xref: /freebsd/contrib/bmake/hash.c (revision d4ae33f0721c1b170fe37d97e395228ffcfb3f80)
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 	if (t == NULL || t->bucketPtr == NULL) {
225 	    return NULL;
226 	}
227 	for (h = 0, p = key; *p;)
228 		h = (h << 5) - h + *p++;
229 	p = key;
230 	for (e = t->bucketPtr[h & t->mask]; e != NULL; e = e->next)
231 		if (e->namehash == h && strcmp(e->name, p) == 0)
232 			return (e);
233 	return NULL;
234 }
235 
236 /*
237  *---------------------------------------------------------
238  *
239  * Hash_CreateEntry --
240  *
241  *	Searches a hash table for an entry corresponding to
242  *	key.  If no entry is found, then one is created.
243  *
244  * Input:
245  *	t		Hash table to search.
246  *	key		A hash key.
247  *	newPtr		Filled in with TRUE if new entry created,
248  *			FALSE otherwise.
249  *
250  * Results:
251  *	The return value is a pointer to the entry.  If *newPtr
252  *	isn't NULL, then *newPtr is filled in with TRUE if a
253  *	new entry was created, and FALSE if an entry already existed
254  *	with the given key.
255  *
256  * Side Effects:
257  *	Memory may be allocated, and the hash buckets may be modified.
258  *---------------------------------------------------------
259  */
260 
261 Hash_Entry *
262 Hash_CreateEntry(Hash_Table *t, const char *key, Boolean *newPtr)
263 {
264 	Hash_Entry *e;
265 	unsigned h;
266 	const char *p;
267 	int keylen;
268 	struct Hash_Entry **hp;
269 
270 	/*
271 	 * Hash the key.  As a side effect, save the length (strlen) of the
272 	 * key in case we need to create the entry.
273 	 */
274 	for (h = 0, p = key; *p;)
275 		h = (h << 5) - h + *p++;
276 	keylen = p - key;
277 	p = key;
278 	for (e = t->bucketPtr[h & t->mask]; e != NULL; e = e->next) {
279 		if (e->namehash == h && strcmp(e->name, p) == 0) {
280 			if (newPtr != NULL)
281 				*newPtr = FALSE;
282 			return (e);
283 		}
284 	}
285 
286 	/*
287 	 * The desired entry isn't there.  Before allocating a new entry,
288 	 * expand the table if necessary (and this changes the resulting
289 	 * bucket chain).
290 	 */
291 	if (t->numEntries >= rebuildLimit * t->size)
292 		RebuildTable(t);
293 	e = bmake_malloc(sizeof(*e) + keylen);
294 	hp = &t->bucketPtr[h & t->mask];
295 	e->next = *hp;
296 	*hp = e;
297 	Hash_SetValue(e, NULL);
298 	e->namehash = h;
299 	(void)strcpy(e->name, p);
300 	t->numEntries++;
301 
302 	if (newPtr != NULL)
303 		*newPtr = TRUE;
304 	return (e);
305 }
306 
307 /*
308  *---------------------------------------------------------
309  *
310  * Hash_DeleteEntry --
311  *
312  * 	Delete the given hash table entry and free memory associated with
313  *	it.
314  *
315  * Results:
316  *	None.
317  *
318  * Side Effects:
319  *	Hash chain that entry lives in is modified and memory is freed.
320  *
321  *---------------------------------------------------------
322  */
323 
324 void
325 Hash_DeleteEntry(Hash_Table *t, Hash_Entry *e)
326 {
327 	Hash_Entry **hp, *p;
328 
329 	if (e == NULL)
330 		return;
331 	for (hp = &t->bucketPtr[e->namehash & t->mask];
332 	     (p = *hp) != NULL; hp = &p->next) {
333 		if (p == e) {
334 			*hp = p->next;
335 			free(p);
336 			t->numEntries--;
337 			return;
338 		}
339 	}
340 	(void)write(2, "bad call to Hash_DeleteEntry\n", 29);
341 	abort();
342 }
343 
344 /*
345  *---------------------------------------------------------
346  *
347  * Hash_EnumFirst --
348  *	This procedure sets things up for a complete search
349  *	of all entries recorded in the hash table.
350  *
351  * Input:
352  *	t		Table to be searched.
353  *	searchPtr	Area in which to keep state about search.
354  *
355  * Results:
356  *	The return value is the address of the first entry in
357  *	the hash table, or NULL if the table is empty.
358  *
359  * Side Effects:
360  *	The information in searchPtr is initialized so that successive
361  *	calls to Hash_Next will return successive HashEntry's
362  *	from the table.
363  *
364  *---------------------------------------------------------
365  */
366 
367 Hash_Entry *
368 Hash_EnumFirst(Hash_Table *t, Hash_Search *searchPtr)
369 {
370 	searchPtr->tablePtr = t;
371 	searchPtr->nextIndex = 0;
372 	searchPtr->hashEntryPtr = NULL;
373 	return Hash_EnumNext(searchPtr);
374 }
375 
376 /*
377  *---------------------------------------------------------
378  *
379  * Hash_EnumNext --
380  *    This procedure returns successive entries in the hash table.
381  *
382  * Input:
383  *	searchPtr	Area used to keep state about search.
384  *
385  * Results:
386  *    The return value is a pointer to the next HashEntry
387  *    in the table, or NULL when the end of the table is
388  *    reached.
389  *
390  * Side Effects:
391  *    The information in searchPtr is modified to advance to the
392  *    next entry.
393  *
394  *---------------------------------------------------------
395  */
396 
397 Hash_Entry *
398 Hash_EnumNext(Hash_Search *searchPtr)
399 {
400 	Hash_Entry *e;
401 	Hash_Table *t = searchPtr->tablePtr;
402 
403 	/*
404 	 * The hashEntryPtr field points to the most recently returned
405 	 * entry, or is nil if we are starting up.  If not nil, we have
406 	 * to start at the next one in the chain.
407 	 */
408 	e = searchPtr->hashEntryPtr;
409 	if (e != NULL)
410 		e = e->next;
411 	/*
412 	 * If the chain ran out, or if we are starting up, we need to
413 	 * find the next nonempty chain.
414 	 */
415 	while (e == NULL) {
416 		if (searchPtr->nextIndex >= t->size)
417 			return NULL;
418 		e = t->bucketPtr[searchPtr->nextIndex++];
419 	}
420 	searchPtr->hashEntryPtr = e;
421 	return (e);
422 }
423 
424 /*
425  *---------------------------------------------------------
426  *
427  * RebuildTable --
428  *	This local routine makes a new hash table that
429  *	is larger than the old one.
430  *
431  * Results:
432  * 	None.
433  *
434  * Side Effects:
435  *	The entire hash table is moved, so any bucket numbers
436  *	from the old table are invalid.
437  *
438  *---------------------------------------------------------
439  */
440 
441 static void
442 RebuildTable(Hash_Table *t)
443 {
444 	Hash_Entry *e, *next = NULL, **hp, **xp;
445 	int i, mask;
446         Hash_Entry **oldhp;
447 	int oldsize;
448 
449 	oldhp = t->bucketPtr;
450 	oldsize = i = t->size;
451 	i <<= 1;
452 	t->size = i;
453 	t->mask = mask = i - 1;
454 	t->bucketPtr = hp = bmake_malloc(sizeof(*hp) * i);
455 	while (--i >= 0)
456 		*hp++ = NULL;
457 	for (hp = oldhp, i = oldsize; --i >= 0;) {
458 		for (e = *hp++; e != NULL; e = next) {
459 			next = e->next;
460 			xp = &t->bucketPtr[e->namehash & mask];
461 			e->next = *xp;
462 			*xp = e;
463 		}
464 	}
465 	free(oldhp);
466 }
467