xref: /illumos-gate/usr/src/lib/libpool/common/dict.c (revision 20a7641f9918de8574b8b3b47dbe35c4bfc78df1)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License, Version 1.0 only
6  * (the "License").  You may not use this file except in compliance
7  * with the License.
8  *
9  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10  * or http://www.opensolaris.org/os/licensing.
11  * See the License for the specific language governing permissions
12  * and limitations under the License.
13  *
14  * When distributing Covered Code, include this CDDL HEADER in each
15  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16  * If applicable, add the following below this CDDL HEADER, with the
17  * fields enclosed by brackets "[]" replaced with your own identifying
18  * information: Portions Copyright [yyyy] [name of copyright owner]
19  *
20  * CDDL HEADER END
21  */
22 /*
23  * Copyright 2003 Sun Microsystems, Inc.  All rights reserved.
24  * Use is subject to license terms.
25  */
26 
27 /*
28  * This file contains a basic dictionary implementation which stores
29  * arbitrary key-value mappings. It is used by libpool to store
30  * information about element pointers (pool_elem_t) in the kernel
31  * provider implementation.
32  */
33 
34 #include <stdio.h>
35 #include <stdlib.h>
36 #include <sys/types.h>
37 #include <unistd.h>
38 #include <string.h>
39 #include <assert.h>
40 
41 #include "dict.h"
42 
43 /*
44  * HASH_64_INIT is the same as the INIT value since it is the value
45  * used by FNV (FNV1_64_INIT). More details on FNV are available at:
46  *
47  * http://www.isthe.com/chongo/tech/comp/fnv/index.html
48  */
49 #define	HASH_64_INIT	(0xcbf29ce484222325ULL) /* Hash initializer */
50 
51 /*
52  * HASH_64_PRIME is a large prime number chosen to minimize hashing
53  * collisions.
54  */
55 #define	HASH_64_PRIME	(0x100000001b3ULL)	/* Large Prime */
56 
57 /*
58  * DICT_SIZE is chosen as it is the nearest prime to 2^9 (512). 512 is
59  * chosen as it is unlikely that this dictionary will contain more
60  * elements than this in normal operation. Of course overflow in each
61  * bucket is acceptable, but if there is too much overflow, then
62  * performance will degrade to that of a list.
63  */
64 #define	DICT_SIZE	509			/* Reasonable prime */
65 
66 /*
67  * Data Types
68  */
69 
70 /*
71  * A key bucket.
72  */
73 typedef struct dict_bucket
74 {
75 	const void		*db_key;	/* key */
76 	void			*db_value;	/* value */
77 	struct dict_bucket	*db_next;	/* next bucket */
78 } dict_bucket_t;
79 
80 /*
81  * A dictionary which holds a mapping between a key and a value.
82  *	dh_change	- detects changes to the dictionary.
83  *	dh_cmp		- comparison function
84  *	dh_hash		- hashing function
85  *	dh_buckets	- key storage
86  *	dh_size		- # of buckets
87  */
88 struct dict_hdl {
89 	uint64_t		dh_change;
90 	int			(*dh_cmp)(const void *, const void *);
91 	uint64_t		(*dh_hash)(const void *);
92 	uint64_t		dh_length;
93 	dict_bucket_t		**dh_buckets;
94 	uint64_t		dh_size;
95 };
96 
97 /*
98  * Utility functions. Mainly used for debugging
99  */
100 
101 #if defined(DEBUG)
102 
103 static void		bit_print_32(unsigned int);
104 static void		bit_print_64(unsigned long long);
105 
106 #endif /* DEBUG */
107 
108 /*
109  * Default functions for hashing and comparing if the user does not specify
110  * these values when creating the dictionary.
111  */
112 static int		cmp_addr(const void *, const void *);
113 static uint64_t		hash_addr(const void *);
114 
115 /*
116  * static functions
117  */
118 
119 #if defined(DEBUG)
120 
121 /*
122  * Print to standard out the bit representation of the supplied value
123  */
124 void
125 bit_print_32(unsigned int v)
126 {
127 #pragma	error_messages(off, E_INTEGER_OVERFLOW_DETECTED)
128 	int i, mask = 1 << 31;
129 #pragma	error_messages(on, E_INTEGER_OVERFLOW_DETECTED)
130 
131 	for (i = 1; i <= 32; i++) {
132 		(void) putchar(((v & mask) == 0) ? '0' : '1');
133 		v <<= 1;
134 		if (i % 8 == 0 && i != 32)
135 			(void) putchar(' ');
136 	}
137 	(void) putchar('\n');
138 }
139 
140 /*
141  * Print to standard out the bit representation of the supplied value
142  */
143 void
144 bit_print_64(unsigned long long v)
145 {
146 #pragma	error_messages(off, E_INTEGER_OVERFLOW_DETECTED)
147 	long long mask = 1ll << 63;
148 #pragma	error_messages(on, E_INTEGER_OVERFLOW_DETECTED)
149 	int i;
150 
151 	for (i = 1; i <= 64; i++) {
152 		(void) putchar(((v & mask) == 0) ? '0' : '1');
153 		v <<= 1;
154 		if (i % 8 == 0 && i != 64)
155 			(void) putchar(' ');
156 	}
157 	(void) putchar('\n');
158 }
159 
160 
161 
162 #endif /* DEBUG */
163 
164 /*
165  * Default comparison function which is used if no comparison function
166  * is supplied when the dictionary is created. The default behaviour
167  * is to compare memory address.
168  */
169 int
170 cmp_addr(const void *x, const void *y)
171 {
172 	return (x != y);
173 }
174 
175 
176 /*
177  * The default hashing function which is used if no hashing function
178  * is provided when the dictionary is created. The default behaviour
179  * is to use the hash_buf() function.
180  */
181 uint64_t
182 hash_addr(const void *key)
183 {
184 	return (hash_buf(&key, sizeof (key)));
185 }
186 
187 
188 /*
189  * public interface
190  */
191 
192 /*
193  * Return a hash which is built by manipulating each byte in the
194  * supplied data. The hash logic follows the approach suggested in the
195  * FNV hash.
196  */
197 uint64_t
198 hash_buf(const void *buf, size_t len)
199 {
200 	uchar_t *start = (uchar_t *)buf;
201 	uchar_t *end = start + len;
202 	uint64_t hash = HASH_64_INIT;
203 
204 	while (start < end) {
205 		hash *= HASH_64_PRIME;
206 		hash ^= (uint64_t)*start++;
207 	}
208 
209 	return (hash);
210 }
211 
212 
213 /*
214  * Return a hash which is built by manipulating each byte in the
215  * supplied string. The hash logic follows the approach suggested in
216  * the FNV hash.
217  */
218 uint64_t
219 hash_str(const char *str)
220 {
221 	uchar_t *p = (uchar_t *)str;
222 	uint64_t hash = HASH_64_INIT;
223 
224 	while (*p) {
225 		hash *= HASH_64_PRIME;
226 		hash ^= (uint64_t)*p++;
227 	}
228 
229 	return (hash);
230 }
231 
232 /*
233  * Return the number of keys held in the supplied dictionary.
234  */
235 uint64_t
236 dict_length(dict_hdl_t *hdl)
237 {
238 	return (hdl->dh_length);
239 }
240 
241 /*
242  * Free the supplied dictionary and all it's associated resource.
243  */
244 void
245 dict_free(dict_hdl_t **hdl)
246 {
247 	if ((*hdl)->dh_length > 0) {
248 		uint64_t i;
249 		for (i = 0; i < (*hdl)->dh_size; i++) {
250 			dict_bucket_t *this, *next;
251 			for (this = (*hdl)->dh_buckets[i]; this != NULL;
252 			    this = next) {
253 				next = this->db_next;
254 				free(this);
255 			}
256 		}
257 	}
258 	free((*hdl)->dh_buckets);
259 	free((*hdl));
260 	*hdl = NULL;
261 }
262 
263 /*
264  * Create a new dictionary using the supplied comparison and hashing
265  * functions. If none are supplied then the defaults are used.
266  */
267 dict_hdl_t *
268 dict_new(int (*cmp)(const void *, const void *),
269     uint64_t (*hash)(const void *))
270 {
271 	dict_hdl_t *hdl;
272 
273 	if ((hdl = calloc(1, sizeof (dict_hdl_t))) == NULL)
274 		return (NULL);
275 	hdl->dh_size = DICT_SIZE;
276 	if ((hdl->dh_buckets = calloc(hdl->dh_size, sizeof (dict_bucket_t *)))
277 	    == NULL) {
278 		free(hdl);
279 		return (NULL);
280 	}
281 	hdl->dh_cmp = cmp ? cmp : cmp_addr;
282 	hdl->dh_hash = hash ? hash : hash_addr;
283 	return (hdl);
284 }
285 
286 /*
287  * Get a value from the hash. Null is returned if the key cannot be
288  * found.
289  */
290 void *
291 dict_get(dict_hdl_t *hdl, const void *key)
292 {
293 	uint64_t i;
294 	dict_bucket_t *bucket;
295 
296 	i = (*hdl->dh_hash)(key)%hdl->dh_size;
297 	for (bucket = hdl->dh_buckets[i]; bucket != NULL;
298 	    bucket = bucket->db_next)
299 		if ((*hdl->dh_cmp)(key, bucket->db_key) == 0)
300 			break;
301 	return (bucket ? bucket->db_value : NULL);
302 }
303 
304 /*
305  * Put an entry into the hash. Null is returned if this key was not
306  * already present, otherwise the previous value is returned.
307  */
308 void *
309 dict_put(dict_hdl_t *hdl, const void *key, void *value)
310 {
311 	uint64_t i;
312 	dict_bucket_t *bucket;
313 	void *prev = NULL;
314 
315 	i = (*hdl->dh_hash)(key)%hdl->dh_size;
316 	for (bucket = hdl->dh_buckets[i]; bucket != NULL;
317 	    bucket = bucket->db_next)
318 		if ((*hdl->dh_cmp)(key, bucket->db_key) == 0)
319 			break;
320 	if (bucket) {
321 		prev = bucket->db_value;
322 	} else {
323 		bucket = malloc(sizeof (dict_bucket_t));
324 		bucket->db_key = key;
325 		bucket->db_next = hdl->dh_buckets[i];
326 		hdl->dh_buckets[i] = bucket;
327 		hdl->dh_length++;
328 	}
329 	hdl->dh_change++;
330 	bucket->db_value = value;
331 	return (prev);
332 }
333 
334 /*
335  * Remove the key/value from the dictionary. The value is returned if
336  * the key is found. NULL is returned if the key cannot be located.
337  */
338 void *
339 dict_remove(dict_hdl_t *hdl, const void *key)
340 {
341 	uint64_t i;
342 	dict_bucket_t	**pbucket;
343 
344 	hdl->dh_change++;
345 	i = (*hdl->dh_hash)(key)%hdl->dh_size;
346 
347 	for (pbucket = &hdl->dh_buckets[i]; *pbucket != NULL;
348 	    pbucket = &(*pbucket)->db_next) {
349 		if ((*hdl->dh_cmp)(key, (*pbucket)->db_key) == 0) {
350 			dict_bucket_t *bucket = *pbucket;
351 			void *value = bucket->db_value;
352 
353 			*pbucket = bucket->db_next;
354 			free(bucket);
355 			hdl->dh_length--;
356 			return (value);
357 		}
358 	}
359 	return (NULL);
360 }
361 
362 /*
363  * For all entries in the dictionary call the user supplied function
364  * (apply) with the key, value and user supplied data. If the
365  * dictionary is modifed while this function is executing, then the
366  * function will fail with an assertion about table modifcation.
367  */
368 void
369 dict_map(dict_hdl_t *hdl, void (*apply)(const void *, void **, void *),
370     void *cl)
371 {
372 	uint64_t i;
373 	dict_bucket_t *bucket = NULL;
374 	uint64_t change_stamp = hdl->dh_change;
375 
376 	for (i = 0; i < hdl->dh_size; i++) {
377 		for (bucket = hdl->dh_buckets[i]; bucket != NULL;
378 		    bucket = bucket->db_next) {
379 			apply(bucket->db_key, &bucket->db_value, cl);
380 			if (hdl->dh_change != change_stamp)
381 				assert(!"table modified illegally");
382 		}
383 	}
384 }
385