xref: /freebsd/contrib/bmake/hash.c (revision 2710751bc309af25c6dea1171781678258e83840)
1 /*	$NetBSD: hash.c,v 1.19 2009/01/24 10:59:09 dsl Exp $	*/
2 
3 /*
4  * Copyright (c) 1988, 1989, 1990 The Regents of the University of California.
5  * All rights reserved.
6  *
7  * This code is derived from software contributed to Berkeley by
8  * Adam de Boor.
9  *
10  * Redistribution and use in source and binary forms, with or without
11  * modification, are permitted provided that the following conditions
12  * are met:
13  * 1. Redistributions of source code must retain the above copyright
14  *    notice, this list of conditions and the following disclaimer.
15  * 2. Redistributions in binary form must reproduce the above copyright
16  *    notice, this list of conditions and the following disclaimer in the
17  *    documentation and/or other materials provided with the distribution.
18  * 3. Neither the name of the University nor the names of its contributors
19  *    may be used to endorse or promote products derived from this software
20  *    without specific prior written permission.
21  *
22  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32  * SUCH DAMAGE.
33  */
34 
35 /*
36  * Copyright (c) 1988, 1989 by Adam de Boor
37  * Copyright (c) 1989 by Berkeley Softworks
38  * All rights reserved.
39  *
40  * This code is derived from software contributed to Berkeley by
41  * Adam de Boor.
42  *
43  * Redistribution and use in source and binary forms, with or without
44  * modification, are permitted provided that the following conditions
45  * are met:
46  * 1. Redistributions of source code must retain the above copyright
47  *    notice, this list of conditions and the following disclaimer.
48  * 2. Redistributions in binary form must reproduce the above copyright
49  *    notice, this list of conditions and the following disclaimer in the
50  *    documentation and/or other materials provided with the distribution.
51  * 3. All advertising materials mentioning features or use of this software
52  *    must display the following acknowledgement:
53  *	This product includes software developed by the University of
54  *	California, Berkeley and its contributors.
55  * 4. Neither the name of the University nor the names of its contributors
56  *    may be used to endorse or promote products derived from this software
57  *    without specific prior written permission.
58  *
59  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
60  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
61  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
62  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
63  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
64  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
65  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
66  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
67  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
68  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
69  * SUCH DAMAGE.
70  */
71 
72 #ifndef MAKE_NATIVE
73 static char rcsid[] = "$NetBSD: hash.c,v 1.19 2009/01/24 10:59:09 dsl Exp $";
74 #else
75 #include <sys/cdefs.h>
76 #ifndef lint
77 #if 0
78 static char sccsid[] = "@(#)hash.c	8.1 (Berkeley) 6/6/93";
79 #else
80 __RCSID("$NetBSD: hash.c,v 1.19 2009/01/24 10:59:09 dsl Exp $");
81 #endif
82 #endif /* not lint */
83 #endif
84 
85 /* hash.c --
86  *
87  * 	This module contains routines to manipulate a hash table.
88  * 	See hash.h for a definition of the structure of the hash
89  * 	table.  Hash tables grow automatically as the amount of
90  * 	information increases.
91  */
92 #include "sprite.h"
93 #include "make.h"
94 #include "hash.h"
95 
96 /*
97  * Forward references to local procedures that are used before they're
98  * defined:
99  */
100 
101 static void RebuildTable(Hash_Table *);
102 
103 /*
104  * The following defines the ratio of # entries to # buckets
105  * at which we rebuild the table to make it larger.
106  */
107 
108 #define rebuildLimit 3
109 
110 /*
111  *---------------------------------------------------------
112  *
113  * Hash_InitTable --
114  *
115  *	This routine just sets up the hash table.
116  *
117  * Input:
118  *	t		Structure to to hold table.
119  *	numBuckets	How many buckets to create for starters. This
120  *			number is rounded up to a power of two.   If
121  *			<= 0, a reasonable default is chosen. The
122  *			table will grow in size later as needed.
123  *
124  * Results:
125  *	None.
126  *
127  * Side Effects:
128  *	Memory is allocated for the initial bucket area.
129  *
130  *---------------------------------------------------------
131  */
132 
133 void
134 Hash_InitTable(Hash_Table *t, int numBuckets)
135 {
136 	int i;
137 	struct Hash_Entry **hp;
138 
139 	/*
140 	 * Round up the size to a power of two.
141 	 */
142 	if (numBuckets <= 0)
143 		i = 16;
144 	else {
145 		for (i = 2; i < numBuckets; i <<= 1)
146 			 continue;
147 	}
148 	t->numEntries = 0;
149 	t->size = i;
150 	t->mask = i - 1;
151 	t->bucketPtr = hp = bmake_malloc(sizeof(*hp) * i);
152 	while (--i >= 0)
153 		*hp++ = NULL;
154 }
155 
156 /*
157  *---------------------------------------------------------
158  *
159  * Hash_DeleteTable --
160  *
161  *	This routine removes everything from a hash table
162  *	and frees up the memory space it occupied (except for
163  *	the space in the Hash_Table structure).
164  *
165  * Results:
166  *	None.
167  *
168  * Side Effects:
169  *	Lots of memory is freed up.
170  *
171  *---------------------------------------------------------
172  */
173 
174 void
175 Hash_DeleteTable(Hash_Table *t)
176 {
177 	struct Hash_Entry **hp, *h, *nexth = NULL;
178 	int i;
179 
180 	for (hp = t->bucketPtr, i = t->size; --i >= 0;) {
181 		for (h = *hp++; h != NULL; h = nexth) {
182 			nexth = h->next;
183 			free(h);
184 		}
185 	}
186 	free(t->bucketPtr);
187 
188 	/*
189 	 * Set up the hash table to cause memory faults on any future access
190 	 * attempts until re-initialization.
191 	 */
192 	t->bucketPtr = NULL;
193 }
194 
195 /*
196  *---------------------------------------------------------
197  *
198  * Hash_FindEntry --
199  *
200  * 	Searches a hash table for an entry corresponding to key.
201  *
202  * Input:
203  *	t		Hash table to search.
204  *	key		A hash key.
205  *
206  * Results:
207  *	The return value is a pointer to the entry for key,
208  *	if key was present in the table.  If key was not
209  *	present, NULL is returned.
210  *
211  * Side Effects:
212  *	None.
213  *
214  *---------------------------------------------------------
215  */
216 
217 Hash_Entry *
218 Hash_FindEntry(Hash_Table *t, const char *key)
219 {
220 	Hash_Entry *e;
221 	unsigned h;
222 	const char *p;
223 
224 	for (h = 0, p = key; *p;)
225 		h = (h << 5) - h + *p++;
226 	p = key;
227 	for (e = t->bucketPtr[h & t->mask]; e != NULL; e = e->next)
228 		if (e->namehash == h && strcmp(e->name, p) == 0)
229 			return (e);
230 	return NULL;
231 }
232 
233 /*
234  *---------------------------------------------------------
235  *
236  * Hash_CreateEntry --
237  *
238  *	Searches a hash table for an entry corresponding to
239  *	key.  If no entry is found, then one is created.
240  *
241  * Input:
242  *	t		Hash table to search.
243  *	key		A hash key.
244  *	newPtr		Filled in with TRUE if new entry created,
245  *			FALSE otherwise.
246  *
247  * Results:
248  *	The return value is a pointer to the entry.  If *newPtr
249  *	isn't NULL, then *newPtr is filled in with TRUE if a
250  *	new entry was created, and FALSE if an entry already existed
251  *	with the given key.
252  *
253  * Side Effects:
254  *	Memory may be allocated, and the hash buckets may be modified.
255  *---------------------------------------------------------
256  */
257 
258 Hash_Entry *
259 Hash_CreateEntry(Hash_Table *t, const char *key, Boolean *newPtr)
260 {
261 	Hash_Entry *e;
262 	unsigned h;
263 	const char *p;
264 	int keylen;
265 	struct Hash_Entry **hp;
266 
267 	/*
268 	 * Hash the key.  As a side effect, save the length (strlen) of the
269 	 * key in case we need to create the entry.
270 	 */
271 	for (h = 0, p = key; *p;)
272 		h = (h << 5) - h + *p++;
273 	keylen = p - key;
274 	p = key;
275 	for (e = t->bucketPtr[h & t->mask]; e != NULL; e = e->next) {
276 		if (e->namehash == h && strcmp(e->name, p) == 0) {
277 			if (newPtr != NULL)
278 				*newPtr = FALSE;
279 			return (e);
280 		}
281 	}
282 
283 	/*
284 	 * The desired entry isn't there.  Before allocating a new entry,
285 	 * expand the table if necessary (and this changes the resulting
286 	 * bucket chain).
287 	 */
288 	if (t->numEntries >= rebuildLimit * t->size)
289 		RebuildTable(t);
290 	e = bmake_malloc(sizeof(*e) + keylen);
291 	hp = &t->bucketPtr[h & t->mask];
292 	e->next = *hp;
293 	*hp = e;
294 	Hash_SetValue(e, NULL);
295 	e->namehash = h;
296 	(void)strcpy(e->name, p);
297 	t->numEntries++;
298 
299 	if (newPtr != NULL)
300 		*newPtr = TRUE;
301 	return (e);
302 }
303 
304 /*
305  *---------------------------------------------------------
306  *
307  * Hash_DeleteEntry --
308  *
309  * 	Delete the given hash table entry and free memory associated with
310  *	it.
311  *
312  * Results:
313  *	None.
314  *
315  * Side Effects:
316  *	Hash chain that entry lives in is modified and memory is freed.
317  *
318  *---------------------------------------------------------
319  */
320 
321 void
322 Hash_DeleteEntry(Hash_Table *t, Hash_Entry *e)
323 {
324 	Hash_Entry **hp, *p;
325 
326 	if (e == NULL)
327 		return;
328 	for (hp = &t->bucketPtr[e->namehash & t->mask];
329 	     (p = *hp) != NULL; hp = &p->next) {
330 		if (p == e) {
331 			*hp = p->next;
332 			free(p);
333 			t->numEntries--;
334 			return;
335 		}
336 	}
337 	(void)write(2, "bad call to Hash_DeleteEntry\n", 29);
338 	abort();
339 }
340 
341 /*
342  *---------------------------------------------------------
343  *
344  * Hash_EnumFirst --
345  *	This procedure sets things up for a complete search
346  *	of all entries recorded in the hash table.
347  *
348  * Input:
349  *	t		Table to be searched.
350  *	searchPtr	Area in which to keep state about search.
351  *
352  * Results:
353  *	The return value is the address of the first entry in
354  *	the hash table, or NULL if the table is empty.
355  *
356  * Side Effects:
357  *	The information in searchPtr is initialized so that successive
358  *	calls to Hash_Next will return successive HashEntry's
359  *	from the table.
360  *
361  *---------------------------------------------------------
362  */
363 
364 Hash_Entry *
365 Hash_EnumFirst(Hash_Table *t, Hash_Search *searchPtr)
366 {
367 	searchPtr->tablePtr = t;
368 	searchPtr->nextIndex = 0;
369 	searchPtr->hashEntryPtr = NULL;
370 	return Hash_EnumNext(searchPtr);
371 }
372 
373 /*
374  *---------------------------------------------------------
375  *
376  * Hash_EnumNext --
377  *    This procedure returns successive entries in the hash table.
378  *
379  * Input:
380  *	searchPtr	Area used to keep state about search.
381  *
382  * Results:
383  *    The return value is a pointer to the next HashEntry
384  *    in the table, or NULL when the end of the table is
385  *    reached.
386  *
387  * Side Effects:
388  *    The information in searchPtr is modified to advance to the
389  *    next entry.
390  *
391  *---------------------------------------------------------
392  */
393 
394 Hash_Entry *
395 Hash_EnumNext(Hash_Search *searchPtr)
396 {
397 	Hash_Entry *e;
398 	Hash_Table *t = searchPtr->tablePtr;
399 
400 	/*
401 	 * The hashEntryPtr field points to the most recently returned
402 	 * entry, or is nil if we are starting up.  If not nil, we have
403 	 * to start at the next one in the chain.
404 	 */
405 	e = searchPtr->hashEntryPtr;
406 	if (e != NULL)
407 		e = e->next;
408 	/*
409 	 * If the chain ran out, or if we are starting up, we need to
410 	 * find the next nonempty chain.
411 	 */
412 	while (e == NULL) {
413 		if (searchPtr->nextIndex >= t->size)
414 			return NULL;
415 		e = t->bucketPtr[searchPtr->nextIndex++];
416 	}
417 	searchPtr->hashEntryPtr = e;
418 	return (e);
419 }
420 
421 /*
422  *---------------------------------------------------------
423  *
424  * RebuildTable --
425  *	This local routine makes a new hash table that
426  *	is larger than the old one.
427  *
428  * Results:
429  * 	None.
430  *
431  * Side Effects:
432  *	The entire hash table is moved, so any bucket numbers
433  *	from the old table are invalid.
434  *
435  *---------------------------------------------------------
436  */
437 
438 static void
439 RebuildTable(Hash_Table *t)
440 {
441 	Hash_Entry *e, *next = NULL, **hp, **xp;
442 	int i, mask;
443         Hash_Entry **oldhp;
444 	int oldsize;
445 
446 	oldhp = t->bucketPtr;
447 	oldsize = i = t->size;
448 	i <<= 1;
449 	t->size = i;
450 	t->mask = mask = i - 1;
451 	t->bucketPtr = hp = bmake_malloc(sizeof(*hp) * i);
452 	while (--i >= 0)
453 		*hp++ = NULL;
454 	for (hp = oldhp, i = oldsize; --i >= 0;) {
455 		for (e = *hp++; e != NULL; e = next) {
456 			next = e->next;
457 			xp = &t->bucketPtr[e->namehash & mask];
458 			e->next = *xp;
459 			*xp = e;
460 		}
461 	}
462 	free(oldhp);
463 }
464