2 * @brief Lightning memory-mapped database library
4 * @mainpage Lightning Memory-Mapped Database Manager (MDB)
6 * @section intro_sec Introduction
7 * MDB is a Btree-based database management library modeled loosely on the
8 * BerkeleyDB API, but much simplified. The entire database is exposed
9 * in a memory map, and all data fetches return data directly
10 * from the mapped memory, so no malloc's or memcpy's occur during
11 * data fetches. As such, the library is extremely simple because it
12 * requires no page caching layer of its own, and it is extremely high
13 * performance and memory-efficient. It is also fully transactional with
14 * full ACID semantics, and when the memory map is read-only, the
15 * database integrity cannot be corrupted by stray pointer writes from
18 * The library is fully thread-aware and supports concurrent read/write
19 * access from multiple processes and threads. Data pages use a copy-on-
20 * write strategy so no active data pages are ever overwritten, which
21 * also provides resistance to corruption and eliminates the need of any
22 * special recovery procedures after a system crash. Writes are fully
23 * serialized; only one write transaction may be active at a time, which
24 * guarantees that writers can never deadlock. The database structure is
25 * multi-versioned so readers run with no locks; writers cannot block
26 * readers, and readers don't block writers.
28 * Unlike other well-known database mechanisms which use either write-ahead
29 * transaction logs or append-only data writes, MDB requires no maintenance
30 * during operation. Both write-ahead loggers and append-only databases
31 * require periodic checkpointing and/or compaction of their log or database
32 * files otherwise they grow without bound. MDB tracks free pages within
33 * the database and re-uses them for new write operations, so the database
34 * size does not grow without bound in normal use.
36 * The memory map can be used as a read-only or read-write map. It is
37 * read-only by default as this provides total immunity to corruption.
38 * Using read-write mode offers much higher write performance, but adds
39 * the possibility for stray application writes thru pointers to silently
40 * corrupt the database. Of course if your application code is known to
41 * be bug-free (...) then this is not an issue.
43 * @section caveats_sec Caveats
44 * Troubleshooting the lock file, plus semaphores on BSD systems:
46 * - A broken lockfile can cause sync issues.
47 * Stale reader transactions left behind by an aborted program
48 * cause further writes to grow the database quickly, and
49 * stale locks can block further operation.
51 * Fix: Terminate all programs using the database, or make
52 * them close it. Next database user will reset the lockfile.
54 * - On BSD systems or others configured with MDB_USE_POSIX_SEM,
55 * startup can fail due to semaphores owned by another userid.
57 * Fix: Open and close the database as the user which owns the
58 * semaphores (likely last user) or as root, while no other
59 * process is using the database.
61 * Restrictions/caveats (in addition to those listed for some functions):
63 * - Only the database owner should normally use the database on
64 * BSD systems or when otherwise configured with MDB_USE_POSIX_SEM.
65 * Multiple users can cause startup to fail later, as noted above.
67 * - A thread can only use one transaction at a time, plus any child
68 * transactions. Each transaction belongs to one thread. See below.
70 * - Use an MDB_env* in the process which opened it, without fork()ing.
72 * - Do not have open an MDB database twice in the same process at
73 * the same time. Not even from a plain open() call - close()ing it
74 * breaks flock() advisory locking.
76 * - Avoid long-lived transactions. Read transactions prevent
77 * reuse of pages freed by newer write transactions, thus the
78 * database can grow quickly. Write transactions prevent
79 * other write transactions, since writes are serialized.
81 * ...when several processes can use a database concurrently:
83 * - Avoid suspending a process with active transactions. These
84 * would then be "long-lived" as above.
86 * - Avoid aborting a process with an active transaction.
87 * The transaction becomes "long-lived" as above until the lockfile
88 * is reset, since the process may not remove it from the lockfile.
90 * - If you do that anyway, close the environment once in a while,
91 * so the lockfile can get reset.
93 * - Do not use MDB databases on remote filesystems, even between
94 * processes on the same host. This breaks flock() on some OSes,
95 * possibly memory map sync, and certainly sync between programs
98 * - Opening a database can fail if another process is opening or
99 * closing it at exactly the same time.
101 * @author Howard Chu, Symas Corporation.
103 * @copyright Copyright 2011-2012 Howard Chu, Symas Corp. All rights reserved.
105 * Redistribution and use in source and binary forms, with or without
106 * modification, are permitted only as authorized by the OpenLDAP
109 * A copy of this license is available in the file LICENSE in the
110 * top-level directory of the distribution or, alternatively, at
111 * <http://www.OpenLDAP.org/license.html>.
114 * This code is derived from btree.c written by Martin Hedenfalk.
116 * Copyright (c) 2009, 2010 Martin Hedenfalk <martin@bzero.se>
118 * Permission to use, copy, modify, and distribute this software for any
119 * purpose with or without fee is hereby granted, provided that the above
120 * copyright notice and this permission notice appear in all copies.
122 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
123 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
124 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
125 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
126 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
127 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
128 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
133 #include <sys/types.h>
139 /** @defgroup mdb MDB API
141 * @brief OpenLDAP Lightning Memory-Mapped Database Manager
143 /** @defgroup Version Version Macros
146 /** Library major version */
147 #define MDB_VERSION_MAJOR 0
148 /** Library minor version */
149 #define MDB_VERSION_MINOR 9
150 /** Library patch version */
151 #define MDB_VERSION_PATCH 5
153 /** Combine args a,b,c into a single integer for easy version comparisons */
154 #define MDB_VERINT(a,b,c) (((a) << 24) | ((b) << 16) | (c))
156 /** The full library version as a single integer */
157 #define MDB_VERSION_FULL \
158 MDB_VERINT(MDB_VERSION_MAJOR,MDB_VERSION_MINOR,MDB_VERSION_PATCH)
160 /** The release date of this library version */
161 #define MDB_VERSION_DATE "December 12, 2012"
163 /** A stringifier for the version info */
164 #define MDB_VERSTR(a,b,c,d) "MDB " #a "." #b "." #c ": (" d ")"
166 /** A helper for the stringifier macro */
167 #define MDB_VERFOO(a,b,c,d) MDB_VERSTR(a,b,c,d)
169 /** The full library version as a C string */
170 #define MDB_VERSION_STRING \
171 MDB_VERFOO(MDB_VERSION_MAJOR,MDB_VERSION_MINOR,MDB_VERSION_PATCH,MDB_VERSION_DATE)
174 /** @brief Opaque structure for a database environment.
176 * A DB environment supports multiple databases, all residing in the same
179 typedef struct MDB_env MDB_env;
181 /** @brief Opaque structure for a transaction handle.
183 * All database operations require a transaction handle. Transactions may be
184 * read-only or read-write.
186 typedef struct MDB_txn MDB_txn;
188 /** @brief A handle for an individual database in the DB environment. */
189 typedef unsigned int MDB_dbi;
191 /** @brief Opaque structure for navigating through a database */
192 typedef struct MDB_cursor MDB_cursor;
194 /** @brief Generic structure used for passing keys and data in and out of the database. */
195 typedef struct MDB_val {
196 size_t mv_size; /**< size of the data item */
197 void *mv_data; /**< address of the data item */
200 /** @brief A callback function used to compare two keys in a database */
201 typedef int (MDB_cmp_func)(const MDB_val *a, const MDB_val *b);
203 /** @brief A callback function used to relocate a position-dependent data item
204 * in a fixed-address database.
206 * The \b newptr gives the item's desired address in
207 * the memory map, and \b oldptr gives its previous address. The item's actual
208 * data resides at the address in \b item. This callback is expected to walk
209 * through the fields of the record in \b item and modify any
210 * values based at the \b oldptr address to be relative to the \b newptr address.
211 * @param[in,out] item The item that is to be relocated.
212 * @param[in] oldptr The previous address.
213 * @param[in] newptr The new address to relocate to.
214 * @param[in] relctx An application-provided context, set by #mdb_set_relctx().
215 * @todo This feature is currently unimplemented.
217 typedef void (MDB_rel_func)(MDB_val *item, void *oldptr, void *newptr, void *relctx);
219 /** @defgroup mdb_env Environment Flags
221 * Values do not overlap Database Flags.
224 /** mmap at a fixed address */
225 #define MDB_FIXEDMAP 0x01
226 /** no environment directory */
227 #define MDB_NOSUBDIR 0x4000
228 /** don't fsync after commit */
229 #define MDB_NOSYNC 0x10000
231 #define MDB_RDONLY 0x20000
232 /** don't fsync metapage after commit */
233 #define MDB_NOMETASYNC 0x40000
234 /** use writable mmap */
235 #define MDB_WRITEMAP 0x80000
236 /** use asynchronous msync */
237 #define MDB_MAPASYNC 0x100000
240 /** @defgroup mdb_dbi_open Database Flags
242 * Values do not overlap Environment Flags.
245 /** use reverse string keys */
246 #define MDB_REVERSEKEY 0x02
247 /** use sorted duplicates */
248 #define MDB_DUPSORT 0x04
249 /** numeric keys in native byte order.
250 * The keys must all be of the same size. */
251 #define MDB_INTEGERKEY 0x08
252 /** with #MDB_DUPSORT, sorted dup items have fixed size */
253 #define MDB_DUPFIXED 0x10
254 /** with #MDB_DUPSORT, dups are numeric in native byte order */
255 #define MDB_INTEGERDUP 0x20
256 /** with #MDB_DUPSORT, use reverse string dups */
257 #define MDB_REVERSEDUP 0x40
258 /** create DB if not already existing */
259 #define MDB_CREATE 0x40000
262 /** @defgroup mdb_put Write Flags
265 /** For put: Don't write if the key already exists. */
266 #define MDB_NOOVERWRITE 0x10
267 /** Only for #MDB_DUPSORT<br>
268 * For put: don't write if the key and data pair already exist.<br>
269 * For mdb_cursor_del: remove all duplicate data items.
271 #define MDB_NODUPDATA 0x20
272 /** For mdb_cursor_put: overwrite the current key/data pair */
273 #define MDB_CURRENT 0x40
274 /** For put: Just reserve space for data, don't copy it. Return a
275 * pointer to the reserved space.
277 #define MDB_RESERVE 0x10000
278 /** Data is being appended, don't split full pages. */
279 #define MDB_APPEND 0x20000
280 /** Duplicate data is being appended, don't split full pages. */
281 #define MDB_APPENDDUP 0x40000
282 /** Store multiple data items in one call. */
283 #define MDB_MULTIPLE 0x80000
286 /** @brief Cursor Get operations.
288 * This is the set of all operations for retrieving data
291 typedef enum MDB_cursor_op {
292 MDB_FIRST, /**< Position at first key/data item */
293 MDB_FIRST_DUP, /**< Position at first data item of current key.
294 Only for #MDB_DUPSORT */
295 MDB_GET_BOTH, /**< Position at key/data pair. Only for #MDB_DUPSORT */
296 MDB_GET_BOTH_RANGE, /**< position at key, nearest data. Only for #MDB_DUPSORT */
297 MDB_GET_CURRENT, /**< Return key/data at current cursor position */
298 MDB_GET_MULTIPLE, /**< Return all the duplicate data items at the current
299 cursor position. Only for #MDB_DUPFIXED */
300 MDB_LAST, /**< Position at last key/data item */
301 MDB_LAST_DUP, /**< Position at last data item of current key.
302 Only for #MDB_DUPSORT */
303 MDB_NEXT, /**< Position at next data item */
304 MDB_NEXT_DUP, /**< Position at next data item of current key.
305 Only for #MDB_DUPSORT */
306 MDB_NEXT_MULTIPLE, /**< Return all duplicate data items at the next
307 cursor position. Only for #MDB_DUPFIXED */
308 MDB_NEXT_NODUP, /**< Position at first data item of next key.
309 Only for #MDB_DUPSORT */
310 MDB_PREV, /**< Position at previous data item */
311 MDB_PREV_DUP, /**< Position at previous data item of current key.
312 Only for #MDB_DUPSORT */
313 MDB_PREV_NODUP, /**< Position at last data item of previous key.
314 Only for #MDB_DUPSORT */
315 MDB_SET, /**< Position at specified key */
316 MDB_SET_KEY, /**< Position at specified key, return key + data */
317 MDB_SET_RANGE /**< Position at first key greater than or equal to specified key. */
320 /** @defgroup errors Return Codes
322 * BerkeleyDB uses -30800 to -30999, we'll go under them
325 /** Successful result */
326 #define MDB_SUCCESS 0
327 /** key/data pair already exists */
328 #define MDB_KEYEXIST (-30799)
329 /** key/data pair not found (EOF) */
330 #define MDB_NOTFOUND (-30798)
331 /** Requested page not found - this usually indicates corruption */
332 #define MDB_PAGE_NOTFOUND (-30797)
333 /** Located page was wrong type */
334 #define MDB_CORRUPTED (-30796)
335 /** Update of meta page failed, probably I/O error */
336 #define MDB_PANIC (-30795)
337 /** Environment version mismatch */
338 #define MDB_VERSION_MISMATCH (-30794)
339 /** File is not a valid MDB file */
340 #define MDB_INVALID (-30793)
341 /** Environment mapsize reached */
342 #define MDB_MAP_FULL (-30792)
343 /** Environment maxdbs reached */
344 #define MDB_DBS_FULL (-30791)
345 /** Environment maxreaders reached */
346 #define MDB_READERS_FULL (-30790)
347 /** Too many TLS keys in use - Windows only */
348 #define MDB_TLS_FULL (-30789)
349 /** Nested txn has too many dirty pages */
350 #define MDB_TXN_FULL (-30788)
351 /** Cursor stack too deep - internal error */
352 #define MDB_CURSOR_FULL (-30787)
353 /** Page has not enough space - internal error */
354 #define MDB_PAGE_FULL (-30786)
355 #define MDB_LAST_ERRCODE MDB_PAGE_FULL
358 /** @brief Statistics for a database in the environment */
359 typedef struct MDB_stat {
360 unsigned int ms_psize; /**< Size of a database page.
361 This is currently the same for all databases. */
362 unsigned int ms_depth; /**< Depth (height) of the B-tree */
363 size_t ms_branch_pages; /**< Number of internal (non-leaf) pages */
364 size_t ms_leaf_pages; /**< Number of leaf pages */
365 size_t ms_overflow_pages; /**< Number of overflow pages */
366 size_t ms_entries; /**< Number of data items */
369 /** @brief Information about the environment */
370 typedef struct MDB_envinfo {
371 void *me_mapaddr; /**< Address of map, if fixed */
372 size_t me_mapsize; /**< Size of the data memory map */
373 size_t me_last_pgno; /**< ID of the last used page */
374 size_t me_last_txnid; /**< ID of the last committed transaction */
375 unsigned int me_maxreaders; /**< maximum number of threads for the environment */
376 unsigned int me_numreaders; /**< maximum number of threads used in the environment */
379 /** @brief Return the mdb library version information.
381 * @param[out] major if non-NULL, the library major version number is copied here
382 * @param[out] minor if non-NULL, the library minor version number is copied here
383 * @param[out] patch if non-NULL, the library patch version number is copied here
384 * @retval "version string" The library version as a string
386 char *mdb_version(int *major, int *minor, int *patch);
388 /** @brief Return a string describing a given error code.
390 * This function is a superset of the ANSI C X3.159-1989 (ANSI C) strerror(3)
391 * function. If the error code is greater than or equal to 0, then the string
392 * returned by the system function strerror(3) is returned. If the error code
393 * is less than 0, an error string corresponding to the MDB library error is
394 * returned. See @ref errors for a list of MDB-specific error codes.
395 * @param[in] err The error code
396 * @retval "error message" The description of the error
398 char *mdb_strerror(int err);
400 /** @brief Create an MDB environment handle.
402 * This function allocates memory for a #MDB_env structure. To release
403 * the allocated memory and discard the handle, call #mdb_env_close().
404 * Before the handle may be used, it must be opened using #mdb_env_open().
405 * Various other options may also need to be set before opening the handle,
406 * e.g. #mdb_env_set_mapsize(), #mdb_env_set_maxreaders(), #mdb_env_set_maxdbs(),
407 * depending on usage requirements.
408 * @param[out] env The address where the new handle will be stored
409 * @return A non-zero error value on failure and 0 on success.
411 int mdb_env_create(MDB_env **env);
413 /** @brief Open an environment handle.
415 * If this function fails, #mdb_env_close() must be called to discard the #MDB_env handle.
416 * @param[in] env An environment handle returned by #mdb_env_create()
417 * @param[in] path The directory in which the database files reside. This
418 * directory must already exist and be writable.
419 * @param[in] flags Special options for this environment. This parameter
420 * must be set to 0 or by bitwise OR'ing together one or more of the
421 * values described here.
422 * Flags set by mdb_env_set_flags() are also used.
425 * use a fixed address for the mmap region. This flag must be specified
426 * when creating the environment, and is stored persistently in the environment.
427 * If successful, the memory map will always reside at the same virtual address
428 * and pointers used to reference data items in the database will be constant
429 * across multiple invocations. This option may not always work, depending on
430 * how the operating system has allocated memory to shared libraries and other uses.
431 * The feature is highly experimental.
433 * By default, MDB creates its environment in a directory whose
434 * pathname is given in \b path, and creates its data and lock files
435 * under that directory. With this option, \b path is used as-is for
436 * the database main data file. The database lock file is the \b path
437 * with "-lock" appended.
439 * Don't perform a synchronous flush after committing a transaction. This means
440 * transactions will exhibit the ACI (atomicity, consistency, and isolation)
441 * properties, but not D (durability); that is database integrity will be
442 * maintained but it is possible some number of the most recently committed
443 * transactions may be undone after a system crash. The number of transactions
444 * at risk is governed by how often the system flushes dirty buffers to disk
445 * and how often #mdb_env_sync() is called. This flag may be changed
446 * at any time using #mdb_env_set_flags().
447 * <li>#MDB_NOMETASYNC
448 * Don't perform a synchronous flush of the meta page after committing
449 * a transaction. This is similar to the #MDB_NOSYNC case, but safer
450 * because the transaction data is still flushed. The meta page for any
451 * transaction N will be flushed by the data flush of transaction N+1.
452 * In case of a system crash, the last committed transaction may be
453 * lost. This flag may be changed at any time using #mdb_env_set_flags().
455 * Open the environment in read-only mode. No write operations will be allowed.
457 * @param[in] mode The UNIX permissions to set on created files. This parameter
458 * is ignored on Windows.
459 * @return A non-zero error value on failure and 0 on success. Some possible
462 * <li>#MDB_VERSION_MISMATCH - the version of the MDB library doesn't match the
463 * version that created the database environment.
464 * <li>EINVAL - the environment file headers are corrupted.
465 * <li>ENOENT - the directory specified by the path parameter doesn't exist.
466 * <li>EACCES - the user didn't have permission to access the environment files.
467 * <li>EAGAIN - the environment was locked by another process.
470 int mdb_env_open(MDB_env *env, const char *path, unsigned int flags, mode_t mode);
472 /** @brief Copy an MDB environment to the specified path.
474 * This function may be used to make a backup of an existing environment.
475 * @param[in] env An environment handle returned by #mdb_env_create(). It
476 * must have already been opened successfully.
477 * @param[in] path The directory in which the copy will reside. This
478 * directory must already exist and be writable but must otherwise be
480 * @return A non-zero error value on failure and 0 on success.
482 int mdb_env_copy(MDB_env *env, const char *path);
484 /** @brief Return statistics about the MDB environment.
486 * @param[in] env An environment handle returned by #mdb_env_create()
487 * @param[out] stat The address of an #MDB_stat structure
488 * where the statistics will be copied
490 int mdb_env_stat(MDB_env *env, MDB_stat *stat);
492 /** @brief Return information about the MDB environment.
494 * @param[in] env An environment handle returned by #mdb_env_create()
495 * @param[out] stat The address of an #MDB_envinfo structure
496 * where the information will be copied
498 int mdb_env_info(MDB_env *env, MDB_envinfo *stat);
500 /** @brief Flush the data buffers to disk.
502 * Data is always written to disk when #mdb_txn_commit() is called,
503 * but the operating system may keep it buffered. MDB always flushes
504 * the OS buffers upon commit as well, unless the environment was
505 * opened with #MDB_NOSYNC.
506 * @param[in] env An environment handle returned by #mdb_env_create()
507 * @param[in] force If non-zero, force a synchronous flush. Otherwise
508 * if the environment has the #MDB_NOSYNC flag set the flushes
509 * will be omitted, and with #MDB_MAPASYNC they will be asynchronous.
510 * @return A non-zero error value on failure and 0 on success. Some possible
513 * <li>EINVAL - an invalid parameter was specified.
514 * <li>EIO - an error occurred during synchronization.
517 int mdb_env_sync(MDB_env *env, int force);
519 /** @brief Close the environment and release the memory map.
521 * Only a single thread may call this function. All transactions, databases,
522 * and cursors must already be closed before calling this function. Attempts to
523 * use any such handles after calling this function will cause a SIGSEGV.
524 * The environment handle will be freed and must not be used again after this call.
525 * @param[in] env An environment handle returned by #mdb_env_create()
527 void mdb_env_close(MDB_env *env);
529 /** @brief Set environment flags.
531 * This may be used to set some flags in addition to those from
532 * #mdb_env_open(), or to unset these flags.
533 * @param[in] env An environment handle returned by #mdb_env_create()
534 * @param[in] flags The flags to change, bitwise OR'ed together
535 * @param[in] onoff A non-zero value sets the flags, zero clears them.
536 * @return A non-zero error value on failure and 0 on success. Some possible
539 * <li>EINVAL - an invalid parameter was specified.
542 int mdb_env_set_flags(MDB_env *env, unsigned int flags, int onoff);
544 /** @brief Get environment flags.
546 * @param[in] env An environment handle returned by #mdb_env_create()
547 * @param[out] flags The address of an integer to store the flags
548 * @return A non-zero error value on failure and 0 on success. Some possible
551 * <li>EINVAL - an invalid parameter was specified.
554 int mdb_env_get_flags(MDB_env *env, unsigned int *flags);
556 /** @brief Return the path that was used in #mdb_env_open().
558 * @param[in] env An environment handle returned by #mdb_env_create()
559 * @param[out] path Address of a string pointer to contain the path. This
560 * is the actual string in the environment, not a copy. It should not be
561 * altered in any way.
562 * @return A non-zero error value on failure and 0 on success. Some possible
565 * <li>EINVAL - an invalid parameter was specified.
568 int mdb_env_get_path(MDB_env *env, const char **path);
570 /** @brief Set the size of the memory map to use for this environment.
572 * The size should be a multiple of the OS page size. The default is
573 * 10485760 bytes. The size of the memory map is also the maximum size
574 * of the database. The value should be chosen as large as possible,
575 * to accommodate future growth of the database.
576 * This function may only be called after #mdb_env_create() and before #mdb_env_open().
577 * The size may be changed by closing and reopening the environment.
578 * Any attempt to set a size smaller than the space already consumed
579 * by the environment will be silently changed to the current size of the used space.
580 * @param[in] env An environment handle returned by #mdb_env_create()
581 * @param[in] size The size in bytes
582 * @return A non-zero error value on failure and 0 on success. Some possible
585 * <li>EINVAL - an invalid parameter was specified, or the environment is already open.
588 int mdb_env_set_mapsize(MDB_env *env, size_t size);
590 /** @brief Set the maximum number of threads for the environment.
592 * This defines the number of slots in the lock table that is used to track readers in the
593 * the environment. The default is 126.
594 * This function may only be called after #mdb_env_create() and before #mdb_env_open().
595 * @param[in] env An environment handle returned by #mdb_env_create()
596 * @param[in] readers The maximum number of threads
597 * @return A non-zero error value on failure and 0 on success. Some possible
600 * <li>EINVAL - an invalid parameter was specified, or the environment is already open.
603 int mdb_env_set_maxreaders(MDB_env *env, unsigned int readers);
605 /** @brief Get the maximum number of threads for the environment.
607 * @param[in] env An environment handle returned by #mdb_env_create()
608 * @param[out] readers Address of an integer to store the number of readers
609 * @return A non-zero error value on failure and 0 on success. Some possible
612 * <li>EINVAL - an invalid parameter was specified.
615 int mdb_env_get_maxreaders(MDB_env *env, unsigned int *readers);
617 /** @brief Set the maximum number of named databases for the environment.
619 * This function is only needed if multiple databases will be used in the
620 * environment. Simpler applications that use the environment as a single
621 * unnamed database can ignore this option.
622 * This function may only be called after #mdb_env_create() and before #mdb_env_open().
623 * @param[in] env An environment handle returned by #mdb_env_create()
624 * @param[in] dbs The maximum number of databases
625 * @return A non-zero error value on failure and 0 on success. Some possible
628 * <li>EINVAL - an invalid parameter was specified, or the environment is already open.
631 int mdb_env_set_maxdbs(MDB_env *env, MDB_dbi dbs);
633 /** @brief Create a transaction for use with the environment.
635 * The transaction handle may be discarded using #mdb_txn_abort() or #mdb_txn_commit().
636 * @note Transactions may not span threads; a transaction must only be used by a
637 * single thread. Also, a thread may only have a single transaction.
638 * @note Cursors may not span transactions; each cursor must be opened and closed
639 * within a single transaction.
640 * @param[in] env An environment handle returned by #mdb_env_create()
641 * @param[in] parent If this parameter is non-NULL, the new transaction
642 * will be a nested transaction, with the transaction indicated by \b parent
643 * as its parent. Transactions may be nested to any level. A parent
644 * transaction may not issue any other operations besides mdb_txn_begin,
645 * mdb_txn_abort, or mdb_txn_commit while it has active child transactions.
646 * @param[in] flags Special options for this transaction. This parameter
647 * must be set to 0 or by bitwise OR'ing together one or more of the
648 * values described here.
651 * This transaction will not perform any write operations.
653 * @param[out] txn Address where the new #MDB_txn handle will be stored
654 * @return A non-zero error value on failure and 0 on success. Some possible
657 * <li>#MDB_PANIC - a fatal error occurred earlier and the environment
659 * <li>ENOMEM - out of memory, or a read-only transaction was requested and
660 * the reader lock table is full. See #mdb_env_set_maxreaders().
663 int mdb_txn_begin(MDB_env *env, MDB_txn *parent, unsigned int flags, MDB_txn **txn);
665 /** @brief Commit all the operations of a transaction into the database.
667 * All cursors opened within the transaction will be closed by this call. The cursors
668 * and transaction handle will be freed and must not be used again after this call.
669 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
670 * @return A non-zero error value on failure and 0 on success. Some possible
673 * <li>EINVAL - an invalid parameter was specified.
674 * <li>ENOSPC - no more disk space.
675 * <li>EIO - a low-level I/O error occurred while writing.
676 * <li>ENOMEM - the transaction is nested and could not be merged into its parent.
679 int mdb_txn_commit(MDB_txn *txn);
681 /** @brief Abandon all the operations of the transaction instead of saving them.
683 * All cursors opened within the transaction will be closed by this call. The cursors
684 * and transaction handle will be freed and must not be used again after this call.
685 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
687 void mdb_txn_abort(MDB_txn *txn);
689 /** @brief Reset a read-only transaction.
691 * This releases the current reader lock but doesn't free the
692 * transaction handle, allowing it to be used again later by #mdb_txn_renew().
693 * It otherwise has the same effect as #mdb_txn_abort() but saves some memory
694 * allocation/deallocation overhead if a thread is going to start a new
695 * read-only transaction again soon.
696 * All cursors opened within the transaction must be closed before the transaction
698 * Reader locks generally don't interfere with writers, but they keep old
699 * versions of database pages allocated. Thus they prevent the old pages
700 * from being reused when writers commit new data, and so under heavy load
701 * the database size may grow much more rapidly than otherwise.
702 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
704 void mdb_txn_reset(MDB_txn *txn);
706 /** @brief Renew a read-only transaction.
708 * This acquires a new reader lock for a transaction handle that had been
709 * released by #mdb_txn_reset(). It must be called before a reset transaction
711 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
712 * @return A non-zero error value on failure and 0 on success. Some possible
715 * <li>#MDB_PANIC - a fatal error occurred earlier and the environment
717 * <li>EINVAL - an invalid parameter was specified.
720 int mdb_txn_renew(MDB_txn *txn);
722 /** Compat with version <= 0.9.4, avoid clash with libmdb from MDB Tools project */
723 #define mdb_open(txn,name,flags,dbi) mdb_dbi_open(txn,name,flags,dbi)
724 /** Compat with version <= 0.9.4, avoid clash with libmdb from MDB Tools project */
725 #define mdb_close(env,dbi) mdb_dbi_close(env,dbi)
727 /** @brief Open a database in the environment.
729 * The database handle may be discarded by calling #mdb_dbi_close(). The
730 * database handle resides in the shared environment, it is not owned
731 * by the given transaction. Only one thread should call this function;
732 * it is not mutex-protected in a read-only transaction.
733 * To use named databases (with name != NULL), #mdb_env_set_maxdbs()
734 * must be called before opening the enviorment.
735 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
736 * @param[in] name The name of the database to open. If only a single
737 * database is needed in the environment, this value may be NULL.
738 * @param[in] flags Special options for this database. This parameter
739 * must be set to 0 or by bitwise OR'ing together one or more of the
740 * values described here.
742 * <li>#MDB_REVERSEKEY
743 * Keys are strings to be compared in reverse order, from the end
744 * of the strings to the beginning. By default, Keys are treated as strings and
745 * compared from beginning to end.
747 * Duplicate keys may be used in the database. (Or, from another perspective,
748 * keys may have multiple data items, stored in sorted order.) By default
749 * keys must be unique and may have only a single data item.
750 * <li>#MDB_INTEGERKEY
751 * Keys are binary integers in native byte order. Setting this option
752 * requires all keys to be the same size, typically sizeof(int)
755 * This flag may only be used in combination with #MDB_DUPSORT. This option
756 * tells the library that the data items for this database are all the same
757 * size, which allows further optimizations in storage and retrieval. When
758 * all data items are the same size, the #MDB_GET_MULTIPLE and #MDB_NEXT_MULTIPLE
759 * cursor operations may be used to retrieve multiple items at once.
760 * <li>#MDB_INTEGERDUP
761 * This option specifies that duplicate data items are also integers, and
762 * should be sorted as such.
763 * <li>#MDB_REVERSEDUP
764 * This option specifies that duplicate data items should be compared as
765 * strings in reverse order.
767 * Create the named database if it doesn't exist. This option is not
768 * allowed in a read-only transaction or a read-only environment.
770 * @param[out] dbi Address where the new #MDB_dbi handle will be stored
771 * @return A non-zero error value on failure and 0 on success. Some possible
774 * <li>#MDB_NOTFOUND - the specified database doesn't exist in the environment
775 * and #MDB_CREATE was not specified.
776 * <li>ENFILE - too many databases have been opened. See #mdb_env_set_maxdbs().
779 int mdb_dbi_open(MDB_txn *txn, const char *name, unsigned int flags, MDB_dbi *dbi);
781 /** @brief Retrieve statistics for a database.
783 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
784 * @param[in] dbi A database handle returned by #mdb_dbi_open()
785 * @param[out] stat The address of an #MDB_stat structure
786 * where the statistics will be copied
787 * @return A non-zero error value on failure and 0 on success. Some possible
790 * <li>EINVAL - an invalid parameter was specified.
793 int mdb_stat(MDB_txn *txn, MDB_dbi dbi, MDB_stat *stat);
795 /** @brief Close a database handle.
797 * This call is not mutex protected. Handles should only be closed by
798 * a single thread, and only if no other threads are going to reference
799 * the database handle any further.
800 * @param[in] env An environment handle returned by #mdb_env_create()
801 * @param[in] dbi A database handle returned by #mdb_dbi_open()
803 void mdb_dbi_close(MDB_env *env, MDB_dbi dbi);
805 /** @brief Delete a database and/or free all its pages.
807 * If the \b del parameter is 1, the DB handle will be closed
808 * and the DB will be deleted.
809 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
810 * @param[in] dbi A database handle returned by #mdb_dbi_open()
811 * @param[in] del 1 to delete the DB from the environment,
812 * 0 to just free its pages.
813 * @return A non-zero error value on failure and 0 on success.
815 int mdb_drop(MDB_txn *txn, MDB_dbi dbi, int del);
817 /** @brief Set a custom key comparison function for a database.
819 * The comparison function is called whenever it is necessary to compare a
820 * key specified by the application with a key currently stored in the database.
821 * If no comparison function is specified, and no special key flags were specified
822 * with #mdb_dbi_open(), the keys are compared lexically, with shorter keys collating
823 * before longer keys.
824 * @warning This function must be called before any data access functions are used,
825 * otherwise data corruption may occur. The same comparison function must be used by every
826 * program accessing the database, every time the database is used.
827 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
828 * @param[in] dbi A database handle returned by #mdb_dbi_open()
829 * @param[in] cmp A #MDB_cmp_func function
830 * @return A non-zero error value on failure and 0 on success. Some possible
833 * <li>EINVAL - an invalid parameter was specified.
836 int mdb_set_compare(MDB_txn *txn, MDB_dbi dbi, MDB_cmp_func *cmp);
838 /** @brief Set a custom data comparison function for a #MDB_DUPSORT database.
840 * This comparison function is called whenever it is necessary to compare a data
841 * item specified by the application with a data item currently stored in the database.
842 * This function only takes effect if the database was opened with the #MDB_DUPSORT
844 * If no comparison function is specified, and no special key flags were specified
845 * with #mdb_dbi_open(), the data items are compared lexically, with shorter items collating
846 * before longer items.
847 * @warning This function must be called before any data access functions are used,
848 * otherwise data corruption may occur. The same comparison function must be used by every
849 * program accessing the database, every time the database is used.
850 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
851 * @param[in] dbi A database handle returned by #mdb_dbi_open()
852 * @param[in] cmp A #MDB_cmp_func function
853 * @return A non-zero error value on failure and 0 on success. Some possible
856 * <li>EINVAL - an invalid parameter was specified.
859 int mdb_set_dupsort(MDB_txn *txn, MDB_dbi dbi, MDB_cmp_func *cmp);
861 /** @brief Set a relocation function for a #MDB_FIXEDMAP database.
863 * @todo The relocation function is called whenever it is necessary to move the data
864 * of an item to a different position in the database (e.g. through tree
865 * balancing operations, shifts as a result of adds or deletes, etc.). It is
866 * intended to allow address/position-dependent data items to be stored in
867 * a database in an environment opened with the #MDB_FIXEDMAP option.
868 * Currently the relocation feature is unimplemented and setting
869 * this function has no effect.
870 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
871 * @param[in] dbi A database handle returned by #mdb_dbi_open()
872 * @param[in] rel A #MDB_rel_func function
873 * @return A non-zero error value on failure and 0 on success. Some possible
876 * <li>EINVAL - an invalid parameter was specified.
879 int mdb_set_relfunc(MDB_txn *txn, MDB_dbi dbi, MDB_rel_func *rel);
881 /** @brief Set a context pointer for a #MDB_FIXEDMAP database's relocation function.
883 * See #mdb_set_relfunc and #MDB_rel_func for more details.
884 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
885 * @param[in] dbi A database handle returned by #mdb_dbi_open()
886 * @param[in] ctx An arbitrary pointer for whatever the application needs.
887 * It will be passed to the callback function set by #mdb_set_relfunc
888 * as its \b relctx parameter whenever the callback is invoked.
889 * @return A non-zero error value on failure and 0 on success. Some possible
892 * <li>EINVAL - an invalid parameter was specified.
895 int mdb_set_relctx(MDB_txn *txn, MDB_dbi dbi, void *ctx);
897 /** @brief Get items from a database.
899 * This function retrieves key/data pairs from the database. The address
900 * and length of the data associated with the specified \b key are returned
901 * in the structure to which \b data refers.
902 * If the database supports duplicate keys (#MDB_DUPSORT) then the
903 * first data item for the key will be returned. Retrieval of other
904 * items requires the use of #mdb_cursor_get().
906 * @note The memory pointed to by the returned values is owned by the
907 * database. The caller need not dispose of the memory, and may not
908 * modify it in any way. For values returned in a read-only transaction
909 * any modification attempts will cause a SIGSEGV.
910 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
911 * @param[in] dbi A database handle returned by #mdb_dbi_open()
912 * @param[in] key The key to search for in the database
913 * @param[out] data The data corresponding to the key
914 * @return A non-zero error value on failure and 0 on success. Some possible
917 * <li>#MDB_NOTFOUND - the key was not in the database.
918 * <li>EINVAL - an invalid parameter was specified.
921 int mdb_get(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data);
923 /** @brief Store items into a database.
925 * This function stores key/data pairs in the database. The default behavior
926 * is to enter the new key/data pair, replacing any previously existing key
927 * if duplicates are disallowed, or adding a duplicate data item if
928 * duplicates are allowed (#MDB_DUPSORT).
929 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
930 * @param[in] dbi A database handle returned by #mdb_dbi_open()
931 * @param[in] key The key to store in the database
932 * @param[in,out] data The data to store
933 * @param[in] flags Special options for this operation. This parameter
934 * must be set to 0 or by bitwise OR'ing together one or more of the
935 * values described here.
937 * <li>#MDB_NODUPDATA - enter the new key/data pair only if it does not
938 * already appear in the database. This flag may only be specified
939 * if the database was opened with #MDB_DUPSORT. The function will
940 * return #MDB_KEYEXIST if the key/data pair already appears in the
942 * <li>#MDB_NOOVERWRITE - enter the new key/data pair only if the key
943 * does not already appear in the database. The function will return
944 * #MDB_KEYEXIST if the key already appears in the database, even if
945 * the database supports duplicates (#MDB_DUPSORT). The \b data
946 * parameter will be set to point to the existing item.
947 * <li>#MDB_RESERVE - reserve space for data of the given size, but
948 * don't copy the given data. Instead, return a pointer to the
949 * reserved space, which the caller can fill in later. This saves
950 * an extra memcpy if the data is being generated later.
951 * <li>#MDB_APPEND - append the given key/data pair to the end of the
952 * database. No key comparisons are performed. This option allows
953 * fast bulk loading when keys are already known to be in the
954 * correct order. Loading unsorted keys with this flag will cause
956 * <li>#MDB_APPENDDUP - as above, but for sorted dup data.
958 * @return A non-zero error value on failure and 0 on success. Some possible
961 * <li>EACCES - an attempt was made to write in a read-only transaction.
962 * <li>EINVAL - an invalid parameter was specified.
963 * <li>ENOMEM - the database is full, see #mdb_env_set_mapsize().
966 int mdb_put(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data,
969 /** @brief Delete items from a database.
971 * This function removes key/data pairs from the database.
972 * If the database does not support sorted duplicate data items
973 * (#MDB_DUPSORT) the data parameter is ignored.
974 * If the database supports sorted duplicates and the data parameter
975 * is NULL, all of the duplicate data items for the key will be
976 * deleted. Otherwise, if the data parameter is non-NULL
977 * only the matching data item will be deleted.
978 * This function will return #MDB_NOTFOUND if the specified key/data
979 * pair is not in the database.
980 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
981 * @param[in] dbi A database handle returned by #mdb_dbi_open()
982 * @param[in] key The key to delete from the database
983 * @param[in] data The data to delete
984 * @return A non-zero error value on failure and 0 on success. Some possible
987 * <li>EACCES - an attempt was made to write in a read-only transaction.
988 * <li>EINVAL - an invalid parameter was specified.
991 int mdb_del(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data);
993 /** @brief Create a cursor handle.
995 * Cursors are associated with a specific transaction and database and
996 * may not span threads.
997 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
998 * @param[in] dbi A database handle returned by #mdb_dbi_open()
999 * @param[out] cursor Address where the new #MDB_cursor handle will be stored
1000 * @return A non-zero error value on failure and 0 on success. Some possible
1003 * <li>EINVAL - an invalid parameter was specified.
1006 int mdb_cursor_open(MDB_txn *txn, MDB_dbi dbi, MDB_cursor **cursor);
1008 /** @brief Close a cursor handle.
1010 * The cursor handle will be freed and must not be used again after this call.
1011 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1013 void mdb_cursor_close(MDB_cursor *cursor);
1015 /** @brief Renew a cursor handle.
1017 * Cursors are associated with a specific transaction and database and
1018 * may not span threads. Cursors that are only used in read-only
1019 * transactions may be re-used, to avoid unnecessary malloc/free overhead.
1020 * The cursor may be associated with a new read-only transaction, and
1021 * referencing the same database handle as it was created with.
1022 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1023 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1024 * @return A non-zero error value on failure and 0 on success. Some possible
1027 * <li>EINVAL - an invalid parameter was specified.
1030 int mdb_cursor_renew(MDB_txn *txn, MDB_cursor *cursor);
1032 /** @brief Return the cursor's transaction handle.
1034 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1036 MDB_txn *mdb_cursor_txn(MDB_cursor *cursor);
1038 /** @brief Return the cursor's database handle.
1040 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1042 MDB_dbi mdb_cursor_dbi(MDB_cursor *cursor);
1044 /** @brief Retrieve by cursor.
1046 * This function retrieves key/data pairs from the database. The address and length
1047 * of the key are returned in the object to which \b key refers (except for the
1048 * case of the #MDB_SET option, in which the \b key object is unchanged), and
1049 * the address and length of the data are returned in the object to which \b data
1051 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1052 * @param[in,out] key The key for a retrieved item
1053 * @param[in,out] data The data of a retrieved item
1054 * @param[in] op A cursor operation #MDB_cursor_op
1055 * @return A non-zero error value on failure and 0 on success. Some possible
1058 * <li>#MDB_NOTFOUND - no matching key found.
1059 * <li>EINVAL - an invalid parameter was specified.
1062 int mdb_cursor_get(MDB_cursor *cursor, MDB_val *key, MDB_val *data,
1065 /** @brief Store by cursor.
1067 * This function stores key/data pairs into the database.
1068 * If the function fails for any reason, the state of the cursor will be
1069 * unchanged. If the function succeeds and an item is inserted into the
1070 * database, the cursor is always positioned to refer to the newly inserted item.
1071 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1072 * @param[in] key The key operated on.
1073 * @param[in] data The data operated on.
1074 * @param[in] flags Options for this operation. This parameter
1075 * must be set to 0 or one of the values described here.
1077 * <li>#MDB_CURRENT - overwrite the data of the key/data pair to which
1078 * the cursor refers with the specified data item. The \b key
1079 * parameter is ignored.
1080 * <li>#MDB_NODUPDATA - enter the new key/data pair only if it does not
1081 * already appear in the database. This flag may only be specified
1082 * if the database was opened with #MDB_DUPSORT. The function will
1083 * return #MDB_KEYEXIST if the key/data pair already appears in the
1085 * <li>#MDB_NOOVERWRITE - enter the new key/data pair only if the key
1086 * does not already appear in the database. The function will return
1087 * #MDB_KEYEXIST if the key already appears in the database, even if
1088 * the database supports duplicates (#MDB_DUPSORT).
1089 * <li>#MDB_RESERVE - reserve space for data of the given size, but
1090 * don't copy the given data. Instead, return a pointer to the
1091 * reserved space, which the caller can fill in later. This saves
1092 * an extra memcpy if the data is being generated later.
1093 * <li>#MDB_APPEND - append the given key/data pair to the end of the
1094 * database. No key comparisons are performed. This option allows
1095 * fast bulk loading when keys are already known to be in the
1096 * correct order. Loading unsorted keys with this flag will cause
1098 * <li>#MDB_APPENDDUP - as above, but for sorted dup data.
1100 * @return A non-zero error value on failure and 0 on success. Some possible
1103 * <li>EACCES - an attempt was made to modify a read-only database.
1104 * <li>EINVAL - an invalid parameter was specified.
1107 int mdb_cursor_put(MDB_cursor *cursor, MDB_val *key, MDB_val *data,
1108 unsigned int flags);
1110 /** @brief Delete current key/data pair
1112 * This function deletes the key/data pair to which the cursor refers.
1113 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1114 * @param[in] flags Options for this operation. This parameter
1115 * must be set to 0 or one of the values described here.
1117 * <li>#MDB_NODUPDATA - delete all of the data items for the current key.
1118 * This flag may only be specified if the database was opened with #MDB_DUPSORT.
1120 * @return A non-zero error value on failure and 0 on success. Some possible
1123 * <li>EACCES - an attempt was made to modify a read-only database.
1124 * <li>EINVAL - an invalid parameter was specified.
1127 int mdb_cursor_del(MDB_cursor *cursor, unsigned int flags);
1129 /** @brief Return count of duplicates for current key.
1131 * This call is only valid on databases that support sorted duplicate
1132 * data items #MDB_DUPSORT.
1133 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1134 * @param[out] countp Address where the count will be stored
1135 * @return A non-zero error value on failure and 0 on success. Some possible
1138 * <li>EINVAL - cursor is not initialized, or an invalid parameter was specified.
1141 int mdb_cursor_count(MDB_cursor *cursor, size_t *countp);
1143 /** @brief Compare two data items according to a particular database.
1145 * This returns a comparison as if the two data items were keys in the
1146 * specified database.
1147 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1148 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1149 * @param[in] a The first item to compare
1150 * @param[in] b The second item to compare
1151 * @return < 0 if a < b, 0 if a == b, > 0 if a > b
1153 int mdb_cmp(MDB_txn *txn, MDB_dbi dbi, const MDB_val *a, const MDB_val *b);
1155 /** @brief Compare two data items according to a particular database.
1157 * This returns a comparison as if the two items were data items of
1158 * a sorted duplicates #MDB_DUPSORT database.
1159 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1160 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1161 * @param[in] a The first item to compare
1162 * @param[in] b The second item to compare
1163 * @return < 0 if a < b, 0 if a == b, > 0 if a > b
1165 int mdb_dcmp(MDB_txn *txn, MDB_dbi dbi, const MDB_val *a, const MDB_val *b);
1171 #endif /* _LMDB_H_ */