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 most systems:
54 * - Windows - automatic
55 * - BSD, systems using SysV semaphores - automatic
56 * - Linux, systems using POSIX mutexes with Robust option - automatic
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_SYSV_SEM or
62 * startup can fail due to semaphores owned by another userid.
64 * Fix: Open and close the database as the user which owns the
65 * semaphores (likely last user) or as root, while no other
66 * process is using the database.
68 * Restrictions/caveats (in addition to those listed for some functions):
70 * - Only the database owner should normally use the database on
71 * BSD systems or when otherwise configured with MDB_USE_POSIX_SEM.
72 * Multiple users can cause startup to fail later, as noted above.
74 * - There is normally no pure read-only mode, since readers need write
75 * access to locks and lock file. Exceptions: On read-only filesystems
76 * or with the #MDB_NOLOCK flag described under #mdb_env_open().
78 * - By default, in versions before 0.9.10, unused portions of the data
79 * file might receive garbage data from memory freed by other code.
80 * (This does not happen when using the #MDB_WRITEMAP flag.) As of
81 * 0.9.10 the default behavior is to initialize such memory before
82 * writing to the data file. Since there may be a slight performance
83 * cost due to this initialization, applications may disable it using
84 * the #MDB_NOMEMINIT flag. Applications handling sensitive data
85 * which must not be written should not use this flag. This flag is
86 * irrelevant when using #MDB_WRITEMAP.
88 * - A thread can only use one transaction at a time, plus any child
89 * transactions. Each transaction belongs to one thread. See below.
90 * The #MDB_NOTLS flag changes this for read-only transactions.
92 * - Use an MDB_env* in the process which opened it, without fork()ing.
94 * - Do not have open an LMDB database twice in the same process at
95 * the same time. Not even from a plain open() call - close()ing it
96 * breaks flock() advisory locking.
98 * - Avoid long-lived transactions. Read transactions prevent
99 * reuse of pages freed by newer write transactions, thus the
100 * database can grow quickly. Write transactions prevent
101 * other write transactions, since writes are serialized.
103 * - Avoid suspending a process with active transactions. These
104 * would then be "long-lived" as above. Also read transactions
105 * suspended when writers commit could sometimes see wrong data.
107 * ...when several processes can use a database concurrently:
109 * - Avoid aborting a process with an active transaction.
110 * The transaction becomes "long-lived" as above until a check
111 * for stale readers is performed or the lockfile is reset,
112 * since the process may not remove it from the lockfile.
114 * This does not apply to write transactions if the system clears
115 * stale writers, see above.
117 * - If you do that anyway, do a periodic check for stale readers. Or
118 * close the environment once in a while, so the lockfile can get reset.
120 * - Do not use LMDB databases on remote filesystems, even between
121 * processes on the same host. This breaks flock() on some OSes,
122 * possibly memory map sync, and certainly sync between programs
123 * on different hosts.
125 * - Opening a database can fail if another process is opening or
126 * closing it at exactly the same time.
128 * @author Howard Chu, Symas Corporation.
130 * @copyright Copyright 2011-2015 Howard Chu, Symas Corp. All rights reserved.
132 * Redistribution and use in source and binary forms, with or without
133 * modification, are permitted only as authorized by the OpenLDAP
136 * A copy of this license is available in the file LICENSE in the
137 * top-level directory of the distribution or, alternatively, at
138 * <http://www.OpenLDAP.org/license.html>.
141 * This code is derived from btree.c written by Martin Hedenfalk.
143 * Copyright (c) 2009, 2010 Martin Hedenfalk <martin@bzero.se>
145 * Permission to use, copy, modify, and distribute this software for any
146 * purpose with or without fee is hereby granted, provided that the above
147 * copyright notice and this permission notice appear in all copies.
149 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
150 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
151 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
152 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
153 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
154 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
155 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
160 #include <sys/types.h>
166 /** Unix permissions for creating files, or dummy definition for Windows */
168 typedef int mdb_mode_t;
170 typedef mode_t mdb_mode_t;
173 /** An abstraction for a file handle.
174 * On POSIX systems file handles are small integers. On Windows
175 * they're opaque pointers.
178 typedef void *mdb_filehandle_t;
180 typedef int mdb_filehandle_t;
183 /** @defgroup mdb LMDB API
185 * @brief OpenLDAP Lightning Memory-Mapped Database Manager
187 /** @defgroup Version Version Macros
190 /** Library major version */
191 #define MDB_VERSION_MAJOR 0
192 /** Library minor version */
193 #define MDB_VERSION_MINOR 9
194 /** Library patch version */
195 #define MDB_VERSION_PATCH 15
197 /** Combine args a,b,c into a single integer for easy version comparisons */
198 #define MDB_VERINT(a,b,c) (((a) << 24) | ((b) << 16) | (c))
200 /** The full library version as a single integer */
201 #define MDB_VERSION_FULL \
202 MDB_VERINT(MDB_VERSION_MAJOR,MDB_VERSION_MINOR,MDB_VERSION_PATCH)
204 /** The release date of this library version */
205 #define MDB_VERSION_DATE "June 19, 2015"
207 /** A stringifier for the version info */
208 #define MDB_VERSTR(a,b,c,d) "LMDB " #a "." #b "." #c ": (" d ")"
210 /** A helper for the stringifier macro */
211 #define MDB_VERFOO(a,b,c,d) MDB_VERSTR(a,b,c,d)
213 /** The full library version as a C string */
214 #define MDB_VERSION_STRING \
215 MDB_VERFOO(MDB_VERSION_MAJOR,MDB_VERSION_MINOR,MDB_VERSION_PATCH,MDB_VERSION_DATE)
218 /** @brief Opaque structure for a database environment.
220 * A DB environment supports multiple databases, all residing in the same
223 typedef struct MDB_env MDB_env;
225 /** @brief Opaque structure for a transaction handle.
227 * All database operations require a transaction handle. Transactions may be
228 * read-only or read-write.
230 typedef struct MDB_txn MDB_txn;
232 /** @brief A handle for an individual database in the DB environment. */
233 typedef unsigned int MDB_dbi;
235 /** @brief Opaque structure for navigating through a database */
236 typedef struct MDB_cursor MDB_cursor;
238 /** @brief Generic structure used for passing keys and data in and out
241 * Values returned from the database are valid only until a subsequent
242 * update operation, or the end of the transaction. Do not modify or
243 * free them, they commonly point into the database itself.
245 * Key sizes must be between 1 and #mdb_env_get_maxkeysize() inclusive.
246 * The same applies to data sizes in databases with the #MDB_DUPSORT flag.
247 * Other data items can in theory be from 0 to 0xffffffff bytes long.
249 typedef struct MDB_val {
250 size_t mv_size; /**< size of the data item */
251 void *mv_data; /**< address of the data item */
254 /** @brief A callback function used to compare two keys in a database */
255 typedef int (MDB_cmp_func)(const MDB_val *a, const MDB_val *b);
257 /** @brief A callback function used to relocate a position-dependent data item
258 * in a fixed-address database.
260 * The \b newptr gives the item's desired address in
261 * the memory map, and \b oldptr gives its previous address. The item's actual
262 * data resides at the address in \b item. This callback is expected to walk
263 * through the fields of the record in \b item and modify any
264 * values based at the \b oldptr address to be relative to the \b newptr address.
265 * @param[in,out] item The item that is to be relocated.
266 * @param[in] oldptr The previous address.
267 * @param[in] newptr The new address to relocate to.
268 * @param[in] relctx An application-provided context, set by #mdb_set_relctx().
269 * @todo This feature is currently unimplemented.
271 typedef void (MDB_rel_func)(MDB_val *item, void *oldptr, void *newptr, void *relctx);
273 /** @defgroup mdb_env Environment Flags
276 /** mmap at a fixed address (experimental) */
277 #define MDB_FIXEDMAP 0x01
278 /** no environment directory */
279 #define MDB_NOSUBDIR 0x4000
280 /** don't fsync after commit */
281 #define MDB_NOSYNC 0x10000
283 #define MDB_RDONLY 0x20000
284 /** don't fsync metapage after commit */
285 #define MDB_NOMETASYNC 0x40000
286 /** use writable mmap */
287 #define MDB_WRITEMAP 0x80000
288 /** use asynchronous msync when #MDB_WRITEMAP is used */
289 #define MDB_MAPASYNC 0x100000
290 /** tie reader locktable slots to #MDB_txn objects instead of to threads */
291 #define MDB_NOTLS 0x200000
292 /** don't do any locking, caller must manage their own locks */
293 #define MDB_NOLOCK 0x400000
294 /** don't do readahead (no effect on Windows) */
295 #define MDB_NORDAHEAD 0x800000
296 /** don't initialize malloc'd memory before writing to datafile */
297 #define MDB_NOMEMINIT 0x1000000
300 /** @defgroup mdb_dbi_open Database Flags
303 /** use reverse string keys */
304 #define MDB_REVERSEKEY 0x02
305 /** use sorted duplicates */
306 #define MDB_DUPSORT 0x04
307 /** numeric keys in native byte order: either unsigned int or size_t.
308 * The keys must all be of the same size. */
309 #define MDB_INTEGERKEY 0x08
310 /** with #MDB_DUPSORT, sorted dup items have fixed size */
311 #define MDB_DUPFIXED 0x10
312 /** with #MDB_DUPSORT, dups are #MDB_INTEGERKEY-style integers */
313 #define MDB_INTEGERDUP 0x20
314 /** with #MDB_DUPSORT, use reverse string dups */
315 #define MDB_REVERSEDUP 0x40
316 /** create DB if not already existing */
317 #define MDB_CREATE 0x40000
320 /** @defgroup mdb_put Write Flags
323 /** For put: Don't write if the key already exists. */
324 #define MDB_NOOVERWRITE 0x10
325 /** Only for #MDB_DUPSORT<br>
326 * For put: don't write if the key and data pair already exist.<br>
327 * For mdb_cursor_del: remove all duplicate data items.
329 #define MDB_NODUPDATA 0x20
330 /** For mdb_cursor_put: overwrite the current key/data pair */
331 #define MDB_CURRENT 0x40
332 /** For put: Just reserve space for data, don't copy it. Return a
333 * pointer to the reserved space.
335 #define MDB_RESERVE 0x10000
336 /** Data is being appended, don't split full pages. */
337 #define MDB_APPEND 0x20000
338 /** Duplicate data is being appended, don't split full pages. */
339 #define MDB_APPENDDUP 0x40000
340 /** Store multiple data items in one call. Only for #MDB_DUPFIXED. */
341 #define MDB_MULTIPLE 0x80000
344 /** @defgroup mdb_copy Copy Flags
347 /** Compacting copy: Omit free space from copy, and renumber all
348 * pages sequentially.
350 #define MDB_CP_COMPACT 0x01
353 /** @brief Cursor Get operations.
355 * This is the set of all operations for retrieving data
358 typedef enum MDB_cursor_op {
359 MDB_FIRST, /**< Position at first key/data item */
360 MDB_FIRST_DUP, /**< Position at first data item of current key.
361 Only for #MDB_DUPSORT */
362 MDB_GET_BOTH, /**< Position at key/data pair. Only for #MDB_DUPSORT */
363 MDB_GET_BOTH_RANGE, /**< position at key, nearest data. Only for #MDB_DUPSORT */
364 MDB_GET_CURRENT, /**< Return key/data at current cursor position */
365 MDB_GET_MULTIPLE, /**< Return key and up to a page of duplicate data items
366 from current cursor position. Move cursor to prepare
367 for #MDB_NEXT_MULTIPLE. Only for #MDB_DUPFIXED */
368 MDB_LAST, /**< Position at last key/data item */
369 MDB_LAST_DUP, /**< Position at last data item of current key.
370 Only for #MDB_DUPSORT */
371 MDB_NEXT, /**< Position at next data item */
372 MDB_NEXT_DUP, /**< Position at next data item of current key.
373 Only for #MDB_DUPSORT */
374 MDB_NEXT_MULTIPLE, /**< Return key and up to a page of duplicate data items
375 from next cursor position. Move cursor to prepare
376 for #MDB_NEXT_MULTIPLE. Only for #MDB_DUPFIXED */
377 MDB_NEXT_NODUP, /**< Position at first data item of next key */
378 MDB_PREV, /**< Position at previous data item */
379 MDB_PREV_DUP, /**< Position at previous data item of current key.
380 Only for #MDB_DUPSORT */
381 MDB_PREV_NODUP, /**< Position at last data item of previous key */
382 MDB_SET, /**< Position at specified key */
383 MDB_SET_KEY, /**< Position at specified key, return key + data */
384 MDB_SET_RANGE /**< Position at first key greater than or equal to specified key. */
387 /** @defgroup errors Return Codes
389 * BerkeleyDB uses -30800 to -30999, we'll go under them
392 /** Successful result */
393 #define MDB_SUCCESS 0
394 /** key/data pair already exists */
395 #define MDB_KEYEXIST (-30799)
396 /** key/data pair not found (EOF) */
397 #define MDB_NOTFOUND (-30798)
398 /** Requested page not found - this usually indicates corruption */
399 #define MDB_PAGE_NOTFOUND (-30797)
400 /** Located page was wrong type */
401 #define MDB_CORRUPTED (-30796)
402 /** Update of meta page failed or environment had fatal error */
403 #define MDB_PANIC (-30795)
404 /** Environment version mismatch */
405 #define MDB_VERSION_MISMATCH (-30794)
406 /** File is not a valid LMDB file */
407 #define MDB_INVALID (-30793)
408 /** Environment mapsize reached */
409 #define MDB_MAP_FULL (-30792)
410 /** Environment maxdbs reached */
411 #define MDB_DBS_FULL (-30791)
412 /** Environment maxreaders reached */
413 #define MDB_READERS_FULL (-30790)
414 /** Too many TLS keys in use - Windows only */
415 #define MDB_TLS_FULL (-30789)
416 /** Txn has too many dirty pages */
417 #define MDB_TXN_FULL (-30788)
418 /** Cursor stack too deep - internal error */
419 #define MDB_CURSOR_FULL (-30787)
420 /** Page has not enough space - internal error */
421 #define MDB_PAGE_FULL (-30786)
422 /** Database contents grew beyond environment mapsize */
423 #define MDB_MAP_RESIZED (-30785)
424 /** Operation and DB incompatible, or DB type changed. This can mean:
426 * <li>The operation expects an #MDB_DUPSORT / #MDB_DUPFIXED database.
427 * <li>Opening a named DB when the unnamed DB has #MDB_DUPSORT / #MDB_INTEGERKEY.
428 * <li>Accessing a data record as a database, or vice versa.
429 * <li>The database was dropped and recreated with different flags.
432 #define MDB_INCOMPATIBLE (-30784)
433 /** Invalid reuse of reader locktable slot */
434 #define MDB_BAD_RSLOT (-30783)
435 /** Transaction cannot recover - it must be aborted */
436 #define MDB_BAD_TXN (-30782)
437 /** Unsupported size of key/DB name/data, or wrong DUPFIXED size */
438 #define MDB_BAD_VALSIZE (-30781)
439 /** The specified DBI was changed unexpectedly */
440 #define MDB_BAD_DBI (-30780)
441 /** The last defined error code */
442 #define MDB_LAST_ERRCODE MDB_BAD_DBI
445 /** @brief Statistics for a database in the environment */
446 typedef struct MDB_stat {
447 unsigned int ms_psize; /**< Size of a database page.
448 This is currently the same for all databases. */
449 unsigned int ms_depth; /**< Depth (height) of the B-tree */
450 size_t ms_branch_pages; /**< Number of internal (non-leaf) pages */
451 size_t ms_leaf_pages; /**< Number of leaf pages */
452 size_t ms_overflow_pages; /**< Number of overflow pages */
453 size_t ms_entries; /**< Number of data items */
456 /** @brief Information about the environment */
457 typedef struct MDB_envinfo {
458 void *me_mapaddr; /**< Address of map, if fixed */
459 size_t me_mapsize; /**< Size of the data memory map */
460 size_t me_last_pgno; /**< ID of the last used page */
461 size_t me_last_txnid; /**< ID of the last committed transaction */
462 unsigned int me_maxreaders; /**< max reader slots in the environment */
463 unsigned int me_numreaders; /**< max reader slots used in the environment */
466 /** @brief Return the LMDB library version information.
468 * @param[out] major if non-NULL, the library major version number is copied here
469 * @param[out] minor if non-NULL, the library minor version number is copied here
470 * @param[out] patch if non-NULL, the library patch version number is copied here
471 * @retval "version string" The library version as a string
473 char *mdb_version(int *major, int *minor, int *patch);
475 /** @brief Return a string describing a given error code.
477 * This function is a superset of the ANSI C X3.159-1989 (ANSI C) strerror(3)
478 * function. If the error code is greater than or equal to 0, then the string
479 * returned by the system function strerror(3) is returned. If the error code
480 * is less than 0, an error string corresponding to the LMDB library error is
481 * returned. See @ref errors for a list of LMDB-specific error codes.
482 * @param[in] err The error code
483 * @retval "error message" The description of the error
485 char *mdb_strerror(int err);
487 /** @brief Create an LMDB environment handle.
489 * This function allocates memory for a #MDB_env structure. To release
490 * the allocated memory and discard the handle, call #mdb_env_close().
491 * Before the handle may be used, it must be opened using #mdb_env_open().
492 * Various other options may also need to be set before opening the handle,
493 * e.g. #mdb_env_set_mapsize(), #mdb_env_set_maxreaders(), #mdb_env_set_maxdbs(),
494 * depending on usage requirements.
495 * @param[out] env The address where the new handle will be stored
496 * @return A non-zero error value on failure and 0 on success.
498 int mdb_env_create(MDB_env **env);
500 /** @brief Open an environment handle.
502 * If this function fails, #mdb_env_close() must be called to discard the #MDB_env handle.
503 * @param[in] env An environment handle returned by #mdb_env_create()
504 * @param[in] path The directory in which the database files reside. This
505 * directory must already exist and be writable.
506 * @param[in] flags Special options for this environment. This parameter
507 * must be set to 0 or by bitwise OR'ing together one or more of the
508 * values described here.
509 * Flags set by mdb_env_set_flags() are also used.
512 * use a fixed address for the mmap region. This flag must be specified
513 * when creating the environment, and is stored persistently in the environment.
514 * If successful, the memory map will always reside at the same virtual address
515 * and pointers used to reference data items in the database will be constant
516 * across multiple invocations. This option may not always work, depending on
517 * how the operating system has allocated memory to shared libraries and other uses.
518 * The feature is highly experimental.
520 * By default, LMDB creates its environment in a directory whose
521 * pathname is given in \b path, and creates its data and lock files
522 * under that directory. With this option, \b path is used as-is for
523 * the database main data file. The database lock file is the \b path
524 * with "-lock" appended.
526 * Open the environment in read-only mode. No write operations will be
527 * allowed. LMDB will still modify the lock file - except on read-only
528 * filesystems, where LMDB does not use locks.
530 * Use a writeable memory map unless MDB_RDONLY is set. This is faster
531 * and uses fewer mallocs, but loses protection from application bugs
532 * like wild pointer writes and other bad updates into the database.
533 * Incompatible with nested transactions.
534 * Do not mix processes with and without MDB_WRITEMAP on the same
535 * environment. This can defeat durability (#mdb_env_sync etc).
536 * <li>#MDB_NOMETASYNC
537 * Flush system buffers to disk only once per transaction, omit the
538 * metadata flush. Defer that until the system flushes files to disk,
539 * or next non-MDB_RDONLY commit or #mdb_env_sync(). This optimization
540 * maintains database integrity, but a system crash may undo the last
541 * committed transaction. I.e. it preserves the ACI (atomicity,
542 * consistency, isolation) but not D (durability) database property.
543 * This flag may be changed at any time using #mdb_env_set_flags().
545 * Don't flush system buffers to disk when committing a transaction.
546 * This optimization means a system crash can corrupt the database or
547 * lose the last transactions if buffers are not yet flushed to disk.
548 * The risk is governed by how often the system flushes dirty buffers
549 * to disk and how often #mdb_env_sync() is called. However, if the
550 * filesystem preserves write order and the #MDB_WRITEMAP flag is not
551 * used, transactions exhibit ACI (atomicity, consistency, isolation)
552 * properties and only lose D (durability). I.e. database integrity
553 * is maintained, but a system crash may undo the final transactions.
554 * Note that (#MDB_NOSYNC | #MDB_WRITEMAP) leaves the system with no
555 * hint for when to write transactions to disk, unless #mdb_env_sync()
556 * is called. (#MDB_MAPASYNC | #MDB_WRITEMAP) may be preferable.
557 * This flag may be changed at any time using #mdb_env_set_flags().
559 * When using #MDB_WRITEMAP, use asynchronous flushes to disk.
560 * As with #MDB_NOSYNC, a system crash can then corrupt the
561 * database or lose the last transactions. Calling #mdb_env_sync()
562 * ensures on-disk database integrity until next commit.
563 * This flag may be changed at any time using #mdb_env_set_flags().
565 * Don't use Thread-Local Storage. Tie reader locktable slots to
566 * #MDB_txn objects instead of to threads. I.e. #mdb_txn_reset() keeps
567 * the slot reseved for the #MDB_txn object. A thread may use parallel
568 * read-only transactions. A read-only transaction may span threads if
569 * the user synchronizes its use. Applications that multiplex many
570 * user threads over individual OS threads need this option. Such an
571 * application must also serialize the write transactions in an OS
572 * thread, since LMDB's write locking is unaware of the user threads.
574 * Don't do any locking. If concurrent access is anticipated, the
575 * caller must manage all concurrency itself. For proper operation
576 * the caller must enforce single-writer semantics, and must ensure
577 * that no readers are using old transactions while a writer is
578 * active. The simplest approach is to use an exclusive lock so that
579 * no readers may be active at all when a writer begins.
581 * Turn off readahead. Most operating systems perform readahead on
582 * read requests by default. This option turns it off if the OS
583 * supports it. Turning it off may help random read performance
584 * when the DB is larger than RAM and system RAM is full.
585 * The option is not implemented on Windows.
587 * Don't initialize malloc'd memory before writing to unused spaces
588 * in the data file. By default, memory for pages written to the data
589 * file is obtained using malloc. While these pages may be reused in
590 * subsequent transactions, freshly malloc'd pages will be initialized
591 * to zeroes before use. This avoids persisting leftover data from other
592 * code (that used the heap and subsequently freed the memory) into the
593 * data file. Note that many other system libraries may allocate
594 * and free memory from the heap for arbitrary uses. E.g., stdio may
595 * use the heap for file I/O buffers. This initialization step has a
596 * modest performance cost so some applications may want to disable
597 * it using this flag. This option can be a problem for applications
598 * which handle sensitive data like passwords, and it makes memory
599 * checkers like Valgrind noisy. This flag is not needed with #MDB_WRITEMAP,
600 * which writes directly to the mmap instead of using malloc for pages. The
601 * initialization is also skipped if #MDB_RESERVE is used; the
602 * caller is expected to overwrite all of the memory that was
603 * reserved in that case.
604 * This flag may be changed at any time using #mdb_env_set_flags().
606 * @param[in] mode The UNIX permissions to set on created files and semaphores.
607 * This parameter is ignored on Windows.
608 * @return A non-zero error value on failure and 0 on success. Some possible
611 * <li>#MDB_VERSION_MISMATCH - the version of the LMDB library doesn't match the
612 * version that created the database environment.
613 * <li>#MDB_INVALID - the environment file headers are corrupted.
614 * <li>ENOENT - the directory specified by the path parameter doesn't exist.
615 * <li>EACCES - the user didn't have permission to access the environment files.
616 * <li>EAGAIN - the environment was locked by another process.
619 int mdb_env_open(MDB_env *env, const char *path, unsigned int flags, mdb_mode_t mode);
621 /** @brief Copy an LMDB environment to the specified path.
623 * This function may be used to make a backup of an existing environment.
624 * No lockfile is created, since it gets recreated at need.
625 * @note This call can trigger significant file size growth if run in
626 * parallel with write transactions, because it employs a read-only
627 * transaction. See long-lived transactions under @ref caveats_sec.
628 * @param[in] env An environment handle returned by #mdb_env_create(). It
629 * must have already been opened successfully.
630 * @param[in] path The directory in which the copy will reside. This
631 * directory must already exist and be writable but must otherwise be
633 * @return A non-zero error value on failure and 0 on success.
635 int mdb_env_copy(MDB_env *env, const char *path);
637 /** @brief Copy an LMDB environment to the specified file descriptor.
639 * This function may be used to make a backup of an existing environment.
640 * No lockfile is created, since it gets recreated at need.
641 * @note This call can trigger significant file size growth if run in
642 * parallel with write transactions, because it employs a read-only
643 * transaction. See long-lived transactions under @ref caveats_sec.
644 * @param[in] env An environment handle returned by #mdb_env_create(). It
645 * must have already been opened successfully.
646 * @param[in] fd The filedescriptor to write the copy to. It must
647 * have already been opened for Write access.
648 * @return A non-zero error value on failure and 0 on success.
650 int mdb_env_copyfd(MDB_env *env, mdb_filehandle_t fd);
652 /** @brief Copy an LMDB environment to the specified path, with options.
654 * This function may be used to make a backup of an existing environment.
655 * No lockfile is created, since it gets recreated at need.
656 * @note This call can trigger significant file size growth if run in
657 * parallel with write transactions, because it employs a read-only
658 * transaction. See long-lived transactions under @ref caveats_sec.
659 * @param[in] env An environment handle returned by #mdb_env_create(). It
660 * must have already been opened successfully.
661 * @param[in] path The directory in which the copy will reside. This
662 * directory must already exist and be writable but must otherwise be
664 * @param[in] flags Special options for this operation. This parameter
665 * must be set to 0 or by bitwise OR'ing together one or more of the
666 * values described here.
668 * <li>#MDB_CP_COMPACT - Perform compaction while copying: omit free
669 * pages and sequentially renumber all pages in output. This option
670 * consumes more CPU and runs more slowly than the default.
672 * @return A non-zero error value on failure and 0 on success.
674 int mdb_env_copy2(MDB_env *env, const char *path, unsigned int flags);
676 /** @brief Copy an LMDB environment to the specified file descriptor,
679 * This function may be used to make a backup of an existing environment.
680 * No lockfile is created, since it gets recreated at need. See
681 * #mdb_env_copy2() for further details.
682 * @note This call can trigger significant file size growth if run in
683 * parallel with write transactions, because it employs a read-only
684 * transaction. See long-lived transactions under @ref caveats_sec.
685 * @param[in] env An environment handle returned by #mdb_env_create(). It
686 * must have already been opened successfully.
687 * @param[in] fd The filedescriptor to write the copy to. It must
688 * have already been opened for Write access.
689 * @param[in] flags Special options for this operation.
690 * See #mdb_env_copy2() for options.
691 * @return A non-zero error value on failure and 0 on success.
693 int mdb_env_copyfd2(MDB_env *env, mdb_filehandle_t fd, unsigned int flags);
695 /** @brief Return statistics about the LMDB environment.
697 * @param[in] env An environment handle returned by #mdb_env_create()
698 * @param[out] stat The address of an #MDB_stat structure
699 * where the statistics will be copied
701 int mdb_env_stat(MDB_env *env, MDB_stat *stat);
703 /** @brief Return information about the LMDB environment.
705 * @param[in] env An environment handle returned by #mdb_env_create()
706 * @param[out] stat The address of an #MDB_envinfo structure
707 * where the information will be copied
709 int mdb_env_info(MDB_env *env, MDB_envinfo *stat);
711 /** @brief Flush the data buffers to disk.
713 * Data is always written to disk when #mdb_txn_commit() is called,
714 * but the operating system may keep it buffered. LMDB always flushes
715 * the OS buffers upon commit as well, unless the environment was
716 * opened with #MDB_NOSYNC or in part #MDB_NOMETASYNC. This call is
717 * not valid if the environment was opened with #MDB_RDONLY.
718 * @param[in] env An environment handle returned by #mdb_env_create()
719 * @param[in] force If non-zero, force a synchronous flush. Otherwise
720 * if the environment has the #MDB_NOSYNC flag set the flushes
721 * will be omitted, and with #MDB_MAPASYNC they will be asynchronous.
722 * @return A non-zero error value on failure and 0 on success. Some possible
725 * <li>EACCES - the environment is read-only.
726 * <li>EINVAL - an invalid parameter was specified.
727 * <li>EIO - an error occurred during synchronization.
730 int mdb_env_sync(MDB_env *env, int force);
732 /** @brief Close the environment and release the memory map.
734 * Only a single thread may call this function. All transactions, databases,
735 * and cursors must already be closed before calling this function. Attempts to
736 * use any such handles after calling this function will cause a SIGSEGV.
737 * The environment handle will be freed and must not be used again after this call.
738 * @param[in] env An environment handle returned by #mdb_env_create()
740 void mdb_env_close(MDB_env *env);
742 /** @brief Set environment flags.
744 * This may be used to set some flags in addition to those from
745 * #mdb_env_open(), or to unset these flags. If several threads
746 * change the flags at the same time, the result is undefined.
747 * @param[in] env An environment handle returned by #mdb_env_create()
748 * @param[in] flags The flags to change, bitwise OR'ed together
749 * @param[in] onoff A non-zero value sets the flags, zero clears them.
750 * @return A non-zero error value on failure and 0 on success. Some possible
753 * <li>EINVAL - an invalid parameter was specified.
756 int mdb_env_set_flags(MDB_env *env, unsigned int flags, int onoff);
758 /** @brief Get environment flags.
760 * @param[in] env An environment handle returned by #mdb_env_create()
761 * @param[out] flags The address of an integer to store the flags
762 * @return A non-zero error value on failure and 0 on success. Some possible
765 * <li>EINVAL - an invalid parameter was specified.
768 int mdb_env_get_flags(MDB_env *env, unsigned int *flags);
770 /** @brief Return the path that was used in #mdb_env_open().
772 * @param[in] env An environment handle returned by #mdb_env_create()
773 * @param[out] path Address of a string pointer to contain the path. This
774 * is the actual string in the environment, not a copy. It should not be
775 * altered in any way.
776 * @return A non-zero error value on failure and 0 on success. Some possible
779 * <li>EINVAL - an invalid parameter was specified.
782 int mdb_env_get_path(MDB_env *env, const char **path);
784 /** @brief Return the filedescriptor for the given environment.
786 * @param[in] env An environment handle returned by #mdb_env_create()
787 * @param[out] fd Address of a mdb_filehandle_t to contain the descriptor.
788 * @return A non-zero error value on failure and 0 on success. Some possible
791 * <li>EINVAL - an invalid parameter was specified.
794 int mdb_env_get_fd(MDB_env *env, mdb_filehandle_t *fd);
796 /** @brief Set the size of the memory map to use for this environment.
798 * The size should be a multiple of the OS page size. The default is
799 * 10485760 bytes. The size of the memory map is also the maximum size
800 * of the database. The value should be chosen as large as possible,
801 * to accommodate future growth of the database.
802 * This function should be called after #mdb_env_create() and before #mdb_env_open().
803 * It may be called at later times if no transactions are active in
804 * this process. Note that the library does not check for this condition,
805 * the caller must ensure it explicitly.
807 * The new size takes effect immediately for the current process but
808 * will not be persisted to any others until a write transaction has been
809 * committed by the current process. Also, only mapsize increases are
810 * persisted into the environment.
812 * If the mapsize is increased by another process, and data has grown
813 * beyond the range of the current mapsize, #mdb_txn_begin() will
814 * return #MDB_MAP_RESIZED. This function may be called with a size
815 * of zero to adopt the new size.
817 * Any attempt to set a size smaller than the space already consumed
818 * by the environment will be silently changed to the current size of the used space.
819 * @param[in] env An environment handle returned by #mdb_env_create()
820 * @param[in] size The size in bytes
821 * @return A non-zero error value on failure and 0 on success. Some possible
824 * <li>EINVAL - an invalid parameter was specified, or the environment has
825 * an active write transaction.
828 int mdb_env_set_mapsize(MDB_env *env, size_t size);
830 /** @brief Set the maximum number of threads/reader slots for the environment.
832 * This defines the number of slots in the lock table that is used to track readers in the
833 * the environment. The default is 126.
834 * Starting a read-only transaction normally ties a lock table slot to the
835 * current thread until the environment closes or the thread exits. If
836 * MDB_NOTLS is in use, #mdb_txn_begin() instead ties the slot to the
837 * MDB_txn object until it or the #MDB_env object is destroyed.
838 * This function may only be called after #mdb_env_create() and before #mdb_env_open().
839 * @param[in] env An environment handle returned by #mdb_env_create()
840 * @param[in] readers The maximum number of reader lock table slots
841 * @return A non-zero error value on failure and 0 on success. Some possible
844 * <li>EINVAL - an invalid parameter was specified, or the environment is already open.
847 int mdb_env_set_maxreaders(MDB_env *env, unsigned int readers);
849 /** @brief Get the maximum number of threads/reader slots for the environment.
851 * @param[in] env An environment handle returned by #mdb_env_create()
852 * @param[out] readers Address of an integer to store the number of readers
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_env_get_maxreaders(MDB_env *env, unsigned int *readers);
861 /** @brief Set the maximum number of named databases for the environment.
863 * This function is only needed if multiple databases will be used in the
864 * environment. Simpler applications that use the environment as a single
865 * unnamed database can ignore this option.
866 * This function may only be called after #mdb_env_create() and before #mdb_env_open().
868 * Currently a moderate number of slots are cheap but a huge number gets
869 * expensive: 7-120 words per transaction, and every #mdb_dbi_open()
870 * does a linear search of the opened slots.
871 * @param[in] env An environment handle returned by #mdb_env_create()
872 * @param[in] dbs The maximum number of databases
873 * @return A non-zero error value on failure and 0 on success. Some possible
876 * <li>EINVAL - an invalid parameter was specified, or the environment is already open.
879 int mdb_env_set_maxdbs(MDB_env *env, MDB_dbi dbs);
881 /** @brief Get the maximum size of keys and #MDB_DUPSORT data we can write.
883 * Depends on the compile-time constant #MDB_MAXKEYSIZE. Default 511.
885 * @param[in] env An environment handle returned by #mdb_env_create()
886 * @return The maximum size of a key we can write
888 int mdb_env_get_maxkeysize(MDB_env *env);
890 /** @brief Set application information associated with the #MDB_env.
892 * @param[in] env An environment handle returned by #mdb_env_create()
893 * @param[in] ctx An arbitrary pointer for whatever the application needs.
894 * @return A non-zero error value on failure and 0 on success.
896 int mdb_env_set_userctx(MDB_env *env, void *ctx);
898 /** @brief Get the application information associated with the #MDB_env.
900 * @param[in] env An environment handle returned by #mdb_env_create()
901 * @return The pointer set by #mdb_env_set_userctx().
903 void *mdb_env_get_userctx(MDB_env *env);
905 /** @brief A callback function for most LMDB assert() failures,
906 * called before printing the message and aborting.
908 * @param[in] env An environment handle returned by #mdb_env_create().
909 * @param[in] msg The assertion message, not including newline.
911 typedef void MDB_assert_func(MDB_env *env, const char *msg);
913 /** Set or reset the assert() callback of the environment.
914 * Disabled if liblmdb is buillt with NDEBUG.
915 * @note This hack should become obsolete as lmdb's error handling matures.
916 * @param[in] env An environment handle returned by #mdb_env_create().
917 * @param[in] func An #MDB_assert_func function, or 0.
918 * @return A non-zero error value on failure and 0 on success.
920 int mdb_env_set_assert(MDB_env *env, MDB_assert_func *func);
922 /** @brief Create a transaction for use with the environment.
924 * The transaction handle may be discarded using #mdb_txn_abort() or #mdb_txn_commit().
925 * @note A transaction and its cursors must only be used by a single
926 * thread, and a thread may only have a single transaction at a time.
927 * If #MDB_NOTLS is in use, this does not apply to read-only transactions.
928 * @note Cursors may not span transactions.
929 * @param[in] env An environment handle returned by #mdb_env_create()
930 * @param[in] parent If this parameter is non-NULL, the new transaction
931 * will be a nested transaction, with the transaction indicated by \b parent
932 * as its parent. Transactions may be nested to any level. A parent
933 * transaction and its cursors may not issue any other operations than
934 * mdb_txn_commit and mdb_txn_abort while it has active child transactions.
935 * @param[in] flags Special options for this transaction. This parameter
936 * must be set to 0 or by bitwise OR'ing together one or more of the
937 * values described here.
940 * This transaction will not perform any write operations.
942 * Don't flush system buffers to disk when committing this transaction.
943 * <li>#MDB_NOMETASYNC
944 * Flush system buffers but omit metadata flush when committing this transaction.
946 * @param[out] txn Address where the new #MDB_txn handle will be stored
947 * @return A non-zero error value on failure and 0 on success. Some possible
950 * <li>#MDB_PANIC - a fatal error occurred earlier and the environment
952 * <li>#MDB_MAP_RESIZED - another process wrote data beyond this MDB_env's
953 * mapsize and this environment's map must be resized as well.
954 * See #mdb_env_set_mapsize().
955 * <li>#MDB_READERS_FULL - a read-only transaction was requested and
956 * the reader lock table is full. See #mdb_env_set_maxreaders().
957 * <li>ENOMEM - out of memory.
960 int mdb_txn_begin(MDB_env *env, MDB_txn *parent, unsigned int flags, MDB_txn **txn);
962 /** @brief Returns the transaction's #MDB_env
964 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
966 MDB_env *mdb_txn_env(MDB_txn *txn);
968 /** @brief Return the transaction's ID.
970 * This returns the identifier associated with this transaction. For a
971 * read-only transaction, this corresponds to the snapshot being read;
972 * concurrent readers will frequently have the same transaction ID.
974 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
975 * @return A transaction ID, valid if input is an active transaction.
977 size_t mdb_txn_id(MDB_txn *txn);
979 /** @brief Commit all the operations of a transaction into the database.
981 * The transaction handle is freed. It and its cursors must not be used
982 * again after this call, except with #mdb_cursor_renew().
983 * @note Earlier documentation incorrectly said all cursors would be freed.
984 * Only write-transactions free cursors.
985 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
986 * @return A non-zero error value on failure and 0 on success. Some possible
989 * <li>EINVAL - an invalid parameter was specified.
990 * <li>ENOSPC - no more disk space.
991 * <li>EIO - a low-level I/O error occurred while writing.
992 * <li>ENOMEM - out of memory.
995 int mdb_txn_commit(MDB_txn *txn);
997 /** @brief Abandon all the operations of the transaction instead of saving them.
999 * The transaction handle is freed. It and its cursors must not be used
1000 * again after this call, except with #mdb_cursor_renew().
1001 * @note Earlier documentation incorrectly said all cursors would be freed.
1002 * Only write-transactions free cursors.
1003 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1005 void mdb_txn_abort(MDB_txn *txn);
1007 /** @brief Reset a read-only transaction.
1009 * Abort the transaction like #mdb_txn_abort(), but keep the transaction
1010 * handle. #mdb_txn_renew() may reuse the handle. This saves allocation
1011 * overhead if the process will start a new read-only transaction soon,
1012 * and also locking overhead if #MDB_NOTLS is in use. The reader table
1013 * lock is released, but the table slot stays tied to its thread or
1014 * #MDB_txn. Use mdb_txn_abort() to discard a reset handle, and to free
1015 * its lock table slot if MDB_NOTLS is in use.
1016 * Cursors opened within the transaction must not be used
1017 * again after this call, except with #mdb_cursor_renew().
1018 * Reader locks generally don't interfere with writers, but they keep old
1019 * versions of database pages allocated. Thus they prevent the old pages
1020 * from being reused when writers commit new data, and so under heavy load
1021 * the database size may grow much more rapidly than otherwise.
1022 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1024 void mdb_txn_reset(MDB_txn *txn);
1026 /** @brief Renew a read-only transaction.
1028 * This acquires a new reader lock for a transaction handle that had been
1029 * released by #mdb_txn_reset(). It must be called before a reset transaction
1030 * may be used again.
1031 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1032 * @return A non-zero error value on failure and 0 on success. Some possible
1035 * <li>#MDB_PANIC - a fatal error occurred earlier and the environment
1036 * must be shut down.
1037 * <li>EINVAL - an invalid parameter was specified.
1040 int mdb_txn_renew(MDB_txn *txn);
1042 /** Compat with version <= 0.9.4, avoid clash with libmdb from MDB Tools project */
1043 #define mdb_open(txn,name,flags,dbi) mdb_dbi_open(txn,name,flags,dbi)
1044 /** Compat with version <= 0.9.4, avoid clash with libmdb from MDB Tools project */
1045 #define mdb_close(env,dbi) mdb_dbi_close(env,dbi)
1047 /** @brief Open a database in the environment.
1049 * A database handle denotes the name and parameters of a database,
1050 * independently of whether such a database exists.
1051 * The database handle may be discarded by calling #mdb_dbi_close().
1052 * The old database handle is returned if the database was already open.
1053 * The handle may only be closed once.
1055 * The database handle will be private to the current transaction until
1056 * the transaction is successfully committed. If the transaction is
1057 * aborted the handle will be closed automatically.
1058 * After a successful commit the handle will reside in the shared
1059 * environment, and may be used by other transactions.
1061 * This function must not be called from multiple concurrent
1062 * transactions in the same process. A transaction that uses
1063 * this function must finish (either commit or abort) before
1064 * any other transaction in the process may use this function.
1066 * To use named databases (with name != NULL), #mdb_env_set_maxdbs()
1067 * must be called before opening the environment. Database names are
1068 * keys in the unnamed database, and may be read but not written.
1070 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1071 * @param[in] name The name of the database to open. If only a single
1072 * database is needed in the environment, this value may be NULL.
1073 * @param[in] flags Special options for this database. This parameter
1074 * must be set to 0 or by bitwise OR'ing together one or more of the
1075 * values described here.
1077 * <li>#MDB_REVERSEKEY
1078 * Keys are strings to be compared in reverse order, from the end
1079 * of the strings to the beginning. By default, Keys are treated as strings and
1080 * compared from beginning to end.
1082 * Duplicate keys may be used in the database. (Or, from another perspective,
1083 * keys may have multiple data items, stored in sorted order.) By default
1084 * keys must be unique and may have only a single data item.
1085 * <li>#MDB_INTEGERKEY
1086 * Keys are binary integers in native byte order, either unsigned int
1087 * or size_t, and will be sorted as such.
1088 * The keys must all be of the same size.
1090 * This flag may only be used in combination with #MDB_DUPSORT. This option
1091 * tells the library that the data items for this database are all the same
1092 * size, which allows further optimizations in storage and retrieval. When
1093 * all data items are the same size, the #MDB_GET_MULTIPLE and #MDB_NEXT_MULTIPLE
1094 * cursor operations may be used to retrieve multiple items at once.
1095 * <li>#MDB_INTEGERDUP
1096 * This option specifies that duplicate data items are binary integers,
1097 * similar to #MDB_INTEGERKEY keys.
1098 * <li>#MDB_REVERSEDUP
1099 * This option specifies that duplicate data items should be compared as
1100 * strings in reverse order.
1102 * Create the named database if it doesn't exist. This option is not
1103 * allowed in a read-only transaction or a read-only environment.
1105 * @param[out] dbi Address where the new #MDB_dbi handle will be stored
1106 * @return A non-zero error value on failure and 0 on success. Some possible
1109 * <li>#MDB_NOTFOUND - the specified database doesn't exist in the environment
1110 * and #MDB_CREATE was not specified.
1111 * <li>#MDB_DBS_FULL - too many databases have been opened. See #mdb_env_set_maxdbs().
1114 int mdb_dbi_open(MDB_txn *txn, const char *name, unsigned int flags, MDB_dbi *dbi);
1116 /** @brief Retrieve statistics for a database.
1118 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1119 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1120 * @param[out] stat The address of an #MDB_stat structure
1121 * where the statistics will be copied
1122 * @return A non-zero error value on failure and 0 on success. Some possible
1125 * <li>EINVAL - an invalid parameter was specified.
1128 int mdb_stat(MDB_txn *txn, MDB_dbi dbi, MDB_stat *stat);
1130 /** @brief Retrieve the DB flags for a database handle.
1132 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1133 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1134 * @param[out] flags Address where the flags will be returned.
1135 * @return A non-zero error value on failure and 0 on success.
1137 int mdb_dbi_flags(MDB_txn *txn, MDB_dbi dbi, unsigned int *flags);
1139 /** @brief Close a database handle. Normally unnecessary. Use with care:
1141 * This call is not mutex protected. Handles should only be closed by
1142 * a single thread, and only if no other threads are going to reference
1143 * the database handle or one of its cursors any further. Do not close
1144 * a handle if an existing transaction has modified its database.
1145 * Doing so can cause misbehavior from database corruption to errors
1146 * like MDB_BAD_VALSIZE (since the DB name is gone).
1148 * Closing a database handle is not necessary, but lets #mdb_dbi_open()
1149 * reuse the handle value. Usually it's better to set a bigger
1150 * #mdb_env_set_maxdbs(), unless that value would be large.
1152 * @param[in] env An environment handle returned by #mdb_env_create()
1153 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1155 void mdb_dbi_close(MDB_env *env, MDB_dbi dbi);
1157 /** @brief Empty or delete+close a database.
1159 * See #mdb_dbi_close() for restrictions about closing the DB handle.
1160 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1161 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1162 * @param[in] del 0 to empty the DB, 1 to delete it from the
1163 * environment and close the DB handle.
1164 * @return A non-zero error value on failure and 0 on success.
1166 int mdb_drop(MDB_txn *txn, MDB_dbi dbi, int del);
1168 /** @brief Set a custom key comparison function for a database.
1170 * The comparison function is called whenever it is necessary to compare a
1171 * key specified by the application with a key currently stored in the database.
1172 * If no comparison function is specified, and no special key flags were specified
1173 * with #mdb_dbi_open(), the keys are compared lexically, with shorter keys collating
1174 * before longer keys.
1175 * @warning This function must be called before any data access functions are used,
1176 * otherwise data corruption may occur. The same comparison function must be used by every
1177 * program accessing the database, every time the database is used.
1178 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1179 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1180 * @param[in] cmp A #MDB_cmp_func function
1181 * @return A non-zero error value on failure and 0 on success. Some possible
1184 * <li>EINVAL - an invalid parameter was specified.
1187 int mdb_set_compare(MDB_txn *txn, MDB_dbi dbi, MDB_cmp_func *cmp);
1189 /** @brief Set a custom data comparison function for a #MDB_DUPSORT database.
1191 * This comparison function is called whenever it is necessary to compare a data
1192 * item specified by the application with a data item currently stored in the database.
1193 * This function only takes effect if the database was opened with the #MDB_DUPSORT
1195 * If no comparison function is specified, and no special key flags were specified
1196 * with #mdb_dbi_open(), the data items are compared lexically, with shorter items collating
1197 * before longer items.
1198 * @warning This function must be called before any data access functions are used,
1199 * otherwise data corruption may occur. The same comparison function must be used by every
1200 * program accessing the database, every time the database is used.
1201 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1202 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1203 * @param[in] cmp A #MDB_cmp_func function
1204 * @return A non-zero error value on failure and 0 on success. Some possible
1207 * <li>EINVAL - an invalid parameter was specified.
1210 int mdb_set_dupsort(MDB_txn *txn, MDB_dbi dbi, MDB_cmp_func *cmp);
1212 /** @brief Set a relocation function for a #MDB_FIXEDMAP database.
1214 * @todo The relocation function is called whenever it is necessary to move the data
1215 * of an item to a different position in the database (e.g. through tree
1216 * balancing operations, shifts as a result of adds or deletes, etc.). It is
1217 * intended to allow address/position-dependent data items to be stored in
1218 * a database in an environment opened with the #MDB_FIXEDMAP option.
1219 * Currently the relocation feature is unimplemented and setting
1220 * this function has no effect.
1221 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1222 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1223 * @param[in] rel A #MDB_rel_func function
1224 * @return A non-zero error value on failure and 0 on success. Some possible
1227 * <li>EINVAL - an invalid parameter was specified.
1230 int mdb_set_relfunc(MDB_txn *txn, MDB_dbi dbi, MDB_rel_func *rel);
1232 /** @brief Set a context pointer for a #MDB_FIXEDMAP database's relocation function.
1234 * See #mdb_set_relfunc and #MDB_rel_func for more details.
1235 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1236 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1237 * @param[in] ctx An arbitrary pointer for whatever the application needs.
1238 * It will be passed to the callback function set by #mdb_set_relfunc
1239 * as its \b relctx parameter whenever the callback is invoked.
1240 * @return A non-zero error value on failure and 0 on success. Some possible
1243 * <li>EINVAL - an invalid parameter was specified.
1246 int mdb_set_relctx(MDB_txn *txn, MDB_dbi dbi, void *ctx);
1248 /** @brief Get items from a database.
1250 * This function retrieves key/data pairs from the database. The address
1251 * and length of the data associated with the specified \b key are returned
1252 * in the structure to which \b data refers.
1253 * If the database supports duplicate keys (#MDB_DUPSORT) then the
1254 * first data item for the key will be returned. Retrieval of other
1255 * items requires the use of #mdb_cursor_get().
1257 * @note The memory pointed to by the returned values is owned by the
1258 * database. The caller need not dispose of the memory, and may not
1259 * modify it in any way. For values returned in a read-only transaction
1260 * any modification attempts will cause a SIGSEGV.
1261 * @note Values returned from the database are valid only until a
1262 * subsequent update operation, or the end of the transaction.
1263 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1264 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1265 * @param[in] key The key to search for in the database
1266 * @param[out] data The data corresponding to the key
1267 * @return A non-zero error value on failure and 0 on success. Some possible
1270 * <li>#MDB_NOTFOUND - the key was not in the database.
1271 * <li>EINVAL - an invalid parameter was specified.
1274 int mdb_get(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data);
1276 /** @brief Store items into a database.
1278 * This function stores key/data pairs in the database. The default behavior
1279 * is to enter the new key/data pair, replacing any previously existing key
1280 * if duplicates are disallowed, or adding a duplicate data item if
1281 * duplicates are allowed (#MDB_DUPSORT).
1282 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1283 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1284 * @param[in] key The key to store in the database
1285 * @param[in,out] data The data to store
1286 * @param[in] flags Special options for this operation. This parameter
1287 * must be set to 0 or by bitwise OR'ing together one or more of the
1288 * values described here.
1290 * <li>#MDB_NODUPDATA - enter the new key/data pair only if it does not
1291 * already appear in the database. This flag may only be specified
1292 * if the database was opened with #MDB_DUPSORT. The function will
1293 * return #MDB_KEYEXIST if the key/data pair already appears in the
1295 * <li>#MDB_NOOVERWRITE - enter the new key/data pair only if the key
1296 * does not already appear in the database. The function will return
1297 * #MDB_KEYEXIST if the key already appears in the database, even if
1298 * the database supports duplicates (#MDB_DUPSORT). The \b data
1299 * parameter will be set to point to the existing item.
1300 * <li>#MDB_RESERVE - reserve space for data of the given size, but
1301 * don't copy the given data. Instead, return a pointer to the
1302 * reserved space, which the caller can fill in later - before
1303 * the next update operation or the transaction ends. This saves
1304 * an extra memcpy if the data is being generated later.
1305 * LMDB does nothing else with this memory, the caller is expected
1306 * to modify all of the space requested.
1307 * <li>#MDB_APPEND - append the given key/data pair to the end of the
1308 * database. This option allows fast bulk loading when keys are
1309 * already known to be in the correct order. Loading unsorted keys
1310 * with this flag will cause a #MDB_KEYEXIST error.
1311 * <li>#MDB_APPENDDUP - as above, but for sorted dup data.
1313 * @return A non-zero error value on failure and 0 on success. Some possible
1316 * <li>#MDB_MAP_FULL - the database is full, see #mdb_env_set_mapsize().
1317 * <li>#MDB_TXN_FULL - the transaction has too many dirty pages.
1318 * <li>EACCES - an attempt was made to write in a read-only transaction.
1319 * <li>EINVAL - an invalid parameter was specified.
1322 int mdb_put(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data,
1323 unsigned int flags);
1325 /** @brief Delete items from a database.
1327 * This function removes key/data pairs from the database.
1328 * If the database does not support sorted duplicate data items
1329 * (#MDB_DUPSORT) the data parameter is ignored.
1330 * If the database supports sorted duplicates and the data parameter
1331 * is NULL, all of the duplicate data items for the key will be
1332 * deleted. Otherwise, if the data parameter is non-NULL
1333 * only the matching data item will be deleted.
1334 * This function will return #MDB_NOTFOUND if the specified key/data
1335 * pair is not in the database.
1336 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1337 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1338 * @param[in] key The key to delete from the database
1339 * @param[in] data The data to delete
1340 * @return A non-zero error value on failure and 0 on success. Some possible
1343 * <li>EACCES - an attempt was made to write in a read-only transaction.
1344 * <li>EINVAL - an invalid parameter was specified.
1347 int mdb_del(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data);
1349 /** @brief Create a cursor handle.
1351 * A cursor is associated with a specific transaction and database.
1352 * A cursor cannot be used when its database handle is closed. Nor
1353 * when its transaction has ended, except with #mdb_cursor_renew().
1354 * It can be discarded with #mdb_cursor_close().
1355 * A cursor in a write-transaction can be closed before its transaction
1356 * ends, and will otherwise be closed when its transaction ends.
1357 * A cursor in a read-only transaction must be closed explicitly, before
1358 * or after its transaction ends. It can be reused with
1359 * #mdb_cursor_renew() before finally closing it.
1360 * @note Earlier documentation said that cursors in every transaction
1361 * were closed when the transaction committed or aborted.
1362 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1363 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1364 * @param[out] cursor Address where the new #MDB_cursor handle will be stored
1365 * @return A non-zero error value on failure and 0 on success. Some possible
1368 * <li>EINVAL - an invalid parameter was specified.
1371 int mdb_cursor_open(MDB_txn *txn, MDB_dbi dbi, MDB_cursor **cursor);
1373 /** @brief Close a cursor handle.
1375 * The cursor handle will be freed and must not be used again after this call.
1376 * Its transaction must still be live if it is a write-transaction.
1377 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1379 void mdb_cursor_close(MDB_cursor *cursor);
1381 /** @brief Renew a cursor handle.
1383 * A cursor is associated with a specific transaction and database.
1384 * Cursors that are only used in read-only
1385 * transactions may be re-used, to avoid unnecessary malloc/free overhead.
1386 * The cursor may be associated with a new read-only transaction, and
1387 * referencing the same database handle as it was created with.
1388 * This may be done whether the previous transaction is live or dead.
1389 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1390 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1391 * @return A non-zero error value on failure and 0 on success. Some possible
1394 * <li>EINVAL - an invalid parameter was specified.
1397 int mdb_cursor_renew(MDB_txn *txn, MDB_cursor *cursor);
1399 /** @brief Return the cursor's transaction handle.
1401 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1403 MDB_txn *mdb_cursor_txn(MDB_cursor *cursor);
1405 /** @brief Return the cursor's database handle.
1407 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1409 MDB_dbi mdb_cursor_dbi(MDB_cursor *cursor);
1411 /** @brief Retrieve by cursor.
1413 * This function retrieves key/data pairs from the database. The address and length
1414 * of the key are returned in the object to which \b key refers (except for the
1415 * case of the #MDB_SET option, in which the \b key object is unchanged), and
1416 * the address and length of the data are returned in the object to which \b data
1418 * See #mdb_get() for restrictions on using the output values.
1419 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1420 * @param[in,out] key The key for a retrieved item
1421 * @param[in,out] data The data of a retrieved item
1422 * @param[in] op A cursor operation #MDB_cursor_op
1423 * @return A non-zero error value on failure and 0 on success. Some possible
1426 * <li>#MDB_NOTFOUND - no matching key found.
1427 * <li>EINVAL - an invalid parameter was specified.
1430 int mdb_cursor_get(MDB_cursor *cursor, MDB_val *key, MDB_val *data,
1433 /** @brief Store by cursor.
1435 * This function stores key/data pairs into the database.
1436 * The cursor is positioned at the new item, or on failure usually near it.
1437 * @note Earlier documentation incorrectly said errors would leave the
1438 * state of the cursor unchanged.
1439 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1440 * @param[in] key The key operated on.
1441 * @param[in] data The data operated on.
1442 * @param[in] flags Options for this operation. This parameter
1443 * must be set to 0 or one of the values described here.
1445 * <li>#MDB_CURRENT - replace the item at the current cursor position.
1446 * The \b key parameter must still be provided, and must match it.
1447 * If using sorted duplicates (#MDB_DUPSORT) the data item must still
1448 * sort into the same place. This is intended to be used when the
1449 * new data is the same size as the old. Otherwise it will simply
1450 * perform a delete of the old record followed by an insert.
1451 * <li>#MDB_NODUPDATA - enter the new key/data pair only if it does not
1452 * already appear in the database. This flag may only be specified
1453 * if the database was opened with #MDB_DUPSORT. The function will
1454 * return #MDB_KEYEXIST if the key/data pair already appears in the
1456 * <li>#MDB_NOOVERWRITE - enter the new key/data pair only if the key
1457 * does not already appear in the database. The function will return
1458 * #MDB_KEYEXIST if the key already appears in the database, even if
1459 * the database supports duplicates (#MDB_DUPSORT).
1460 * <li>#MDB_RESERVE - reserve space for data of the given size, but
1461 * don't copy the given data. Instead, return a pointer to the
1462 * reserved space, which the caller can fill in later. This saves
1463 * an extra memcpy if the data is being generated later.
1464 * <li>#MDB_APPEND - append the given key/data pair to the end of the
1465 * database. No key comparisons are performed. This option allows
1466 * fast bulk loading when keys are already known to be in the
1467 * correct order. Loading unsorted keys with this flag will cause
1469 * <li>#MDB_APPENDDUP - as above, but for sorted dup data.
1470 * <li>#MDB_MULTIPLE - store multiple contiguous data elements in a
1471 * single request. This flag may only be specified if the database
1472 * was opened with #MDB_DUPFIXED. The \b data argument must be an
1473 * array of two MDB_vals. The mv_size of the first MDB_val must be
1474 * the size of a single data element. The mv_data of the first MDB_val
1475 * must point to the beginning of the array of contiguous data elements.
1476 * The mv_size of the second MDB_val must be the count of the number
1477 * of data elements to store. On return this field will be set to
1478 * the count of the number of elements actually written. The mv_data
1479 * of the second MDB_val is unused.
1481 * @return A non-zero error value on failure and 0 on success. Some possible
1484 * <li>#MDB_MAP_FULL - the database is full, see #mdb_env_set_mapsize().
1485 * <li>#MDB_TXN_FULL - the transaction has too many dirty pages.
1486 * <li>EACCES - an attempt was made to write in a read-only transaction.
1487 * <li>EINVAL - an invalid parameter was specified.
1490 int mdb_cursor_put(MDB_cursor *cursor, MDB_val *key, MDB_val *data,
1491 unsigned int flags);
1493 /** @brief Delete current key/data pair
1495 * This function deletes the key/data pair to which the cursor refers.
1496 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1497 * @param[in] flags Options for this operation. This parameter
1498 * must be set to 0 or one of the values described here.
1500 * <li>#MDB_NODUPDATA - delete all of the data items for the current key.
1501 * This flag may only be specified if the database was opened with #MDB_DUPSORT.
1503 * @return A non-zero error value on failure and 0 on success. Some possible
1506 * <li>EACCES - an attempt was made to write in a read-only transaction.
1507 * <li>EINVAL - an invalid parameter was specified.
1510 int mdb_cursor_del(MDB_cursor *cursor, unsigned int flags);
1512 /** @brief Return count of duplicates for current key.
1514 * This call is only valid on databases that support sorted duplicate
1515 * data items #MDB_DUPSORT.
1516 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1517 * @param[out] countp Address where the count will be stored
1518 * @return A non-zero error value on failure and 0 on success. Some possible
1521 * <li>EINVAL - cursor is not initialized, or an invalid parameter was specified.
1524 int mdb_cursor_count(MDB_cursor *cursor, size_t *countp);
1526 /** @brief Compare two data items according to a particular database.
1528 * This returns a comparison as if the two data items were keys in the
1529 * specified database.
1530 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1531 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1532 * @param[in] a The first item to compare
1533 * @param[in] b The second item to compare
1534 * @return < 0 if a < b, 0 if a == b, > 0 if a > b
1536 int mdb_cmp(MDB_txn *txn, MDB_dbi dbi, const MDB_val *a, const MDB_val *b);
1538 /** @brief Compare two data items according to a particular database.
1540 * This returns a comparison as if the two items were data items of
1541 * the specified database. The database must have the #MDB_DUPSORT flag.
1542 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1543 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1544 * @param[in] a The first item to compare
1545 * @param[in] b The second item to compare
1546 * @return < 0 if a < b, 0 if a == b, > 0 if a > b
1548 int mdb_dcmp(MDB_txn *txn, MDB_dbi dbi, const MDB_val *a, const MDB_val *b);
1550 /** @brief A callback function used to print a message from the library.
1552 * @param[in] msg The string to be printed.
1553 * @param[in] ctx An arbitrary context pointer for the callback.
1554 * @return < 0 on failure, >= 0 on success.
1556 typedef int (MDB_msg_func)(const char *msg, void *ctx);
1558 /** @brief Dump the entries in the reader lock table.
1560 * @param[in] env An environment handle returned by #mdb_env_create()
1561 * @param[in] func A #MDB_msg_func function
1562 * @param[in] ctx Anything the message function needs
1563 * @return < 0 on failure, >= 0 on success.
1565 int mdb_reader_list(MDB_env *env, MDB_msg_func *func, void *ctx);
1567 /** @brief Check for stale entries in the reader lock table.
1569 * @param[in] env An environment handle returned by #mdb_env_create()
1570 * @param[out] dead Number of stale slots that were cleared
1571 * @return 0 on success, non-zero on failure.
1573 int mdb_reader_check(MDB_env *env, int *dead);
1579 /** @page tools LMDB Command Line Tools
1580 The following describes the command line tools that are available for LMDB.
1587 #endif /* _LMDB_H_ */