| 1 | /**************************************************************************** |
| 2 | ** |
| 3 | ** Copyright (C) 2016 The Qt Company Ltd. |
| 4 | ** Contact: https://www.qt.io/licensing/ |
| 5 | ** |
| 6 | ** This file is part of the test suite of the Qt Toolkit. |
| 7 | ** |
| 8 | ** $QT_BEGIN_LICENSE:GPL-EXCEPT$ |
| 9 | ** Commercial License Usage |
| 10 | ** Licensees holding valid commercial Qt licenses may use this file in |
| 11 | ** accordance with the commercial license agreement provided with the |
| 12 | ** Software or, alternatively, in accordance with the terms contained in |
| 13 | ** a written agreement between you and The Qt Company. For licensing terms |
| 14 | ** and conditions see https://www.qt.io/terms-conditions. For further |
| 15 | ** information use the contact form at https://www.qt.io/contact-us. |
| 16 | ** |
| 17 | ** GNU General Public License Usage |
| 18 | ** Alternatively, this file may be used under the terms of the GNU |
| 19 | ** General Public License version 3 as published by the Free Software |
| 20 | ** Foundation with exceptions as appearing in the file LICENSE.GPL3-EXCEPT |
| 21 | ** included in the packaging of this file. Please review the following |
| 22 | ** information to ensure the GNU General Public License requirements will |
| 23 | ** be met: https://www.gnu.org/licenses/gpl-3.0.html. |
| 24 | ** |
| 25 | ** $QT_END_LICENSE$ |
| 26 | ** |
| 27 | ****************************************************************************/ |
| 28 | |
| 29 | |
| 30 | /* |
| 31 | These functions are based on: |
| 32 | |
| 33 | ------------------------------------------------------------------------------- |
| 34 | lookup3.c, by Bob Jenkins, May 2006, Public Domain. |
| 35 | |
| 36 | These are functions for producing 32-bit hashes for hash table lookup. |
| 37 | hashword(), hashlittle(), hashlittle2(), hashbig(), mix(), and final() |
| 38 | are externally useful functions. Routines to test the hash are included |
| 39 | if SELF_TEST is defined. You can use this free for any purpose. It's in |
| 40 | the public domain. It has no warranty. |
| 41 | |
| 42 | You probably want to use hashlittle(). hashlittle() and hashbig() |
| 43 | hash byte arrays. hashlittle() is is faster than hashbig() on |
| 44 | little-endian machines. Intel and AMD are little-endian machines. |
| 45 | On second thought, you probably want hashlittle2(), which is identical to |
| 46 | hashlittle() except it returns two 32-bit hashes for the price of one. |
| 47 | You could implement hashbig2() if you wanted but I haven't bothered here. |
| 48 | |
| 49 | If you want to find a hash of, say, exactly 7 integers, do |
| 50 | a = i1; b = i2; c = i3; |
| 51 | mix(a,b,c); |
| 52 | a += i4; b += i5; c += i6; |
| 53 | mix(a,b,c); |
| 54 | a += i7; |
| 55 | final(a,b,c); |
| 56 | then use c as the hash value. If you have a variable length array of |
| 57 | 4-byte integers to hash, use hashword(). If you have a byte array (like |
| 58 | a character string), use hashlittle(). If you have several byte arrays, or |
| 59 | a mix of things, see the comments above hashlittle(). |
| 60 | |
| 61 | Why is this so big? I read 12 bytes at a time into 3 4-byte integers, |
| 62 | then mix those integers. This is fast (you can do a lot more thorough |
| 63 | mixing with 12*3 instructions on 3 integers than you can with 3 instructions |
| 64 | on 1 byte), but shoehorning those bytes into integers efficiently is messy. |
| 65 | ------------------------------------------------------------------------------- |
| 66 | */ |
| 67 | |
| 68 | #include <QtGlobal> |
| 69 | |
| 70 | #if Q_BYTE_ORDER == Q_BIG_ENDIAN |
| 71 | # define HASH_LITTLE_ENDIAN 0 |
| 72 | # define HASH_BIG_ENDIAN 1 |
| 73 | #else |
| 74 | # define HASH_LITTLE_ENDIAN 1 |
| 75 | # define HASH_BIG_ENDIAN 0 |
| 76 | #endif |
| 77 | |
| 78 | #define hashsize(n) ((quint32)1<<(n)) |
| 79 | #define hashmask(n) (hashsize(n)-1) |
| 80 | #define rot(x,k) (((x)<<(k)) | ((x)>>(32-(k)))) |
| 81 | |
| 82 | /* |
| 83 | ------------------------------------------------------------------------------- |
| 84 | mix -- mix 3 32-bit values reversibly. |
| 85 | |
| 86 | This is reversible, so any information in (a,b,c) before mix() is |
| 87 | still in (a,b,c) after mix(). |
| 88 | |
| 89 | If four pairs of (a,b,c) inputs are run through mix(), or through |
| 90 | mix() in reverse, there are at least 32 bits of the output that |
| 91 | are sometimes the same for one pair and different for another pair. |
| 92 | This was tested for: |
| 93 | * pairs that differed by one bit, by two bits, in any combination |
| 94 | of top bits of (a,b,c), or in any combination of bottom bits of |
| 95 | (a,b,c). |
| 96 | * "differ" is defined as +, -, ^, or ~^. For + and -, I transformed |
| 97 | the output delta to a Gray code (a^(a>>1)) so a string of 1's (as |
| 98 | is commonly produced by subtraction) look like a single 1-bit |
| 99 | difference. |
| 100 | * the base values were pseudorandom, all zero but one bit set, or |
| 101 | all zero plus a counter that starts at zero. |
| 102 | |
| 103 | Some k values for my "a-=c; a^=rot(c,k); c+=b;" arrangement that |
| 104 | satisfy this are |
| 105 | 4 6 8 16 19 4 |
| 106 | 9 15 3 18 27 15 |
| 107 | 14 9 3 7 17 3 |
| 108 | Well, "9 15 3 18 27 15" didn't quite get 32 bits diffing |
| 109 | for "differ" defined as + with a one-bit base and a two-bit delta. I |
| 110 | used http://burtleburtle.net/bob/hash/avalanche.html to choose |
| 111 | the operations, constants, and arrangements of the variables. |
| 112 | |
| 113 | This does not achieve avalanche. There are input bits of (a,b,c) |
| 114 | that fail to affect some output bits of (a,b,c), especially of a. The |
| 115 | most thoroughly mixed value is c, but it doesn't really even achieve |
| 116 | avalanche in c. |
| 117 | |
| 118 | This allows some parallelism. Read-after-writes are good at doubling |
| 119 | the number of bits affected, so the goal of mixing pulls in the opposite |
| 120 | direction as the goal of parallelism. I did what I could. Rotates |
| 121 | seem to cost as much as shifts on every machine I could lay my hands |
| 122 | on, and rotates are much kinder to the top and bottom bits, so I used |
| 123 | rotates. |
| 124 | ------------------------------------------------------------------------------- |
| 125 | */ |
| 126 | #define mix(a,b,c) \ |
| 127 | { \ |
| 128 | a -= c; a ^= rot(c, 4); c += b; \ |
| 129 | b -= a; b ^= rot(a, 6); a += c; \ |
| 130 | c -= b; c ^= rot(b, 8); b += a; \ |
| 131 | a -= c; a ^= rot(c,16); c += b; \ |
| 132 | b -= a; b ^= rot(a,19); a += c; \ |
| 133 | c -= b; c ^= rot(b, 4); b += a; \ |
| 134 | } |
| 135 | |
| 136 | /* |
| 137 | ------------------------------------------------------------------------------- |
| 138 | final -- final mixing of 3 32-bit values (a,b,c) into c |
| 139 | |
| 140 | Pairs of (a,b,c) values differing in only a few bits will usually |
| 141 | produce values of c that look totally different. This was tested for |
| 142 | * pairs that differed by one bit, by two bits, in any combination |
| 143 | of top bits of (a,b,c), or in any combination of bottom bits of |
| 144 | (a,b,c). |
| 145 | * "differ" is defined as +, -, ^, or ~^. For + and -, I transformed |
| 146 | the output delta to a Gray code (a^(a>>1)) so a string of 1's (as |
| 147 | is commonly produced by subtraction) look like a single 1-bit |
| 148 | difference. |
| 149 | * the base values were pseudorandom, all zero but one bit set, or |
| 150 | all zero plus a counter that starts at zero. |
| 151 | |
| 152 | These constants passed: |
| 153 | 14 11 25 16 4 14 24 |
| 154 | 12 14 25 16 4 14 24 |
| 155 | and these came close: |
| 156 | 4 8 15 26 3 22 24 |
| 157 | 10 8 15 26 3 22 24 |
| 158 | 11 8 15 26 3 22 24 |
| 159 | ------------------------------------------------------------------------------- |
| 160 | */ |
| 161 | #define final(a,b,c) \ |
| 162 | { \ |
| 163 | c ^= b; c -= rot(b,14); \ |
| 164 | a ^= c; a -= rot(c,11); \ |
| 165 | b ^= a; b -= rot(a,25); \ |
| 166 | c ^= b; c -= rot(b,16); \ |
| 167 | a ^= c; a -= rot(c,4); \ |
| 168 | b ^= a; b -= rot(a,14); \ |
| 169 | c ^= b; c -= rot(b,24); \ |
| 170 | } |
| 171 | |
| 172 | /* |
| 173 | -------------------------------------------------------------------- |
| 174 | This works on all machines. To be useful, it requires |
| 175 | -- that the key be an array of quint32's, and |
| 176 | -- that the length be the number of quint32's in the key |
| 177 | |
| 178 | The function hashword() is identical to hashlittle() on little-endian |
| 179 | machines, and identical to hashbig() on big-endian machines, |
| 180 | except that the length has to be measured in quint32s rather than in |
| 181 | bytes. hashlittle() is more complicated than hashword() only because |
| 182 | hashlittle() has to dance around fitting the key bytes into registers. |
| 183 | -------------------------------------------------------------------- |
| 184 | */ |
| 185 | quint32 hashword( |
| 186 | const quint32 *k, /* the key, an array of quint32 values */ |
| 187 | size_t length, /* the length of the key, in quint32s */ |
| 188 | quint32 initval) /* the previous hash, or an arbitrary value */ |
| 189 | { |
| 190 | quint32 a,b,c; |
| 191 | |
| 192 | /* Set up the internal state */ |
| 193 | a = b = c = 0xdeadbeef + (((quint32)length)<<2) + initval; |
| 194 | |
| 195 | /*------------------------------------------------- handle most of the key */ |
| 196 | while (length > 3) |
| 197 | { |
| 198 | a += k[0]; |
| 199 | b += k[1]; |
| 200 | c += k[2]; |
| 201 | mix(a,b,c); |
| 202 | length -= 3; |
| 203 | k += 3; |
| 204 | } |
| 205 | |
| 206 | /*------------------------------------------- handle the last 3 quint32's */ |
| 207 | switch(length) /* all the case statements fall through */ |
| 208 | { |
| 209 | case 3 : c+=k[2]; |
| 210 | Q_FALLTHROUGH(); |
| 211 | case 2 : b+=k[1]; |
| 212 | Q_FALLTHROUGH(); |
| 213 | case 1 : a+=k[0]; |
| 214 | final(a,b,c); |
| 215 | Q_FALLTHROUGH(); |
| 216 | case 0: /* case 0: nothing left to add */ |
| 217 | break; |
| 218 | } |
| 219 | /*------------------------------------------------------ report the result */ |
| 220 | return c; |
| 221 | } |
| 222 | |
| 223 | |
| 224 | /* |
| 225 | -------------------------------------------------------------------- |
| 226 | hashword2() -- same as hashword(), but take two seeds and return two |
| 227 | 32-bit values. pc and pb must both be nonnull, and *pc and *pb must |
| 228 | both be initialized with seeds. If you pass in (*pb)==0, the output |
| 229 | (*pc) will be the same as the return value from hashword(). |
| 230 | -------------------------------------------------------------------- |
| 231 | */ |
| 232 | void hashword2 ( |
| 233 | const quint32 *k, /* the key, an array of quint32 values */ |
| 234 | size_t length, /* the length of the key, in quint32s */ |
| 235 | quint32 *pc, /* IN: seed OUT: primary hash value */ |
| 236 | quint32 *pb) /* IN: more seed OUT: secondary hash value */ |
| 237 | { |
| 238 | quint32 a,b,c; |
| 239 | |
| 240 | /* Set up the internal state */ |
| 241 | a = b = c = 0xdeadbeef + ((quint32)(length<<2)) + *pc; |
| 242 | c += *pb; |
| 243 | |
| 244 | /*------------------------------------------------- handle most of the key */ |
| 245 | while (length > 3) |
| 246 | { |
| 247 | a += k[0]; |
| 248 | b += k[1]; |
| 249 | c += k[2]; |
| 250 | mix(a,b,c); |
| 251 | length -= 3; |
| 252 | k += 3; |
| 253 | } |
| 254 | |
| 255 | /*------------------------------------------- handle the last 3 quint32's */ |
| 256 | switch(length) /* all the case statements fall through */ |
| 257 | { |
| 258 | case 3 : c+=k[2]; |
| 259 | Q_FALLTHROUGH(); |
| 260 | case 2 : b+=k[1]; |
| 261 | Q_FALLTHROUGH(); |
| 262 | case 1 : a+=k[0]; |
| 263 | final(a,b,c); |
| 264 | Q_FALLTHROUGH(); |
| 265 | case 0: /* case 0: nothing left to add */ |
| 266 | break; |
| 267 | } |
| 268 | /*------------------------------------------------------ report the result */ |
| 269 | *pc=c; *pb=b; |
| 270 | } |
| 271 | |
| 272 | |
| 273 | /* |
| 274 | ------------------------------------------------------------------------------- |
| 275 | hashlittle() -- hash a variable-length key into a 32-bit value |
| 276 | k : the key (the unaligned variable-length array of bytes) |
| 277 | length : the length of the key, counting by bytes |
| 278 | initval : can be any 4-byte value |
| 279 | Returns a 32-bit value. Every bit of the key affects every bit of |
| 280 | the return value. Two keys differing by one or two bits will have |
| 281 | totally different hash values. |
| 282 | |
| 283 | The best hash table sizes are powers of 2. There is no need to do |
| 284 | mod a prime (mod is sooo slow!). If you need less than 32 bits, |
| 285 | use a bitmask. For example, if you need only 10 bits, do |
| 286 | h = (h & hashmask(10)); |
| 287 | In which case, the hash table should have hashsize(10) elements. |
| 288 | |
| 289 | If you are hashing n strings (quint8 **)k, do it like this: |
| 290 | for (i=0, h=0; i<n; ++i) h = hashlittle( k[i], len[i], h); |
| 291 | |
| 292 | By Bob Jenkins, 2006. bob_jenkins@burtleburtle.net. You may use this |
| 293 | code any way you wish, private, educational, or commercial. It's free. |
| 294 | |
| 295 | Use for hash table lookup, or anything where one collision in 2^^32 is |
| 296 | acceptable. Do NOT use for cryptographic purposes. |
| 297 | ------------------------------------------------------------------------------- |
| 298 | */ |
| 299 | |
| 300 | quint32 hashlittle( const void *key, size_t length, quint32 initval) |
| 301 | { |
| 302 | quint32 a,b,c; /* internal state */ |
| 303 | union { const void *ptr; size_t i; } u; /* needed for Mac Powerbook G4 */ |
| 304 | |
| 305 | /* Set up the internal state */ |
| 306 | a = b = c = 0xdeadbeef + ((quint32)length) + initval; |
| 307 | |
| 308 | u.ptr = key; |
| 309 | if (HASH_LITTLE_ENDIAN && ((u.i & 0x3) == 0)) { |
| 310 | const quint32 *k = (const quint32 *)key; /* read 32-bit chunks */ |
| 311 | |
| 312 | /*------ all but last block: aligned reads and affect 32 bits of (a,b,c) */ |
| 313 | while (length > 12) |
| 314 | { |
| 315 | a += k[0]; |
| 316 | b += k[1]; |
| 317 | c += k[2]; |
| 318 | mix(a,b,c); |
| 319 | length -= 12; |
| 320 | k += 3; |
| 321 | } |
| 322 | |
| 323 | /*----------------------------- handle the last (probably partial) block */ |
| 324 | /* |
| 325 | * "k[2]&0xffffff" actually reads beyond the end of the string, but |
| 326 | * then masks off the part it's not allowed to read. Because the |
| 327 | * string is aligned, the masked-off tail is in the same word as the |
| 328 | * rest of the string. Every machine with memory protection I've seen |
| 329 | * does it on word boundaries, so is OK with this. But VALGRIND will |
| 330 | * still catch it and complain. The masking trick does make the hash |
| 331 | * noticably faster for short strings (like English words). |
| 332 | */ |
| 333 | #ifndef VALGRIND |
| 334 | |
| 335 | switch(length) |
| 336 | { |
| 337 | case 12: c+=k[2]; b+=k[1]; a+=k[0]; break; |
| 338 | case 11: c+=k[2]&0xffffff; b+=k[1]; a+=k[0]; break; |
| 339 | case 10: c+=k[2]&0xffff; b+=k[1]; a+=k[0]; break; |
| 340 | case 9 : c+=k[2]&0xff; b+=k[1]; a+=k[0]; break; |
| 341 | case 8 : b+=k[1]; a+=k[0]; break; |
| 342 | case 7 : b+=k[1]&0xffffff; a+=k[0]; break; |
| 343 | case 6 : b+=k[1]&0xffff; a+=k[0]; break; |
| 344 | case 5 : b+=k[1]&0xff; a+=k[0]; break; |
| 345 | case 4 : a+=k[0]; break; |
| 346 | case 3 : a+=k[0]&0xffffff; break; |
| 347 | case 2 : a+=k[0]&0xffff; break; |
| 348 | case 1 : a+=k[0]&0xff; break; |
| 349 | case 0 : return c; /* zero length strings require no mixing */ |
| 350 | } |
| 351 | |
| 352 | #else /* make valgrind happy */ |
| 353 | |
| 354 | const quint8 *k8 = (const quint8 *)k; |
| 355 | switch(length) |
| 356 | { |
| 357 | case 12: c+=k[2]; b+=k[1]; a+=k[0]; break; |
| 358 | case 11: c+=((quint32)k8[10])<<16; |
| 359 | Q_FALLTHROUGH(); |
| 360 | case 10: c+=((quint32)k8[9])<<8; |
| 361 | Q_FALLTHROUGH(); |
| 362 | case 9 : c+=k8[8]; |
| 363 | Q_FALLTHROUGH(); |
| 364 | case 8 : b+=k[1]; a+=k[0]; break; |
| 365 | case 7 : b+=((quint32)k8[6])<<16; |
| 366 | Q_FALLTHROUGH(); |
| 367 | case 6 : b+=((quint32)k8[5])<<8; |
| 368 | Q_FALLTHROUGH(); |
| 369 | case 5 : b+=k8[4]; |
| 370 | Q_FALLTHROUGH(); |
| 371 | case 4 : a+=k[0]; break; |
| 372 | case 3 : a+=((quint32)k8[2])<<16; |
| 373 | Q_FALLTHROUGH(); |
| 374 | case 2 : a+=((quint32)k8[1])<<8; |
| 375 | Q_FALLTHROUGH(); |
| 376 | case 1 : a+=k8[0]; break; |
| 377 | case 0 : return c; |
| 378 | } |
| 379 | |
| 380 | #endif /* !valgrind */ |
| 381 | |
| 382 | } else if (HASH_LITTLE_ENDIAN && ((u.i & 0x1) == 0)) { |
| 383 | const quint16 *k = (const quint16 *)key; /* read 16-bit chunks */ |
| 384 | const quint8 *k8; |
| 385 | |
| 386 | /*--------------- all but last block: aligned reads and different mixing */ |
| 387 | while (length > 12) |
| 388 | { |
| 389 | a += k[0] + (((quint32)k[1])<<16); |
| 390 | b += k[2] + (((quint32)k[3])<<16); |
| 391 | c += k[4] + (((quint32)k[5])<<16); |
| 392 | mix(a,b,c); |
| 393 | length -= 12; |
| 394 | k += 6; |
| 395 | } |
| 396 | |
| 397 | /*----------------------------- handle the last (probably partial) block */ |
| 398 | k8 = (const quint8 *)k; |
| 399 | switch(length) |
| 400 | { |
| 401 | case 12: c+=k[4]+(((quint32)k[5])<<16); |
| 402 | b+=k[2]+(((quint32)k[3])<<16); |
| 403 | a+=k[0]+(((quint32)k[1])<<16); |
| 404 | break; |
| 405 | case 11: c+=((quint32)k8[10])<<16; |
| 406 | Q_FALLTHROUGH(); |
| 407 | case 10: c+=k[4]; |
| 408 | b+=k[2]+(((quint32)k[3])<<16); |
| 409 | a+=k[0]+(((quint32)k[1])<<16); |
| 410 | break; |
| 411 | case 9 : c+=k8[8]; |
| 412 | Q_FALLTHROUGH(); |
| 413 | case 8 : b+=k[2]+(((quint32)k[3])<<16); |
| 414 | a+=k[0]+(((quint32)k[1])<<16); |
| 415 | break; |
| 416 | case 7 : b+=((quint32)k8[6])<<16; |
| 417 | Q_FALLTHROUGH(); |
| 418 | case 6 : b+=k[2]; |
| 419 | a+=k[0]+(((quint32)k[1])<<16); |
| 420 | break; |
| 421 | case 5 : b+=k8[4]; |
| 422 | Q_FALLTHROUGH(); |
| 423 | case 4 : a+=k[0]+(((quint32)k[1])<<16); |
| 424 | break; |
| 425 | case 3 : a+=((quint32)k8[2])<<16; |
| 426 | Q_FALLTHROUGH(); |
| 427 | case 2 : a+=k[0]; |
| 428 | break; |
| 429 | case 1 : a+=k8[0]; |
| 430 | break; |
| 431 | case 0 : return c; /* zero length requires no mixing */ |
| 432 | } |
| 433 | |
| 434 | } else { /* need to read the key one byte at a time */ |
| 435 | const quint8 *k = (const quint8 *)key; |
| 436 | |
| 437 | /*--------------- all but the last block: affect some 32 bits of (a,b,c) */ |
| 438 | while (length > 12) |
| 439 | { |
| 440 | a += k[0]; |
| 441 | a += ((quint32)k[1])<<8; |
| 442 | a += ((quint32)k[2])<<16; |
| 443 | a += ((quint32)k[3])<<24; |
| 444 | b += k[4]; |
| 445 | b += ((quint32)k[5])<<8; |
| 446 | b += ((quint32)k[6])<<16; |
| 447 | b += ((quint32)k[7])<<24; |
| 448 | c += k[8]; |
| 449 | c += ((quint32)k[9])<<8; |
| 450 | c += ((quint32)k[10])<<16; |
| 451 | c += ((quint32)k[11])<<24; |
| 452 | mix(a,b,c); |
| 453 | length -= 12; |
| 454 | k += 12; |
| 455 | } |
| 456 | |
| 457 | /*-------------------------------- last block: affect all 32 bits of (c) */ |
| 458 | switch(length) /* all the case statements fall through */ |
| 459 | { |
| 460 | case 12: c+=((quint32)k[11])<<24; |
| 461 | Q_FALLTHROUGH(); |
| 462 | case 11: c+=((quint32)k[10])<<16; |
| 463 | Q_FALLTHROUGH(); |
| 464 | case 10: c+=((quint32)k[9])<<8; |
| 465 | Q_FALLTHROUGH(); |
| 466 | case 9 : c+=k[8]; |
| 467 | Q_FALLTHROUGH(); |
| 468 | case 8 : b+=((quint32)k[7])<<24; |
| 469 | Q_FALLTHROUGH(); |
| 470 | case 7 : b+=((quint32)k[6])<<16; |
| 471 | Q_FALLTHROUGH(); |
| 472 | case 6 : b+=((quint32)k[5])<<8; |
| 473 | Q_FALLTHROUGH(); |
| 474 | case 5 : b+=k[4]; |
| 475 | Q_FALLTHROUGH(); |
| 476 | case 4 : a+=((quint32)k[3])<<24; |
| 477 | Q_FALLTHROUGH(); |
| 478 | case 3 : a+=((quint32)k[2])<<16; |
| 479 | Q_FALLTHROUGH(); |
| 480 | case 2 : a+=((quint32)k[1])<<8; |
| 481 | Q_FALLTHROUGH(); |
| 482 | case 1 : a+=k[0]; |
| 483 | break; |
| 484 | case 0 : return c; |
| 485 | } |
| 486 | } |
| 487 | |
| 488 | final(a,b,c); |
| 489 | return c; |
| 490 | } |
| 491 | |
| 492 | |
| 493 | /* |
| 494 | * hashlittle2: return 2 32-bit hash values |
| 495 | * |
| 496 | * This is identical to hashlittle(), except it returns two 32-bit hash |
| 497 | * values instead of just one. This is good enough for hash table |
| 498 | * lookup with 2^^64 buckets, or if you want a second hash if you're not |
| 499 | * happy with the first, or if you want a probably-unique 64-bit ID for |
| 500 | * the key. *pc is better mixed than *pb, so use *pc first. If you want |
| 501 | * a 64-bit value do something like "*pc + (((uint64_t)*pb)<<32)". |
| 502 | */ |
| 503 | void hashlittle2( |
| 504 | const void *key, /* the key to hash */ |
| 505 | size_t length, /* length of the key */ |
| 506 | quint32 *pc, /* IN: primary initval, OUT: primary hash */ |
| 507 | quint32 *pb) /* IN: secondary initval, OUT: secondary hash */ |
| 508 | { |
| 509 | quint32 a,b,c; /* internal state */ |
| 510 | union { const void *ptr; size_t i; } u; /* needed for Mac Powerbook G4 */ |
| 511 | |
| 512 | /* Set up the internal state */ |
| 513 | a = b = c = 0xdeadbeef + ((quint32)length) + *pc; |
| 514 | c += *pb; |
| 515 | |
| 516 | u.ptr = key; |
| 517 | if (HASH_LITTLE_ENDIAN && ((u.i & 0x3) == 0)) { |
| 518 | const quint32 *k = (const quint32 *)key; /* read 32-bit chunks */ |
| 519 | |
| 520 | /*------ all but last block: aligned reads and affect 32 bits of (a,b,c) */ |
| 521 | while (length > 12) |
| 522 | { |
| 523 | a += k[0]; |
| 524 | b += k[1]; |
| 525 | c += k[2]; |
| 526 | mix(a,b,c); |
| 527 | length -= 12; |
| 528 | k += 3; |
| 529 | } |
| 530 | |
| 531 | /*----------------------------- handle the last (probably partial) block */ |
| 532 | /* |
| 533 | * "k[2]&0xffffff" actually reads beyond the end of the string, but |
| 534 | * then masks off the part it's not allowed to read. Because the |
| 535 | * string is aligned, the masked-off tail is in the same word as the |
| 536 | * rest of the string. Every machine with memory protection I've seen |
| 537 | * does it on word boundaries, so is OK with this. But VALGRIND will |
| 538 | * still catch it and complain. The masking trick does make the hash |
| 539 | * noticably faster for short strings (like English words). |
| 540 | */ |
| 541 | #ifndef VALGRIND |
| 542 | |
| 543 | switch(length) |
| 544 | { |
| 545 | case 12: c+=k[2]; b+=k[1]; a+=k[0]; break; |
| 546 | case 11: c+=k[2]&0xffffff; b+=k[1]; a+=k[0]; break; |
| 547 | case 10: c+=k[2]&0xffff; b+=k[1]; a+=k[0]; break; |
| 548 | case 9 : c+=k[2]&0xff; b+=k[1]; a+=k[0]; break; |
| 549 | case 8 : b+=k[1]; a+=k[0]; break; |
| 550 | case 7 : b+=k[1]&0xffffff; a+=k[0]; break; |
| 551 | case 6 : b+=k[1]&0xffff; a+=k[0]; break; |
| 552 | case 5 : b+=k[1]&0xff; a+=k[0]; break; |
| 553 | case 4 : a+=k[0]; break; |
| 554 | case 3 : a+=k[0]&0xffffff; break; |
| 555 | case 2 : a+=k[0]&0xffff; break; |
| 556 | case 1 : a+=k[0]&0xff; break; |
| 557 | case 0 : *pc=c; *pb=b; return; /* zero length strings require no mixing */ |
| 558 | } |
| 559 | |
| 560 | #else /* make valgrind happy */ |
| 561 | |
| 562 | const quint8 *k8 = (const quint8 *)k; |
| 563 | switch(length) |
| 564 | { |
| 565 | case 12: c+=k[2]; b+=k[1]; a+=k[0]; break; |
| 566 | case 11: c+=((quint32)k8[10])<<16; |
| 567 | Q_FALLTHROUGH(); |
| 568 | case 10: c+=((quint32)k8[9])<<8; |
| 569 | Q_FALLTHROUGH(); |
| 570 | case 9 : c+=k8[8]; |
| 571 | Q_FALLTHROUGH(); |
| 572 | case 8 : b+=k[1]; a+=k[0]; break; |
| 573 | case 7 : b+=((quint32)k8[6])<<16; |
| 574 | Q_FALLTHROUGH(); |
| 575 | case 6 : b+=((quint32)k8[5])<<8; |
| 576 | Q_FALLTHROUGH(); |
| 577 | case 5 : b+=k8[4]; |
| 578 | Q_FALLTHROUGH(); |
| 579 | case 4 : a+=k[0]; break; |
| 580 | case 3 : a+=((quint32)k8[2])<<16; |
| 581 | Q_FALLTHROUGH(); |
| 582 | case 2 : a+=((quint32)k8[1])<<8; |
| 583 | Q_FALLTHROUGH(); |
| 584 | case 1 : a+=k8[0]; break; |
| 585 | case 0 : *pc=c; *pb=b; return; /* zero length strings require no mixing */ |
| 586 | } |
| 587 | |
| 588 | #endif /* !valgrind */ |
| 589 | |
| 590 | } else if (HASH_LITTLE_ENDIAN && ((u.i & 0x1) == 0)) { |
| 591 | const quint16 *k = (const quint16 *)key; /* read 16-bit chunks */ |
| 592 | const quint8 *k8; |
| 593 | |
| 594 | /*--------------- all but last block: aligned reads and different mixing */ |
| 595 | while (length > 12) |
| 596 | { |
| 597 | a += k[0] + (((quint32)k[1])<<16); |
| 598 | b += k[2] + (((quint32)k[3])<<16); |
| 599 | c += k[4] + (((quint32)k[5])<<16); |
| 600 | mix(a,b,c); |
| 601 | length -= 12; |
| 602 | k += 6; |
| 603 | } |
| 604 | |
| 605 | /*----------------------------- handle the last (probably partial) block */ |
| 606 | k8 = (const quint8 *)k; |
| 607 | switch(length) |
| 608 | { |
| 609 | case 12: c+=k[4]+(((quint32)k[5])<<16); |
| 610 | b+=k[2]+(((quint32)k[3])<<16); |
| 611 | a+=k[0]+(((quint32)k[1])<<16); |
| 612 | break; |
| 613 | case 11: c+=((quint32)k8[10])<<16; |
| 614 | Q_FALLTHROUGH(); |
| 615 | case 10: c+=k[4]; |
| 616 | b+=k[2]+(((quint32)k[3])<<16); |
| 617 | a+=k[0]+(((quint32)k[1])<<16); |
| 618 | break; |
| 619 | case 9 : c+=k8[8]; |
| 620 | Q_FALLTHROUGH(); |
| 621 | case 8 : b+=k[2]+(((quint32)k[3])<<16); |
| 622 | a+=k[0]+(((quint32)k[1])<<16); |
| 623 | break; |
| 624 | case 7 : b+=((quint32)k8[6])<<16; |
| 625 | Q_FALLTHROUGH(); |
| 626 | case 6 : b+=k[2]; |
| 627 | a+=k[0]+(((quint32)k[1])<<16); |
| 628 | break; |
| 629 | case 5 : b+=k8[4]; |
| 630 | Q_FALLTHROUGH(); |
| 631 | case 4 : a+=k[0]+(((quint32)k[1])<<16); |
| 632 | break; |
| 633 | case 3 : a+=((quint32)k8[2])<<16; |
| 634 | Q_FALLTHROUGH(); |
| 635 | case 2 : a+=k[0]; |
| 636 | break; |
| 637 | case 1 : a+=k8[0]; |
| 638 | break; |
| 639 | case 0 : *pc=c; *pb=b; return; /* zero length strings require no mixing */ |
| 640 | } |
| 641 | |
| 642 | } else { /* need to read the key one byte at a time */ |
| 643 | const quint8 *k = (const quint8 *)key; |
| 644 | |
| 645 | /*--------------- all but the last block: affect some 32 bits of (a,b,c) */ |
| 646 | while (length > 12) |
| 647 | { |
| 648 | a += k[0]; |
| 649 | a += ((quint32)k[1])<<8; |
| 650 | a += ((quint32)k[2])<<16; |
| 651 | a += ((quint32)k[3])<<24; |
| 652 | b += k[4]; |
| 653 | b += ((quint32)k[5])<<8; |
| 654 | b += ((quint32)k[6])<<16; |
| 655 | b += ((quint32)k[7])<<24; |
| 656 | c += k[8]; |
| 657 | c += ((quint32)k[9])<<8; |
| 658 | c += ((quint32)k[10])<<16; |
| 659 | c += ((quint32)k[11])<<24; |
| 660 | mix(a,b,c); |
| 661 | length -= 12; |
| 662 | k += 12; |
| 663 | } |
| 664 | |
| 665 | /*-------------------------------- last block: affect all 32 bits of (c) */ |
| 666 | switch(length) /* all the case statements fall through */ |
| 667 | { |
| 668 | case 12: c+=((quint32)k[11])<<24; |
| 669 | Q_FALLTHROUGH(); |
| 670 | case 11: c+=((quint32)k[10])<<16; |
| 671 | Q_FALLTHROUGH(); |
| 672 | case 10: c+=((quint32)k[9])<<8; |
| 673 | Q_FALLTHROUGH(); |
| 674 | case 9 : c+=k[8]; |
| 675 | Q_FALLTHROUGH(); |
| 676 | case 8 : b+=((quint32)k[7])<<24; |
| 677 | Q_FALLTHROUGH(); |
| 678 | case 7 : b+=((quint32)k[6])<<16; |
| 679 | Q_FALLTHROUGH(); |
| 680 | case 6 : b+=((quint32)k[5])<<8; |
| 681 | Q_FALLTHROUGH(); |
| 682 | case 5 : b+=k[4]; |
| 683 | Q_FALLTHROUGH(); |
| 684 | case 4 : a+=((quint32)k[3])<<24; |
| 685 | Q_FALLTHROUGH(); |
| 686 | case 3 : a+=((quint32)k[2])<<16; |
| 687 | Q_FALLTHROUGH(); |
| 688 | case 2 : a+=((quint32)k[1])<<8; |
| 689 | Q_FALLTHROUGH(); |
| 690 | case 1 : a+=k[0]; |
| 691 | break; |
| 692 | case 0 : *pc=c; *pb=b; return; /* zero length strings require no mixing */ |
| 693 | } |
| 694 | } |
| 695 | |
| 696 | final(a,b,c); |
| 697 | *pc=c; *pb=b; |
| 698 | } |
| 699 | |
| 700 | |
| 701 | |
| 702 | /* |
| 703 | * hashbig(): |
| 704 | * This is the same as hashword() on big-endian machines. It is different |
| 705 | * from hashlittle() on all machines. hashbig() takes advantage of |
| 706 | * big-endian byte ordering. |
| 707 | */ |
| 708 | quint32 hashbig( const void *key, size_t length, quint32 initval) |
| 709 | { |
| 710 | quint32 a,b,c; |
| 711 | union { const void *ptr; size_t i; } u; /* to cast key to (size_t) happily */ |
| 712 | |
| 713 | /* Set up the internal state */ |
| 714 | a = b = c = 0xdeadbeef + ((quint32)length) + initval; |
| 715 | |
| 716 | u.ptr = key; |
| 717 | if (HASH_BIG_ENDIAN && ((u.i & 0x3) == 0)) { |
| 718 | const quint32 *k = (const quint32 *)key; /* read 32-bit chunks */ |
| 719 | |
| 720 | /*------ all but last block: aligned reads and affect 32 bits of (a,b,c) */ |
| 721 | while (length > 12) |
| 722 | { |
| 723 | a += k[0]; |
| 724 | b += k[1]; |
| 725 | c += k[2]; |
| 726 | mix(a,b,c); |
| 727 | length -= 12; |
| 728 | k += 3; |
| 729 | } |
| 730 | |
| 731 | /*----------------------------- handle the last (probably partial) block */ |
| 732 | /* |
| 733 | * "k[2]<<8" actually reads beyond the end of the string, but |
| 734 | * then shifts out the part it's not allowed to read. Because the |
| 735 | * string is aligned, the illegal read is in the same word as the |
| 736 | * rest of the string. Every machine with memory protection I've seen |
| 737 | * does it on word boundaries, so is OK with this. But VALGRIND will |
| 738 | * still catch it and complain. The masking trick does make the hash |
| 739 | * noticably faster for short strings (like English words). |
| 740 | */ |
| 741 | #ifndef VALGRIND |
| 742 | |
| 743 | switch(length) |
| 744 | { |
| 745 | case 12: c+=k[2]; b+=k[1]; a+=k[0]; break; |
| 746 | case 11: c+=k[2]&0xffffff00; b+=k[1]; a+=k[0]; break; |
| 747 | case 10: c+=k[2]&0xffff0000; b+=k[1]; a+=k[0]; break; |
| 748 | case 9 : c+=k[2]&0xff000000; b+=k[1]; a+=k[0]; break; |
| 749 | case 8 : b+=k[1]; a+=k[0]; break; |
| 750 | case 7 : b+=k[1]&0xffffff00; a+=k[0]; break; |
| 751 | case 6 : b+=k[1]&0xffff0000; a+=k[0]; break; |
| 752 | case 5 : b+=k[1]&0xff000000; a+=k[0]; break; |
| 753 | case 4 : a+=k[0]; break; |
| 754 | case 3 : a+=k[0]&0xffffff00; break; |
| 755 | case 2 : a+=k[0]&0xffff0000; break; |
| 756 | case 1 : a+=k[0]&0xff000000; break; |
| 757 | case 0 : return c; /* zero length strings require no mixing */ |
| 758 | } |
| 759 | |
| 760 | #else /* make valgrind happy */ |
| 761 | |
| 762 | const quint8 *k8 = (const quint8 *)k; |
| 763 | switch(length) /* all the case statements fall through */ |
| 764 | { |
| 765 | case 12: c+=k[2]; b+=k[1]; a+=k[0]; break; |
| 766 | case 11: c+=((quint32)k8[10])<<8; |
| 767 | Q_FALLTHROUGH(); |
| 768 | case 10: c+=((quint32)k8[9])<<16; |
| 769 | Q_FALLTHROUGH(); |
| 770 | case 9 : c+=((quint32)k8[8])<<24; |
| 771 | Q_FALLTHROUGH(); |
| 772 | case 8 : b+=k[1]; a+=k[0]; break; |
| 773 | case 7 : b+=((quint32)k8[6])<<8; |
| 774 | Q_FALLTHROUGH(); |
| 775 | case 6 : b+=((quint32)k8[5])<<16; |
| 776 | Q_FALLTHROUGH(); |
| 777 | case 5 : b+=((quint32)k8[4])<<24; |
| 778 | Q_FALLTHROUGH(); |
| 779 | case 4 : a+=k[0]; break; |
| 780 | case 3 : a+=((quint32)k8[2])<<8; |
| 781 | Q_FALLTHROUGH(); |
| 782 | case 2 : a+=((quint32)k8[1])<<16; |
| 783 | Q_FALLTHROUGH(); |
| 784 | case 1 : a+=((quint32)k8[0])<<24; break; |
| 785 | case 0 : return c; |
| 786 | } |
| 787 | |
| 788 | #endif /* !VALGRIND */ |
| 789 | |
| 790 | } else { /* need to read the key one byte at a time */ |
| 791 | const quint8 *k = (const quint8 *)key; |
| 792 | |
| 793 | /*--------------- all but the last block: affect some 32 bits of (a,b,c) */ |
| 794 | while (length > 12) |
| 795 | { |
| 796 | a += ((quint32)k[0])<<24; |
| 797 | a += ((quint32)k[1])<<16; |
| 798 | a += ((quint32)k[2])<<8; |
| 799 | a += ((quint32)k[3]); |
| 800 | b += ((quint32)k[4])<<24; |
| 801 | b += ((quint32)k[5])<<16; |
| 802 | b += ((quint32)k[6])<<8; |
| 803 | b += ((quint32)k[7]); |
| 804 | c += ((quint32)k[8])<<24; |
| 805 | c += ((quint32)k[9])<<16; |
| 806 | c += ((quint32)k[10])<<8; |
| 807 | c += ((quint32)k[11]); |
| 808 | mix(a,b,c); |
| 809 | length -= 12; |
| 810 | k += 12; |
| 811 | } |
| 812 | |
| 813 | /*-------------------------------- last block: affect all 32 bits of (c) */ |
| 814 | switch(length) /* all the case statements fall through */ |
| 815 | { |
| 816 | case 12: c+=k[11]; |
| 817 | Q_FALLTHROUGH(); |
| 818 | case 11: c+=((quint32)k[10])<<8; |
| 819 | Q_FALLTHROUGH(); |
| 820 | case 10: c+=((quint32)k[9])<<16; |
| 821 | Q_FALLTHROUGH(); |
| 822 | case 9 : c+=((quint32)k[8])<<24; |
| 823 | Q_FALLTHROUGH(); |
| 824 | case 8 : b+=k[7]; |
| 825 | Q_FALLTHROUGH(); |
| 826 | case 7 : b+=((quint32)k[6])<<8; |
| 827 | Q_FALLTHROUGH(); |
| 828 | case 6 : b+=((quint32)k[5])<<16; |
| 829 | Q_FALLTHROUGH(); |
| 830 | case 5 : b+=((quint32)k[4])<<24; |
| 831 | Q_FALLTHROUGH(); |
| 832 | case 4 : a+=k[3]; |
| 833 | Q_FALLTHROUGH(); |
| 834 | case 3 : a+=((quint32)k[2])<<8; |
| 835 | Q_FALLTHROUGH(); |
| 836 | case 2 : a+=((quint32)k[1])<<16; |
| 837 | Q_FALLTHROUGH(); |
| 838 | case 1 : a+=((quint32)k[0])<<24; |
| 839 | break; |
| 840 | case 0 : return c; |
| 841 | } |
| 842 | } |
| 843 | |
| 844 | final(a,b,c); |
| 845 | return c; |
| 846 | } |
| 847 | |