2 * @brief Lightning memory-mapped database library
4 * @mainpage Lightning Memory-Mapped Database Manager (LMDB)
6 * @section intro_sec Introduction
7 * LMDB 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, LMDB 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. LMDB 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: Check for stale readers periodically, using the
52 * #mdb_reader_check function or the \ref mdb_stat_1 "mdb_stat" tool.
53 * Stale writers will be cleared automatically on some systems:
54 * - Windows - automatic
55 * - Linux, systems using POSIX mutexes with Robust option - automatic
56 * - not on BSD, systems using POSIX semaphores.
57 * Otherwise just make all programs using the database close it;
58 * the lockfile is always reset on first open of the environment.
60 * - On BSD systems or others configured with MDB_USE_POSIX_SEM,
61 * startup can fail due to semaphores owned by another userid.
63 * Fix: Open and close the database as the user which owns the
64 * semaphores (likely last user) or as root, while no other
65 * process is using the database.
67 * Restrictions/caveats (in addition to those listed for some functions):
69 * - Only the database owner should normally use the database on
70 * BSD systems or when otherwise configured with MDB_USE_POSIX_SEM.
71 * Multiple users can cause startup to fail later, as noted above.
73 * - There is normally no pure read-only mode, since readers need write
74 * access to locks and lock file. Exceptions: On read-only filesystems
75 * or with the #MDB_NOLOCK flag described under #mdb_env_open().
77 * - By default, in versions before 0.9.10, unused portions of the data
78 * file might receive garbage data from memory freed by other code.
79 * (This does not happen when using the #MDB_WRITEMAP flag.) As of
80 * 0.9.10 the default behavior is to initialize such memory before
81 * writing to the data file. Since there may be a slight performance
82 * cost due to this initialization, applications may disable it using
83 * the #MDB_NOMEMINIT flag. Applications handling sensitive data
84 * which must not be written should not use this flag. This flag is
85 * irrelevant when using #MDB_WRITEMAP.
87 * - A thread can only use one transaction at a time, plus any child
88 * transactions. Each transaction belongs to one thread. See below.
89 * The #MDB_NOTLS flag changes this for read-only transactions.
91 * - Use an MDB_env* in the process which opened it, without fork()ing.
93 * - Do not have open an LMDB database twice in the same process at
94 * the same time. Not even from a plain open() call - close()ing it
95 * breaks flock() advisory locking.
97 * - Avoid long-lived transactions. Read transactions prevent
98 * reuse of pages freed by newer write transactions, thus the
99 * database can grow quickly. Write transactions prevent
100 * other write transactions, since writes are serialized.
102 * - Avoid suspending a process with active transactions. These
103 * would then be "long-lived" as above. Also read transactions
104 * suspended when writers commit could sometimes see wrong data.
106 * ...when several processes can use a database concurrently:
108 * - Avoid aborting a process with an active transaction.
109 * The transaction becomes "long-lived" as above until a check
110 * for stale readers is performed or the lockfile is reset,
111 * since the process may not remove it from the lockfile.
113 * This does not apply to write transactions if the system clears
114 * stale writers, see above.
116 * - If you do that anyway, do a periodic check for stale readers. Or
117 * close the environment once in a while, so the lockfile can get reset.
119 * - Do not use LMDB databases on remote filesystems, even between
120 * processes on the same host. This breaks flock() on some OSes,
121 * possibly memory map sync, and certainly sync between programs
122 * on different hosts.
124 * - Opening a database can fail if another process is opening or
125 * closing it at exactly the same time.
127 * @author Howard Chu, Symas Corporation.
129 * @copyright Copyright 2011-2015 Howard Chu, Symas Corp. All rights reserved.
131 * Redistribution and use in source and binary forms, with or without
132 * modification, are permitted only as authorized by the OpenLDAP
135 * A copy of this license is available in the file LICENSE in the
136 * top-level directory of the distribution or, alternatively, at
137 * <http://www.OpenLDAP.org/license.html>.
140 * This code is derived from btree.c written by Martin Hedenfalk.
142 * Copyright (c) 2009, 2010 Martin Hedenfalk <martin@bzero.se>
144 * Permission to use, copy, modify, and distribute this software for any
145 * purpose with or without fee is hereby granted, provided that the above
146 * copyright notice and this permission notice appear in all copies.
148 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
149 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
150 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
151 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
152 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
153 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
154 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
159 #include <sys/types.h>
165 /** Unix permissions for creating files, or dummy definition for Windows */
167 typedef int mdb_mode_t;
169 typedef mode_t mdb_mode_t;
172 /** An abstraction for a file handle.
173 * On POSIX systems file handles are small integers. On Windows
174 * they're opaque pointers.
177 typedef void *mdb_filehandle_t;
179 typedef int mdb_filehandle_t;
182 /** @defgroup mdb LMDB API
184 * @brief OpenLDAP Lightning Memory-Mapped Database Manager
186 /** @defgroup Version Version Macros
189 /** Library major version */
190 #define MDB_VERSION_MAJOR 0
191 /** Library minor version */
192 #define MDB_VERSION_MINOR 9
193 /** Library patch version */
194 #define MDB_VERSION_PATCH 16
196 /** Combine args a,b,c into a single integer for easy version comparisons */
197 #define MDB_VERINT(a,b,c) (((a) << 24) | ((b) << 16) | (c))
199 /** The full library version as a single integer */
200 #define MDB_VERSION_FULL \
201 MDB_VERINT(MDB_VERSION_MAJOR,MDB_VERSION_MINOR,MDB_VERSION_PATCH)
203 /** The release date of this library version */
204 #define MDB_VERSION_DATE "August 14, 2015"
206 /** A stringifier for the version info */
207 #define MDB_VERSTR(a,b,c,d) "LMDB " #a "." #b "." #c ": (" d ")"
209 /** A helper for the stringifier macro */
210 #define MDB_VERFOO(a,b,c,d) MDB_VERSTR(a,b,c,d)
212 /** The full library version as a C string */
213 #define MDB_VERSION_STRING \
214 MDB_VERFOO(MDB_VERSION_MAJOR,MDB_VERSION_MINOR,MDB_VERSION_PATCH,MDB_VERSION_DATE)
217 /** @brief Opaque structure for a database environment.
219 * A DB environment supports multiple databases, all residing in the same
222 typedef struct MDB_env MDB_env;
224 /** @brief Opaque structure for a transaction handle.
226 * All database operations require a transaction handle. Transactions may be
227 * read-only or read-write.
229 typedef struct MDB_txn MDB_txn;
231 /** @brief A handle for an individual database in the DB environment. */
232 typedef unsigned int MDB_dbi;
234 /** @brief Opaque structure for navigating through a database */
235 typedef struct MDB_cursor MDB_cursor;
237 /** @brief Generic structure used for passing keys and data in and out
240 * Values returned from the database are valid only until a subsequent
241 * update operation, or the end of the transaction. Do not modify or
242 * free them, they commonly point into the database itself.
244 * Key sizes must be between 1 and #mdb_env_get_maxkeysize() inclusive.
245 * The same applies to data sizes in databases with the #MDB_DUPSORT flag.
246 * Other data items can in theory be from 0 to 0xffffffff bytes long.
248 typedef struct MDB_val {
249 size_t mv_size; /**< size of the data item */
250 void *mv_data; /**< address of the data item */
253 /** @brief A callback function used to compare two keys in a database */
254 typedef int (MDB_cmp_func)(const MDB_val *a, const MDB_val *b);
256 /** @brief A callback function used to relocate a position-dependent data item
257 * in a fixed-address database.
259 * The \b newptr gives the item's desired address in
260 * the memory map, and \b oldptr gives its previous address. The item's actual
261 * data resides at the address in \b item. This callback is expected to walk
262 * through the fields of the record in \b item and modify any
263 * values based at the \b oldptr address to be relative to the \b newptr address.
264 * @param[in,out] item The item that is to be relocated.
265 * @param[in] oldptr The previous address.
266 * @param[in] newptr The new address to relocate to.
267 * @param[in] relctx An application-provided context, set by #mdb_set_relctx().
268 * @todo This feature is currently unimplemented.
270 typedef void (MDB_rel_func)(MDB_val *item, void *oldptr, void *newptr, void *relctx);
272 /** @defgroup mdb_env Environment Flags
275 /** mmap at a fixed address (experimental) */
276 #define MDB_FIXEDMAP 0x01
277 /** no environment directory */
278 #define MDB_NOSUBDIR 0x4000
279 /** don't fsync after commit */
280 #define MDB_NOSYNC 0x10000
282 #define MDB_RDONLY 0x20000
283 /** don't fsync metapage after commit */
284 #define MDB_NOMETASYNC 0x40000
285 /** use writable mmap */
286 #define MDB_WRITEMAP 0x80000
287 /** use asynchronous msync when #MDB_WRITEMAP is used */
288 #define MDB_MAPASYNC 0x100000
289 /** tie reader locktable slots to #MDB_txn objects instead of to threads */
290 #define MDB_NOTLS 0x200000
291 /** don't do any locking, caller must manage their own locks */
292 #define MDB_NOLOCK 0x400000
293 /** don't do readahead (no effect on Windows) */
294 #define MDB_NORDAHEAD 0x800000
295 /** don't initialize malloc'd memory before writing to datafile */
296 #define MDB_NOMEMINIT 0x1000000
299 /** @defgroup mdb_dbi_open Database Flags
302 /** use reverse string keys */
303 #define MDB_REVERSEKEY 0x02
304 /** use sorted duplicates */
305 #define MDB_DUPSORT 0x04
306 /** numeric keys in native byte order: either unsigned int or size_t.
307 * The keys must all be of the same size. */
308 #define MDB_INTEGERKEY 0x08
309 /** with #MDB_DUPSORT, sorted dup items have fixed size */
310 #define MDB_DUPFIXED 0x10
311 /** with #MDB_DUPSORT, dups are #MDB_INTEGERKEY-style integers */
312 #define MDB_INTEGERDUP 0x20
313 /** with #MDB_DUPSORT, use reverse string dups */
314 #define MDB_REVERSEDUP 0x40
315 /** create DB if not already existing */
316 #define MDB_CREATE 0x40000
319 /** @defgroup mdb_put Write Flags
322 /** For put: Don't write if the key already exists. */
323 #define MDB_NOOVERWRITE 0x10
324 /** Only for #MDB_DUPSORT<br>
325 * For put: don't write if the key and data pair already exist.<br>
326 * For mdb_cursor_del: remove all duplicate data items.
328 #define MDB_NODUPDATA 0x20
329 /** For mdb_cursor_put: overwrite the current key/data pair */
330 #define MDB_CURRENT 0x40
331 /** For put: Just reserve space for data, don't copy it. Return a
332 * pointer to the reserved space.
334 #define MDB_RESERVE 0x10000
335 /** Data is being appended, don't split full pages. */
336 #define MDB_APPEND 0x20000
337 /** Duplicate data is being appended, don't split full pages. */
338 #define MDB_APPENDDUP 0x40000
339 /** Store multiple data items in one call. Only for #MDB_DUPFIXED. */
340 #define MDB_MULTIPLE 0x80000
343 /** @defgroup mdb_copy Copy Flags
346 /** Compacting copy: Omit free space from copy, and renumber all
347 * pages sequentially.
349 #define MDB_CP_COMPACT 0x01
352 /** @brief Cursor Get operations.
354 * This is the set of all operations for retrieving data
357 typedef enum MDB_cursor_op {
358 MDB_FIRST, /**< Position at first key/data item */
359 MDB_FIRST_DUP, /**< Position at first data item of current key.
360 Only for #MDB_DUPSORT */
361 MDB_GET_BOTH, /**< Position at key/data pair. Only for #MDB_DUPSORT */
362 MDB_GET_BOTH_RANGE, /**< position at key, nearest data. Only for #MDB_DUPSORT */
363 MDB_GET_CURRENT, /**< Return key/data at current cursor position */
364 MDB_GET_MULTIPLE, /**< Return key and up to a page of duplicate data items
365 from current cursor position. Move cursor to prepare
366 for #MDB_NEXT_MULTIPLE. Only for #MDB_DUPFIXED */
367 MDB_LAST, /**< Position at last key/data item */
368 MDB_LAST_DUP, /**< Position at last data item of current key.
369 Only for #MDB_DUPSORT */
370 MDB_NEXT, /**< Position at next data item */
371 MDB_NEXT_DUP, /**< Position at next data item of current key.
372 Only for #MDB_DUPSORT */
373 MDB_NEXT_MULTIPLE, /**< Return key and up to a page of duplicate data items
374 from next cursor position. Move cursor to prepare
375 for #MDB_NEXT_MULTIPLE. Only for #MDB_DUPFIXED */
376 MDB_NEXT_NODUP, /**< Position at first data item of next key */
377 MDB_PREV, /**< Position at previous data item */
378 MDB_PREV_DUP, /**< Position at previous data item of current key.
379 Only for #MDB_DUPSORT */
380 MDB_PREV_NODUP, /**< Position at last data item of previous key */
381 MDB_SET, /**< Position at specified key */
382 MDB_SET_KEY, /**< Position at specified key, return key + data */
383 MDB_SET_RANGE /**< Position at first key greater than or equal to specified key. */
386 /** @defgroup errors Return Codes
388 * BerkeleyDB uses -30800 to -30999, we'll go under them
391 /** Successful result */
392 #define MDB_SUCCESS 0
393 /** key/data pair already exists */
394 #define MDB_KEYEXIST (-30799)
395 /** key/data pair not found (EOF) */
396 #define MDB_NOTFOUND (-30798)
397 /** Requested page not found - this usually indicates corruption */
398 #define MDB_PAGE_NOTFOUND (-30797)
399 /** Located page was wrong type */
400 #define MDB_CORRUPTED (-30796)
401 /** Update of meta page failed or environment had fatal error */
402 #define MDB_PANIC (-30795)
403 /** Environment version mismatch */
404 #define MDB_VERSION_MISMATCH (-30794)
405 /** File is not a valid LMDB file */
406 #define MDB_INVALID (-30793)
407 /** Environment mapsize reached */
408 #define MDB_MAP_FULL (-30792)
409 /** Environment maxdbs reached */
410 #define MDB_DBS_FULL (-30791)
411 /** Environment maxreaders reached */
412 #define MDB_READERS_FULL (-30790)
413 /** Too many TLS keys in use - Windows only */
414 #define MDB_TLS_FULL (-30789)
415 /** Txn has too many dirty pages */
416 #define MDB_TXN_FULL (-30788)
417 /** Cursor stack too deep - internal error */
418 #define MDB_CURSOR_FULL (-30787)
419 /** Page has not enough space - internal error */
420 #define MDB_PAGE_FULL (-30786)
421 /** Database contents grew beyond environment mapsize */
422 #define MDB_MAP_RESIZED (-30785)
423 /** Operation and DB incompatible, or DB type changed. This can mean:
425 * <li>The operation expects an #MDB_DUPSORT / #MDB_DUPFIXED database.
426 * <li>Opening a named DB when the unnamed DB has #MDB_DUPSORT / #MDB_INTEGERKEY.
427 * <li>Accessing a data record as a database, or vice versa.
428 * <li>The database was dropped and recreated with different flags.
431 #define MDB_INCOMPATIBLE (-30784)
432 /** Invalid reuse of reader locktable slot */
433 #define MDB_BAD_RSLOT (-30783)
434 /** Transaction must abort, has a child, or is invalid */
435 #define MDB_BAD_TXN (-30782)
436 /** Unsupported size of key/DB name/data, or wrong DUPFIXED size */
437 #define MDB_BAD_VALSIZE (-30781)
438 /** The specified DBI was changed unexpectedly */
439 #define MDB_BAD_DBI (-30780)
440 /** The last defined error code */
441 #define MDB_LAST_ERRCODE MDB_BAD_DBI
444 /** @brief Statistics for a database in the environment */
445 typedef struct MDB_stat {
446 unsigned int ms_psize; /**< Size of a database page.
447 This is currently the same for all databases. */
448 unsigned int ms_depth; /**< Depth (height) of the B-tree */
449 size_t ms_branch_pages; /**< Number of internal (non-leaf) pages */
450 size_t ms_leaf_pages; /**< Number of leaf pages */
451 size_t ms_overflow_pages; /**< Number of overflow pages */
452 size_t ms_entries; /**< Number of data items */
455 /** @brief Information about the environment */
456 typedef struct MDB_envinfo {
457 void *me_mapaddr; /**< Address of map, if fixed */
458 size_t me_mapsize; /**< Size of the data memory map */
459 size_t me_last_pgno; /**< ID of the last used page */
460 size_t me_last_txnid; /**< ID of the last committed transaction */
461 unsigned int me_maxreaders; /**< max reader slots in the environment */
462 unsigned int me_numreaders; /**< max reader slots used in the environment */
465 /** @brief Return the LMDB library version information.
467 * @param[out] major if non-NULL, the library major version number is copied here
468 * @param[out] minor if non-NULL, the library minor version number is copied here
469 * @param[out] patch if non-NULL, the library patch version number is copied here
470 * @retval "version string" The library version as a string
472 char *mdb_version(int *major, int *minor, int *patch);
474 /** @brief Return a string describing a given error code.
476 * This function is a superset of the ANSI C X3.159-1989 (ANSI C) strerror(3)
477 * function. If the error code is greater than or equal to 0, then the string
478 * returned by the system function strerror(3) is returned. If the error code
479 * is less than 0, an error string corresponding to the LMDB library error is
480 * returned. See @ref errors for a list of LMDB-specific error codes.
481 * @param[in] err The error code
482 * @retval "error message" The description of the error
484 char *mdb_strerror(int err);
486 /** @brief Create an LMDB environment handle.
488 * This function allocates memory for a #MDB_env structure. To release
489 * the allocated memory and discard the handle, call #mdb_env_close().
490 * Before the handle may be used, it must be opened using #mdb_env_open().
491 * Various other options may also need to be set before opening the handle,
492 * e.g. #mdb_env_set_mapsize(), #mdb_env_set_maxreaders(), #mdb_env_set_maxdbs(),
493 * depending on usage requirements.
494 * @param[out] env The address where the new handle will be stored
495 * @return A non-zero error value on failure and 0 on success.
497 int mdb_env_create(MDB_env **env);
499 /** @brief Open an environment handle.
501 * If this function fails, #mdb_env_close() must be called to discard the #MDB_env handle.
502 * @param[in] env An environment handle returned by #mdb_env_create()
503 * @param[in] path The directory in which the database files reside. This
504 * directory must already exist and be writable.
505 * @param[in] flags Special options for this environment. This parameter
506 * must be set to 0 or by bitwise OR'ing together one or more of the
507 * values described here.
508 * Flags set by mdb_env_set_flags() are also used.
511 * use a fixed address for the mmap region. This flag must be specified
512 * when creating the environment, and is stored persistently in the environment.
513 * If successful, the memory map will always reside at the same virtual address
514 * and pointers used to reference data items in the database will be constant
515 * across multiple invocations. This option may not always work, depending on
516 * how the operating system has allocated memory to shared libraries and other uses.
517 * The feature is highly experimental.
519 * By default, LMDB creates its environment in a directory whose
520 * pathname is given in \b path, and creates its data and lock files
521 * under that directory. With this option, \b path is used as-is for
522 * the database main data file. The database lock file is the \b path
523 * with "-lock" appended.
525 * Open the environment in read-only mode. No write operations will be
526 * allowed. LMDB will still modify the lock file - except on read-only
527 * filesystems, where LMDB does not use locks.
529 * Use a writeable memory map unless MDB_RDONLY is set. This is faster
530 * and uses fewer mallocs, but loses protection from application bugs
531 * like wild pointer writes and other bad updates into the database.
532 * Incompatible with nested transactions.
533 * Do not mix processes with and without MDB_WRITEMAP on the same
534 * environment. This can defeat durability (#mdb_env_sync etc).
535 * <li>#MDB_NOMETASYNC
536 * Flush system buffers to disk only once per transaction, omit the
537 * metadata flush. Defer that until the system flushes files to disk,
538 * or next non-MDB_RDONLY commit or #mdb_env_sync(). This optimization
539 * maintains database integrity, but a system crash may undo the last
540 * committed transaction. I.e. it preserves the ACI (atomicity,
541 * consistency, isolation) but not D (durability) database property.
542 * This flag may be changed at any time using #mdb_env_set_flags().
544 * Don't flush system buffers to disk when committing a transaction.
545 * This optimization means a system crash can corrupt the database or
546 * lose the last transactions if buffers are not yet flushed to disk.
547 * The risk is governed by how often the system flushes dirty buffers
548 * to disk and how often #mdb_env_sync() is called. However, if the
549 * filesystem preserves write order and the #MDB_WRITEMAP flag is not
550 * used, transactions exhibit ACI (atomicity, consistency, isolation)
551 * properties and only lose D (durability). I.e. database integrity
552 * is maintained, but a system crash may undo the final transactions.
553 * Note that (#MDB_NOSYNC | #MDB_WRITEMAP) leaves the system with no
554 * hint for when to write transactions to disk, unless #mdb_env_sync()
555 * is called. (#MDB_MAPASYNC | #MDB_WRITEMAP) may be preferable.
556 * This flag may be changed at any time using #mdb_env_set_flags().
558 * When using #MDB_WRITEMAP, use asynchronous flushes to disk.
559 * As with #MDB_NOSYNC, a system crash can then corrupt the
560 * database or lose the last transactions. Calling #mdb_env_sync()
561 * ensures on-disk database integrity until next commit.
562 * This flag may be changed at any time using #mdb_env_set_flags().
564 * Don't use Thread-Local Storage. Tie reader locktable slots to
565 * #MDB_txn objects instead of to threads. I.e. #mdb_txn_reset() keeps
566 * the slot reseved for the #MDB_txn object. A thread may use parallel
567 * read-only transactions. A read-only transaction may span threads if
568 * the user synchronizes its use. Applications that multiplex many
569 * user threads over individual OS threads need this option. Such an
570 * application must also serialize the write transactions in an OS
571 * thread, since LMDB's write locking is unaware of the user threads.
573 * Don't do any locking. If concurrent access is anticipated, the
574 * caller must manage all concurrency itself. For proper operation
575 * the caller must enforce single-writer semantics, and must ensure
576 * that no readers are using old transactions while a writer is
577 * active. The simplest approach is to use an exclusive lock so that
578 * no readers may be active at all when a writer begins.
580 * Turn off readahead. Most operating systems perform readahead on
581 * read requests by default. This option turns it off if the OS
582 * supports it. Turning it off may help random read performance
583 * when the DB is larger than RAM and system RAM is full.
584 * The option is not implemented on Windows.
586 * Don't initialize malloc'd memory before writing to unused spaces
587 * in the data file. By default, memory for pages written to the data
588 * file is obtained using malloc. While these pages may be reused in
589 * subsequent transactions, freshly malloc'd pages will be initialized
590 * to zeroes before use. This avoids persisting leftover data from other
591 * code (that used the heap and subsequently freed the memory) into the
592 * data file. Note that many other system libraries may allocate
593 * and free memory from the heap for arbitrary uses. E.g., stdio may
594 * use the heap for file I/O buffers. This initialization step has a
595 * modest performance cost so some applications may want to disable
596 * it using this flag. This option can be a problem for applications
597 * which handle sensitive data like passwords, and it makes memory
598 * checkers like Valgrind noisy. This flag is not needed with #MDB_WRITEMAP,
599 * which writes directly to the mmap instead of using malloc for pages. The
600 * initialization is also skipped if #MDB_RESERVE is used; the
601 * caller is expected to overwrite all of the memory that was
602 * reserved in that case.
603 * This flag may be changed at any time using #mdb_env_set_flags().
605 * @param[in] mode The UNIX permissions to set on created files and semaphores.
606 * This parameter is ignored on Windows.
607 * @return A non-zero error value on failure and 0 on success. Some possible
610 * <li>#MDB_VERSION_MISMATCH - the version of the LMDB library doesn't match the
611 * version that created the database environment.
612 * <li>#MDB_INVALID - the environment file headers are corrupted.
613 * <li>ENOENT - the directory specified by the path parameter doesn't exist.
614 * <li>EACCES - the user didn't have permission to access the environment files.
615 * <li>EAGAIN - the environment was locked by another process.
618 int mdb_env_open(MDB_env *env, const char *path, unsigned int flags, mdb_mode_t mode);
620 /** @brief Copy an LMDB environment to the specified path.
622 * This function may be used to make a backup of an existing environment.
623 * No lockfile is created, since it gets recreated at need.
624 * @note This call can trigger significant file size growth if run in
625 * parallel with write transactions, because it employs a read-only
626 * transaction. See long-lived transactions under @ref caveats_sec.
627 * @param[in] env An environment handle returned by #mdb_env_create(). It
628 * must have already been opened successfully.
629 * @param[in] path The directory in which the copy will reside. This
630 * directory must already exist and be writable but must otherwise be
632 * @return A non-zero error value on failure and 0 on success.
634 int mdb_env_copy(MDB_env *env, const char *path);
636 /** @brief Copy an LMDB environment to the specified file descriptor.
638 * This function may be used to make a backup of an existing environment.
639 * No lockfile is created, since it gets recreated at need.
640 * @note This call can trigger significant file size growth if run in
641 * parallel with write transactions, because it employs a read-only
642 * transaction. See long-lived transactions under @ref caveats_sec.
643 * @param[in] env An environment handle returned by #mdb_env_create(). It
644 * must have already been opened successfully.
645 * @param[in] fd The filedescriptor to write the copy to. It must
646 * have already been opened for Write access.
647 * @return A non-zero error value on failure and 0 on success.
649 int mdb_env_copyfd(MDB_env *env, mdb_filehandle_t fd);
651 /** @brief Copy an LMDB environment to the specified path, with options.
653 * This function may be used to make a backup of an existing environment.
654 * No lockfile is created, since it gets recreated at need.
655 * @note This call can trigger significant file size growth if run in
656 * parallel with write transactions, because it employs a read-only
657 * transaction. See long-lived transactions under @ref caveats_sec.
658 * @param[in] env An environment handle returned by #mdb_env_create(). It
659 * must have already been opened successfully.
660 * @param[in] path The directory in which the copy will reside. This
661 * directory must already exist and be writable but must otherwise be
663 * @param[in] flags Special options for this operation. This parameter
664 * must be set to 0 or by bitwise OR'ing together one or more of the
665 * values described here.
667 * <li>#MDB_CP_COMPACT - Perform compaction while copying: omit free
668 * pages and sequentially renumber all pages in output. This option
669 * consumes more CPU and runs more slowly than the default.
671 * @return A non-zero error value on failure and 0 on success.
673 int mdb_env_copy2(MDB_env *env, const char *path, unsigned int flags);
675 /** @brief Copy an LMDB environment to the specified file descriptor,
678 * This function may be used to make a backup of an existing environment.
679 * No lockfile is created, since it gets recreated at need. See
680 * #mdb_env_copy2() for further details.
681 * @note This call can trigger significant file size growth if run in
682 * parallel with write transactions, because it employs a read-only
683 * transaction. See long-lived transactions under @ref caveats_sec.
684 * @param[in] env An environment handle returned by #mdb_env_create(). It
685 * must have already been opened successfully.
686 * @param[in] fd The filedescriptor to write the copy to. It must
687 * have already been opened for Write access.
688 * @param[in] flags Special options for this operation.
689 * See #mdb_env_copy2() for options.
690 * @return A non-zero error value on failure and 0 on success.
692 int mdb_env_copyfd2(MDB_env *env, mdb_filehandle_t fd, unsigned int flags);
694 /** @brief Return statistics about the LMDB environment.
696 * @param[in] env An environment handle returned by #mdb_env_create()
697 * @param[out] stat The address of an #MDB_stat structure
698 * where the statistics will be copied
700 int mdb_env_stat(MDB_env *env, MDB_stat *stat);
702 /** @brief Return information about the LMDB environment.
704 * @param[in] env An environment handle returned by #mdb_env_create()
705 * @param[out] stat The address of an #MDB_envinfo structure
706 * where the information will be copied
708 int mdb_env_info(MDB_env *env, MDB_envinfo *stat);
710 /** @brief Flush the data buffers to disk.
712 * Data is always written to disk when #mdb_txn_commit() is called,
713 * but the operating system may keep it buffered. LMDB always flushes
714 * the OS buffers upon commit as well, unless the environment was
715 * opened with #MDB_NOSYNC or in part #MDB_NOMETASYNC. This call is
716 * not valid if the environment was opened with #MDB_RDONLY.
717 * @param[in] env An environment handle returned by #mdb_env_create()
718 * @param[in] force If non-zero, force a synchronous flush. Otherwise
719 * if the environment has the #MDB_NOSYNC flag set the flushes
720 * will be omitted, and with #MDB_MAPASYNC they will be asynchronous.
721 * @return A non-zero error value on failure and 0 on success. Some possible
724 * <li>EACCES - the environment is read-only.
725 * <li>EINVAL - an invalid parameter was specified.
726 * <li>EIO - an error occurred during synchronization.
729 int mdb_env_sync(MDB_env *env, int force);
731 /** @brief Close the environment and release the memory map.
733 * Only a single thread may call this function. All transactions, databases,
734 * and cursors must already be closed before calling this function. Attempts to
735 * use any such handles after calling this function will cause a SIGSEGV.
736 * The environment handle will be freed and must not be used again after this call.
737 * @param[in] env An environment handle returned by #mdb_env_create()
739 void mdb_env_close(MDB_env *env);
741 /** @brief Set environment flags.
743 * This may be used to set some flags in addition to those from
744 * #mdb_env_open(), or to unset these flags. If several threads
745 * change the flags at the same time, the result is undefined.
746 * @param[in] env An environment handle returned by #mdb_env_create()
747 * @param[in] flags The flags to change, bitwise OR'ed together
748 * @param[in] onoff A non-zero value sets the flags, zero clears them.
749 * @return A non-zero error value on failure and 0 on success. Some possible
752 * <li>EINVAL - an invalid parameter was specified.
755 int mdb_env_set_flags(MDB_env *env, unsigned int flags, int onoff);
757 /** @brief Get environment flags.
759 * @param[in] env An environment handle returned by #mdb_env_create()
760 * @param[out] flags The address of an integer to store the flags
761 * @return A non-zero error value on failure and 0 on success. Some possible
764 * <li>EINVAL - an invalid parameter was specified.
767 int mdb_env_get_flags(MDB_env *env, unsigned int *flags);
769 /** @brief Return the path that was used in #mdb_env_open().
771 * @param[in] env An environment handle returned by #mdb_env_create()
772 * @param[out] path Address of a string pointer to contain the path. This
773 * is the actual string in the environment, not a copy. It should not be
774 * altered in any way.
775 * @return A non-zero error value on failure and 0 on success. Some possible
778 * <li>EINVAL - an invalid parameter was specified.
781 int mdb_env_get_path(MDB_env *env, const char **path);
783 /** @brief Return the filedescriptor for the given environment.
785 * @param[in] env An environment handle returned by #mdb_env_create()
786 * @param[out] fd Address of a mdb_filehandle_t to contain the descriptor.
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_env_get_fd(MDB_env *env, mdb_filehandle_t *fd);
795 /** @brief Set the size of the memory map to use for this environment.
797 * The size should be a multiple of the OS page size. The default is
798 * 10485760 bytes. The size of the memory map is also the maximum size
799 * of the database. The value should be chosen as large as possible,
800 * to accommodate future growth of the database.
801 * This function should be called after #mdb_env_create() and before #mdb_env_open().
802 * It may be called at later times if no transactions are active in
803 * this process. Note that the library does not check for this condition,
804 * the caller must ensure it explicitly.
806 * The new size takes effect immediately for the current process but
807 * will not be persisted to any others until a write transaction has been
808 * committed by the current process. Also, only mapsize increases are
809 * persisted into the environment.
811 * If the mapsize is increased by another process, and data has grown
812 * beyond the range of the current mapsize, #mdb_txn_begin() will
813 * return #MDB_MAP_RESIZED. This function may be called with a size
814 * of zero to adopt the new size.
816 * Any attempt to set a size smaller than the space already consumed
817 * by the environment will be silently changed to the current size of the used space.
818 * @param[in] env An environment handle returned by #mdb_env_create()
819 * @param[in] size The size in bytes
820 * @return A non-zero error value on failure and 0 on success. Some possible
823 * <li>EINVAL - an invalid parameter was specified, or the environment has
824 * an active write transaction.
827 int mdb_env_set_mapsize(MDB_env *env, size_t size);
829 /** @brief Set the maximum number of threads/reader slots for the environment.
831 * This defines the number of slots in the lock table that is used to track readers in the
832 * the environment. The default is 126.
833 * Starting a read-only transaction normally ties a lock table slot to the
834 * current thread until the environment closes or the thread exits. If
835 * MDB_NOTLS is in use, #mdb_txn_begin() instead ties the slot to the
836 * MDB_txn object until it or the #MDB_env object is destroyed.
837 * This function may only be called after #mdb_env_create() and before #mdb_env_open().
838 * @param[in] env An environment handle returned by #mdb_env_create()
839 * @param[in] readers The maximum number of reader lock table slots
840 * @return A non-zero error value on failure and 0 on success. Some possible
843 * <li>EINVAL - an invalid parameter was specified, or the environment is already open.
846 int mdb_env_set_maxreaders(MDB_env *env, unsigned int readers);
848 /** @brief Get the maximum number of threads/reader slots for the environment.
850 * @param[in] env An environment handle returned by #mdb_env_create()
851 * @param[out] readers Address of an integer to store the number of readers
852 * @return A non-zero error value on failure and 0 on success. Some possible
855 * <li>EINVAL - an invalid parameter was specified.
858 int mdb_env_get_maxreaders(MDB_env *env, unsigned int *readers);
860 /** @brief Set the maximum number of named databases for the environment.
862 * This function is only needed if multiple databases will be used in the
863 * environment. Simpler applications that use the environment as a single
864 * unnamed database can ignore this option.
865 * This function may only be called after #mdb_env_create() and before #mdb_env_open().
867 * Currently a moderate number of slots are cheap but a huge number gets
868 * expensive: 7-120 words per transaction, and every #mdb_dbi_open()
869 * does a linear search of the opened slots.
870 * @param[in] env An environment handle returned by #mdb_env_create()
871 * @param[in] dbs The maximum number of databases
872 * @return A non-zero error value on failure and 0 on success. Some possible
875 * <li>EINVAL - an invalid parameter was specified, or the environment is already open.
878 int mdb_env_set_maxdbs(MDB_env *env, MDB_dbi dbs);
880 /** @brief Get the maximum size of keys and #MDB_DUPSORT data we can write.
882 * Depends on the compile-time constant #MDB_MAXKEYSIZE. Default 511.
884 * @param[in] env An environment handle returned by #mdb_env_create()
885 * @return The maximum size of a key we can write
887 int mdb_env_get_maxkeysize(MDB_env *env);
889 /** @brief Set application information associated with the #MDB_env.
891 * @param[in] env An environment handle returned by #mdb_env_create()
892 * @param[in] ctx An arbitrary pointer for whatever the application needs.
893 * @return A non-zero error value on failure and 0 on success.
895 int mdb_env_set_userctx(MDB_env *env, void *ctx);
897 /** @brief Get the application information associated with the #MDB_env.
899 * @param[in] env An environment handle returned by #mdb_env_create()
900 * @return The pointer set by #mdb_env_set_userctx().
902 void *mdb_env_get_userctx(MDB_env *env);
904 /** @brief A callback function for most LMDB assert() failures,
905 * called before printing the message and aborting.
907 * @param[in] env An environment handle returned by #mdb_env_create().
908 * @param[in] msg The assertion message, not including newline.
910 typedef void MDB_assert_func(MDB_env *env, const char *msg);
912 /** Set or reset the assert() callback of the environment.
913 * Disabled if liblmdb is buillt with NDEBUG.
914 * @note This hack should become obsolete as lmdb's error handling matures.
915 * @param[in] env An environment handle returned by #mdb_env_create().
916 * @param[in] func An #MDB_assert_func function, or 0.
917 * @return A non-zero error value on failure and 0 on success.
919 int mdb_env_set_assert(MDB_env *env, MDB_assert_func *func);
921 /** @brief Create a transaction for use with the environment.
923 * The transaction handle may be discarded using #mdb_txn_abort() or #mdb_txn_commit().
924 * @note A transaction and its cursors must only be used by a single
925 * thread, and a thread may only have a single transaction at a time.
926 * If #MDB_NOTLS is in use, this does not apply to read-only transactions.
927 * @note Cursors may not span transactions.
928 * @param[in] env An environment handle returned by #mdb_env_create()
929 * @param[in] parent If this parameter is non-NULL, the new transaction
930 * will be a nested transaction, with the transaction indicated by \b parent
931 * as its parent. Transactions may be nested to any level. A parent
932 * transaction and its cursors may not issue any other operations than
933 * mdb_txn_commit and mdb_txn_abort while it has active child transactions.
934 * @param[in] flags Special options for this transaction. This parameter
935 * must be set to 0 or by bitwise OR'ing together one or more of the
936 * values described here.
939 * This transaction will not perform any write operations.
941 * @param[out] txn Address where the new #MDB_txn handle will be stored
942 * @return A non-zero error value on failure and 0 on success. Some possible
945 * <li>#MDB_PANIC - a fatal error occurred earlier and the environment
947 * <li>#MDB_MAP_RESIZED - another process wrote data beyond this MDB_env's
948 * mapsize and this environment's map must be resized as well.
949 * See #mdb_env_set_mapsize().
950 * <li>#MDB_READERS_FULL - a read-only transaction was requested and
951 * the reader lock table is full. See #mdb_env_set_maxreaders().
952 * <li>ENOMEM - out of memory.
955 int mdb_txn_begin(MDB_env *env, MDB_txn *parent, unsigned int flags, MDB_txn **txn);
957 /** @brief Returns the transaction's #MDB_env
959 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
961 MDB_env *mdb_txn_env(MDB_txn *txn);
963 /** @brief Return the transaction's ID.
965 * This returns the identifier associated with this transaction. For a
966 * read-only transaction, this corresponds to the snapshot being read;
967 * concurrent readers will frequently have the same transaction ID.
969 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
970 * @return A transaction ID, valid if input is an active transaction.
972 size_t mdb_txn_id(MDB_txn *txn);
974 /** @brief Commit all the operations of a transaction into the database.
976 * The transaction handle is freed. It and its cursors must not be used
977 * again after this call, except with #mdb_cursor_renew().
978 * @note Earlier documentation incorrectly said all cursors would be freed.
979 * Only write-transactions free cursors.
980 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
981 * @return A non-zero error value on failure and 0 on success. Some possible
984 * <li>EINVAL - an invalid parameter was specified.
985 * <li>ENOSPC - no more disk space.
986 * <li>EIO - a low-level I/O error occurred while writing.
987 * <li>ENOMEM - out of memory.
990 int mdb_txn_commit(MDB_txn *txn);
992 /** @brief Abandon all the operations of the transaction instead of saving them.
994 * The transaction handle is freed. It and its cursors must not be used
995 * again after this call, except with #mdb_cursor_renew().
996 * @note Earlier documentation incorrectly said all cursors would be freed.
997 * Only write-transactions free cursors.
998 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1000 void mdb_txn_abort(MDB_txn *txn);
1002 /** @brief Reset a read-only transaction.
1004 * Abort the transaction like #mdb_txn_abort(), but keep the transaction
1005 * handle. #mdb_txn_renew() may reuse the handle. This saves allocation
1006 * overhead if the process will start a new read-only transaction soon,
1007 * and also locking overhead if #MDB_NOTLS is in use. The reader table
1008 * lock is released, but the table slot stays tied to its thread or
1009 * #MDB_txn. Use mdb_txn_abort() to discard a reset handle, and to free
1010 * its lock table slot if MDB_NOTLS is in use.
1011 * Cursors opened within the transaction must not be used
1012 * again after this call, except with #mdb_cursor_renew().
1013 * Reader locks generally don't interfere with writers, but they keep old
1014 * versions of database pages allocated. Thus they prevent the old pages
1015 * from being reused when writers commit new data, and so under heavy load
1016 * the database size may grow much more rapidly than otherwise.
1017 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1019 void mdb_txn_reset(MDB_txn *txn);
1021 /** @brief Renew a read-only transaction.
1023 * This acquires a new reader lock for a transaction handle that had been
1024 * released by #mdb_txn_reset(). It must be called before a reset transaction
1025 * may be used again.
1026 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1027 * @return A non-zero error value on failure and 0 on success. Some possible
1030 * <li>#MDB_PANIC - a fatal error occurred earlier and the environment
1031 * must be shut down.
1032 * <li>EINVAL - an invalid parameter was specified.
1035 int mdb_txn_renew(MDB_txn *txn);
1037 /** Compat with version <= 0.9.4, avoid clash with libmdb from MDB Tools project */
1038 #define mdb_open(txn,name,flags,dbi) mdb_dbi_open(txn,name,flags,dbi)
1039 /** Compat with version <= 0.9.4, avoid clash with libmdb from MDB Tools project */
1040 #define mdb_close(env,dbi) mdb_dbi_close(env,dbi)
1042 /** @brief Open a database in the environment.
1044 * A database handle denotes the name and parameters of a database,
1045 * independently of whether such a database exists.
1046 * The database handle may be discarded by calling #mdb_dbi_close().
1047 * The old database handle is returned if the database was already open.
1048 * The handle may only be closed once.
1050 * The database handle will be private to the current transaction until
1051 * the transaction is successfully committed. If the transaction is
1052 * aborted the handle will be closed automatically.
1053 * After a successful commit the handle will reside in the shared
1054 * environment, and may be used by other transactions.
1056 * This function must not be called from multiple concurrent
1057 * transactions in the same process. A transaction that uses
1058 * this function must finish (either commit or abort) before
1059 * any other transaction in the process may use this function.
1061 * To use named databases (with name != NULL), #mdb_env_set_maxdbs()
1062 * must be called before opening the environment. Database names are
1063 * keys in the unnamed database, and may be read but not written.
1065 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1066 * @param[in] name The name of the database to open. If only a single
1067 * database is needed in the environment, this value may be NULL.
1068 * @param[in] flags Special options for this database. This parameter
1069 * must be set to 0 or by bitwise OR'ing together one or more of the
1070 * values described here.
1072 * <li>#MDB_REVERSEKEY
1073 * Keys are strings to be compared in reverse order, from the end
1074 * of the strings to the beginning. By default, Keys are treated as strings and
1075 * compared from beginning to end.
1077 * Duplicate keys may be used in the database. (Or, from another perspective,
1078 * keys may have multiple data items, stored in sorted order.) By default
1079 * keys must be unique and may have only a single data item.
1080 * <li>#MDB_INTEGERKEY
1081 * Keys are binary integers in native byte order, either unsigned int
1082 * or size_t, and will be sorted as such.
1083 * The keys must all be of the same size.
1085 * This flag may only be used in combination with #MDB_DUPSORT. This option
1086 * tells the library that the data items for this database are all the same
1087 * size, which allows further optimizations in storage and retrieval. When
1088 * all data items are the same size, the #MDB_GET_MULTIPLE and #MDB_NEXT_MULTIPLE
1089 * cursor operations may be used to retrieve multiple items at once.
1090 * <li>#MDB_INTEGERDUP
1091 * This option specifies that duplicate data items are binary integers,
1092 * similar to #MDB_INTEGERKEY keys.
1093 * <li>#MDB_REVERSEDUP
1094 * This option specifies that duplicate data items should be compared as
1095 * strings in reverse order.
1097 * Create the named database if it doesn't exist. This option is not
1098 * allowed in a read-only transaction or a read-only environment.
1100 * @param[out] dbi Address where the new #MDB_dbi handle will be stored
1101 * @return A non-zero error value on failure and 0 on success. Some possible
1104 * <li>#MDB_NOTFOUND - the specified database doesn't exist in the environment
1105 * and #MDB_CREATE was not specified.
1106 * <li>#MDB_DBS_FULL - too many databases have been opened. See #mdb_env_set_maxdbs().
1109 int mdb_dbi_open(MDB_txn *txn, const char *name, unsigned int flags, MDB_dbi *dbi);
1111 /** @brief Retrieve statistics for a database.
1113 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1114 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1115 * @param[out] stat The address of an #MDB_stat structure
1116 * where the statistics will be copied
1117 * @return A non-zero error value on failure and 0 on success. Some possible
1120 * <li>EINVAL - an invalid parameter was specified.
1123 int mdb_stat(MDB_txn *txn, MDB_dbi dbi, MDB_stat *stat);
1125 /** @brief Retrieve the DB flags for a database handle.
1127 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1128 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1129 * @param[out] flags Address where the flags will be returned.
1130 * @return A non-zero error value on failure and 0 on success.
1132 int mdb_dbi_flags(MDB_txn *txn, MDB_dbi dbi, unsigned int *flags);
1134 /** @brief Close a database handle. Normally unnecessary. Use with care:
1136 * This call is not mutex protected. Handles should only be closed by
1137 * a single thread, and only if no other threads are going to reference
1138 * the database handle or one of its cursors any further. Do not close
1139 * a handle if an existing transaction has modified its database.
1140 * Doing so can cause misbehavior from database corruption to errors
1141 * like MDB_BAD_VALSIZE (since the DB name is gone).
1143 * Closing a database handle is not necessary, but lets #mdb_dbi_open()
1144 * reuse the handle value. Usually it's better to set a bigger
1145 * #mdb_env_set_maxdbs(), unless that value would be large.
1147 * @param[in] env An environment handle returned by #mdb_env_create()
1148 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1150 void mdb_dbi_close(MDB_env *env, MDB_dbi dbi);
1152 /** @brief Empty or delete+close a database.
1154 * See #mdb_dbi_close() for restrictions about closing the DB handle.
1155 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1156 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1157 * @param[in] del 0 to empty the DB, 1 to delete it from the
1158 * environment and close the DB handle.
1159 * @return A non-zero error value on failure and 0 on success.
1161 int mdb_drop(MDB_txn *txn, MDB_dbi dbi, int del);
1163 /** @brief Set a custom key comparison function for a database.
1165 * The comparison function is called whenever it is necessary to compare a
1166 * key specified by the application with a key currently stored in the database.
1167 * If no comparison function is specified, and no special key flags were specified
1168 * with #mdb_dbi_open(), the keys are compared lexically, with shorter keys collating
1169 * before longer keys.
1170 * @warning This function must be called before any data access functions are used,
1171 * otherwise data corruption may occur. The same comparison function must be used by every
1172 * program accessing the database, every time the database is used.
1173 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1174 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1175 * @param[in] cmp A #MDB_cmp_func function
1176 * @return A non-zero error value on failure and 0 on success. Some possible
1179 * <li>EINVAL - an invalid parameter was specified.
1182 int mdb_set_compare(MDB_txn *txn, MDB_dbi dbi, MDB_cmp_func *cmp);
1184 /** @brief Set a custom data comparison function for a #MDB_DUPSORT database.
1186 * This comparison function is called whenever it is necessary to compare a data
1187 * item specified by the application with a data item currently stored in the database.
1188 * This function only takes effect if the database was opened with the #MDB_DUPSORT
1190 * If no comparison function is specified, and no special key flags were specified
1191 * with #mdb_dbi_open(), the data items are compared lexically, with shorter items collating
1192 * before longer items.
1193 * @warning This function must be called before any data access functions are used,
1194 * otherwise data corruption may occur. The same comparison function must be used by every
1195 * program accessing the database, every time the database is used.
1196 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1197 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1198 * @param[in] cmp A #MDB_cmp_func function
1199 * @return A non-zero error value on failure and 0 on success. Some possible
1202 * <li>EINVAL - an invalid parameter was specified.
1205 int mdb_set_dupsort(MDB_txn *txn, MDB_dbi dbi, MDB_cmp_func *cmp);
1207 /** @brief Set a relocation function for a #MDB_FIXEDMAP database.
1209 * @todo The relocation function is called whenever it is necessary to move the data
1210 * of an item to a different position in the database (e.g. through tree
1211 * balancing operations, shifts as a result of adds or deletes, etc.). It is
1212 * intended to allow address/position-dependent data items to be stored in
1213 * a database in an environment opened with the #MDB_FIXEDMAP option.
1214 * Currently the relocation feature is unimplemented and setting
1215 * this function has no effect.
1216 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1217 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1218 * @param[in] rel A #MDB_rel_func function
1219 * @return A non-zero error value on failure and 0 on success. Some possible
1222 * <li>EINVAL - an invalid parameter was specified.
1225 int mdb_set_relfunc(MDB_txn *txn, MDB_dbi dbi, MDB_rel_func *rel);
1227 /** @brief Set a context pointer for a #MDB_FIXEDMAP database's relocation function.
1229 * See #mdb_set_relfunc and #MDB_rel_func for more details.
1230 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1231 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1232 * @param[in] ctx An arbitrary pointer for whatever the application needs.
1233 * It will be passed to the callback function set by #mdb_set_relfunc
1234 * as its \b relctx parameter whenever the callback is invoked.
1235 * @return A non-zero error value on failure and 0 on success. Some possible
1238 * <li>EINVAL - an invalid parameter was specified.
1241 int mdb_set_relctx(MDB_txn *txn, MDB_dbi dbi, void *ctx);
1243 /** @brief Get items from a database.
1245 * This function retrieves key/data pairs from the database. The address
1246 * and length of the data associated with the specified \b key are returned
1247 * in the structure to which \b data refers.
1248 * If the database supports duplicate keys (#MDB_DUPSORT) then the
1249 * first data item for the key will be returned. Retrieval of other
1250 * items requires the use of #mdb_cursor_get().
1252 * @note The memory pointed to by the returned values is owned by the
1253 * database. The caller need not dispose of the memory, and may not
1254 * modify it in any way. For values returned in a read-only transaction
1255 * any modification attempts will cause a SIGSEGV.
1256 * @note Values returned from the database are valid only until a
1257 * subsequent update operation, or the end of the transaction.
1258 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1259 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1260 * @param[in] key The key to search for in the database
1261 * @param[out] data The data corresponding to the key
1262 * @return A non-zero error value on failure and 0 on success. Some possible
1265 * <li>#MDB_NOTFOUND - the key was not in the database.
1266 * <li>EINVAL - an invalid parameter was specified.
1269 int mdb_get(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data);
1271 /** @brief Store items into a database.
1273 * This function stores key/data pairs in the database. The default behavior
1274 * is to enter the new key/data pair, replacing any previously existing key
1275 * if duplicates are disallowed, or adding a duplicate data item if
1276 * duplicates are allowed (#MDB_DUPSORT).
1277 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1278 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1279 * @param[in] key The key to store in the database
1280 * @param[in,out] data The data to store
1281 * @param[in] flags Special options for this operation. This parameter
1282 * must be set to 0 or by bitwise OR'ing together one or more of the
1283 * values described here.
1285 * <li>#MDB_NODUPDATA - enter the new key/data pair only if it does not
1286 * already appear in the database. This flag may only be specified
1287 * if the database was opened with #MDB_DUPSORT. The function will
1288 * return #MDB_KEYEXIST if the key/data pair already appears in the
1290 * <li>#MDB_NOOVERWRITE - enter the new key/data pair only if the key
1291 * does not already appear in the database. The function will return
1292 * #MDB_KEYEXIST if the key already appears in the database, even if
1293 * the database supports duplicates (#MDB_DUPSORT). The \b data
1294 * parameter will be set to point to the existing item.
1295 * <li>#MDB_RESERVE - reserve space for data of the given size, but
1296 * don't copy the given data. Instead, return a pointer to the
1297 * reserved space, which the caller can fill in later - before
1298 * the next update operation or the transaction ends. This saves
1299 * an extra memcpy if the data is being generated later.
1300 * LMDB does nothing else with this memory, the caller is expected
1301 * to modify all of the space requested. This flag must not be
1302 * specified if the database was opened with #MDB_DUPSORT.
1303 * <li>#MDB_APPEND - append the given key/data pair to the end of the
1304 * database. This option allows fast bulk loading when keys are
1305 * already known to be in the correct order. Loading unsorted keys
1306 * with this flag will cause a #MDB_KEYEXIST error.
1307 * <li>#MDB_APPENDDUP - as above, but for sorted dup data.
1309 * @return A non-zero error value on failure and 0 on success. Some possible
1312 * <li>#MDB_MAP_FULL - the database is full, see #mdb_env_set_mapsize().
1313 * <li>#MDB_TXN_FULL - the transaction has too many dirty pages.
1314 * <li>EACCES - an attempt was made to write in a read-only transaction.
1315 * <li>EINVAL - an invalid parameter was specified.
1318 int mdb_put(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data,
1319 unsigned int flags);
1321 /** @brief Delete items from a database.
1323 * This function removes key/data pairs from the database.
1324 * If the database does not support sorted duplicate data items
1325 * (#MDB_DUPSORT) the data parameter is ignored.
1326 * If the database supports sorted duplicates and the data parameter
1327 * is NULL, all of the duplicate data items for the key will be
1328 * deleted. Otherwise, if the data parameter is non-NULL
1329 * only the matching data item will be deleted.
1330 * This function will return #MDB_NOTFOUND if the specified key/data
1331 * pair is not in the database.
1332 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1333 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1334 * @param[in] key The key to delete from the database
1335 * @param[in] data The data to delete
1336 * @return A non-zero error value on failure and 0 on success. Some possible
1339 * <li>EACCES - an attempt was made to write in a read-only transaction.
1340 * <li>EINVAL - an invalid parameter was specified.
1343 int mdb_del(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data);
1345 /** @brief Create a cursor handle.
1347 * A cursor is associated with a specific transaction and database.
1348 * A cursor cannot be used when its database handle is closed. Nor
1349 * when its transaction has ended, except with #mdb_cursor_renew().
1350 * It can be discarded with #mdb_cursor_close().
1351 * A cursor in a write-transaction can be closed before its transaction
1352 * ends, and will otherwise be closed when its transaction ends.
1353 * A cursor in a read-only transaction must be closed explicitly, before
1354 * or after its transaction ends. It can be reused with
1355 * #mdb_cursor_renew() before finally closing it.
1356 * @note Earlier documentation said that cursors in every transaction
1357 * were closed when the transaction committed or aborted.
1358 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1359 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1360 * @param[out] cursor Address where the new #MDB_cursor handle will be stored
1361 * @return A non-zero error value on failure and 0 on success. Some possible
1364 * <li>EINVAL - an invalid parameter was specified.
1367 int mdb_cursor_open(MDB_txn *txn, MDB_dbi dbi, MDB_cursor **cursor);
1369 /** @brief Close a cursor handle.
1371 * The cursor handle will be freed and must not be used again after this call.
1372 * Its transaction must still be live if it is a write-transaction.
1373 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1375 void mdb_cursor_close(MDB_cursor *cursor);
1377 /** @brief Renew a cursor handle.
1379 * A cursor is associated with a specific transaction and database.
1380 * Cursors that are only used in read-only
1381 * transactions may be re-used, to avoid unnecessary malloc/free overhead.
1382 * The cursor may be associated with a new read-only transaction, and
1383 * referencing the same database handle as it was created with.
1384 * This may be done whether the previous transaction is live or dead.
1385 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1386 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1387 * @return A non-zero error value on failure and 0 on success. Some possible
1390 * <li>EINVAL - an invalid parameter was specified.
1393 int mdb_cursor_renew(MDB_txn *txn, MDB_cursor *cursor);
1395 /** @brief Return the cursor's transaction handle.
1397 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1399 MDB_txn *mdb_cursor_txn(MDB_cursor *cursor);
1401 /** @brief Return the cursor's database handle.
1403 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1405 MDB_dbi mdb_cursor_dbi(MDB_cursor *cursor);
1407 /** @brief Retrieve by cursor.
1409 * This function retrieves key/data pairs from the database. The address and length
1410 * of the key are returned in the object to which \b key refers (except for the
1411 * case of the #MDB_SET option, in which the \b key object is unchanged), and
1412 * the address and length of the data are returned in the object to which \b data
1414 * See #mdb_get() for restrictions on using the output values.
1415 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1416 * @param[in,out] key The key for a retrieved item
1417 * @param[in,out] data The data of a retrieved item
1418 * @param[in] op A cursor operation #MDB_cursor_op
1419 * @return A non-zero error value on failure and 0 on success. Some possible
1422 * <li>#MDB_NOTFOUND - no matching key found.
1423 * <li>EINVAL - an invalid parameter was specified.
1426 int mdb_cursor_get(MDB_cursor *cursor, MDB_val *key, MDB_val *data,
1429 /** @brief Store by cursor.
1431 * This function stores key/data pairs into the database.
1432 * The cursor is positioned at the new item, or on failure usually near it.
1433 * @note Earlier documentation incorrectly said errors would leave the
1434 * state of the cursor unchanged.
1435 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1436 * @param[in] key The key operated on.
1437 * @param[in] data The data operated on.
1438 * @param[in] flags Options for this operation. This parameter
1439 * must be set to 0 or one of the values described here.
1441 * <li>#MDB_CURRENT - replace the item at the current cursor position.
1442 * The \b key parameter must still be provided, and must match it.
1443 * If using sorted duplicates (#MDB_DUPSORT) the data item must still
1444 * sort into the same place. This is intended to be used when the
1445 * new data is the same size as the old. Otherwise it will simply
1446 * perform a delete of the old record followed by an insert.
1447 * <li>#MDB_NODUPDATA - enter the new key/data pair only if it does not
1448 * already appear in the database. This flag may only be specified
1449 * if the database was opened with #MDB_DUPSORT. The function will
1450 * return #MDB_KEYEXIST if the key/data pair already appears in the
1452 * <li>#MDB_NOOVERWRITE - enter the new key/data pair only if the key
1453 * does not already appear in the database. The function will return
1454 * #MDB_KEYEXIST if the key already appears in the database, even if
1455 * the database supports duplicates (#MDB_DUPSORT).
1456 * <li>#MDB_RESERVE - reserve space for data of the given size, but
1457 * don't copy the given data. Instead, return a pointer to the
1458 * reserved space, which the caller can fill in later. This saves
1459 * an extra memcpy if the data is being generated later. This flag
1460 * must not be specified if the database was opened with #MDB_DUPSORT.
1461 * <li>#MDB_APPEND - append the given key/data pair to the end of the
1462 * database. No key comparisons are performed. This option allows
1463 * fast bulk loading when keys are already known to be in the
1464 * correct order. Loading unsorted keys with this flag will cause
1465 * a #MDB_KEYEXIST error.
1466 * <li>#MDB_APPENDDUP - as above, but for sorted dup data.
1467 * <li>#MDB_MULTIPLE - store multiple contiguous data elements in a
1468 * single request. This flag may only be specified if the database
1469 * was opened with #MDB_DUPFIXED. The \b data argument must be an
1470 * array of two MDB_vals. The mv_size of the first MDB_val must be
1471 * the size of a single data element. The mv_data of the first MDB_val
1472 * must point to the beginning of the array of contiguous data elements.
1473 * The mv_size of the second MDB_val must be the count of the number
1474 * of data elements to store. On return this field will be set to
1475 * the count of the number of elements actually written. The mv_data
1476 * of the second MDB_val is unused.
1478 * @return A non-zero error value on failure and 0 on success. Some possible
1481 * <li>#MDB_MAP_FULL - the database is full, see #mdb_env_set_mapsize().
1482 * <li>#MDB_TXN_FULL - the transaction has too many dirty pages.
1483 * <li>EACCES - an attempt was made to write in a read-only transaction.
1484 * <li>EINVAL - an invalid parameter was specified.
1487 int mdb_cursor_put(MDB_cursor *cursor, MDB_val *key, MDB_val *data,
1488 unsigned int flags);
1490 /** @brief Delete current key/data pair
1492 * This function deletes the key/data pair to which the cursor refers.
1493 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1494 * @param[in] flags Options for this operation. This parameter
1495 * must be set to 0 or one of the values described here.
1497 * <li>#MDB_NODUPDATA - delete all of the data items for the current key.
1498 * This flag may only be specified if the database was opened with #MDB_DUPSORT.
1500 * @return A non-zero error value on failure and 0 on success. Some possible
1503 * <li>EACCES - an attempt was made to write in a read-only transaction.
1504 * <li>EINVAL - an invalid parameter was specified.
1507 int mdb_cursor_del(MDB_cursor *cursor, unsigned int flags);
1509 /** @brief Return count of duplicates for current key.
1511 * This call is only valid on databases that support sorted duplicate
1512 * data items #MDB_DUPSORT.
1513 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1514 * @param[out] countp Address where the count will be stored
1515 * @return A non-zero error value on failure and 0 on success. Some possible
1518 * <li>EINVAL - cursor is not initialized, or an invalid parameter was specified.
1521 int mdb_cursor_count(MDB_cursor *cursor, size_t *countp);
1523 /** @brief Compare two data items according to a particular database.
1525 * This returns a comparison as if the two data items were keys in the
1526 * specified database.
1527 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1528 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1529 * @param[in] a The first item to compare
1530 * @param[in] b The second item to compare
1531 * @return < 0 if a < b, 0 if a == b, > 0 if a > b
1533 int mdb_cmp(MDB_txn *txn, MDB_dbi dbi, const MDB_val *a, const MDB_val *b);
1535 /** @brief Compare two data items according to a particular database.
1537 * This returns a comparison as if the two items were data items of
1538 * the specified database. The database must have the #MDB_DUPSORT flag.
1539 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1540 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1541 * @param[in] a The first item to compare
1542 * @param[in] b The second item to compare
1543 * @return < 0 if a < b, 0 if a == b, > 0 if a > b
1545 int mdb_dcmp(MDB_txn *txn, MDB_dbi dbi, const MDB_val *a, const MDB_val *b);
1547 /** @brief A callback function used to print a message from the library.
1549 * @param[in] msg The string to be printed.
1550 * @param[in] ctx An arbitrary context pointer for the callback.
1551 * @return < 0 on failure, >= 0 on success.
1553 typedef int (MDB_msg_func)(const char *msg, void *ctx);
1555 /** @brief Dump the entries in the reader lock table.
1557 * @param[in] env An environment handle returned by #mdb_env_create()
1558 * @param[in] func A #MDB_msg_func function
1559 * @param[in] ctx Anything the message function needs
1560 * @return < 0 on failure, >= 0 on success.
1562 int mdb_reader_list(MDB_env *env, MDB_msg_func *func, void *ctx);
1564 /** @brief Check for stale entries in the reader lock table.
1566 * @param[in] env An environment handle returned by #mdb_env_create()
1567 * @param[out] dead Number of stale slots that were cleared
1568 * @return 0 on success, non-zero on failure.
1570 int mdb_reader_check(MDB_env *env, int *dead);
1576 /** @page tools LMDB Command Line Tools
1577 The following describes the command line tools that are available for LMDB.
1584 #endif /* _LMDB_H_ */