xref: /freebsd/contrib/bmake/hash.c (revision cab6a39d7b343596a5823e65c0f7b426551ec22d)
1 /*	$NetBSD: hash.c,v 1.64 2021/04/11 12:46:54 rillig 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 /* Hash tables with string keys. */
73 
74 #include "make.h"
75 
76 /*	"@(#)hash.c	8.1 (Berkeley) 6/6/93"	*/
77 MAKE_RCSID("$NetBSD: hash.c,v 1.64 2021/04/11 12:46:54 rillig Exp $");
78 
79 /*
80  * The ratio of # entries to # buckets at which we rebuild the table to
81  * make it larger.
82  */
83 #define rebuildLimit 3
84 
85 /* This hash function matches Gosling's Emacs and java.lang.String. */
86 static unsigned int
87 Hash_String(const char *key, size_t *out_keylen)
88 {
89 	unsigned int h;
90 	const char *p;
91 
92 	h = 0;
93 	for (p = key; *p != '\0'; p++)
94 		h = 31 * h + (unsigned char)*p;
95 
96 	if (out_keylen != NULL)
97 		*out_keylen = (size_t)(p - key);
98 	return h;
99 }
100 
101 /* This hash function matches Gosling's Emacs and java.lang.String. */
102 unsigned int
103 Hash_Substring(Substring key)
104 {
105 	unsigned int h;
106 	const char *p;
107 
108 	h = 0;
109 	for (p = key.start; p != key.end; p++)
110 		h = 31 * h + (unsigned char)*p;
111 	return h;
112 }
113 
114 static HashEntry *
115 HashTable_Find(HashTable *t, unsigned int h, const char *key)
116 {
117 	HashEntry *e;
118 	unsigned int chainlen = 0;
119 
120 #ifdef DEBUG_HASH_LOOKUP
121 	DEBUG4(HASH, "%s: %p h=%08x key=%s\n", __func__, t, h, key);
122 #endif
123 
124 	for (e = t->buckets[h & t->bucketsMask]; e != NULL; e = e->next) {
125 		chainlen++;
126 		if (e->key_hash == h && strcmp(e->key, key) == 0)
127 			break;
128 	}
129 
130 	if (chainlen > t->maxchain)
131 		t->maxchain = chainlen;
132 
133 	return e;
134 }
135 
136 static bool
137 HashEntry_KeyEquals(const HashEntry *he, Substring key)
138 {
139 	const char *heKey, *p;
140 
141 	heKey = he->key;
142 	for (p = key.start; p != key.end; p++, heKey++)
143 		if (*p != *heKey || *heKey == '\0')
144 			return false;
145 	return *heKey == '\0';
146 }
147 
148 static HashEntry *
149 HashTable_FindEntryBySubstring(HashTable *t, Substring key, unsigned int h)
150 {
151 	HashEntry *e;
152 	unsigned int chainlen = 0;
153 
154 #ifdef DEBUG_HASH_LOOKUP
155 	DEBUG4(HASH, "%s: %p h=%08x key=%.*s\n", __func__, t, h,
156 	    (int)Substring_Length(key), key.start);
157 #endif
158 
159 	for (e = t->buckets[h & t->bucketsMask]; e != NULL; e = e->next) {
160 		chainlen++;
161 		if (e->key_hash == h && HashEntry_KeyEquals(e, key))
162 			break;
163 	}
164 
165 	if (chainlen > t->maxchain)
166 		t->maxchain = chainlen;
167 
168 	return e;
169 }
170 
171 /* Set up the hash table. */
172 void
173 HashTable_Init(HashTable *t)
174 {
175 	unsigned int n = 16, i;
176 	HashEntry **buckets = bmake_malloc(sizeof *buckets * n);
177 	for (i = 0; i < n; i++)
178 		buckets[i] = NULL;
179 
180 	t->buckets = buckets;
181 	t->bucketsSize = n;
182 	t->numEntries = 0;
183 	t->bucketsMask = n - 1;
184 	t->maxchain = 0;
185 }
186 
187 /*
188  * Remove everything from the hash table and free up the memory for the keys
189  * of the hash table, but not for the values associated to these keys.
190  */
191 void
192 HashTable_Done(HashTable *t)
193 {
194 	HashEntry **buckets = t->buckets;
195 	size_t i, n = t->bucketsSize;
196 
197 	for (i = 0; i < n; i++) {
198 		HashEntry *he = buckets[i];
199 		while (he != NULL) {
200 			HashEntry *next = he->next;
201 			free(he);
202 			he = next;
203 		}
204 	}
205 
206 	free(t->buckets);
207 #ifdef CLEANUP
208 	t->buckets = NULL;
209 #endif
210 }
211 
212 /* Find the entry corresponding to the key, or return NULL. */
213 HashEntry *
214 HashTable_FindEntry(HashTable *t, const char *key)
215 {
216 	unsigned int h = Hash_String(key, NULL);
217 	return HashTable_Find(t, h, key);
218 }
219 
220 /* Find the value corresponding to the key, or return NULL. */
221 void *
222 HashTable_FindValue(HashTable *t, const char *key)
223 {
224 	HashEntry *he = HashTable_FindEntry(t, key);
225 	return he != NULL ? he->value : NULL;
226 }
227 
228 /*
229  * Find the value corresponding to the key and the precomputed hash,
230  * or return NULL.
231  */
232 void *
233 HashTable_FindValueBySubstringHash(HashTable *t, Substring key, unsigned int h)
234 {
235 	HashEntry *he = HashTable_FindEntryBySubstring(t, key, h);
236 	return he != NULL ? he->value : NULL;
237 }
238 
239 /*
240  * Make the hash table larger. Any bucket numbers from the old table become
241  * invalid; the hash codes stay valid though.
242  */
243 static void
244 HashTable_Enlarge(HashTable *t)
245 {
246 	unsigned int oldSize = t->bucketsSize;
247 	HashEntry **oldBuckets = t->buckets;
248 	unsigned int newSize = 2 * oldSize;
249 	unsigned int newMask = newSize - 1;
250 	HashEntry **newBuckets = bmake_malloc(sizeof *newBuckets * newSize);
251 	size_t i;
252 
253 	for (i = 0; i < newSize; i++)
254 		newBuckets[i] = NULL;
255 
256 	for (i = 0; i < oldSize; i++) {
257 		HashEntry *he = oldBuckets[i];
258 		while (he != NULL) {
259 			HashEntry *next = he->next;
260 			he->next = newBuckets[he->key_hash & newMask];
261 			newBuckets[he->key_hash & newMask] = he;
262 			he = next;
263 		}
264 	}
265 
266 	free(oldBuckets);
267 
268 	t->bucketsSize = newSize;
269 	t->bucketsMask = newMask;
270 	t->buckets = newBuckets;
271 	DEBUG5(HASH, "%s: %p size=%d entries=%d maxchain=%d\n",
272 	    __func__, (void *)t, t->bucketsSize, t->numEntries, t->maxchain);
273 	t->maxchain = 0;
274 }
275 
276 /*
277  * Find or create an entry corresponding to the key.
278  * Return in out_isNew whether a new entry has been created.
279  */
280 HashEntry *
281 HashTable_CreateEntry(HashTable *t, const char *key, bool *out_isNew)
282 {
283 	size_t keylen;
284 	unsigned int h = Hash_String(key, &keylen);
285 	HashEntry *he = HashTable_Find(t, h, key);
286 
287 	if (he != NULL) {
288 		if (out_isNew != NULL)
289 			*out_isNew = false;
290 		return he;
291 	}
292 
293 	if (t->numEntries >= rebuildLimit * t->bucketsSize)
294 		HashTable_Enlarge(t);
295 
296 	he = bmake_malloc(sizeof *he + keylen);
297 	he->value = NULL;
298 	he->key_hash = h;
299 	memcpy(he->key, key, keylen + 1);
300 
301 	he->next = t->buckets[h & t->bucketsMask];
302 	t->buckets[h & t->bucketsMask] = he;
303 	t->numEntries++;
304 
305 	if (out_isNew != NULL)
306 		*out_isNew = true;
307 	return he;
308 }
309 
310 HashEntry *
311 HashTable_Set(HashTable *t, const char *key, void *value)
312 {
313 	HashEntry *he = HashTable_CreateEntry(t, key, NULL);
314 	HashEntry_Set(he, value);
315 	return he;
316 }
317 
318 /* Delete the entry from the table and free the associated memory. */
319 void
320 HashTable_DeleteEntry(HashTable *t, HashEntry *he)
321 {
322 	HashEntry **ref = &t->buckets[he->key_hash & t->bucketsMask];
323 	HashEntry *p;
324 
325 	for (; (p = *ref) != NULL; ref = &p->next) {
326 		if (p == he) {
327 			*ref = p->next;
328 			free(p);
329 			t->numEntries--;
330 			return;
331 		}
332 	}
333 	abort();
334 }
335 
336 /* Set things up for iterating over all entries in the hash table. */
337 void
338 HashIter_Init(HashIter *hi, HashTable *t)
339 {
340 	hi->table = t;
341 	hi->nextBucket = 0;
342 	hi->entry = NULL;
343 }
344 
345 /*
346  * Return the next entry in the hash table, or NULL if the end of the table
347  * is reached.
348  */
349 HashEntry *
350 HashIter_Next(HashIter *hi)
351 {
352 	HashTable *t = hi->table;
353 	HashEntry *he = hi->entry;
354 	HashEntry **buckets = t->buckets;
355 	unsigned int bucketsSize = t->bucketsSize;
356 
357 	if (he != NULL)
358 		he = he->next;	/* skip the most recently returned entry */
359 
360 	while (he == NULL) {	/* find the next nonempty chain */
361 		if (hi->nextBucket >= bucketsSize)
362 			return NULL;
363 		he = buckets[hi->nextBucket++];
364 	}
365 	hi->entry = he;
366 	return he;
367 }
368 
369 void
370 HashTable_DebugStats(HashTable *t, const char *name)
371 {
372 	DEBUG4(HASH, "HashTable %s: size=%u numEntries=%u maxchain=%u\n",
373 	       name, t->bucketsSize, t->numEntries, t->maxchain);
374 }
375