hash.c revision 138232
1/* 2 * Copyright (c) 1988, 1989, 1990, 1993 3 * The Regents of the University of California. All rights reserved. 4 * Copyright (c) 1988, 1989 by Adam de Boor 5 * Copyright (c) 1989 by Berkeley Softworks 6 * All rights reserved. 7 * 8 * This code is derived from software contributed to Berkeley by 9 * Adam de Boor. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. All advertising materials mentioning features or use of this software 20 * must display the following acknowledgement: 21 * This product includes software developed by the University of 22 * California, Berkeley and its contributors. 23 * 4. Neither the name of the University nor the names of its contributors 24 * may be used to endorse or promote products derived from this software 25 * without specific prior written permission. 26 * 27 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 28 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 29 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 30 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 31 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 32 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 33 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 34 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 35 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 36 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 37 * SUCH DAMAGE. 38 * 39 * @(#)hash.c 8.1 (Berkeley) 6/6/93 40 */ 41 42#include <sys/cdefs.h> 43__FBSDID("$FreeBSD: head/usr.bin/make/hash.c 138232 2004-11-30 17:46:29Z harti $"); 44 45/* hash.c -- 46 * 47 * This module contains routines to manipulate a hash table. 48 * See hash.h for a definition of the structure of the hash 49 * table. Hash tables grow automatically as the amount of 50 * information increases. 51 */ 52#include <unistd.h> 53#include "sprite.h" 54#include "make.h" 55#include "hash.h" 56 57/* 58 * Forward references to local procedures that are used before they're 59 * defined: 60 */ 61 62static void RebuildTable(Hash_Table *); 63 64/* 65 * The following defines the ratio of # entries to # buckets 66 * at which we rebuild the table to make it larger. 67 */ 68 69#define rebuildLimit 8 70 71/* 72 *--------------------------------------------------------- 73 * 74 * Hash_InitTable -- 75 * 76 * Set up the hash table t with a given number of buckets, or a 77 * reasonable default if the number requested is less than or 78 * equal to zero. Hash tables will grow in size as needed. 79 * 80 * 81 * Results: 82 * None. 83 * 84 * Side Effects: 85 * Memory is allocated for the initial bucket area. 86 * 87 *--------------------------------------------------------- 88 */ 89 90void 91Hash_InitTable(Hash_Table *t, int numBuckets) 92{ 93 int i; 94 struct Hash_Entry **hp; 95 96 /* 97 * Round up the size to a power of two. 98 */ 99 if (numBuckets <= 0) 100 i = 16; 101 else { 102 for (i = 2; i < numBuckets; i <<= 1) 103 continue; 104 } 105 t->numEntries = 0; 106 t->size = i; 107 t->mask = i - 1; 108 t->bucketPtr = hp = (struct Hash_Entry **)emalloc(sizeof(*hp) * i); 109 while (--i >= 0) 110 *hp++ = NULL; 111} 112 113/* 114 *--------------------------------------------------------- 115 * 116 * Hash_DeleteTable -- 117 * 118 * This routine removes everything from a hash table 119 * and frees up the memory space it occupied (except for 120 * the space in the Hash_Table structure). 121 * 122 * Results: 123 * None. 124 * 125 * Side Effects: 126 * Lots of memory is freed up. 127 * 128 *--------------------------------------------------------- 129 */ 130 131void 132Hash_DeleteTable(Hash_Table *t) 133{ 134 struct Hash_Entry **hp, *h, *nexth = NULL; 135 int i; 136 137 for (hp = t->bucketPtr, i = t->size; --i >= 0;) { 138 for (h = *hp++; h != NULL; h = nexth) { 139 nexth = h->next; 140 free((char *)h); 141 } 142 } 143 free((char *)t->bucketPtr); 144 145 /* 146 * Set up the hash table to cause memory faults on any future access 147 * attempts until re-initialization. 148 */ 149 t->bucketPtr = NULL; 150} 151 152/* 153 *--------------------------------------------------------- 154 * 155 * Hash_FindEntry -- 156 * 157 * Searches a hash table for an entry corresponding to key. 158 * 159 * Results: 160 * The return value is a pointer to the entry for key, 161 * if key was present in the table. If key was not 162 * present, NULL is returned. 163 * 164 * Side Effects: 165 * None. 166 * 167 *--------------------------------------------------------- 168 */ 169 170Hash_Entry * 171Hash_FindEntry(Hash_Table *t, char *key) 172{ 173 Hash_Entry *e; 174 unsigned h; 175 char *p; 176 177 for (h = 0, p = key; *p;) 178 h = (h << 5) - h + *p++; 179 p = key; 180 for (e = t->bucketPtr[h & t->mask]; e != NULL; e = e->next) 181 if (e->namehash == h && strcmp(e->name, p) == 0) 182 return (e); 183 return (NULL); 184} 185 186/* 187 *--------------------------------------------------------- 188 * 189 * Hash_CreateEntry -- 190 * 191 * Searches a hash table for an entry corresponding to 192 * key. If no entry is found, then one is created. 193 * 194 * Results: 195 * The return value is a pointer to the entry. If *newPtr 196 * isn't NULL, then *newPtr is filled in with TRUE if a 197 * new entry was created, and FALSE if an entry already existed 198 * with the given key. 199 * 200 * Side Effects: 201 * Memory may be allocated, and the hash buckets may be modified. 202 *--------------------------------------------------------- 203 */ 204 205Hash_Entry * 206Hash_CreateEntry(Hash_Table *t, char *key, Boolean *newPtr) 207{ 208 Hash_Entry *e; 209 unsigned int h; 210 char *p; 211 int keylen; 212 struct Hash_Entry **hp; 213 214 /* 215 * Hash the key. As a side effect, save the length (strlen) of the 216 * key in case we need to create the entry. 217 */ 218 for (h = 0, p = key; *p;) 219 h = (h << 5) - h + *p++; 220 keylen = p - key; 221 p = key; 222 for (e = t->bucketPtr[h & t->mask]; e != NULL; e = e->next) { 223 if (e->namehash == h && strcmp(e->name, p) == 0) { 224 if (newPtr != NULL) 225 *newPtr = FALSE; 226 return (e); 227 } 228 } 229 230 /* 231 * The desired entry isn't there. Before allocating a new entry, 232 * expand the table if necessary (and this changes the resulting 233 * bucket chain). 234 */ 235 if (t->numEntries >= rebuildLimit * t->size) 236 RebuildTable(t); 237 e = (Hash_Entry *)emalloc(sizeof(*e) + keylen); 238 hp = &t->bucketPtr[h & t->mask]; 239 e->next = *hp; 240 *hp = e; 241 e->clientData = NULL; 242 e->namehash = h; 243 strcpy(e->name, p); 244 t->numEntries++; 245 246 if (newPtr != NULL) 247 *newPtr = TRUE; 248 return (e); 249} 250 251/* 252 *--------------------------------------------------------- 253 * 254 * Hash_DeleteEntry -- 255 * 256 * Delete the given hash table entry and free memory associated with 257 * it. 258 * 259 * Results: 260 * None. 261 * 262 * Side Effects: 263 * Hash chain that entry lives in is modified and memory is freed. 264 * 265 *--------------------------------------------------------- 266 */ 267 268void 269Hash_DeleteEntry(Hash_Table *t, Hash_Entry *e) 270{ 271 Hash_Entry **hp, *p; 272 273 if (e == NULL) 274 return; 275 for (hp = &t->bucketPtr[e->namehash & t->mask]; 276 (p = *hp) != NULL; hp = &p->next) { 277 if (p == e) { 278 *hp = p->next; 279 free((char *)p); 280 t->numEntries--; 281 return; 282 } 283 } 284 write(STDERR_FILENO, "bad call to Hash_DeleteEntry\n", 29); 285 abort(); 286} 287 288/* 289 *--------------------------------------------------------- 290 * 291 * Hash_EnumFirst -- 292 * This procedure sets things up for a complete search 293 * of all entries recorded in the hash table. 294 * 295 * Results: 296 * The return value is the address of the first entry in 297 * the hash table, or NULL if the table is empty. 298 * 299 * Side Effects: 300 * The information in searchPtr is initialized so that successive 301 * calls to Hash_Next will return successive HashEntry's 302 * from the table. 303 * 304 *--------------------------------------------------------- 305 */ 306 307Hash_Entry * 308Hash_EnumFirst(Hash_Table *t, Hash_Search *searchPtr) 309{ 310 311 searchPtr->tablePtr = t; 312 searchPtr->nextIndex = 0; 313 searchPtr->hashEntryPtr = NULL; 314 return (Hash_EnumNext(searchPtr)); 315} 316 317/* 318 *--------------------------------------------------------- 319 * 320 * Hash_EnumNext -- 321 * This procedure returns successive entries in the hash table. 322 * 323 * Results: 324 * The return value is a pointer to the next HashEntry 325 * in the table, or NULL when the end of the table is 326 * reached. 327 * 328 * Side Effects: 329 * The information in searchPtr is modified to advance to the 330 * next entry. 331 * 332 *--------------------------------------------------------- 333 */ 334 335Hash_Entry * 336Hash_EnumNext(Hash_Search *searchPtr) 337{ 338 Hash_Entry *e; 339 Hash_Table *t = searchPtr->tablePtr; 340 341 /* 342 * The hashEntryPtr field points to the most recently returned 343 * entry, or is NULL if we are starting up. If not NULL, we have 344 * to start at the next one in the chain. 345 */ 346 e = searchPtr->hashEntryPtr; 347 if (e != NULL) 348 e = e->next; 349 /* 350 * If the chain ran out, or if we are starting up, we need to 351 * find the next nonempty chain. 352 */ 353 while (e == NULL) { 354 if (searchPtr->nextIndex >= t->size) 355 return (NULL); 356 e = t->bucketPtr[searchPtr->nextIndex++]; 357 } 358 searchPtr->hashEntryPtr = e; 359 return (e); 360} 361 362/* 363 *--------------------------------------------------------- 364 * 365 * RebuildTable -- 366 * This local routine makes a new hash table that 367 * is larger than the old one. 368 * 369 * Results: 370 * None. 371 * 372 * Side Effects: 373 * The entire hash table is moved, so any bucket numbers 374 * from the old table are invalid. 375 * 376 *--------------------------------------------------------- 377 */ 378 379static void 380RebuildTable(Hash_Table *t) 381{ 382 Hash_Entry *e, *next = NULL, **hp, **xp; 383 int i, mask; 384 Hash_Entry **oldhp; 385 int oldsize; 386 387 oldhp = t->bucketPtr; 388 oldsize = i = t->size; 389 i <<= 1; 390 t->size = i; 391 t->mask = mask = i - 1; 392 t->bucketPtr = hp = (struct Hash_Entry **)emalloc(sizeof(*hp) * i); 393 while (--i >= 0) 394 *hp++ = NULL; 395 for (hp = oldhp, i = oldsize; --i >= 0;) { 396 for (e = *hp++; e != NULL; e = next) { 397 next = e->next; 398 xp = &t->bucketPtr[e->namehash & mask]; 399 e->next = *xp; 400 *xp = e; 401 } 402 } 403 free((char *)oldhp); 404} 405