=f[l]=g[l]=h[l]=x[l]=y[l]=~((uint32_t)0);
821 /*---- check that every output bit is affected by that input bit */
822 for (k=0; k<MAXPAIR; k+=2)
825 /* keys have one bit different */
826 for (l=0; l<hlen+1; ++l) {a[l] = b[l] = (uint8_t)0;}
827 /* have a and b be two keys differing in only one bit */
830 c[0] = hashlittle(a, hlen, m);
832 b[i] ^= ((k+1)>>(8-j));
833 d[0] = hashlittle(b, hlen, m);
834 /* check every bit is 1, 0, set, and not set at least once */
835 for (l=0; l<HASHSTATE; ++l)
838 f[l] &= ~(c[l]^d[l]);
843 if (e[l]|f[l]|g[l]|h[l]|x[l]|y[l]) finished=0;
850 printf("Some bit didn't change: ");
851 printf("%.8x %.8x %.8x %.8x %.8x %.8x ",
852 e[0],f[0],g[0],h[0],x[0],y[0]);
853 printf("i %d j %d m %d len %d\n", i, j, m, hlen);
855 if (z==MAXPAIR) goto done;
862 printf("Mix success %2d bytes %2d initvals ",i,m);
863 printf("required %d trials\n", z/2);
869 /* Check for reading beyond the end of the buffer and alignment problems */
872 uint8_t buf[MAXLEN+20], *b;
874 uint8_t q[] = "This is the time for all good men to come to the aid of their country...";
876 uint8_t qq[] = "xThis is the time for all good men to come to the aid of their country...";
878 uint8_t qqq[] = "xxThis is the time for all good men to come to the aid of their country...";
880 uint8_t qqqq[] = "xxxThis is the time for all good men to come to the aid of their country...";
884 printf("Endianness. These lines should all be the same (for values filled in):\n");
885 printf("%.8x %.8x %.8x\n",
886 hashword((const uint32_t *)q, (sizeof(q)-1)/4, 13),
887 hashword((const uint32_t *)q, (sizeof(q)-5)/4, 13),
888 hashword((const uint32_t *)q, (sizeof(q)-9)/4, 13));
890 printf("%.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x\n",
891 hashlittle(p, sizeof(q)-1, 13), hashlittle(p, sizeof(q)-2, 13),
892 hashlittle(p, sizeof(q)-3, 13), hashlittle(p, sizeof(q)-4, 13),
893 hashlittle(p, sizeof(q)-5, 13), hashlittle(p, sizeof(q)-6, 13),
894 hashlittle(p, sizeof(q)-7, 13), hashlittle(p, sizeof(q)-8, 13),
895 hashlittle(p, sizeof(q)-9, 13), hashlittle(p, sizeof(q)-10, 13),
896 hashlittle(p, sizeof(q)-11, 13), hashlittle(p, sizeof(q)-12, 13));
898 printf("%.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x\n",
899 hashlittle(p, sizeof(q)-1, 13), hashlittle(p, sizeof(q)-2, 13),
900 hashlittle(p, sizeof(q)-3, 13), hashlittle(p, sizeof(q)-4, 13),
901 hashlittle(p, sizeof(q)-5, 13), hashlittle(p, sizeof(q)-6, 13),
902 hashlittle(p, sizeof(q)-7, 13), hashlittle(p, sizeof(q)-8, 13),
903 hashlittle(p, sizeof(q)-9, 13), hashlittle(p, sizeof(q)-10, 13),
904 hashlittle(p, sizeof(q)-11, 13), hashlittle(p, sizeof(q)-12, 13));
906 printf("%.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x\n",
907 hashlittle(p, sizeof(q)-1, 13), hashlittle(p, sizeof(q)-2, 13),
908 hashlittle(p, sizeof(q)-3, 13), hashlittle(p, sizeof(q)-4, 13),
909 hashlittle(p, sizeof(q)-5, 13), hashlittle(p, sizeof(q)-6, 13),
910 hashlittle(p, sizeof(q)-7, 13), hashlittle(p, sizeof(q)-8, 13),
911 hashlittle(p, sizeof(q)-9, 13), hashlittle(p, sizeof(q)-10, 13),
912 hashlittle(p, sizeof(q)-11, 13), hashlittle(p, sizeof(q)-12, 13));
914 printf("%.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x %.8x\n",
915 hashlittle(p, sizeof(q)-1, 13), hashlittle(p, sizeof(q)-2, 13),
916 hashlittle(p, sizeof(q)-3, 13), hashlittle(p, sizeof(q)-4, 13),
917 hashlittle(p, sizeof(q)-5, 13), hashlittle(p, sizeof(q)-6, 13),
918 hashlittle(p, sizeof(q)-7, 13), hashlittle(p, sizeof(q)-8, 13),
919 hashlittle(p, sizeof(q)-9, 13), hashlittle(p, sizeof(q)-10, 13),
920 hashlittle(p, sizeof(q)-11, 13), hashlittle(p, sizeof(q)-12, 13));
923 /* check that hashlittle2 and hashlittle produce the same results */
925 hashlittle2(q, sizeof(q), &i, &j);
926 if (hashlittle(q, sizeof(q), 47) != i)
927 printf("hashlittle2 and hashlittle mismatch\n");
929 /* check that hashword2 and hashword produce the same results */
932 hashword2(&len, 1, &i, &j);
933 if (hashword(&len, 1, 47) != i)
934 printf("hashword2 and hashword mismatch %x %x\n",
935 i, hashword(&len, 1, 47));
937 /* check hashlittle doesn't read before or after the ends of the string */
938 for (h=0, b=buf+1; h<8; ++h, ++b)
940 for (i=0; i<MAXLEN; ++i)
943 for (j=0; j<i; ++j) *(b+j)=0;
945 /* these should all be equal */
946 ref = hashlittle(b, len, (uint32_t)1);
949 x = hashlittle(b, len, (uint32_t)1);
950 y = hashlittle(b, len, (uint32_t)1);
951 if ((ref != x) || (ref != y))
953 printf("alignment error: %.8x %.8x %.8x %d %d\n",ref,x,y,
960 /* check for problems with nulls */
964 uint32_t h,i,state[HASHSTATE];
968 for (i=0; i<HASHSTATE; ++i) state[i] = 1;
969 printf("These should all be different\n");
970 for (i=0, h=0; i<8; ++i)
972 h = hashlittle(buf, 0, h);
973 printf("%2ld 0-byte strings, hash is %.8x\n", i, h);
980 driver1(); /* test that the key is hashed: used for timings */
981 driver2(); /* test that whole key is hashed thoroughly */
982 driver3(); /* test that nothing but the key is hashed */
983 driver4(); /* test hashing multiple buffers (all buffers are null) */
987 #endif /* SELF_TEST */