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 * If this is your first time using a transactional embedded key/value
44 * store, you may find the \ref starting page to be helpful.
46 * @section caveats_sec Caveats
47 * Troubleshooting the lock file, plus semaphores on BSD systems:
49 * - A broken lockfile can cause sync issues.
50 * Stale reader transactions left behind by an aborted program
51 * cause further writes to grow the database quickly, and
52 * stale locks can block further operation.
54 * Fix: Check for stale readers periodically, using the
55 * #mdb_reader_check function or the \ref mdb_stat_1 "mdb_stat" tool.
56 * Stale writers will be cleared automatically on most systems:
57 * - Windows - automatic
58 * - BSD, systems using SysV semaphores - automatic
59 * - Linux, systems using POSIX mutexes with Robust option - automatic
60 * Otherwise just make all programs using the database close it;
61 * the lockfile is always reset on first open of the environment.
63 * - On BSD systems or others configured with MDB_USE_SYSV_SEM or
65 * startup can fail due to semaphores owned by another userid.
67 * Fix: Open and close the database as the user which owns the
68 * semaphores (likely last user) or as root, while no other
69 * process is using the database.
71 * Restrictions/caveats (in addition to those listed for some functions):
73 * - Only the database owner should normally use the database on
74 * BSD systems or when otherwise configured with MDB_USE_POSIX_SEM.
75 * Multiple users can cause startup to fail later, as noted above.
77 * - There is normally no pure read-only mode, since readers need write
78 * access to locks and lock file. Exceptions: On read-only filesystems
79 * or with the #MDB_NOLOCK flag described under #mdb_env_open().
81 * - An LMDB configuration will often reserve considerable \b unused
82 * memory address space and maybe file size for future growth.
83 * This does not use actual memory or disk space, but users may need
84 * to understand the difference so they won't be scared off.
86 * - By default, in versions before 0.9.10, unused portions of the data
87 * file might receive garbage data from memory freed by other code.
88 * (This does not happen when using the #MDB_WRITEMAP flag.) As of
89 * 0.9.10 the default behavior is to initialize such memory before
90 * writing to the data file. Since there may be a slight performance
91 * cost due to this initialization, applications may disable it using
92 * the #MDB_NOMEMINIT flag. Applications handling sensitive data
93 * which must not be written should not use this flag. This flag is
94 * irrelevant when using #MDB_WRITEMAP.
96 * - A thread can only use one transaction at a time, plus any child
97 * transactions. Each transaction belongs to one thread. See below.
98 * The #MDB_NOTLS flag changes this for read-only transactions.
100 * - Use an MDB_env* in the process which opened it, without fork()ing.
102 * - Do not have open an LMDB database twice in the same process at
103 * the same time. Not even from a plain open() call - close()ing it
104 * breaks flock() advisory locking.
106 * - Avoid long-lived transactions. Read transactions prevent
107 * reuse of pages freed by newer write transactions, thus the
108 * database can grow quickly. Write transactions prevent
109 * other write transactions, since writes are serialized.
111 * - Avoid suspending a process with active transactions. These
112 * would then be "long-lived" as above. Also read transactions
113 * suspended when writers commit could sometimes see wrong data.
115 * ...when several processes can use a database concurrently:
117 * - Avoid aborting a process with an active transaction.
118 * The transaction becomes "long-lived" as above until a check
119 * for stale readers is performed or the lockfile is reset,
120 * since the process may not remove it from the lockfile.
122 * This does not apply to write transactions if the system clears
123 * stale writers, see above.
125 * - If you do that anyway, do a periodic check for stale readers. Or
126 * close the environment once in a while, so the lockfile can get reset.
128 * - Do not use LMDB databases on remote filesystems, even between
129 * processes on the same host. This breaks flock() on some OSes,
130 * possibly memory map sync, and certainly sync between programs
131 * on different hosts.
133 * - Opening a database can fail if another process is opening or
134 * closing it at exactly the same time.
136 * @author Howard Chu, Symas Corporation.
138 * @copyright Copyright 2011-2016 Howard Chu, Symas Corp. All rights reserved.
140 * Redistribution and use in source and binary forms, with or without
141 * modification, are permitted only as authorized by the OpenLDAP
144 * A copy of this license is available in the file LICENSE in the
145 * top-level directory of the distribution or, alternatively, at
146 * <http://www.OpenLDAP.org/license.html>.
149 * This code is derived from btree.c written by Martin Hedenfalk.
151 * Copyright (c) 2009, 2010 Martin Hedenfalk <martin@bzero.se>
153 * Permission to use, copy, modify, and distribute this software for any
154 * purpose with or without fee is hereby granted, provided that the above
155 * copyright notice and this permission notice appear in all copies.
157 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
158 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
159 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
160 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
161 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
162 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
163 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
168 #include <sys/types.h>
169 #include <inttypes.h>
176 /** Unix permissions for creating files, or dummy definition for Windows */
178 typedef int mdb_mode_t;
180 typedef mode_t mdb_mode_t;
184 # define MDB_FMT_Z "I"
186 # define MDB_FMT_Z "z" /**< printf/scanf format modifier for size_t */
189 #if !defined(MDB_VL32) || SIZE_MAX > 0xffffffffU
190 typedef size_t mdb_size_t;
191 # define MDB_SIZE_MAX SIZE_MAX /**< max #mdb_size_t */
192 # define MDB_FMT_Y MDB_FMT_Z /**< Obsolescent, see #MDB_PRIz()/#MDB_SCNz() */
193 /* TODO: For VL32, use uint64_t (trivial) and therefore PRI<c>64 (big patch) */
194 #elif defined(_WIN32)
195 typedef unsigned __int64 mdb_size_t;
196 # define MDB_SIZE_MAX _UI64_MAX
197 # define MDB_FMT_Y "I64"
198 #elif defined(ULLONG_MAX)
199 typedef unsigned long long mdb_size_t;
200 # define MDB_SIZE_MAX ULLONG_MAX
201 # define MDB_FMT_Y "ll"
203 # error "Found no acceptable integer type for mdb_size_t"
206 # define mdb_env_create mdb_env_create_vl32 /**< Prevent mixing with non-VL32 builds */
209 /** #mdb_size_t printf formats, \b t = one of [diouxX] without quotes */
210 #define MDB_PRIz(t) MDB_FMT_Y #t
211 /** #mdb_size_t scanf formats, \b t = one of [dioux] without quotes */
212 #define MDB_SCNz(t) MDB_FMT_Y #t
214 /** An abstraction for a file handle.
215 * On POSIX systems file handles are small integers. On Windows
216 * they're opaque pointers.
219 typedef void *mdb_filehandle_t;
221 typedef int mdb_filehandle_t;
224 /** @defgroup mdb LMDB API
226 * @brief OpenLDAP Lightning Memory-Mapped Database Manager
228 /** @defgroup Version Version Macros
231 /** Library major version */
232 #define MDB_VERSION_MAJOR 0
233 /** Library minor version */
234 #define MDB_VERSION_MINOR 9
235 /** Library patch version */
236 #define MDB_VERSION_PATCH 70
238 /** Combine args a,b,c into a single integer for easy version comparisons */
239 #define MDB_VERINT(a,b,c) (((a) << 24) | ((b) << 16) | (c))
241 /** The full library version as a single integer */
242 #define MDB_VERSION_FULL \
243 MDB_VERINT(MDB_VERSION_MAJOR,MDB_VERSION_MINOR,MDB_VERSION_PATCH)
245 /** The release date of this library version */
246 #define MDB_VERSION_DATE "December 19, 2015"
248 /** A stringifier for the version info */
249 #define MDB_VERSTR(a,b,c,d) "LMDB " #a "." #b "." #c ": (" d ")"
251 /** A helper for the stringifier macro */
252 #define MDB_VERFOO(a,b,c,d) MDB_VERSTR(a,b,c,d)
254 /** The full library version as a C string */
255 #define MDB_VERSION_STRING \
256 MDB_VERFOO(MDB_VERSION_MAJOR,MDB_VERSION_MINOR,MDB_VERSION_PATCH,MDB_VERSION_DATE)
259 /** @brief Opaque structure for a database environment.
261 * A DB environment supports multiple databases, all residing in the same
264 typedef struct MDB_env MDB_env;
266 /** @brief Opaque structure for a transaction handle.
268 * All database operations require a transaction handle. Transactions may be
269 * read-only or read-write.
271 typedef struct MDB_txn MDB_txn;
273 /** @brief A handle for an individual database in the DB environment. */
274 typedef unsigned int MDB_dbi;
276 /** @brief Opaque structure for navigating through a database */
277 typedef struct MDB_cursor MDB_cursor;
279 /** @brief Generic structure used for passing keys and data in and out
282 * Values returned from the database are valid only until a subsequent
283 * update operation, or the end of the transaction. Do not modify or
284 * free them, they commonly point into the database itself.
286 * Key sizes must be between 1 and #mdb_env_get_maxkeysize() inclusive.
287 * The same applies to data sizes in databases with the #MDB_DUPSORT flag.
288 * Other data items can in theory be from 0 to 0xffffffff bytes long.
290 typedef struct MDB_val {
291 size_t mv_size; /**< size of the data item */
292 void *mv_data; /**< address of the data item */
295 /** @brief A callback function used to compare two keys in a database */
296 typedef int (MDB_cmp_func)(const MDB_val *a, const MDB_val *b);
298 /** @brief A callback function used to relocate a position-dependent data item
299 * in a fixed-address database.
301 * The \b newptr gives the item's desired address in
302 * the memory map, and \b oldptr gives its previous address. The item's actual
303 * data resides at the address in \b item. This callback is expected to walk
304 * through the fields of the record in \b item and modify any
305 * values based at the \b oldptr address to be relative to the \b newptr address.
306 * @param[in,out] item The item that is to be relocated.
307 * @param[in] oldptr The previous address.
308 * @param[in] newptr The new address to relocate to.
309 * @param[in] relctx An application-provided context, set by #mdb_set_relctx().
310 * @todo This feature is currently unimplemented.
312 typedef void (MDB_rel_func)(MDB_val *item, void *oldptr, void *newptr, void *relctx);
314 /** @defgroup mdb_env Environment Flags
317 /** mmap at a fixed address (experimental) */
318 #define MDB_FIXEDMAP 0x01
319 /** no environment directory */
320 #define MDB_NOSUBDIR 0x4000
321 /** don't fsync after commit */
322 #define MDB_NOSYNC 0x10000
324 #define MDB_RDONLY 0x20000
325 /** don't fsync metapage after commit */
326 #define MDB_NOMETASYNC 0x40000
327 /** use writable mmap */
328 #define MDB_WRITEMAP 0x80000
329 /** use asynchronous msync when #MDB_WRITEMAP is used */
330 #define MDB_MAPASYNC 0x100000
331 /** tie reader locktable slots to #MDB_txn objects instead of to threads */
332 #define MDB_NOTLS 0x200000
333 /** don't do any locking, caller must manage their own locks */
334 #define MDB_NOLOCK 0x400000
335 /** don't do readahead (no effect on Windows) */
336 #define MDB_NORDAHEAD 0x800000
337 /** don't initialize malloc'd memory before writing to datafile */
338 #define MDB_NOMEMINIT 0x1000000
341 /** @defgroup mdb_dbi_open Database Flags
344 /** use reverse string keys */
345 #define MDB_REVERSEKEY 0x02
346 /** use sorted duplicates */
347 #define MDB_DUPSORT 0x04
348 /** numeric keys in native byte order, either unsigned int or #mdb_size_t.
349 * (lmdb expects 32-bit int <= size_t <= 32/64-bit mdb_size_t.)
350 * The keys must all be of the same size. */
351 #define MDB_INTEGERKEY 0x08
352 /** with #MDB_DUPSORT, sorted dup items have fixed size */
353 #define MDB_DUPFIXED 0x10
354 /** with #MDB_DUPSORT, dups are #MDB_INTEGERKEY-style integers */
355 #define MDB_INTEGERDUP 0x20
356 /** with #MDB_DUPSORT, use reverse string dups */
357 #define MDB_REVERSEDUP 0x40
358 /** create DB if not already existing */
359 #define MDB_CREATE 0x40000
362 /** @defgroup mdb_put Write Flags
365 /** For put: Don't write if the key already exists. */
366 #define MDB_NOOVERWRITE 0x10
367 /** Only for #MDB_DUPSORT<br>
368 * For put: don't write if the key and data pair already exist.<br>
369 * For mdb_cursor_del: remove all duplicate data items.
371 #define MDB_NODUPDATA 0x20
372 /** For mdb_cursor_put: overwrite the current key/data pair */
373 #define MDB_CURRENT 0x40
374 /** For put: Just reserve space for data, don't copy it. Return a
375 * pointer to the reserved space.
377 #define MDB_RESERVE 0x10000
378 /** Data is being appended, don't split full pages. */
379 #define MDB_APPEND 0x20000
380 /** Duplicate data is being appended, don't split full pages. */
381 #define MDB_APPENDDUP 0x40000
382 /** Store multiple data items in one call. Only for #MDB_DUPFIXED. */
383 #define MDB_MULTIPLE 0x80000
386 /** @defgroup mdb_copy Copy Flags
389 /** Compacting copy: Omit free space from copy, and renumber all
390 * pages sequentially.
392 #define MDB_CP_COMPACT 0x01
395 /** @brief Cursor Get operations.
397 * This is the set of all operations for retrieving data
400 typedef enum MDB_cursor_op {
401 MDB_FIRST, /**< Position at first key/data item */
402 MDB_FIRST_DUP, /**< Position at first data item of current key.
403 Only for #MDB_DUPSORT */
404 MDB_GET_BOTH, /**< Position at key/data pair. Only for #MDB_DUPSORT */
405 MDB_GET_BOTH_RANGE, /**< position at key, nearest data. Only for #MDB_DUPSORT */
406 MDB_GET_CURRENT, /**< Return key/data at current cursor position */
407 MDB_GET_MULTIPLE, /**< Return key and up to a page of duplicate data items
408 from current cursor position. Move cursor to prepare
409 for #MDB_NEXT_MULTIPLE. Only for #MDB_DUPFIXED */
410 MDB_LAST, /**< Position at last key/data item */
411 MDB_LAST_DUP, /**< Position at last data item of current key.
412 Only for #MDB_DUPSORT */
413 MDB_NEXT, /**< Position at next data item */
414 MDB_NEXT_DUP, /**< Position at next data item of current key.
415 Only for #MDB_DUPSORT */
416 MDB_NEXT_MULTIPLE, /**< Return key and up to a page of duplicate data items
417 from next cursor position. Move cursor to prepare
418 for #MDB_NEXT_MULTIPLE. Only for #MDB_DUPFIXED */
419 MDB_NEXT_NODUP, /**< Position at first data item of next key */
420 MDB_PREV, /**< Position at previous data item */
421 MDB_PREV_DUP, /**< Position at previous data item of current key.
422 Only for #MDB_DUPSORT */
423 MDB_PREV_NODUP, /**< Position at last data item of previous key */
424 MDB_SET, /**< Position at specified key */
425 MDB_SET_KEY, /**< Position at specified key, return key + data */
426 MDB_SET_RANGE, /**< Position at first key greater than or equal to specified key. */
427 MDB_PREV_MULTIPLE /**< Position at previous page and return key and up to
428 a page of duplicate data items. Only for #MDB_DUPFIXED */
431 /** @defgroup errors Return Codes
433 * BerkeleyDB uses -30800 to -30999, we'll go under them
436 /** Successful result */
437 #define MDB_SUCCESS 0
438 /** key/data pair already exists */
439 #define MDB_KEYEXIST (-30799)
440 /** key/data pair not found (EOF) */
441 #define MDB_NOTFOUND (-30798)
442 /** Requested page not found - this usually indicates corruption */
443 #define MDB_PAGE_NOTFOUND (-30797)
444 /** Located page was wrong type */
445 #define MDB_CORRUPTED (-30796)
446 /** Update of meta page failed or environment had fatal error */
447 #define MDB_PANIC (-30795)
448 /** Environment version mismatch */
449 #define MDB_VERSION_MISMATCH (-30794)
450 /** File is not a valid LMDB file */
451 #define MDB_INVALID (-30793)
452 /** Environment mapsize reached */
453 #define MDB_MAP_FULL (-30792)
454 /** Environment maxdbs reached */
455 #define MDB_DBS_FULL (-30791)
456 /** Environment maxreaders reached */
457 #define MDB_READERS_FULL (-30790)
458 /** Too many TLS keys in use - Windows only */
459 #define MDB_TLS_FULL (-30789)
460 /** Txn has too many dirty pages */
461 #define MDB_TXN_FULL (-30788)
462 /** Cursor stack too deep - internal error */
463 #define MDB_CURSOR_FULL (-30787)
464 /** Page has not enough space - internal error */
465 #define MDB_PAGE_FULL (-30786)
466 /** Database contents grew beyond environment mapsize */
467 #define MDB_MAP_RESIZED (-30785)
468 /** Operation and DB incompatible, or DB type changed. This can mean:
470 * <li>The operation expects an #MDB_DUPSORT / #MDB_DUPFIXED database.
471 * <li>Opening a named DB when the unnamed DB has #MDB_DUPSORT / #MDB_INTEGERKEY.
472 * <li>Accessing a data record as a database, or vice versa.
473 * <li>The database was dropped and recreated with different flags.
476 #define MDB_INCOMPATIBLE (-30784)
477 /** Invalid reuse of reader locktable slot */
478 #define MDB_BAD_RSLOT (-30783)
479 /** Transaction must abort, has a child, or is invalid */
480 #define MDB_BAD_TXN (-30782)
481 /** Unsupported size of key/DB name/data, or wrong DUPFIXED size */
482 #define MDB_BAD_VALSIZE (-30781)
483 /** The specified DBI was changed unexpectedly */
484 #define MDB_BAD_DBI (-30780)
485 /** Unexpected problem - txn should abort */
486 #define MDB_PROBLEM (-30779)
487 /** The last defined error code */
488 #define MDB_LAST_ERRCODE MDB_PROBLEM
491 /** @brief Statistics for a database in the environment */
492 typedef struct MDB_stat {
493 unsigned int ms_psize; /**< Size of a database page.
494 This is currently the same for all databases. */
495 unsigned int ms_depth; /**< Depth (height) of the B-tree */
496 mdb_size_t ms_branch_pages; /**< Number of internal (non-leaf) pages */
497 mdb_size_t ms_leaf_pages; /**< Number of leaf pages */
498 mdb_size_t ms_overflow_pages; /**< Number of overflow pages */
499 mdb_size_t ms_entries; /**< Number of data items */
502 /** @brief Information about the environment */
503 typedef struct MDB_envinfo {
504 void *me_mapaddr; /**< Address of map, if fixed */
505 mdb_size_t me_mapsize; /**< Size of the data memory map */
506 mdb_size_t me_last_pgno; /**< ID of the last used page */
507 mdb_size_t me_last_txnid; /**< ID of the last committed transaction */
508 unsigned int me_maxreaders; /**< max reader slots in the environment */
509 unsigned int me_numreaders; /**< max reader slots used in the environment */
512 /** @brief Return the LMDB library version information.
514 * @param[out] major if non-NULL, the library major version number is copied here
515 * @param[out] minor if non-NULL, the library minor version number is copied here
516 * @param[out] patch if non-NULL, the library patch version number is copied here
517 * @retval "version string" The library version as a string
519 char *mdb_version(int *major, int *minor, int *patch);
521 /** @brief Return a string describing a given error code.
523 * This function is a superset of the ANSI C X3.159-1989 (ANSI C) strerror(3)
524 * function. If the error code is greater than or equal to 0, then the string
525 * returned by the system function strerror(3) is returned. If the error code
526 * is less than 0, an error string corresponding to the LMDB library error is
527 * returned. See @ref errors for a list of LMDB-specific error codes.
528 * @param[in] err The error code
529 * @retval "error message" The description of the error
531 char *mdb_strerror(int err);
533 /** @brief Create an LMDB environment handle.
535 * This function allocates memory for a #MDB_env structure. To release
536 * the allocated memory and discard the handle, call #mdb_env_close().
537 * Before the handle may be used, it must be opened using #mdb_env_open().
538 * Various other options may also need to be set before opening the handle,
539 * e.g. #mdb_env_set_mapsize(), #mdb_env_set_maxreaders(), #mdb_env_set_maxdbs(),
540 * depending on usage requirements.
541 * @param[out] env The address where the new handle will be stored
542 * @return A non-zero error value on failure and 0 on success.
544 int mdb_env_create(MDB_env **env);
546 /** @brief Open an environment handle.
548 * If this function fails, #mdb_env_close() must be called to discard the #MDB_env handle.
549 * @param[in] env An environment handle returned by #mdb_env_create()
550 * @param[in] path The directory in which the database files reside. This
551 * directory must already exist and be writable.
552 * @param[in] flags Special options for this environment. This parameter
553 * must be set to 0 or by bitwise OR'ing together one or more of the
554 * values described here.
555 * Flags set by mdb_env_set_flags() are also used.
558 * use a fixed address for the mmap region. This flag must be specified
559 * when creating the environment, and is stored persistently in the environment.
560 * If successful, the memory map will always reside at the same virtual address
561 * and pointers used to reference data items in the database will be constant
562 * across multiple invocations. This option may not always work, depending on
563 * how the operating system has allocated memory to shared libraries and other uses.
564 * The feature is highly experimental.
566 * By default, LMDB creates its environment in a directory whose
567 * pathname is given in \b path, and creates its data and lock files
568 * under that directory. With this option, \b path is used as-is for
569 * the database main data file. The database lock file is the \b path
570 * with "-lock" appended.
572 * Open the environment in read-only mode. No write operations will be
573 * allowed. LMDB will still modify the lock file - except on read-only
574 * filesystems, where LMDB does not use locks.
576 * Use a writeable memory map unless MDB_RDONLY is set. This uses
577 * fewer mallocs but loses protection from application bugs
578 * like wild pointer writes and other bad updates into the database.
579 * This may be slightly faster for DBs that fit entirely in RAM, but
580 * is slower for DBs larger than RAM.
581 * Incompatible with nested transactions.
582 * Do not mix processes with and without MDB_WRITEMAP on the same
583 * environment. This can defeat durability (#mdb_env_sync etc).
584 * <li>#MDB_NOMETASYNC
585 * Flush system buffers to disk only once per transaction, omit the
586 * metadata flush. Defer that until the system flushes files to disk,
587 * or next non-MDB_RDONLY commit or #mdb_env_sync(). This optimization
588 * maintains database integrity, but a system crash may undo the last
589 * committed transaction. I.e. it preserves the ACI (atomicity,
590 * consistency, isolation) but not D (durability) database property.
591 * This flag may be changed at any time using #mdb_env_set_flags().
593 * Don't flush system buffers to disk when committing a transaction.
594 * This optimization means a system crash can corrupt the database or
595 * lose the last transactions if buffers are not yet flushed to disk.
596 * The risk is governed by how often the system flushes dirty buffers
597 * to disk and how often #mdb_env_sync() is called. However, if the
598 * filesystem preserves write order and the #MDB_WRITEMAP flag is not
599 * used, transactions exhibit ACI (atomicity, consistency, isolation)
600 * properties and only lose D (durability). I.e. database integrity
601 * is maintained, but a system crash may undo the final transactions.
602 * Note that (#MDB_NOSYNC | #MDB_WRITEMAP) leaves the system with no
603 * hint for when to write transactions to disk, unless #mdb_env_sync()
604 * is called. (#MDB_MAPASYNC | #MDB_WRITEMAP) may be preferable.
605 * This flag may be changed at any time using #mdb_env_set_flags().
607 * When using #MDB_WRITEMAP, use asynchronous flushes to disk.
608 * As with #MDB_NOSYNC, a system crash can then corrupt the
609 * database or lose the last transactions. Calling #mdb_env_sync()
610 * ensures on-disk database integrity until next commit.
611 * This flag may be changed at any time using #mdb_env_set_flags().
613 * Don't use Thread-Local Storage. Tie reader locktable slots to
614 * #MDB_txn objects instead of to threads. I.e. #mdb_txn_reset() keeps
615 * the slot reseved for the #MDB_txn object. A thread may use parallel
616 * read-only transactions. A read-only transaction may span threads if
617 * the user synchronizes its use. Applications that multiplex many
618 * user threads over individual OS threads need this option. Such an
619 * application must also serialize the write transactions in an OS
620 * thread, since LMDB's write locking is unaware of the user threads.
622 * Don't do any locking. If concurrent access is anticipated, the
623 * caller must manage all concurrency itself. For proper operation
624 * the caller must enforce single-writer semantics, and must ensure
625 * that no readers are using old transactions while a writer is
626 * active. The simplest approach is to use an exclusive lock so that
627 * no readers may be active at all when a writer begins.
629 * Turn off readahead. Most operating systems perform readahead on
630 * read requests by default. This option turns it off if the OS
631 * supports it. Turning it off may help random read performance
632 * when the DB is larger than RAM and system RAM is full.
633 * The option is not implemented on Windows.
635 * Don't initialize malloc'd memory before writing to unused spaces
636 * in the data file. By default, memory for pages written to the data
637 * file is obtained using malloc. While these pages may be reused in
638 * subsequent transactions, freshly malloc'd pages will be initialized
639 * to zeroes before use. This avoids persisting leftover data from other
640 * code (that used the heap and subsequently freed the memory) into the
641 * data file. Note that many other system libraries may allocate
642 * and free memory from the heap for arbitrary uses. E.g., stdio may
643 * use the heap for file I/O buffers. This initialization step has a
644 * modest performance cost so some applications may want to disable
645 * it using this flag. This option can be a problem for applications
646 * which handle sensitive data like passwords, and it makes memory
647 * checkers like Valgrind noisy. This flag is not needed with #MDB_WRITEMAP,
648 * which writes directly to the mmap instead of using malloc for pages. The
649 * initialization is also skipped if #MDB_RESERVE is used; the
650 * caller is expected to overwrite all of the memory that was
651 * reserved in that case.
652 * This flag may be changed at any time using #mdb_env_set_flags().
654 * @param[in] mode The UNIX permissions to set on created files and semaphores.
655 * This parameter is ignored on Windows.
656 * @return A non-zero error value on failure and 0 on success. Some possible
659 * <li>#MDB_VERSION_MISMATCH - the version of the LMDB library doesn't match the
660 * version that created the database environment.
661 * <li>#MDB_INVALID - the environment file headers are corrupted.
662 * <li>ENOENT - the directory specified by the path parameter doesn't exist.
663 * <li>EACCES - the user didn't have permission to access the environment files.
664 * <li>EAGAIN - the environment was locked by another process.
667 int mdb_env_open(MDB_env *env, const char *path, unsigned int flags, mdb_mode_t mode);
669 /** @brief Copy an LMDB environment to the specified path.
671 * This function may be used to make a backup of an existing environment.
672 * No lockfile is created, since it gets recreated at need.
673 * @note This call can trigger significant file size growth if run in
674 * parallel with write transactions, because it employs a read-only
675 * transaction. See long-lived transactions under @ref caveats_sec.
676 * @param[in] env An environment handle returned by #mdb_env_create(). It
677 * must have already been opened successfully.
678 * @param[in] path The directory in which the copy will reside. This
679 * directory must already exist and be writable but must otherwise be
681 * @return A non-zero error value on failure and 0 on success.
683 int mdb_env_copy(MDB_env *env, const char *path);
685 /** @brief Copy an LMDB environment to the specified file descriptor.
687 * This function may be used to make a backup of an existing environment.
688 * No lockfile is created, since it gets recreated at need.
689 * @note This call can trigger significant file size growth if run in
690 * parallel with write transactions, because it employs a read-only
691 * transaction. See long-lived transactions under @ref caveats_sec.
692 * @param[in] env An environment handle returned by #mdb_env_create(). It
693 * must have already been opened successfully.
694 * @param[in] fd The filedescriptor to write the copy to. It must
695 * have already been opened for Write access.
696 * @return A non-zero error value on failure and 0 on success.
698 int mdb_env_copyfd(MDB_env *env, mdb_filehandle_t fd);
700 /** @brief Copy an LMDB environment to the specified path, with options.
702 * This function may be used to make a backup of an existing environment.
703 * No lockfile is created, since it gets recreated at need.
704 * @note This call can trigger significant file size growth if run in
705 * parallel with write transactions, because it employs a read-only
706 * transaction. See long-lived transactions under @ref caveats_sec.
707 * @param[in] env An environment handle returned by #mdb_env_create(). It
708 * must have already been opened successfully.
709 * @param[in] path The directory in which the copy will reside. This
710 * directory must already exist and be writable but must otherwise be
712 * @param[in] flags Special options for this operation. This parameter
713 * must be set to 0 or by bitwise OR'ing together one or more of the
714 * values described here.
716 * <li>#MDB_CP_COMPACT - Perform compaction while copying: omit free
717 * pages and sequentially renumber all pages in output. This option
718 * consumes more CPU and runs more slowly than the default.
719 * Currently it fails if the environment has suffered a page leak.
721 * @return A non-zero error value on failure and 0 on success.
723 int mdb_env_copy2(MDB_env *env, const char *path, unsigned int flags);
725 /** @brief Copy an LMDB environment to the specified file descriptor,
728 * This function may be used to make a backup of an existing environment.
729 * No lockfile is created, since it gets recreated at need. See
730 * #mdb_env_copy2() for further details.
731 * @note This call can trigger significant file size growth if run in
732 * parallel with write transactions, because it employs a read-only
733 * transaction. See long-lived transactions under @ref caveats_sec.
734 * @param[in] env An environment handle returned by #mdb_env_create(). It
735 * must have already been opened successfully.
736 * @param[in] fd The filedescriptor to write the copy to. It must
737 * have already been opened for Write access.
738 * @param[in] flags Special options for this operation.
739 * See #mdb_env_copy2() for options.
740 * @return A non-zero error value on failure and 0 on success.
742 int mdb_env_copyfd2(MDB_env *env, mdb_filehandle_t fd, unsigned int flags);
744 /** @brief Return statistics about the LMDB environment.
746 * @param[in] env An environment handle returned by #mdb_env_create()
747 * @param[out] stat The address of an #MDB_stat structure
748 * where the statistics will be copied
750 int mdb_env_stat(MDB_env *env, MDB_stat *stat);
752 /** @brief Return information about the LMDB environment.
754 * @param[in] env An environment handle returned by #mdb_env_create()
755 * @param[out] stat The address of an #MDB_envinfo structure
756 * where the information will be copied
758 int mdb_env_info(MDB_env *env, MDB_envinfo *stat);
760 /** @brief Flush the data buffers to disk.
762 * Data is always written to disk when #mdb_txn_commit() is called,
763 * but the operating system may keep it buffered. LMDB always flushes
764 * the OS buffers upon commit as well, unless the environment was
765 * opened with #MDB_NOSYNC or in part #MDB_NOMETASYNC. This call is
766 * not valid if the environment was opened with #MDB_RDONLY.
767 * @param[in] env An environment handle returned by #mdb_env_create()
768 * @param[in] force If non-zero, force a synchronous flush. Otherwise
769 * if the environment has the #MDB_NOSYNC flag set the flushes
770 * will be omitted, and with #MDB_MAPASYNC they will be asynchronous.
771 * @return A non-zero error value on failure and 0 on success. Some possible
774 * <li>EACCES - the environment is read-only.
775 * <li>EINVAL - an invalid parameter was specified.
776 * <li>EIO - an error occurred during synchronization.
779 int mdb_env_sync(MDB_env *env, int force);
781 /** @brief Close the environment and release the memory map.
783 * Only a single thread may call this function. All transactions, databases,
784 * and cursors must already be closed before calling this function. Attempts to
785 * use any such handles after calling this function will cause a SIGSEGV.
786 * The environment handle will be freed and must not be used again after this call.
787 * @param[in] env An environment handle returned by #mdb_env_create()
789 void mdb_env_close(MDB_env *env);
791 /** @brief Set environment flags.
793 * This may be used to set some flags in addition to those from
794 * #mdb_env_open(), or to unset these flags. If several threads
795 * change the flags at the same time, the result is undefined.
796 * @param[in] env An environment handle returned by #mdb_env_create()
797 * @param[in] flags The flags to change, bitwise OR'ed together
798 * @param[in] onoff A non-zero value sets the flags, zero clears them.
799 * @return A non-zero error value on failure and 0 on success. Some possible
802 * <li>EINVAL - an invalid parameter was specified.
805 int mdb_env_set_flags(MDB_env *env, unsigned int flags, int onoff);
807 /** @brief Get environment flags.
809 * @param[in] env An environment handle returned by #mdb_env_create()
810 * @param[out] flags The address of an integer to store the flags
811 * @return A non-zero error value on failure and 0 on success. Some possible
814 * <li>EINVAL - an invalid parameter was specified.
817 int mdb_env_get_flags(MDB_env *env, unsigned int *flags);
819 /** @brief Return the path that was used in #mdb_env_open().
821 * @param[in] env An environment handle returned by #mdb_env_create()
822 * @param[out] path Address of a string pointer to contain the path. This
823 * is the actual string in the environment, not a copy. It should not be
824 * altered in any way.
825 * @return A non-zero error value on failure and 0 on success. Some possible
828 * <li>EINVAL - an invalid parameter was specified.
831 int mdb_env_get_path(MDB_env *env, const char **path);
833 /** @brief Return the filedescriptor for the given environment.
835 * @param[in] env An environment handle returned by #mdb_env_create()
836 * @param[out] fd Address of a mdb_filehandle_t to contain the descriptor.
837 * @return A non-zero error value on failure and 0 on success. Some possible
840 * <li>EINVAL - an invalid parameter was specified.
843 int mdb_env_get_fd(MDB_env *env, mdb_filehandle_t *fd);
845 /** @brief Set the size of the memory map to use for this environment.
847 * The size should be a multiple of the OS page size. The default is
848 * 10485760 bytes. The size of the memory map is also the maximum size
849 * of the database. The value should be chosen as large as possible,
850 * to accommodate future growth of the database.
851 * This function should be called after #mdb_env_create() and before #mdb_env_open().
852 * It may be called at later times if no transactions are active in
853 * this process. Note that the library does not check for this condition,
854 * the caller must ensure it explicitly.
856 * The new size takes effect immediately for the current process but
857 * will not be persisted to any others until a write transaction has been
858 * committed by the current process. Also, only mapsize increases are
859 * persisted into the environment.
861 * If the mapsize is increased by another process, and data has grown
862 * beyond the range of the current mapsize, #mdb_txn_begin() will
863 * return #MDB_MAP_RESIZED. This function may be called with a size
864 * of zero to adopt the new size.
866 * Any attempt to set a size smaller than the space already consumed
867 * by the environment will be silently changed to the current size of the used space.
868 * @param[in] env An environment handle returned by #mdb_env_create()
869 * @param[in] size The size in bytes
870 * @return A non-zero error value on failure and 0 on success. Some possible
873 * <li>EINVAL - an invalid parameter was specified, or the environment has
874 * an active write transaction.
877 int mdb_env_set_mapsize(MDB_env *env, mdb_size_t size);
879 /** @brief Set the maximum number of threads/reader slots for the environment.
881 * This defines the number of slots in the lock table that is used to track readers in the
882 * the environment. The default is 126.
883 * Starting a read-only transaction normally ties a lock table slot to the
884 * current thread until the environment closes or the thread exits. If
885 * MDB_NOTLS is in use, #mdb_txn_begin() instead ties the slot to the
886 * MDB_txn object until it or the #MDB_env object is destroyed.
887 * This function may only be called after #mdb_env_create() and before #mdb_env_open().
888 * @param[in] env An environment handle returned by #mdb_env_create()
889 * @param[in] readers The maximum number of reader lock table slots
890 * @return A non-zero error value on failure and 0 on success. Some possible
893 * <li>EINVAL - an invalid parameter was specified, or the environment is already open.
896 int mdb_env_set_maxreaders(MDB_env *env, unsigned int readers);
898 /** @brief Get the maximum number of threads/reader slots for the environment.
900 * @param[in] env An environment handle returned by #mdb_env_create()
901 * @param[out] readers Address of an integer to store the number of readers
902 * @return A non-zero error value on failure and 0 on success. Some possible
905 * <li>EINVAL - an invalid parameter was specified.
908 int mdb_env_get_maxreaders(MDB_env *env, unsigned int *readers);
910 /** @brief Set the maximum number of named databases for the environment.
912 * This function is only needed if multiple databases will be used in the
913 * environment. Simpler applications that use the environment as a single
914 * unnamed database can ignore this option.
915 * This function may only be called after #mdb_env_create() and before #mdb_env_open().
917 * Currently a moderate number of slots are cheap but a huge number gets
918 * expensive: 7-120 words per transaction, and every #mdb_dbi_open()
919 * does a linear search of the opened slots.
920 * @param[in] env An environment handle returned by #mdb_env_create()
921 * @param[in] dbs The maximum number of databases
922 * @return A non-zero error value on failure and 0 on success. Some possible
925 * <li>EINVAL - an invalid parameter was specified, or the environment is already open.
928 int mdb_env_set_maxdbs(MDB_env *env, MDB_dbi dbs);
930 /** @brief Get the maximum size of keys and #MDB_DUPSORT data we can write.
932 * Depends on the compile-time constant #MDB_MAXKEYSIZE. Default 511.
934 * @param[in] env An environment handle returned by #mdb_env_create()
935 * @return The maximum size of a key we can write
937 int mdb_env_get_maxkeysize(MDB_env *env);
939 /** @brief Set application information associated with the #MDB_env.
941 * @param[in] env An environment handle returned by #mdb_env_create()
942 * @param[in] ctx An arbitrary pointer for whatever the application needs.
943 * @return A non-zero error value on failure and 0 on success.
945 int mdb_env_set_userctx(MDB_env *env, void *ctx);
947 /** @brief Get the application information associated with the #MDB_env.
949 * @param[in] env An environment handle returned by #mdb_env_create()
950 * @return The pointer set by #mdb_env_set_userctx().
952 void *mdb_env_get_userctx(MDB_env *env);
954 /** @brief A callback function for most LMDB assert() failures,
955 * called before printing the message and aborting.
957 * @param[in] env An environment handle returned by #mdb_env_create().
958 * @param[in] msg The assertion message, not including newline.
960 typedef void MDB_assert_func(MDB_env *env, const char *msg);
962 /** Set or reset the assert() callback of the environment.
963 * Disabled if liblmdb is buillt with NDEBUG.
964 * @note This hack should become obsolete as lmdb's error handling matures.
965 * @param[in] env An environment handle returned by #mdb_env_create().
966 * @param[in] func An #MDB_assert_func function, or 0.
967 * @return A non-zero error value on failure and 0 on success.
969 int mdb_env_set_assert(MDB_env *env, MDB_assert_func *func);
971 /** @brief Create a transaction for use with the environment.
973 * The transaction handle may be discarded using #mdb_txn_abort() or #mdb_txn_commit().
974 * @note A transaction and its cursors must only be used by a single
975 * thread, and a thread may only have a single transaction at a time.
976 * If #MDB_NOTLS is in use, this does not apply to read-only transactions.
977 * @note Cursors may not span transactions.
978 * @param[in] env An environment handle returned by #mdb_env_create()
979 * @param[in] parent If this parameter is non-NULL, the new transaction
980 * will be a nested transaction, with the transaction indicated by \b parent
981 * as its parent. Transactions may be nested to any level. A parent
982 * transaction and its cursors may not issue any other operations than
983 * mdb_txn_commit and mdb_txn_abort while it has active child transactions.
984 * @param[in] flags Special options for this transaction. This parameter
985 * must be set to 0 or by bitwise OR'ing together one or more of the
986 * values described here.
989 * This transaction will not perform any write operations.
991 * Don't flush system buffers to disk when committing this transaction.
992 * <li>#MDB_NOMETASYNC
993 * Flush system buffers but omit metadata flush when committing this transaction.
995 * @param[out] txn Address where the new #MDB_txn handle will be stored
996 * @return A non-zero error value on failure and 0 on success. Some possible
999 * <li>#MDB_PANIC - a fatal error occurred earlier and the environment
1000 * must be shut down.
1001 * <li>#MDB_MAP_RESIZED - another process wrote data beyond this MDB_env's
1002 * mapsize and this environment's map must be resized as well.
1003 * See #mdb_env_set_mapsize().
1004 * <li>#MDB_READERS_FULL - a read-only transaction was requested and
1005 * the reader lock table is full. See #mdb_env_set_maxreaders().
1006 * <li>ENOMEM - out of memory.
1009 int mdb_txn_begin(MDB_env *env, MDB_txn *parent, unsigned int flags, MDB_txn **txn);
1011 /** @brief Returns the transaction's #MDB_env
1013 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1015 MDB_env *mdb_txn_env(MDB_txn *txn);
1017 /** @brief Return the transaction's ID.
1019 * This returns the identifier associated with this transaction. For a
1020 * read-only transaction, this corresponds to the snapshot being read;
1021 * concurrent readers will frequently have the same transaction ID.
1023 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1024 * @return A transaction ID, valid if input is an active transaction.
1026 mdb_size_t mdb_txn_id(MDB_txn *txn);
1028 /** @brief Commit all the operations of a transaction into the database.
1030 * The transaction handle is freed. It and its cursors must not be used
1031 * again after this call, except with #mdb_cursor_renew().
1032 * @note Earlier documentation incorrectly said all cursors would be freed.
1033 * Only write-transactions free cursors.
1034 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1035 * @return A non-zero error value on failure and 0 on success. Some possible
1038 * <li>EINVAL - an invalid parameter was specified.
1039 * <li>ENOSPC - no more disk space.
1040 * <li>EIO - a low-level I/O error occurred while writing.
1041 * <li>ENOMEM - out of memory.
1044 int mdb_txn_commit(MDB_txn *txn);
1046 /** @brief Abandon all the operations of the transaction instead of saving them.
1048 * The transaction handle is freed. It and its cursors must not be used
1049 * again after this call, except with #mdb_cursor_renew().
1050 * @note Earlier documentation incorrectly said all cursors would be freed.
1051 * Only write-transactions free cursors.
1052 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1054 void mdb_txn_abort(MDB_txn *txn);
1056 /** @brief Reset a read-only transaction.
1058 * Abort the transaction like #mdb_txn_abort(), but keep the transaction
1059 * handle. #mdb_txn_renew() may reuse the handle. This saves allocation
1060 * overhead if the process will start a new read-only transaction soon,
1061 * and also locking overhead if #MDB_NOTLS is in use. The reader table
1062 * lock is released, but the table slot stays tied to its thread or
1063 * #MDB_txn. Use mdb_txn_abort() to discard a reset handle, and to free
1064 * its lock table slot if MDB_NOTLS is in use.
1065 * Cursors opened within the transaction must not be used
1066 * again after this call, except with #mdb_cursor_renew().
1067 * Reader locks generally don't interfere with writers, but they keep old
1068 * versions of database pages allocated. Thus they prevent the old pages
1069 * from being reused when writers commit new data, and so under heavy load
1070 * the database size may grow much more rapidly than otherwise.
1071 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1073 void mdb_txn_reset(MDB_txn *txn);
1075 /** @brief Renew a read-only transaction.
1077 * This acquires a new reader lock for a transaction handle that had been
1078 * released by #mdb_txn_reset(). It must be called before a reset transaction
1079 * may be used again.
1080 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1081 * @return A non-zero error value on failure and 0 on success. Some possible
1084 * <li>#MDB_PANIC - a fatal error occurred earlier and the environment
1085 * must be shut down.
1086 * <li>EINVAL - an invalid parameter was specified.
1089 int mdb_txn_renew(MDB_txn *txn);
1091 /** Compat with version <= 0.9.4, avoid clash with libmdb from MDB Tools project */
1092 #define mdb_open(txn,name,flags,dbi) mdb_dbi_open(txn,name,flags,dbi)
1093 /** Compat with version <= 0.9.4, avoid clash with libmdb from MDB Tools project */
1094 #define mdb_close(env,dbi) mdb_dbi_close(env,dbi)
1096 /** @brief Open a database in the environment.
1098 * A database handle denotes the name and parameters of a database,
1099 * independently of whether such a database exists.
1100 * The database handle may be discarded by calling #mdb_dbi_close().
1101 * The old database handle is returned if the database was already open.
1102 * The handle may only be closed once.
1104 * The database handle will be private to the current transaction until
1105 * the transaction is successfully committed. If the transaction is
1106 * aborted the handle will be closed automatically.
1107 * After a successful commit the handle will reside in the shared
1108 * environment, and may be used by other transactions.
1110 * This function must not be called from multiple concurrent
1111 * transactions in the same process. A transaction that uses
1112 * this function must finish (either commit or abort) before
1113 * any other transaction in the process may use this function.
1115 * To use named databases (with name != NULL), #mdb_env_set_maxdbs()
1116 * must be called before opening the environment. Database names are
1117 * keys in the unnamed database, and may be read but not written.
1119 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1120 * @param[in] name The name of the database to open. If only a single
1121 * database is needed in the environment, this value may be NULL.
1122 * @param[in] flags Special options for this database. This parameter
1123 * must be set to 0 or by bitwise OR'ing together one or more of the
1124 * values described here.
1126 * <li>#MDB_REVERSEKEY
1127 * Keys are strings to be compared in reverse order, from the end
1128 * of the strings to the beginning. By default, Keys are treated as strings and
1129 * compared from beginning to end.
1131 * Duplicate keys may be used in the database. (Or, from another perspective,
1132 * keys may have multiple data items, stored in sorted order.) By default
1133 * keys must be unique and may have only a single data item.
1134 * <li>#MDB_INTEGERKEY
1135 * Keys are binary integers in native byte order, either unsigned int
1136 * or #mdb_size_t, and will be sorted as such.
1137 * (lmdb expects 32-bit int <= size_t <= 32/64-bit mdb_size_t.)
1138 * The keys must all be of the same size.
1140 * This flag may only be used in combination with #MDB_DUPSORT. This option
1141 * tells the library that the data items for this database are all the same
1142 * size, which allows further optimizations in storage and retrieval. When
1143 * all data items are the same size, the #MDB_GET_MULTIPLE and #MDB_NEXT_MULTIPLE
1144 * cursor operations may be used to retrieve multiple items at once.
1145 * <li>#MDB_INTEGERDUP
1146 * This option specifies that duplicate data items are binary integers,
1147 * similar to #MDB_INTEGERKEY keys.
1148 * <li>#MDB_REVERSEDUP
1149 * This option specifies that duplicate data items should be compared as
1150 * strings in reverse order.
1152 * Create the named database if it doesn't exist. This option is not
1153 * allowed in a read-only transaction or a read-only environment.
1155 * @param[out] dbi Address where the new #MDB_dbi handle will be stored
1156 * @return A non-zero error value on failure and 0 on success. Some possible
1159 * <li>#MDB_NOTFOUND - the specified database doesn't exist in the environment
1160 * and #MDB_CREATE was not specified.
1161 * <li>#MDB_DBS_FULL - too many databases have been opened. See #mdb_env_set_maxdbs().
1164 int mdb_dbi_open(MDB_txn *txn, const char *name, unsigned int flags, MDB_dbi *dbi);
1166 /** @brief Retrieve statistics for a database.
1168 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1169 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1170 * @param[out] stat The address of an #MDB_stat structure
1171 * where the statistics will be copied
1172 * @return A non-zero error value on failure and 0 on success. Some possible
1175 * <li>EINVAL - an invalid parameter was specified.
1178 int mdb_stat(MDB_txn *txn, MDB_dbi dbi, MDB_stat *stat);
1180 /** @brief Retrieve the DB flags for a database handle.
1182 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1183 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1184 * @param[out] flags Address where the flags will be returned.
1185 * @return A non-zero error value on failure and 0 on success.
1187 int mdb_dbi_flags(MDB_txn *txn, MDB_dbi dbi, unsigned int *flags);
1189 /** @brief Close a database handle. Normally unnecessary. Use with care:
1191 * This call is not mutex protected. Handles should only be closed by
1192 * a single thread, and only if no other threads are going to reference
1193 * the database handle or one of its cursors any further. Do not close
1194 * a handle if an existing transaction has modified its database.
1195 * Doing so can cause misbehavior from database corruption to errors
1196 * like MDB_BAD_VALSIZE (since the DB name is gone).
1198 * Closing a database handle is not necessary, but lets #mdb_dbi_open()
1199 * reuse the handle value. Usually it's better to set a bigger
1200 * #mdb_env_set_maxdbs(), unless that value would be large.
1202 * @param[in] env An environment handle returned by #mdb_env_create()
1203 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1205 void mdb_dbi_close(MDB_env *env, MDB_dbi dbi);
1207 /** @brief Empty or delete+close a database.
1209 * See #mdb_dbi_close() for restrictions about closing the DB handle.
1210 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1211 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1212 * @param[in] del 0 to empty the DB, 1 to delete it from the
1213 * environment and close the DB handle.
1214 * @return A non-zero error value on failure and 0 on success.
1216 int mdb_drop(MDB_txn *txn, MDB_dbi dbi, int del);
1218 /** @brief Set a custom key comparison function for a database.
1220 * The comparison function is called whenever it is necessary to compare a
1221 * key specified by the application with a key currently stored in the database.
1222 * If no comparison function is specified, and no special key flags were specified
1223 * with #mdb_dbi_open(), the keys are compared lexically, with shorter keys collating
1224 * before longer keys.
1225 * @warning This function must be called before any data access functions are used,
1226 * otherwise data corruption may occur. The same comparison function must be used by every
1227 * program accessing the database, every time the database is used.
1228 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1229 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1230 * @param[in] cmp A #MDB_cmp_func function
1231 * @return A non-zero error value on failure and 0 on success. Some possible
1234 * <li>EINVAL - an invalid parameter was specified.
1237 int mdb_set_compare(MDB_txn *txn, MDB_dbi dbi, MDB_cmp_func *cmp);
1239 /** @brief Set a custom data comparison function for a #MDB_DUPSORT database.
1241 * This comparison function is called whenever it is necessary to compare a data
1242 * item specified by the application with a data item currently stored in the database.
1243 * This function only takes effect if the database was opened with the #MDB_DUPSORT
1245 * If no comparison function is specified, and no special key flags were specified
1246 * with #mdb_dbi_open(), the data items are compared lexically, with shorter items collating
1247 * before longer items.
1248 * @warning This function must be called before any data access functions are used,
1249 * otherwise data corruption may occur. The same comparison function must be used by every
1250 * program accessing the database, every time the database is used.
1251 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1252 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1253 * @param[in] cmp A #MDB_cmp_func function
1254 * @return A non-zero error value on failure and 0 on success. Some possible
1257 * <li>EINVAL - an invalid parameter was specified.
1260 int mdb_set_dupsort(MDB_txn *txn, MDB_dbi dbi, MDB_cmp_func *cmp);
1262 /** @brief Set a relocation function for a #MDB_FIXEDMAP database.
1264 * @todo The relocation function is called whenever it is necessary to move the data
1265 * of an item to a different position in the database (e.g. through tree
1266 * balancing operations, shifts as a result of adds or deletes, etc.). It is
1267 * intended to allow address/position-dependent data items to be stored in
1268 * a database in an environment opened with the #MDB_FIXEDMAP option.
1269 * Currently the relocation feature is unimplemented and setting
1270 * this function has no effect.
1271 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1272 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1273 * @param[in] rel A #MDB_rel_func function
1274 * @return A non-zero error value on failure and 0 on success. Some possible
1277 * <li>EINVAL - an invalid parameter was specified.
1280 int mdb_set_relfunc(MDB_txn *txn, MDB_dbi dbi, MDB_rel_func *rel);
1282 /** @brief Set a context pointer for a #MDB_FIXEDMAP database's relocation function.
1284 * See #mdb_set_relfunc and #MDB_rel_func for more details.
1285 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1286 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1287 * @param[in] ctx An arbitrary pointer for whatever the application needs.
1288 * It will be passed to the callback function set by #mdb_set_relfunc
1289 * as its \b relctx parameter whenever the callback is invoked.
1290 * @return A non-zero error value on failure and 0 on success. Some possible
1293 * <li>EINVAL - an invalid parameter was specified.
1296 int mdb_set_relctx(MDB_txn *txn, MDB_dbi dbi, void *ctx);
1298 /** @brief Get items from a database.
1300 * This function retrieves key/data pairs from the database. The address
1301 * and length of the data associated with the specified \b key are returned
1302 * in the structure to which \b data refers.
1303 * If the database supports duplicate keys (#MDB_DUPSORT) then the
1304 * first data item for the key will be returned. Retrieval of other
1305 * items requires the use of #mdb_cursor_get().
1307 * @note The memory pointed to by the returned values is owned by the
1308 * database. The caller need not dispose of the memory, and may not
1309 * modify it in any way. For values returned in a read-only transaction
1310 * any modification attempts will cause a SIGSEGV.
1311 * @note Values returned from the database are valid only until a
1312 * subsequent update operation, or the end of the transaction.
1313 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1314 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1315 * @param[in] key The key to search for in the database
1316 * @param[out] data The data corresponding to the key
1317 * @return A non-zero error value on failure and 0 on success. Some possible
1320 * <li>#MDB_NOTFOUND - the key was not in the database.
1321 * <li>EINVAL - an invalid parameter was specified.
1324 int mdb_get(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data);
1326 /** @brief Store items into a database.
1328 * This function stores key/data pairs in the database. The default behavior
1329 * is to enter the new key/data pair, replacing any previously existing key
1330 * if duplicates are disallowed, or adding a duplicate data item if
1331 * duplicates are allowed (#MDB_DUPSORT).
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 store in the database
1335 * @param[in,out] data The data to store
1336 * @param[in] flags Special options for this operation. This parameter
1337 * must be set to 0 or by bitwise OR'ing together one or more of the
1338 * values described here.
1340 * <li>#MDB_NODUPDATA - enter the new key/data pair only if it does not
1341 * already appear in the database. This flag may only be specified
1342 * if the database was opened with #MDB_DUPSORT. The function will
1343 * return #MDB_KEYEXIST if the key/data pair already appears in the
1345 * <li>#MDB_NOOVERWRITE - enter the new key/data pair only if the key
1346 * does not already appear in the database. The function will return
1347 * #MDB_KEYEXIST if the key already appears in the database, even if
1348 * the database supports duplicates (#MDB_DUPSORT). The \b data
1349 * parameter will be set to point to the existing item.
1350 * <li>#MDB_RESERVE - reserve space for data of the given size, but
1351 * don't copy the given data. Instead, return a pointer to the
1352 * reserved space, which the caller can fill in later - before
1353 * the next update operation or the transaction ends. This saves
1354 * an extra memcpy if the data is being generated later.
1355 * LMDB does nothing else with this memory, the caller is expected
1356 * to modify all of the space requested. This flag must not be
1357 * specified if the database was opened with #MDB_DUPSORT.
1358 * <li>#MDB_APPEND - append the given key/data pair to the end of the
1359 * database. This option allows fast bulk loading when keys are
1360 * already known to be in the correct order. Loading unsorted keys
1361 * with this flag will cause a #MDB_KEYEXIST error.
1362 * <li>#MDB_APPENDDUP - as above, but for sorted dup data.
1364 * @return A non-zero error value on failure and 0 on success. Some possible
1367 * <li>#MDB_MAP_FULL - the database is full, see #mdb_env_set_mapsize().
1368 * <li>#MDB_TXN_FULL - the transaction has too many dirty pages.
1369 * <li>EACCES - an attempt was made to write in a read-only transaction.
1370 * <li>EINVAL - an invalid parameter was specified.
1373 int mdb_put(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data,
1374 unsigned int flags);
1376 /** @brief Delete items from a database.
1378 * This function removes key/data pairs from the database.
1379 * If the database does not support sorted duplicate data items
1380 * (#MDB_DUPSORT) the data parameter is ignored.
1381 * If the database supports sorted duplicates and the data parameter
1382 * is NULL, all of the duplicate data items for the key will be
1383 * deleted. Otherwise, if the data parameter is non-NULL
1384 * only the matching data item will be deleted.
1385 * This function will return #MDB_NOTFOUND if the specified key/data
1386 * pair is not in the database.
1387 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1388 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1389 * @param[in] key The key to delete from the database
1390 * @param[in] data The data to delete
1391 * @return A non-zero error value on failure and 0 on success. Some possible
1394 * <li>EACCES - an attempt was made to write in a read-only transaction.
1395 * <li>EINVAL - an invalid parameter was specified.
1398 int mdb_del(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data);
1400 /** @brief Create a cursor handle.
1402 * A cursor is associated with a specific transaction and database.
1403 * A cursor cannot be used when its database handle is closed. Nor
1404 * when its transaction has ended, except with #mdb_cursor_renew().
1405 * It can be discarded with #mdb_cursor_close().
1406 * A cursor in a write-transaction can be closed before its transaction
1407 * ends, and will otherwise be closed when its transaction ends.
1408 * A cursor in a read-only transaction must be closed explicitly, before
1409 * or after its transaction ends. It can be reused with
1410 * #mdb_cursor_renew() before finally closing it.
1411 * @note Earlier documentation said that cursors in every transaction
1412 * were closed when the transaction committed or aborted.
1413 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1414 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1415 * @param[out] cursor Address where the new #MDB_cursor handle will be stored
1416 * @return A non-zero error value on failure and 0 on success. Some possible
1419 * <li>EINVAL - an invalid parameter was specified.
1422 int mdb_cursor_open(MDB_txn *txn, MDB_dbi dbi, MDB_cursor **cursor);
1424 /** @brief Close a cursor handle.
1426 * The cursor handle will be freed and must not be used again after this call.
1427 * Its transaction must still be live if it is a write-transaction.
1428 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1430 void mdb_cursor_close(MDB_cursor *cursor);
1432 /** @brief Renew a cursor handle.
1434 * A cursor is associated with a specific transaction and database.
1435 * Cursors that are only used in read-only
1436 * transactions may be re-used, to avoid unnecessary malloc/free overhead.
1437 * The cursor may be associated with a new read-only transaction, and
1438 * referencing the same database handle as it was created with.
1439 * This may be done whether the previous transaction is live or dead.
1440 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1441 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1442 * @return A non-zero error value on failure and 0 on success. Some possible
1445 * <li>EINVAL - an invalid parameter was specified.
1448 int mdb_cursor_renew(MDB_txn *txn, MDB_cursor *cursor);
1450 /** @brief Return the cursor's transaction handle.
1452 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1454 MDB_txn *mdb_cursor_txn(MDB_cursor *cursor);
1456 /** @brief Return the cursor's database handle.
1458 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1460 MDB_dbi mdb_cursor_dbi(MDB_cursor *cursor);
1462 /** @brief Retrieve by cursor.
1464 * This function retrieves key/data pairs from the database. The address and length
1465 * of the key are returned in the object to which \b key refers (except for the
1466 * case of the #MDB_SET option, in which the \b key object is unchanged), and
1467 * the address and length of the data are returned in the object to which \b data
1469 * See #mdb_get() for restrictions on using the output values.
1470 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1471 * @param[in,out] key The key for a retrieved item
1472 * @param[in,out] data The data of a retrieved item
1473 * @param[in] op A cursor operation #MDB_cursor_op
1474 * @return A non-zero error value on failure and 0 on success. Some possible
1477 * <li>#MDB_NOTFOUND - no matching key found.
1478 * <li>EINVAL - an invalid parameter was specified.
1481 int mdb_cursor_get(MDB_cursor *cursor, MDB_val *key, MDB_val *data,
1484 /** @brief Store by cursor.
1486 * This function stores key/data pairs into the database.
1487 * The cursor is positioned at the new item, or on failure usually near it.
1488 * @note Earlier documentation incorrectly said errors would leave the
1489 * state of the cursor unchanged.
1490 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1491 * @param[in] key The key operated on.
1492 * @param[in] data The data operated on.
1493 * @param[in] flags Options for this operation. This parameter
1494 * must be set to 0 or one of the values described here.
1496 * <li>#MDB_CURRENT - replace the item at the current cursor position.
1497 * The \b key parameter must still be provided, and must match it.
1498 * If using sorted duplicates (#MDB_DUPSORT) the data item must still
1499 * sort into the same place. This is intended to be used when the
1500 * new data is the same size as the old. Otherwise it will simply
1501 * perform a delete of the old record followed by an insert.
1502 * <li>#MDB_NODUPDATA - enter the new key/data pair only if it does not
1503 * already appear in the database. This flag may only be specified
1504 * if the database was opened with #MDB_DUPSORT. The function will
1505 * return #MDB_KEYEXIST if the key/data pair already appears in the
1507 * <li>#MDB_NOOVERWRITE - enter the new key/data pair only if the key
1508 * does not already appear in the database. The function will return
1509 * #MDB_KEYEXIST if the key already appears in the database, even if
1510 * the database supports duplicates (#MDB_DUPSORT).
1511 * <li>#MDB_RESERVE - reserve space for data of the given size, but
1512 * don't copy the given data. Instead, return a pointer to the
1513 * reserved space, which the caller can fill in later - before
1514 * the next update operation or the transaction ends. This saves
1515 * an extra memcpy if the data is being generated later. This flag
1516 * must not be specified if the database was opened with #MDB_DUPSORT.
1517 * <li>#MDB_APPEND - append the given key/data pair to the end of the
1518 * database. No key comparisons are performed. This option allows
1519 * fast bulk loading when keys are already known to be in the
1520 * correct order. Loading unsorted keys with this flag will cause
1521 * a #MDB_KEYEXIST error.
1522 * <li>#MDB_APPENDDUP - as above, but for sorted dup data.
1523 * <li>#MDB_MULTIPLE - store multiple contiguous data elements in a
1524 * single request. This flag may only be specified if the database
1525 * was opened with #MDB_DUPFIXED. The \b data argument must be an
1526 * array of two MDB_vals. The mv_size of the first MDB_val must be
1527 * the size of a single data element. The mv_data of the first MDB_val
1528 * must point to the beginning of the array of contiguous data elements.
1529 * The mv_size of the second MDB_val must be the count of the number
1530 * of data elements to store. On return this field will be set to
1531 * the count of the number of elements actually written. The mv_data
1532 * of the second MDB_val is unused.
1534 * @return A non-zero error value on failure and 0 on success. Some possible
1537 * <li>#MDB_MAP_FULL - the database is full, see #mdb_env_set_mapsize().
1538 * <li>#MDB_TXN_FULL - the transaction has too many dirty pages.
1539 * <li>EACCES - an attempt was made to write in a read-only transaction.
1540 * <li>EINVAL - an invalid parameter was specified.
1543 int mdb_cursor_put(MDB_cursor *cursor, MDB_val *key, MDB_val *data,
1544 unsigned int flags);
1546 /** @brief Delete current key/data pair
1548 * This function deletes the key/data pair to which the cursor refers.
1549 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1550 * @param[in] flags Options for this operation. This parameter
1551 * must be set to 0 or one of the values described here.
1553 * <li>#MDB_NODUPDATA - delete all of the data items for the current key.
1554 * This flag may only be specified if the database was opened with #MDB_DUPSORT.
1556 * @return A non-zero error value on failure and 0 on success. Some possible
1559 * <li>EACCES - an attempt was made to write in a read-only transaction.
1560 * <li>EINVAL - an invalid parameter was specified.
1563 int mdb_cursor_del(MDB_cursor *cursor, unsigned int flags);
1565 /** @brief Return count of duplicates for current key.
1567 * This call is only valid on databases that support sorted duplicate
1568 * data items #MDB_DUPSORT.
1569 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1570 * @param[out] countp Address where the count will be stored
1571 * @return A non-zero error value on failure and 0 on success. Some possible
1574 * <li>EINVAL - cursor is not initialized, or an invalid parameter was specified.
1577 int mdb_cursor_count(MDB_cursor *cursor, mdb_size_t *countp);
1579 /** @brief Compare two data items according to a particular database.
1581 * This returns a comparison as if the two data items were keys in the
1582 * specified database.
1583 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1584 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1585 * @param[in] a The first item to compare
1586 * @param[in] b The second item to compare
1587 * @return < 0 if a < b, 0 if a == b, > 0 if a > b
1589 int mdb_cmp(MDB_txn *txn, MDB_dbi dbi, const MDB_val *a, const MDB_val *b);
1591 /** @brief Compare two data items according to a particular database.
1593 * This returns a comparison as if the two items were data items of
1594 * the specified database. The database must have the #MDB_DUPSORT flag.
1595 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1596 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1597 * @param[in] a The first item to compare
1598 * @param[in] b The second item to compare
1599 * @return < 0 if a < b, 0 if a == b, > 0 if a > b
1601 int mdb_dcmp(MDB_txn *txn, MDB_dbi dbi, const MDB_val *a, const MDB_val *b);
1603 /** @brief A callback function used to print a message from the library.
1605 * @param[in] msg The string to be printed.
1606 * @param[in] ctx An arbitrary context pointer for the callback.
1607 * @return < 0 on failure, >= 0 on success.
1609 typedef int (MDB_msg_func)(const char *msg, void *ctx);
1611 /** @brief Dump the entries in the reader lock table.
1613 * @param[in] env An environment handle returned by #mdb_env_create()
1614 * @param[in] func A #MDB_msg_func function
1615 * @param[in] ctx Anything the message function needs
1616 * @return < 0 on failure, >= 0 on success.
1618 int mdb_reader_list(MDB_env *env, MDB_msg_func *func, void *ctx);
1620 /** @brief Check for stale entries in the reader lock table.
1622 * @param[in] env An environment handle returned by #mdb_env_create()
1623 * @param[out] dead Number of stale slots that were cleared
1624 * @return 0 on success, non-zero on failure.
1626 int mdb_reader_check(MDB_env *env, int *dead);
1632 /** @page tools LMDB Command Line Tools
1633 The following describes the command line tools that are available for LMDB.
1640 #endif /* _LMDB_H_ */