summaryrefslogtreecommitdiff
path: root/tdb/common/traverse.c
blob: 69c81e6e98fd27b7a4d8d86c7c6f620e64cd552a (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
 /* 
   Unix SMB/CIFS implementation.

   trivial database library

   Copyright (C) Andrew Tridgell              1999-2005
   Copyright (C) Paul `Rusty' Russell		   2000
   Copyright (C) Jeremy Allison			   2000-2003
   
     ** NOTE! The following LGPL license applies to the tdb
     ** library. This does NOT imply that all of Samba is released
     ** under the LGPL
   
   This library is free software; you can redistribute it and/or
   modify it under the terms of the GNU Lesser General Public
   License as published by the Free Software Foundation; either
   version 3 of the License, or (at your option) any later version.

   This library is distributed in the hope that it will be useful,
   but WITHOUT ANY WARRANTY; without even the implied warranty of
   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
   Lesser General Public License for more details.

   You should have received a copy of the GNU Lesser General Public
   License along with this library; if not, see <http://www.gnu.org/licenses/>.
*/

#include "tdb_private.h"

/* Uses traverse lock: 0 = finish, -1 = error, other = record offset */
static int tdb_next_lock(struct tdb_context *tdb, struct tdb_traverse_lock *tlock,
			 struct list_struct *rec)
{
	int want_next = (tlock->off != 0);

	/* Lock each chain from the start one. */
	for (; tlock->hash < tdb->header.hash_size; tlock->hash++) {
		if (!tlock->off && tlock->hash != 0) {
			/* this is an optimisation for the common case where
			   the hash chain is empty, which is particularly
			   common for the use of tdb with ldb, where large
			   hashes are used. In that case we spend most of our
			   time in tdb_brlock(), locking empty hash chains.
			   
			   To avoid this, we do an unlocked pre-check to see
			   if the hash chain is empty before starting to look
			   inside it. If it is empty then we can avoid that
			   hash chain. If it isn't empty then we can't believe
			   the value we get back, as we read it without a
			   lock, so instead we get the lock and re-fetch the
			   value below.
			   
			   Notice that not doing this optimisation on the
			   first hash chain is critical. We must guarantee
			   that we have done at least one fcntl lock at the
			   start of a search to guarantee that memory is
			   coherent on SMP systems. If records are added by
			   others during the search then thats OK, and we
			   could possibly miss those with this trick, but we
			   could miss them anyway without this trick, so the
			   semantics don't change.
			   
			   With a non-indexed ldb search this trick gains us a
			   factor of around 80 in speed on a linux 2.6.x
			   system (testing using ldbtest).
			*/
			tdb->methods->next_hash_chain(tdb, &tlock->hash);
			if (tlock->hash == tdb->header.hash_size) {
				continue;
			}
		}

		if (tdb_lock(tdb, tlock->hash, tlock->lock_rw) == -1)
			return -1;

		/* No previous record?  Start at top of chain. */
		if (!tlock->off) {
			if (tdb_ofs_read(tdb, TDB_HASH_TOP(tlock->hash),
				     &tlock->off) == -1)
				goto fail;
		} else {
			/* Otherwise unlock the previous record. */
			if (tdb_unlock_record(tdb, tlock->off) != 0)
				goto fail;
		}

		if (want_next) {
			/* We have offset of old record: grab next */
			if (tdb_rec_read(tdb, tlock->off, rec) == -1)
				goto fail;
			tlock->off = rec->next;
		}

		/* Iterate through chain */
		while( tlock->off) {
			tdb_off_t current;
			if (tdb_rec_read(tdb, tlock->off, rec) == -1)
				goto fail;

			/* Detect infinite loops. From "Shlomi Yaakobovich" <Shlomi@exanet.com>. */
			if (tlock->off == rec->next) {
				TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_next_lock: loop detected.\n"));
				goto fail;
			}

			if (!TDB_DEAD(rec)) {
				/* Woohoo: we found one! */
				if (tdb_lock_record(tdb, tlock->off) != 0)
					goto fail;
				return tlock->off;
			}

			/* Try to clean dead ones from old traverses */
			current = tlock->off;
			tlock->off = rec->next;
			if (!(tdb->read_only || tdb->traverse_read) && 
			    tdb_do_delete(tdb, current, rec) != 0)
				goto fail;
		}
		tdb_unlock(tdb, tlock->hash, tlock->lock_rw);
		want_next = 0;
	}
	/* We finished iteration without finding anything */
	return TDB_ERRCODE(TDB_SUCCESS, 0);

 fail:
	tlock->off = 0;
	if (tdb_unlock(tdb, tlock->hash, tlock->lock_rw) != 0)
		TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_next_lock: On error unlock failed!\n"));
	return -1;
}

/* traverse the entire database - calling fn(tdb, key, data) on each element.
   return -1 on error or the record count traversed
   if fn is NULL then it is not called
   a non-zero return value from fn() indicates that the traversal should stop
  */
static int tdb_traverse_internal(struct tdb_context *tdb, 
				 tdb_traverse_func fn, void *private_data,
				 struct tdb_traverse_lock *tl)
{
	TDB_DATA key, dbuf;
	struct list_struct rec;
	int ret, count = 0;

	/* This was in the initializaton, above, but the IRIX compiler
	 * did not like it.  crh
	 */
	tl->next = tdb->travlocks.next;

	/* fcntl locks don't stack: beware traverse inside traverse */
	tdb->travlocks.next = tl;

	/* tdb_next_lock places locks on the record returned, and its chain */
	while ((ret = tdb_next_lock(tdb, tl, &rec)) > 0) {
		count++;
		/* now read the full record */
		key.dptr = tdb_alloc_read(tdb, tl->off + sizeof(rec), 
					  rec.key_len + rec.data_len);
		if (!key.dptr) {
			ret = -1;
			if (tdb_unlock(tdb, tl->hash, tl->lock_rw) != 0)
				goto out;
			if (tdb_unlock_record(tdb, tl->off) != 0)
				TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_traverse: key.dptr == NULL and unlock_record failed!\n"));
			goto out;
		}
		key.dsize = rec.key_len;
		dbuf.dptr = key.dptr + rec.key_len;
		dbuf.dsize = rec.data_len;

		/* Drop chain lock, call out */
		if (tdb_unlock(tdb, tl->hash, tl->lock_rw) != 0) {
			ret = -1;
			SAFE_FREE(key.dptr);
			goto out;
		}
		if (fn && fn(tdb, key, dbuf, private_data)) {
			/* They want us to terminate traversal */
			ret = count;
			if (tdb_unlock_record(tdb, tl->off) != 0) {
				TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_traverse: unlock_record failed!\n"));;
				ret = -1;
			}
			SAFE_FREE(key.dptr);
			goto out;
		}
		SAFE_FREE(key.dptr);
	}
out:
	tdb->travlocks.next = tl->next;
	if (ret < 0)
		return -1;
	else
		return count;
}


/*
  a write style traverse - temporarily marks the db read only
*/
int tdb_traverse_read(struct tdb_context *tdb, 
		      tdb_traverse_func fn, void *private_data)
{
	struct tdb_traverse_lock tl = { NULL, 0, 0, F_RDLCK };
	int ret;
	bool in_transaction = (tdb->transaction != NULL);

	/* we need to get a read lock on the transaction lock here to
	   cope with the lock ordering semantics of solaris10 */
	if (!in_transaction) {
		if (tdb_transaction_lock(tdb, F_RDLCK)) {
			return -1;
		}
	}

	tdb->traverse_read++;
	ret = tdb_traverse_internal(tdb, fn, private_data, &tl);
	tdb->traverse_read--;

	if (!in_transaction) {
		tdb_transaction_unlock(tdb);
	}

	return ret;
}

/*
  a write style traverse - needs to get the transaction lock to
  prevent deadlocks

  WARNING: The data buffer given to the callback fn does NOT meet the
  alignment restrictions malloc gives you.
*/
int tdb_traverse(struct tdb_context *tdb, 
		 tdb_traverse_func fn, void *private_data)
{
	struct tdb_traverse_lock tl = { NULL, 0, 0, F_WRLCK };
	int ret;
	bool in_transaction = (tdb->transaction != NULL);

	if (tdb->read_only || tdb->traverse_read) {
		return tdb_traverse_read(tdb, fn, private_data);
	}
	
	if (!in_transaction) {
		if (tdb_transaction_lock(tdb, F_WRLCK)) {
			return -1;
		}
	}

	tdb->traverse_write++;
	ret = tdb_traverse_internal(tdb, fn, private_data, &tl);
	tdb->traverse_write--;

	if (!in_transaction) {
		tdb_transaction_unlock(tdb);
	}

	return ret;
}


/* find the first entry in the database and return its key */
TDB_DATA tdb_firstkey(struct tdb_context *tdb)
{
	TDB_DATA key;
	struct list_struct rec;

	/* release any old lock */
	if (tdb_unlock_record(tdb, tdb->travlocks.off) != 0)
		return tdb_null;
	tdb->travlocks.off = tdb->travlocks.hash = 0;
	tdb->travlocks.lock_rw = F_RDLCK;

	/* Grab first record: locks chain and returned record. */
	if (tdb_next_lock(tdb, &tdb->travlocks, &rec) <= 0)
		return tdb_null;
	/* now read the key */
	key.dsize = rec.key_len;
	key.dptr =tdb_alloc_read(tdb,tdb->travlocks.off+sizeof(rec),key.dsize);

	/* Unlock the hash chain of the record we just read. */
	if (tdb_unlock(tdb, tdb->travlocks.hash, tdb->travlocks.lock_rw) != 0)
		TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_firstkey: error occurred while tdb_unlocking!\n"));
	return key;
}

/* find the next entry in the database, returning its key */
TDB_DATA tdb_nextkey(struct tdb_context *tdb, TDB_DATA oldkey)
{
	uint32_t oldhash;
	TDB_DATA key = tdb_null;
	struct list_struct rec;
	unsigned char *k = NULL;

	/* Is locked key the old key?  If so, traverse will be reliable. */
	if (tdb->travlocks.off) {
		if (tdb_lock(tdb,tdb->travlocks.hash,tdb->travlocks.lock_rw))
			return tdb_null;
		if (tdb_rec_read(tdb, tdb->travlocks.off, &rec) == -1
		    || !(k = tdb_alloc_read(tdb,tdb->travlocks.off+sizeof(rec),
					    rec.key_len))
		    || memcmp(k, oldkey.dptr, oldkey.dsize) != 0) {
			/* No, it wasn't: unlock it and start from scratch */
			if (tdb_unlock_record(tdb, tdb->travlocks.off) != 0) {
				SAFE_FREE(k);
				return tdb_null;
			}
			if (tdb_unlock(tdb, tdb->travlocks.hash, tdb->travlocks.lock_rw) != 0) {
				SAFE_FREE(k);
				return tdb_null;
			}
			tdb->travlocks.off = 0;
		}

		SAFE_FREE(k);
	}

	if (!tdb->travlocks.off) {
		/* No previous element: do normal find, and lock record */
		tdb->travlocks.off = tdb_find_lock_hash(tdb, oldkey, tdb->hash_fn(&oldkey), tdb->travlocks.lock_rw, &rec);
		if (!tdb->travlocks.off)
			return tdb_null;
		tdb->travlocks.hash = BUCKET(rec.full_hash);
		if (tdb_lock_record(tdb, tdb->travlocks.off) != 0) {
			TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_nextkey: lock_record failed (%s)!\n", strerror(errno)));
			return tdb_null;
		}
	}
	oldhash = tdb->travlocks.hash;

	/* Grab next record: locks chain and returned record,
	   unlocks old record */
	if (tdb_next_lock(tdb, &tdb->travlocks, &rec) > 0) {
		key.dsize = rec.key_len;
		key.dptr = tdb_alloc_read(tdb, tdb->travlocks.off+sizeof(rec),
					  key.dsize);
		/* Unlock the chain of this new record */
		if (tdb_unlock(tdb, tdb->travlocks.hash, tdb->travlocks.lock_rw) != 0)
			TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_nextkey: WARNING tdb_unlock failed!\n"));
	}
	/* Unlock the chain of old record */
	if (tdb_unlock(tdb, BUCKET(oldhash), tdb->travlocks.lock_rw) != 0)
		TDB_LOG((tdb, TDB_DEBUG_FATAL, "tdb_nextkey: WARNING tdb_unlock failed!\n"));
	return key;
}