usr.bin/make/hash.c

   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  * $FreeBSD: src/usr.bin/make/hash.c,v 1.24 2005/02/01 10:50:35 harti Exp $
  41  * $DragonFly: src/usr.bin/make/hash.c,v 1.21 2005/08/04 00:19:04 okumoto Exp $
  42  */
  43
  44 /* hash.c --
  45  *
  46  *      This module contains routines to manipulate a hash table.
  47  *      See hash.h for a definition of the structure of the hash
  48  *      table.  Hash tables grow automatically as the amount of
  49  *      information increases.
  50  */
  51
  52 #include <stdlib.h>
  53 #include <string.h>
  54 #include <unistd.h>
  55
  56 #include "hash.h"
  57 #include "util.h"
  58
  59 /*
  60  * Forward references to local procedures that are used before they're
  61  * defined:
  62  */
  63 static void RebuildTable(Hash_Table *);
  64
  65 /*
  66  * The following defines the ratio of # entries to # buckets
  67  * at which we rebuild the table to make it larger.
  68  */
  69
  70 #define rebuildLimit 8
  71
  72 /*
  73  *---------------------------------------------------------
  74  *
  75  * Hash_InitTable --
  76  *
  77  *      This routine just sets up the hash table.
  78  *
  79  * Input:
  80  *      t               Structure to to hold table.
  81  *      numBuckets      How many buckets to create for starters. This
  82  *                      number is rounded up to a power of two.   If
  83  *                      <= 0, a reasonable default is chosen. The
  84  *                      table will grow in size later as needed.
  85  *
  86  * Results:
  87  *      None.
  88  *
  89  * Side Effects:
  90  *      Memory is allocated for the initial bucket area.
  91  *
  92  *---------------------------------------------------------
  93  */
  94 void
  95 Hash_InitTable(Hash_Table *t, int numBuckets)
  96 {
  97         int i;
  98         struct Hash_Entry **hp;
  99
 100         /*
 101          * Round up the size to a power of two.
 102          */
 103         if (numBuckets <= 0)
 104                 i = 16;
 105         else {
 106                 for (i = 2; i < numBuckets; i <<= 1)
 107                          continue;
 108         }
 109         t->numEntries = 0;
 110         t->size = i;
 111         t->mask = i - 1;
 112         t->bucketPtr = hp = emalloc(sizeof(*hp) * i);
 113         while (--i >= 0)
 114                 *hp++ = NULL;
 115 }
 116
 117 /*
 118  *---------------------------------------------------------
 119  *
 120  * Hash_DeleteTable --
 121  *
 122  *      This routine removes everything from a hash table
 123  *      and frees up the memory space it occupied (except for
 124  *      the space in the Hash_Table structure).
 125  *
 126  * Results:
 127  *      None.
 128  *
 129  * Side Effects:
 130  *      Lots of memory is freed up.
 131  *
 132  *---------------------------------------------------------
 133  */
 134 void
 135 Hash_DeleteTable(Hash_Table *t)
 136 {
 137         struct Hash_Entry **hp, *h, *nexth = NULL;
 138         int i;
 139
 140         for (hp = t->bucketPtr, i = t->size; --i >= 0;) {
 141                 for (h = *hp++; h != NULL; h = nexth) {
 142                         nexth = h->next;
 143                         free(h);
 144                 }
 145         }
 146         free(t->bucketPtr);
 147
 148         /*
 149          * Set up the hash table to cause memory faults on any future access
 150          * attempts until re-initialization.
 151          */
 152         t->bucketPtr = NULL;
 153 }
 154
 155 /*
 156  *---------------------------------------------------------
 157  *
 158  * Hash_FindEntry --
 159  *
 160  *      Searches a hash table for an entry corresponding to key.
 161  *
 162  * Input:
 163  *      t               Hash table to search.
 164  *      key             A hash key.
 165  *
 166  * Results:
 167  *      The return value is a pointer to the entry for key,
 168  *      if key was present in the table.  If key was not
 169  *      present, NULL is returned.
 170  *
 171  * Side Effects:
 172  *      None.
 173  *
 174  *---------------------------------------------------------
 175  */
 176 Hash_Entry *
 177 Hash_FindEntry(const Hash_Table *t, const char *key)
 178 {
 179         Hash_Entry *e;
 180         unsigned h;
 181         const char *p;
 182
 183         for (h = 0, p = key; *p;)
 184                 h = (h << 5) - h + *p++;
 185         p = key;
 186         for (e = t->bucketPtr[h & t->mask]; e != NULL; e = e->next)
 187                 if (e->namehash == h && strcmp(e->name, p) == 0)
 188                         return (e);
 189         return (NULL);
 190 }
 191
 192 /*
 193  *---------------------------------------------------------
 194  *
 195  * Hash_CreateEntry --
 196  *
 197  *      Searches a hash table for an entry corresponding to
 198  *      key.  If no entry is found, then one is created.
 199  *
 200  * Input:
 201  *      t               Hash table to search.
 202  *      key             A hash key.
 203  *      newPtr          Filled in with true if new entry created,
 204  *                      FALSE otherwise.
 205  *
 206  * Results:
 207  *      The return value is a pointer to the entry.  If *newPtr
 208  *      isn't NULL, then *newPtr is filled in with true if a
 209  *      new entry was created, and false if an entry already existed
 210  *      with the given key.
 211  *
 212  * Side Effects:
 213  *      Memory may be allocated, and the hash buckets may be modified.
 214  *---------------------------------------------------------
 215  */
 216 Hash_Entry *
 217 Hash_CreateEntry(Hash_Table *t, const char *key, bool *newPtr)
 218 {
 219         Hash_Entry *e;
 220         unsigned int h;
 221         const char *p;
 222         int keylen;
 223         struct Hash_Entry **hp;
 224
 225         /*
 226          * Hash the key.  As a side effect, save the length (strlen) of the
 227          * key in case we need to create the entry.
 228          */
 229         for (h = 0, p = key; *p;)
 230                 h = (h << 5) - h + *p++;
 231         keylen = p - key;
 232         p = key;
 233         for (e = t->bucketPtr[h & t->mask]; e != NULL; e = e->next) {
 234                 if (e->namehash == h && strcmp(e->name, p) == 0) {
 235                         if (newPtr != NULL)
 236                                 *newPtr = false;
 237                         return (e);
 238                 }
 239         }
 240
 241         /*
 242          * The desired entry isn't there.  Before allocating a new entry,
 243          * expand the table if necessary (and this changes the resulting
 244          * bucket chain).
 245          */
 246         if (t->numEntries >= rebuildLimit * t->size)
 247                 RebuildTable(t);
 248         e = emalloc(sizeof(*e) + keylen);
 249         hp = &t->bucketPtr[h & t->mask];
 250         e->next = *hp;
 251         *hp = e;
 252         e->clientData = NULL;
 253         e->namehash = h;
 254         strcpy(e->name, p);
 255         t->numEntries++;
 256
 257         if (newPtr != NULL)
 258                 *newPtr = true;
 259         return (e);
 260 }
 261
 262 /*
 263  *---------------------------------------------------------
 264  *
 265  * Hash_DeleteEntry --
 266  *
 267  *      Delete the given hash table entry and free memory associated with
 268  *      it.
 269  *
 270  * Results:
 271  *      None.
 272  *
 273  * Side Effects:
 274  *      Hash chain that entry lives in is modified and memory is freed.
 275  *
 276  *---------------------------------------------------------
 277  */
 278 void
 279 Hash_DeleteEntry(Hash_Table *t, Hash_Entry *e)
 280 {
 281         Hash_Entry **hp, *p;
 282
 283         if (e == NULL)
 284                 return;
 285         for (hp = &t->bucketPtr[e->namehash & t->mask];
 286              (p = *hp) != NULL; hp = &p->next) {
 287                 if (p == e) {
 288                         *hp = p->next;
 289                         free(p);
 290                         t->numEntries--;
 291                         return;
 292                 }
 293         }
 294         write(STDERR_FILENO, "bad call to Hash_DeleteEntry\n", 29);
 295         abort();
 296 }
 297
 298 /*
 299  *---------------------------------------------------------
 300  *
 301  * Hash_EnumFirst --
 302  *      This procedure sets things up for a complete search
 303  *      of all entries recorded in the hash table.
 304  *
 305  * Input:
 306  *      t               Table to be searched.
 307  *      searchPtr       Area in which to keep state about search.
 308  *
 309  * Results:
 310  *      The return value is the address of the first entry in
 311  *      the hash table, or NULL if the table is empty.
 312  *
 313  * Side Effects:
 314  *      The information in searchPtr is initialized so that successive
 315  *      calls to Hash_Next will return successive HashEntry's
 316  *      from the table.
 317  *
 318  *---------------------------------------------------------
 319  */
 320 Hash_Entry *
 321 Hash_EnumFirst(const Hash_Table *t, Hash_Search *searchPtr)
 322 {
 323
 324         searchPtr->tablePtr = t;
 325         searchPtr->nextIndex = 0;
 326         searchPtr->hashEntryPtr = NULL;
 327         return (Hash_EnumNext(searchPtr));
 328 }
 329
 330 /*
 331  *---------------------------------------------------------
 332  *
 333  * Hash_EnumNext --
 334  *    This procedure returns successive entries in the hash table.
 335  *
 336  * Input:
 337  *      searchPtr       Area used to keep state about search.
 338  *
 339  * Results:
 340  *    The return value is a pointer to the next HashEntry
 341  *    in the table, or NULL when the end of the table is
 342  *    reached.
 343  *
 344  * Side Effects:
 345  *    The information in searchPtr is modified to advance to the
 346  *    next entry.
 347  *
 348  *---------------------------------------------------------
 349  */
 350 Hash_Entry *
 351 Hash_EnumNext(Hash_Search *searchPtr)
 352 {
 353         Hash_Entry *e;
 354         const Hash_Table *t = searchPtr->tablePtr;
 355
 356         /*
 357          * The hashEntryPtr field points to the most recently returned
 358          * entry, or is NULL if we are starting up.  If not NULL, we have
 359          * to start at the next one in the chain.
 360          */
 361         e = searchPtr->hashEntryPtr;
 362         if (e != NULL)
 363                 e = e->next;
 364         /*
 365          * If the chain ran out, or if we are starting up, we need to
 366          * find the next nonempty chain.
 367          */
 368         while (e == NULL) {
 369                 if (searchPtr->nextIndex >= t->size)
 370                         return (NULL);
 371                 e = t->bucketPtr[searchPtr->nextIndex++];
 372         }
 373         searchPtr->hashEntryPtr = e;
 374         return (e);
 375 }
 376
 377 /*
 378  *---------------------------------------------------------
 379  *
 380  * RebuildTable --
 381  *      This local routine makes a new hash table that
 382  *      is larger than the old one.
 383  *
 384  * Results:
 385  *      None.
 386  *
 387  * Side Effects:
 388  *      The entire hash table is moved, so any bucket numbers
 389  *      from the old table are invalid.
 390  *
 391  *---------------------------------------------------------
 392  */
 393 static void
 394 RebuildTable(Hash_Table *t)
 395 {
 396         Hash_Entry *e, *next = NULL, **hp, **xp;
 397         int i, mask;
 398         Hash_Entry **oldhp;
 399         int oldsize;
 400
 401         oldhp = t->bucketPtr;
 402         oldsize = i = t->size;
 403         i <<= 1;
 404         t->size = i;
 405         t->mask = mask = i - 1;
 406         t->bucketPtr = hp = emalloc(sizeof(*hp) * i);
 407         while (--i >= 0)
 408                 *hp++ = NULL;
 409         for (hp = oldhp, i = oldsize; --i >= 0;) {
 410                 for (e = *hp++; e != NULL; e = next) {
 411                         next = e->next;
 412                         xp = &t->bucketPtr[e->namehash & mask];
 413                         e->next = *xp;
 414                         *xp = e;
 415                 }
 416         }
 417         free(oldhp);
 418 }