2 * @brief Lightning memory-mapped database library
4 * @mainpage Lightning Memory-Mapped Database Manager (MDB)
6 * @section intro_sec Introduction
7 * MDB is a Btree-based database management library modeled loosely on the
8 * BerkeleyDB API, but much simplified. The entire database is exposed
9 * in a memory map, and all data fetches return data directly
10 * from the mapped memory, so no malloc's or memcpy's occur during
11 * data fetches. As such, the library is extremely simple because it
12 * requires no page caching layer of its own, and it is extremely high
13 * performance and memory-efficient. It is also fully transactional with
14 * full ACID semantics, and when the memory map is read-only, the
15 * database integrity cannot be corrupted by stray pointer writes from
18 * The library is fully thread-aware and supports concurrent read/write
19 * access from multiple processes and threads. Data pages use a copy-on-
20 * write strategy so no active data pages are ever overwritten, which
21 * also provides resistance to corruption and eliminates the need of any
22 * special recovery procedures after a system crash. Writes are fully
23 * serialized; only one write transaction may be active at a time, which
24 * guarantees that writers can never deadlock. The database structure is
25 * multi-versioned so readers run with no locks; writers cannot block
26 * readers, and readers don't block writers.
28 * Unlike other well-known database mechanisms which use either write-ahead
29 * transaction logs or append-only data writes, MDB requires no maintenance
30 * during operation. Both write-ahead loggers and append-only databases
31 * require periodic checkpointing and/or compaction of their log or database
32 * files otherwise they grow without bound. MDB tracks free pages within
33 * the database and re-uses them for new write operations, so the database
34 * size does not grow without bound in normal use.
36 * The memory map can be used as a read-only or read-write map. It is
37 * read-only by default as this provides total immunity to corruption.
38 * Using read-write mode offers much higher write performance, but adds
39 * the possibility for stray application writes thru pointers to silently
40 * corrupt the database. Of course if your application code is known to
41 * be bug-free (...) then this is not an issue.
43 * @section caveats_sec Caveats
44 * Troubleshooting the lock file, plus semaphores on BSD systems:
46 * - A broken lockfile can cause sync issues.
47 * Stale reader transactions left behind by an aborted program
48 * cause further writes to grow the database quickly, and
49 * stale locks can block further operation.
51 * Fix: Check for stale readers periodically, using the
52 * #mdb_reader_check function or the mdb_stat tool. Or just
53 * make all programs using the database close it; the lockfile
54 * is always reset on first open of the environment.
56 * - On BSD systems or others configured with MDB_USE_POSIX_SEM,
57 * startup can fail due to semaphores owned by another userid.
59 * Fix: Open and close the database as the user which owns the
60 * semaphores (likely last user) or as root, while no other
61 * process is using the database.
63 * Restrictions/caveats (in addition to those listed for some functions):
65 * - Only the database owner should normally use the database on
66 * BSD systems or when otherwise configured with MDB_USE_POSIX_SEM.
67 * Multiple users can cause startup to fail later, as noted above.
69 * - There is normally no pure read-only mode, since readers need write
70 * access to locks and lock file. Exceptions: On read-only filesystems
71 * or with the #MDB_NOLOCK flag described under #mdb_env_open().
73 * - By default, in versions before 0.9.10, unused portions of the data
74 * file might receive garbage data from memory freed by other code.
75 * (This does not happen when using the #MDB_WRITEMAP flag.) As of
76 * 0.9.10 the default behavior is to initialize such memory before
77 * writing to the data file. Since there may be a slight performance
78 * cost due to this initialization, applications may disable it using
79 * the #MDB_NOMEMINIT flag. Applications handling sensitive data
80 * which must not be written should not use this flag. This flag is
81 * irrelevant when using #MDB_WRITEMAP.
83 * - A thread can only use one transaction at a time, plus any child
84 * transactions. Each transaction belongs to one thread. See below.
85 * The #MDB_NOTLS flag changes this for read-only transactions.
87 * - Use an MDB_env* in the process which opened it, without fork()ing.
89 * - Do not have open an MDB database twice in the same process at
90 * the same time. Not even from a plain open() call - close()ing it
91 * breaks flock() advisory locking.
93 * - Avoid long-lived transactions. Read transactions prevent
94 * reuse of pages freed by newer write transactions, thus the
95 * database can grow quickly. Write transactions prevent
96 * other write transactions, since writes are serialized.
98 * - Avoid suspending a process with active transactions. These
99 * would then be "long-lived" as above. Also read transactions
100 * suspended when writers commit could sometimes see wrong data.
102 * ...when several processes can use a database concurrently:
104 * - Avoid aborting a process with an active transaction.
105 * The transaction becomes "long-lived" as above until a check
106 * for stale readers is performed or the lockfile is reset,
107 * since the process may not remove it from the lockfile.
109 * - If you do that anyway, do a periodic check for stale readers. Or
110 * close the environment once in a while, so the lockfile can get reset.
112 * - Do not use MDB databases on remote filesystems, even between
113 * processes on the same host. This breaks flock() on some OSes,
114 * possibly memory map sync, and certainly sync between programs
115 * on different hosts.
117 * - Opening a database can fail if another process is opening or
118 * closing it at exactly the same time.
120 * @author Howard Chu, Symas Corporation.
122 * @copyright Copyright 2011-2013 Howard Chu, Symas Corp. All rights reserved.
124 * Redistribution and use in source and binary forms, with or without
125 * modification, are permitted only as authorized by the OpenLDAP
128 * A copy of this license is available in the file LICENSE in the
129 * top-level directory of the distribution or, alternatively, at
130 * <http://www.OpenLDAP.org/license.html>.
133 * This code is derived from btree.c written by Martin Hedenfalk.
135 * Copyright (c) 2009, 2010 Martin Hedenfalk <martin@bzero.se>
137 * Permission to use, copy, modify, and distribute this software for any
138 * purpose with or without fee is hereby granted, provided that the above
139 * copyright notice and this permission notice appear in all copies.
141 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
142 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
143 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
144 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
145 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
146 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
147 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
152 #include <sys/types.h>
158 /** Unix permissions for creating files, or dummy definition for Windows */
160 typedef int mdb_mode_t;
162 typedef mode_t mdb_mode_t;
165 /** An abstraction for a file handle.
166 * On POSIX systems file handles are small integers. On Windows
167 * they're opaque pointers.
170 typedef void *mdb_filehandle_t;
172 typedef int mdb_filehandle_t;
175 /** @defgroup mdb MDB API
177 * @brief OpenLDAP Lightning Memory-Mapped Database Manager
179 /** @defgroup Version Version Macros
182 /** Library major version */
183 #define MDB_VERSION_MAJOR 0
184 /** Library minor version */
185 #define MDB_VERSION_MINOR 9
186 /** Library patch version */
187 #define MDB_VERSION_PATCH 10
189 /** Combine args a,b,c into a single integer for easy version comparisons */
190 #define MDB_VERINT(a,b,c) (((a) << 24) | ((b) << 16) | (c))
192 /** The full library version as a single integer */
193 #define MDB_VERSION_FULL \
194 MDB_VERINT(MDB_VERSION_MAJOR,MDB_VERSION_MINOR,MDB_VERSION_PATCH)
196 /** The release date of this library version */
197 #define MDB_VERSION_DATE "November 11, 2013"
199 /** A stringifier for the version info */
200 #define MDB_VERSTR(a,b,c,d) "MDB " #a "." #b "." #c ": (" d ")"
202 /** A helper for the stringifier macro */
203 #define MDB_VERFOO(a,b,c,d) MDB_VERSTR(a,b,c,d)
205 /** The full library version as a C string */
206 #define MDB_VERSION_STRING \
207 MDB_VERFOO(MDB_VERSION_MAJOR,MDB_VERSION_MINOR,MDB_VERSION_PATCH,MDB_VERSION_DATE)
210 /** @brief Opaque structure for a database environment.
212 * A DB environment supports multiple databases, all residing in the same
215 typedef struct MDB_env MDB_env;
217 /** @brief Opaque structure for a transaction handle.
219 * All database operations require a transaction handle. Transactions may be
220 * read-only or read-write.
222 typedef struct MDB_txn MDB_txn;
224 /** @brief A handle for an individual database in the DB environment. */
225 typedef unsigned int MDB_dbi;
227 /** @brief Opaque structure for navigating through a database */
228 typedef struct MDB_cursor MDB_cursor;
230 /** @brief Generic structure used for passing keys and data in and out
233 * Values returned from the database are valid only until a subsequent
234 * update operation, or the end of the transaction. Do not modify or
235 * free them, they commonly point into the database itself.
237 * Key sizes must be between 1 and #mdb_env_get_maxkeysize() inclusive.
238 * The same applies to data sizes in databases with the #MDB_DUPSORT flag.
239 * Other data items can in theory be from 0 to 0xffffffff bytes long.
241 typedef struct MDB_val {
242 size_t mv_size; /**< size of the data item */
243 void *mv_data; /**< address of the data item */
246 /** @brief A callback function used to compare two keys in a database */
247 typedef int (MDB_cmp_func)(const MDB_val *a, const MDB_val *b);
249 /** @brief A callback function used to relocate a position-dependent data item
250 * in a fixed-address database.
252 * The \b newptr gives the item's desired address in
253 * the memory map, and \b oldptr gives its previous address. The item's actual
254 * data resides at the address in \b item. This callback is expected to walk
255 * through the fields of the record in \b item and modify any
256 * values based at the \b oldptr address to be relative to the \b newptr address.
257 * @param[in,out] item The item that is to be relocated.
258 * @param[in] oldptr The previous address.
259 * @param[in] newptr The new address to relocate to.
260 * @param[in] relctx An application-provided context, set by #mdb_set_relctx().
261 * @todo This feature is currently unimplemented.
263 typedef void (MDB_rel_func)(MDB_val *item, void *oldptr, void *newptr, void *relctx);
265 /** @defgroup mdb_env Environment Flags
268 /** mmap at a fixed address (experimental) */
269 #define MDB_FIXEDMAP 0x01
270 /** no environment directory */
271 #define MDB_NOSUBDIR 0x4000
272 /** don't fsync after commit */
273 #define MDB_NOSYNC 0x10000
275 #define MDB_RDONLY 0x20000
276 /** don't fsync metapage after commit */
277 #define MDB_NOMETASYNC 0x40000
278 /** use writable mmap */
279 #define MDB_WRITEMAP 0x80000
280 /** use asynchronous msync when #MDB_WRITEMAP is used */
281 #define MDB_MAPASYNC 0x100000
282 /** tie reader locktable slots to #MDB_txn objects instead of to threads */
283 #define MDB_NOTLS 0x200000
284 /** don't do any locking, caller must manage their own locks */
285 #define MDB_NOLOCK 0x400000
286 /** don't do readahead (no effect on Windows) */
287 #define MDB_NORDAHEAD 0x800000
288 /** don't initialize malloc'd memory before writing to datafile */
289 #define MDB_NOMEMINIT 0x1000000
292 /** @defgroup mdb_dbi_open Database Flags
295 /** use reverse string keys */
296 #define MDB_REVERSEKEY 0x02
297 /** use sorted duplicates */
298 #define MDB_DUPSORT 0x04
299 /** numeric keys in native byte order.
300 * The keys must all be of the same size. */
301 #define MDB_INTEGERKEY 0x08
302 /** with #MDB_DUPSORT, sorted dup items have fixed size */
303 #define MDB_DUPFIXED 0x10
304 /** with #MDB_DUPSORT, dups are numeric in native byte order */
305 #define MDB_INTEGERDUP 0x20
306 /** with #MDB_DUPSORT, use reverse string dups */
307 #define MDB_REVERSEDUP 0x40
308 /** create DB if not already existing */
309 #define MDB_CREATE 0x40000
312 /** @defgroup mdb_put Write Flags
315 /** For put: Don't write if the key already exists. */
316 #define MDB_NOOVERWRITE 0x10
317 /** Only for #MDB_DUPSORT<br>
318 * For put: don't write if the key and data pair already exist.<br>
319 * For mdb_cursor_del: remove all duplicate data items.
321 #define MDB_NODUPDATA 0x20
322 /** For mdb_cursor_put: overwrite the current key/data pair */
323 #define MDB_CURRENT 0x40
324 /** For put: Just reserve space for data, don't copy it. Return a
325 * pointer to the reserved space.
327 #define MDB_RESERVE 0x10000
328 /** Data is being appended, don't split full pages. */
329 #define MDB_APPEND 0x20000
330 /** Duplicate data is being appended, don't split full pages. */
331 #define MDB_APPENDDUP 0x40000
332 /** Store multiple data items in one call. Only for #MDB_DUPFIXED. */
333 #define MDB_MULTIPLE 0x80000
336 /** @brief Cursor Get operations.
338 * This is the set of all operations for retrieving data
341 typedef enum MDB_cursor_op {
342 MDB_FIRST, /**< Position at first key/data item */
343 MDB_FIRST_DUP, /**< Position at first data item of current key.
344 Only for #MDB_DUPSORT */
345 MDB_GET_BOTH, /**< Position at key/data pair. Only for #MDB_DUPSORT */
346 MDB_GET_BOTH_RANGE, /**< position at key, nearest data. Only for #MDB_DUPSORT */
347 MDB_GET_CURRENT, /**< Return key/data at current cursor position */
348 MDB_GET_MULTIPLE, /**< Return all the duplicate data items at the current
349 cursor position. Only for #MDB_DUPFIXED */
350 MDB_LAST, /**< Position at last key/data item */
351 MDB_LAST_DUP, /**< Position at last data item of current key.
352 Only for #MDB_DUPSORT */
353 MDB_NEXT, /**< Position at next data item */
354 MDB_NEXT_DUP, /**< Position at next data item of current key.
355 Only for #MDB_DUPSORT */
356 MDB_NEXT_MULTIPLE, /**< Return all duplicate data items at the next
357 cursor position. Only for #MDB_DUPFIXED */
358 MDB_NEXT_NODUP, /**< Position at first data item of next key */
359 MDB_PREV, /**< Position at previous data item */
360 MDB_PREV_DUP, /**< Position at previous data item of current key.
361 Only for #MDB_DUPSORT */
362 MDB_PREV_NODUP, /**< Position at last data item of previous key */
363 MDB_SET, /**< Position at specified key */
364 MDB_SET_KEY, /**< Position at specified key, return key + data */
365 MDB_SET_RANGE /**< Position at first key greater than or equal to specified key. */
368 /** @defgroup errors Return Codes
370 * BerkeleyDB uses -30800 to -30999, we'll go under them
373 /** Successful result */
374 #define MDB_SUCCESS 0
375 /** key/data pair already exists */
376 #define MDB_KEYEXIST (-30799)
377 /** key/data pair not found (EOF) */
378 #define MDB_NOTFOUND (-30798)
379 /** Requested page not found - this usually indicates corruption */
380 #define MDB_PAGE_NOTFOUND (-30797)
381 /** Located page was wrong type */
382 #define MDB_CORRUPTED (-30796)
383 /** Update of meta page failed, probably I/O error */
384 #define MDB_PANIC (-30795)
385 /** Environment version mismatch */
386 #define MDB_VERSION_MISMATCH (-30794)
387 /** File is not a valid MDB file */
388 #define MDB_INVALID (-30793)
389 /** Environment mapsize reached */
390 #define MDB_MAP_FULL (-30792)
391 /** Environment maxdbs reached */
392 #define MDB_DBS_FULL (-30791)
393 /** Environment maxreaders reached */
394 #define MDB_READERS_FULL (-30790)
395 /** Too many TLS keys in use - Windows only */
396 #define MDB_TLS_FULL (-30789)
397 /** Txn has too many dirty pages */
398 #define MDB_TXN_FULL (-30788)
399 /** Cursor stack too deep - internal error */
400 #define MDB_CURSOR_FULL (-30787)
401 /** Page has not enough space - internal error */
402 #define MDB_PAGE_FULL (-30786)
403 /** Database contents grew beyond environment mapsize */
404 #define MDB_MAP_RESIZED (-30785)
405 /** MDB_INCOMPATIBLE: Operation and DB incompatible, or DB flags changed */
406 #define MDB_INCOMPATIBLE (-30784)
407 /** Invalid reuse of reader locktable slot */
408 #define MDB_BAD_RSLOT (-30783)
409 /** Transaction cannot recover - it must be aborted */
410 #define MDB_BAD_TXN (-30782)
411 /** Too big key/data, key is empty, or wrong DUPFIXED size */
412 #define MDB_BAD_VALSIZE (-30781)
413 #define MDB_LAST_ERRCODE MDB_BAD_VALSIZE
416 /** @brief Statistics for a database in the environment */
417 typedef struct MDB_stat {
418 unsigned int ms_psize; /**< Size of a database page.
419 This is currently the same for all databases. */
420 unsigned int ms_depth; /**< Depth (height) of the B-tree */
421 size_t ms_branch_pages; /**< Number of internal (non-leaf) pages */
422 size_t ms_leaf_pages; /**< Number of leaf pages */
423 size_t ms_overflow_pages; /**< Number of overflow pages */
424 size_t ms_entries; /**< Number of data items */
427 /** @brief Information about the environment */
428 typedef struct MDB_envinfo {
429 void *me_mapaddr; /**< Address of map, if fixed */
430 size_t me_mapsize; /**< Size of the data memory map */
431 size_t me_last_pgno; /**< ID of the last used page */
432 size_t me_last_txnid; /**< ID of the last committed transaction */
433 unsigned int me_maxreaders; /**< max reader slots in the environment */
434 unsigned int me_numreaders; /**< max reader slots used in the environment */
437 /** @brief Return the mdb library version information.
439 * @param[out] major if non-NULL, the library major version number is copied here
440 * @param[out] minor if non-NULL, the library minor version number is copied here
441 * @param[out] patch if non-NULL, the library patch version number is copied here
442 * @retval "version string" The library version as a string
444 char *mdb_version(int *major, int *minor, int *patch);
446 /** @brief Return a string describing a given error code.
448 * This function is a superset of the ANSI C X3.159-1989 (ANSI C) strerror(3)
449 * function. If the error code is greater than or equal to 0, then the string
450 * returned by the system function strerror(3) is returned. If the error code
451 * is less than 0, an error string corresponding to the MDB library error is
452 * returned. See @ref errors for a list of MDB-specific error codes.
453 * @param[in] err The error code
454 * @retval "error message" The description of the error
456 char *mdb_strerror(int err);
458 /** @brief Create an MDB environment handle.
460 * This function allocates memory for a #MDB_env structure. To release
461 * the allocated memory and discard the handle, call #mdb_env_close().
462 * Before the handle may be used, it must be opened using #mdb_env_open().
463 * Various other options may also need to be set before opening the handle,
464 * e.g. #mdb_env_set_mapsize(), #mdb_env_set_maxreaders(), #mdb_env_set_maxdbs(),
465 * depending on usage requirements.
466 * @param[out] env The address where the new handle will be stored
467 * @return A non-zero error value on failure and 0 on success.
469 int mdb_env_create(MDB_env **env);
471 /** @brief Open an environment handle.
473 * If this function fails, #mdb_env_close() must be called to discard the #MDB_env handle.
474 * @param[in] env An environment handle returned by #mdb_env_create()
475 * @param[in] path The directory in which the database files reside. This
476 * directory must already exist and be writable.
477 * @param[in] flags Special options for this environment. This parameter
478 * must be set to 0 or by bitwise OR'ing together one or more of the
479 * values described here.
480 * Flags set by mdb_env_set_flags() are also used.
483 * use a fixed address for the mmap region. This flag must be specified
484 * when creating the environment, and is stored persistently in the environment.
485 * If successful, the memory map will always reside at the same virtual address
486 * and pointers used to reference data items in the database will be constant
487 * across multiple invocations. This option may not always work, depending on
488 * how the operating system has allocated memory to shared libraries and other uses.
489 * The feature is highly experimental.
491 * By default, MDB creates its environment in a directory whose
492 * pathname is given in \b path, and creates its data and lock files
493 * under that directory. With this option, \b path is used as-is for
494 * the database main data file. The database lock file is the \b path
495 * with "-lock" appended.
497 * Open the environment in read-only mode. No write operations will be
498 * allowed. MDB will still modify the lock file - except on read-only
499 * filesystems, where MDB does not use locks.
501 * Use a writeable memory map unless MDB_RDONLY is set. This is faster
502 * and uses fewer mallocs, but loses protection from application bugs
503 * like wild pointer writes and other bad updates into the database.
504 * Incompatible with nested transactions.
505 * Processes with and without MDB_WRITEMAP on the same environment do
506 * not cooperate well.
507 * <li>#MDB_NOMETASYNC
508 * Flush system buffers to disk only once per transaction, omit the
509 * metadata flush. Defer that until the system flushes files to disk,
510 * or next non-MDB_RDONLY commit or #mdb_env_sync(). This optimization
511 * maintains database integrity, but a system crash may undo the last
512 * committed transaction. I.e. it preserves the ACI (atomicity,
513 * consistency, isolation) but not D (durability) database property.
514 * This flag may be changed at any time using #mdb_env_set_flags().
516 * Don't flush system buffers to disk when committing a transaction.
517 * This optimization means a system crash can corrupt the database or
518 * lose the last transactions if buffers are not yet flushed to disk.
519 * The risk is governed by how often the system flushes dirty buffers
520 * to disk and how often #mdb_env_sync() is called. However, if the
521 * filesystem preserves write order and the #MDB_WRITEMAP flag is not
522 * used, transactions exhibit ACI (atomicity, consistency, isolation)
523 * properties and only lose D (durability). I.e. database integrity
524 * is maintained, but a system crash may undo the final transactions.
525 * Note that (#MDB_NOSYNC | #MDB_WRITEMAP) leaves the system with no
526 * hint for when to write transactions to disk, unless #mdb_env_sync()
527 * is called. (#MDB_MAPASYNC | #MDB_WRITEMAP) may be preferable.
528 * This flag may be changed at any time using #mdb_env_set_flags().
530 * When using #MDB_WRITEMAP, use asynchronous flushes to disk.
531 * As with #MDB_NOSYNC, a system crash can then corrupt the
532 * database or lose the last transactions. Calling #mdb_env_sync()
533 * ensures on-disk database integrity until next commit.
534 * This flag may be changed at any time using #mdb_env_set_flags().
536 * Don't use Thread-Local Storage. Tie reader locktable slots to
537 * #MDB_txn objects instead of to threads. I.e. #mdb_txn_reset() keeps
538 * the slot reseved for the #MDB_txn object. A thread may use parallel
539 * read-only transactions. A read-only transaction may span threads if
540 * the user synchronizes its use. Applications that multiplex many
541 * user threads over individual OS threads need this option. Such an
542 * application must also serialize the write transactions in an OS
543 * thread, since MDB's write locking is unaware of the user threads.
545 * Don't do any locking. If concurrent access is anticipated, the
546 * caller must manage all concurrency itself. For proper operation
547 * the caller must enforce single-writer semantics, and must ensure
548 * that no readers are using old transactions while a writer is
549 * active. The simplest approach is to use an exclusive lock so that
550 * no readers may be active at all when a writer begins.
552 * Turn off readahead. Most operating systems perform readahead on
553 * read requests by default. This option turns it off if the OS
554 * supports it. Turning it off may help random read performance
555 * when the DB is larger than RAM and system RAM is full.
556 * The option is not implemented on Windows.
558 * Don't initialize malloc'd memory before writing to unused spaces
559 * in the data file. By default, memory for pages written to the data
560 * file is obtained using malloc. While these pages may be reused in
561 * subsequent transactions, freshly malloc'd pages will be initialized
562 * to zeroes before use. This avoids persisting leftover data from other
563 * code (that used the heap and subsequently freed the memory) into the
564 * data file. Note that many other system libraries may allocate
565 * and free memory from the heap for arbitrary uses. E.g., stdio may
566 * use the heap for file I/O buffers. This initialization step has a
567 * modest performance cost so some applications may want to disable
568 * it using this flag. This option can be a problem for applications
569 * which handle sensitive data like passwords, and it makes memory
570 * checkers like Valgrind noisy. This flag is not needed with #MDB_WRITEMAP,
571 * which writes directly to the mmap instead of using malloc for pages. The
572 * initialization is also skipped if #MDB_RESERVE is used; the
573 * caller is expected to overwrite all of the memory that was
574 * reserved in that case.
575 * This flag may be changed at any time using #mdb_env_set_flags().
577 * @param[in] mode The UNIX permissions to set on created files. This parameter
578 * is ignored on Windows.
579 * @return A non-zero error value on failure and 0 on success. Some possible
582 * <li>#MDB_VERSION_MISMATCH - the version of the MDB library doesn't match the
583 * version that created the database environment.
584 * <li>#MDB_INVALID - the environment file headers are corrupted.
585 * <li>ENOENT - the directory specified by the path parameter doesn't exist.
586 * <li>EACCES - the user didn't have permission to access the environment files.
587 * <li>EAGAIN - the environment was locked by another process.
590 int mdb_env_open(MDB_env *env, const char *path, unsigned int flags, mdb_mode_t mode);
592 /** @brief Copy an MDB environment to the specified path.
594 * This function may be used to make a backup of an existing environment.
595 * No lockfile is created, since it gets recreated at need.
596 * @note This call can trigger significant file size growth if run in
597 * parallel with write transactions, because it employs a read-only
598 * transaction. See long-lived transactions under @ref caveats_sec.
599 * @param[in] env An environment handle returned by #mdb_env_create(). It
600 * must have already been opened successfully.
601 * @param[in] path The directory in which the copy will reside. This
602 * directory must already exist and be writable but must otherwise be
604 * @return A non-zero error value on failure and 0 on success.
606 int mdb_env_copy(MDB_env *env, const char *path);
608 /** @brief Copy an MDB environment to the specified file descriptor.
610 * This function may be used to make a backup of an existing environment.
611 * No lockfile is created, since it gets recreated at need.
612 * @note This call can trigger significant file size growth if run in
613 * parallel with write transactions, because it employs a read-only
614 * transaction. See long-lived transactions under @ref caveats_sec.
615 * @param[in] env An environment handle returned by #mdb_env_create(). It
616 * must have already been opened successfully.
617 * @param[in] fd The filedescriptor to write the copy to. It must
618 * have already been opened for Write access.
619 * @return A non-zero error value on failure and 0 on success.
621 int mdb_env_copyfd(MDB_env *env, mdb_filehandle_t fd);
623 /** @brief Return statistics about the MDB environment.
625 * @param[in] env An environment handle returned by #mdb_env_create()
626 * @param[out] stat The address of an #MDB_stat structure
627 * where the statistics will be copied
629 int mdb_env_stat(MDB_env *env, MDB_stat *stat);
631 /** @brief Return information about the MDB environment.
633 * @param[in] env An environment handle returned by #mdb_env_create()
634 * @param[out] stat The address of an #MDB_envinfo structure
635 * where the information will be copied
637 int mdb_env_info(MDB_env *env, MDB_envinfo *stat);
639 /** @brief Flush the data buffers to disk.
641 * Data is always written to disk when #mdb_txn_commit() is called,
642 * but the operating system may keep it buffered. MDB always flushes
643 * the OS buffers upon commit as well, unless the environment was
644 * opened with #MDB_NOSYNC or in part #MDB_NOMETASYNC.
645 * @param[in] env An environment handle returned by #mdb_env_create()
646 * @param[in] force If non-zero, force a synchronous flush. Otherwise
647 * if the environment has the #MDB_NOSYNC flag set the flushes
648 * will be omitted, and with #MDB_MAPASYNC they will be asynchronous.
649 * @return A non-zero error value on failure and 0 on success. Some possible
652 * <li>EINVAL - an invalid parameter was specified.
653 * <li>EIO - an error occurred during synchronization.
656 int mdb_env_sync(MDB_env *env, int force);
658 /** @brief Close the environment and release the memory map.
660 * Only a single thread may call this function. All transactions, databases,
661 * and cursors must already be closed before calling this function. Attempts to
662 * use any such handles after calling this function will cause a SIGSEGV.
663 * The environment handle will be freed and must not be used again after this call.
664 * @param[in] env An environment handle returned by #mdb_env_create()
666 void mdb_env_close(MDB_env *env);
668 /** @brief Set environment flags.
670 * This may be used to set some flags in addition to those from
671 * #mdb_env_open(), or to unset these flags.
672 * @param[in] env An environment handle returned by #mdb_env_create()
673 * @param[in] flags The flags to change, bitwise OR'ed together
674 * @param[in] onoff A non-zero value sets the flags, zero clears them.
675 * @return A non-zero error value on failure and 0 on success. Some possible
678 * <li>EINVAL - an invalid parameter was specified.
681 int mdb_env_set_flags(MDB_env *env, unsigned int flags, int onoff);
683 /** @brief Get environment flags.
685 * @param[in] env An environment handle returned by #mdb_env_create()
686 * @param[out] flags The address of an integer to store the flags
687 * @return A non-zero error value on failure and 0 on success. Some possible
690 * <li>EINVAL - an invalid parameter was specified.
693 int mdb_env_get_flags(MDB_env *env, unsigned int *flags);
695 /** @brief Return the path that was used in #mdb_env_open().
697 * @param[in] env An environment handle returned by #mdb_env_create()
698 * @param[out] path Address of a string pointer to contain the path. This
699 * is the actual string in the environment, not a copy. It should not be
700 * altered in any way.
701 * @return A non-zero error value on failure and 0 on success. Some possible
704 * <li>EINVAL - an invalid parameter was specified.
707 int mdb_env_get_path(MDB_env *env, const char **path);
709 /** @brief Return the filedescriptor for the given environment.
711 * @param[in] env An environment handle returned by #mdb_env_create()
712 * @param[out] fd Address of a mdb_filehandle_t to contain the descriptor.
713 * @return A non-zero error value on failure and 0 on success. Some possible
716 * <li>EINVAL - an invalid parameter was specified.
719 int mdb_env_get_fd(MDB_env *env, mdb_filehandle_t *fd);
721 /** @brief Set the size of the memory map to use for this environment.
723 * The size should be a multiple of the OS page size. The default is
724 * 10485760 bytes. The size of the memory map is also the maximum size
725 * of the database. The value should be chosen as large as possible,
726 * to accommodate future growth of the database.
727 * This function should be called after #mdb_env_create() and before #mdb_env_open().
728 * It may be called at later times if no transactions are active in
729 * this process. Note that the library does not check for this condition,
730 * the caller must ensure it explicitly.
732 * If the mapsize is changed by another process, #mdb_txn_begin() will
733 * return #MDB_MAP_RESIZED. This function may be called with a size
734 * of zero to adopt the new size.
736 * Any attempt to set a size smaller than the space already consumed
737 * by the environment will be silently changed to the current size of the used space.
738 * @param[in] env An environment handle returned by #mdb_env_create()
739 * @param[in] size The size in bytes
740 * @return A non-zero error value on failure and 0 on success. Some possible
743 * <li>EINVAL - an invalid parameter was specified, or the environment has
744 * an active write transaction.
747 int mdb_env_set_mapsize(MDB_env *env, size_t size);
749 /** @brief Set the maximum number of threads/reader slots for the environment.
751 * This defines the number of slots in the lock table that is used to track readers in the
752 * the environment. The default is 126.
753 * Starting a read-only transaction normally ties a lock table slot to the
754 * current thread until the environment closes or the thread exits. If
755 * MDB_NOTLS is in use, #mdb_txn_begin() instead ties the slot to the
756 * MDB_txn object until it or the #MDB_env object is destroyed.
757 * This function may only be called after #mdb_env_create() and before #mdb_env_open().
758 * @param[in] env An environment handle returned by #mdb_env_create()
759 * @param[in] readers The maximum number of reader lock table slots
760 * @return A non-zero error value on failure and 0 on success. Some possible
763 * <li>EINVAL - an invalid parameter was specified, or the environment is already open.
766 int mdb_env_set_maxreaders(MDB_env *env, unsigned int readers);
768 /** @brief Get the maximum number of threads/reader slots for the environment.
770 * @param[in] env An environment handle returned by #mdb_env_create()
771 * @param[out] readers Address of an integer to store the number of readers
772 * @return A non-zero error value on failure and 0 on success. Some possible
775 * <li>EINVAL - an invalid parameter was specified.
778 int mdb_env_get_maxreaders(MDB_env *env, unsigned int *readers);
780 /** @brief Set the maximum number of named databases for the environment.
782 * This function is only needed if multiple databases will be used in the
783 * environment. Simpler applications that use the environment as a single
784 * unnamed database can ignore this option.
785 * This function may only be called after #mdb_env_create() and before #mdb_env_open().
786 * @param[in] env An environment handle returned by #mdb_env_create()
787 * @param[in] dbs The maximum number of databases
788 * @return A non-zero error value on failure and 0 on success. Some possible
791 * <li>EINVAL - an invalid parameter was specified, or the environment is already open.
794 int mdb_env_set_maxdbs(MDB_env *env, MDB_dbi dbs);
796 /** @brief Get the maximum size of keys and #MDB_DUPSORT data we can write.
798 * Depends on the compile-time constant #MDB_MAXKEYSIZE. Default 511.
800 * @param[in] env An environment handle returned by #mdb_env_create()
801 * @return The maximum size of a key we can write
803 int mdb_env_get_maxkeysize(MDB_env *env);
805 /** @brief Set application information associated with the #MDB_env.
807 * @param[in] env An environment handle returned by #mdb_env_create()
808 * @param[in] ctx An arbitrary pointer for whatever the application needs.
809 * @return A non-zero error value on failure and 0 on success.
811 int mdb_env_set_userctx(MDB_env *env, void *ctx);
813 /** @brief Get the application information associated with the #MDB_env.
815 * @param[in] env An environment handle returned by #mdb_env_create()
816 * @return The pointer set by #mdb_env_set_userctx().
818 void *mdb_env_get_userctx(MDB_env *env);
820 /** @brief Create a transaction for use with the environment.
822 * The transaction handle may be discarded using #mdb_txn_abort() or #mdb_txn_commit().
823 * @note A transaction and its cursors must only be used by a single
824 * thread, and a thread may only have a single transaction at a time.
825 * If #MDB_NOTLS is in use, this does not apply to read-only transactions.
826 * @note Cursors may not span transactions.
827 * @param[in] env An environment handle returned by #mdb_env_create()
828 * @param[in] parent If this parameter is non-NULL, the new transaction
829 * will be a nested transaction, with the transaction indicated by \b parent
830 * as its parent. Transactions may be nested to any level. A parent
831 * transaction and its cursors may not issue any other operations than
832 * mdb_txn_commit and mdb_txn_abort while it has active child transactions.
833 * @param[in] flags Special options for this transaction. This parameter
834 * must be set to 0 or by bitwise OR'ing together one or more of the
835 * values described here.
838 * This transaction will not perform any write operations.
840 * @param[out] txn Address where the new #MDB_txn handle will be stored
841 * @return A non-zero error value on failure and 0 on success. Some possible
844 * <li>#MDB_PANIC - a fatal error occurred earlier and the environment
846 * <li>#MDB_MAP_RESIZED - another process wrote data beyond this MDB_env's
847 * mapsize and this environment's map must be resized as well.
848 * See #mdb_env_set_mapsize().
849 * <li>#MDB_READERS_FULL - a read-only transaction was requested and
850 * the reader lock table is full. See #mdb_env_set_maxreaders().
851 * <li>ENOMEM - out of memory.
854 int mdb_txn_begin(MDB_env *env, MDB_txn *parent, unsigned int flags, MDB_txn **txn);
856 /** @brief Returns the transaction's #MDB_env
858 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
860 MDB_env *mdb_txn_env(MDB_txn *txn);
862 /** @brief Commit all the operations of a transaction into the database.
864 * The transaction handle is freed. It and its cursors must not be used
865 * again after this call, except with #mdb_cursor_renew().
866 * @note Earlier documentation incorrectly said all cursors would be freed.
867 * Only write-transactions free cursors.
868 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
869 * @return A non-zero error value on failure and 0 on success. Some possible
872 * <li>EINVAL - an invalid parameter was specified.
873 * <li>ENOSPC - no more disk space.
874 * <li>EIO - a low-level I/O error occurred while writing.
875 * <li>ENOMEM - out of memory.
878 int mdb_txn_commit(MDB_txn *txn);
880 /** @brief Abandon all the operations of the transaction instead of saving them.
882 * The transaction handle is freed. It and its cursors must not be used
883 * again after this call, except with #mdb_cursor_renew().
884 * @note Earlier documentation incorrectly said all cursors would be freed.
885 * Only write-transactions free cursors.
886 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
888 void mdb_txn_abort(MDB_txn *txn);
890 /** @brief Reset a read-only transaction.
892 * Abort the transaction like #mdb_txn_abort(), but keep the transaction
893 * handle. #mdb_txn_renew() may reuse the handle. This saves allocation
894 * overhead if the process will start a new read-only transaction soon,
895 * and also locking overhead if #MDB_NOTLS is in use. The reader table
896 * lock is released, but the table slot stays tied to its thread or
897 * #MDB_txn. Use mdb_txn_abort() to discard a reset handle, and to free
898 * its lock table slot if MDB_NOTLS is in use.
899 * Cursors opened within the transaction must not be used
900 * again after this call, except with #mdb_cursor_renew().
901 * Reader locks generally don't interfere with writers, but they keep old
902 * versions of database pages allocated. Thus they prevent the old pages
903 * from being reused when writers commit new data, and so under heavy load
904 * the database size may grow much more rapidly than otherwise.
905 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
907 void mdb_txn_reset(MDB_txn *txn);
909 /** @brief Renew a read-only transaction.
911 * This acquires a new reader lock for a transaction handle that had been
912 * released by #mdb_txn_reset(). It must be called before a reset transaction
914 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
915 * @return A non-zero error value on failure and 0 on success. Some possible
918 * <li>#MDB_PANIC - a fatal error occurred earlier and the environment
920 * <li>EINVAL - an invalid parameter was specified.
923 int mdb_txn_renew(MDB_txn *txn);
925 /** Compat with version <= 0.9.4, avoid clash with libmdb from MDB Tools project */
926 #define mdb_open(txn,name,flags,dbi) mdb_dbi_open(txn,name,flags,dbi)
927 /** Compat with version <= 0.9.4, avoid clash with libmdb from MDB Tools project */
928 #define mdb_close(env,dbi) mdb_dbi_close(env,dbi)
930 /** @brief Open a database in the environment.
932 * A database handle denotes the name and parameters of a database,
933 * independently of whether such a database exists.
934 * The database handle may be discarded by calling #mdb_dbi_close().
935 * The old database handle is returned if the database was already open.
936 * The handle must only be closed once.
937 * The database handle will be private to the current transaction until
938 * the transaction is successfully committed. If the transaction is
939 * aborted the handle will be closed automatically.
940 * After a successful commit the
941 * handle will reside in the shared environment, and may be used
942 * by other transactions. This function must not be called from
943 * multiple concurrent transactions. A transaction that uses this function
944 * must finish (either commit or abort) before any other transaction may
947 * To use named databases (with name != NULL), #mdb_env_set_maxdbs()
948 * must be called before opening the environment.
949 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
950 * @param[in] name The name of the database to open. If only a single
951 * database is needed in the environment, this value may be NULL.
952 * @param[in] flags Special options for this database. This parameter
953 * must be set to 0 or by bitwise OR'ing together one or more of the
954 * values described here.
956 * <li>#MDB_REVERSEKEY
957 * Keys are strings to be compared in reverse order, from the end
958 * of the strings to the beginning. By default, Keys are treated as strings and
959 * compared from beginning to end.
961 * Duplicate keys may be used in the database. (Or, from another perspective,
962 * keys may have multiple data items, stored in sorted order.) By default
963 * keys must be unique and may have only a single data item.
964 * <li>#MDB_INTEGERKEY
965 * Keys are binary integers in native byte order. Setting this option
966 * requires all keys to be the same size, typically sizeof(int)
969 * This flag may only be used in combination with #MDB_DUPSORT. This option
970 * tells the library that the data items for this database are all the same
971 * size, which allows further optimizations in storage and retrieval. When
972 * all data items are the same size, the #MDB_GET_MULTIPLE and #MDB_NEXT_MULTIPLE
973 * cursor operations may be used to retrieve multiple items at once.
974 * <li>#MDB_INTEGERDUP
975 * This option specifies that duplicate data items are also integers, and
976 * should be sorted as such.
977 * <li>#MDB_REVERSEDUP
978 * This option specifies that duplicate data items should be compared as
979 * strings in reverse order.
981 * Create the named database if it doesn't exist. This option is not
982 * allowed in a read-only transaction or a read-only environment.
984 * @param[out] dbi Address where the new #MDB_dbi handle will be stored
985 * @return A non-zero error value on failure and 0 on success. Some possible
988 * <li>#MDB_NOTFOUND - the specified database doesn't exist in the environment
989 * and #MDB_CREATE was not specified.
990 * <li>#MDB_DBS_FULL - too many databases have been opened. See #mdb_env_set_maxdbs().
993 int mdb_dbi_open(MDB_txn *txn, const char *name, unsigned int flags, MDB_dbi *dbi);
995 /** @brief Retrieve statistics for a database.
997 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
998 * @param[in] dbi A database handle returned by #mdb_dbi_open()
999 * @param[out] stat The address of an #MDB_stat structure
1000 * where the statistics will be copied
1001 * @return A non-zero error value on failure and 0 on success. Some possible
1004 * <li>EINVAL - an invalid parameter was specified.
1007 int mdb_stat(MDB_txn *txn, MDB_dbi dbi, MDB_stat *stat);
1009 /** @brief Retrieve the DB flags for a database handle.
1011 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1012 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1013 * @param[out] flags Address where the flags will be returned.
1014 * @return A non-zero error value on failure and 0 on success.
1016 int mdb_dbi_flags(MDB_txn *txn, MDB_dbi dbi, unsigned int *flags);
1018 /** @brief Close a database handle.
1020 * This call is not mutex protected. Handles should only be closed by
1021 * a single thread, and only if no other threads are going to reference
1022 * the database handle or one of its cursors any further. Do not close
1023 * a handle if an existing transaction has modified its database.
1024 * @param[in] env An environment handle returned by #mdb_env_create()
1025 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1027 void mdb_dbi_close(MDB_env *env, MDB_dbi dbi);
1029 /** @brief Empty or delete+close a database.
1031 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1032 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1033 * @param[in] del 0 to empty the DB, 1 to delete it from the
1034 * environment and close the DB handle.
1035 * @return A non-zero error value on failure and 0 on success.
1037 int mdb_drop(MDB_txn *txn, MDB_dbi dbi, int del);
1039 /** @brief Set a custom key comparison function for a database.
1041 * The comparison function is called whenever it is necessary to compare a
1042 * key specified by the application with a key currently stored in the database.
1043 * If no comparison function is specified, and no special key flags were specified
1044 * with #mdb_dbi_open(), the keys are compared lexically, with shorter keys collating
1045 * before longer keys.
1046 * @warning This function must be called before any data access functions are used,
1047 * otherwise data corruption may occur. The same comparison function must be used by every
1048 * program accessing the database, every time the database is used.
1049 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1050 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1051 * @param[in] cmp A #MDB_cmp_func function
1052 * @return A non-zero error value on failure and 0 on success. Some possible
1055 * <li>EINVAL - an invalid parameter was specified.
1058 int mdb_set_compare(MDB_txn *txn, MDB_dbi dbi, MDB_cmp_func *cmp);
1060 /** @brief Set a custom data comparison function for a #MDB_DUPSORT database.
1062 * This comparison function is called whenever it is necessary to compare a data
1063 * item specified by the application with a data item currently stored in the database.
1064 * This function only takes effect if the database was opened with the #MDB_DUPSORT
1066 * If no comparison function is specified, and no special key flags were specified
1067 * with #mdb_dbi_open(), the data items are compared lexically, with shorter items collating
1068 * before longer items.
1069 * @warning This function must be called before any data access functions are used,
1070 * otherwise data corruption may occur. The same comparison function must be used by every
1071 * program accessing the database, every time the database is used.
1072 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1073 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1074 * @param[in] cmp A #MDB_cmp_func function
1075 * @return A non-zero error value on failure and 0 on success. Some possible
1078 * <li>EINVAL - an invalid parameter was specified.
1081 int mdb_set_dupsort(MDB_txn *txn, MDB_dbi dbi, MDB_cmp_func *cmp);
1083 /** @brief Set a relocation function for a #MDB_FIXEDMAP database.
1085 * @todo The relocation function is called whenever it is necessary to move the data
1086 * of an item to a different position in the database (e.g. through tree
1087 * balancing operations, shifts as a result of adds or deletes, etc.). It is
1088 * intended to allow address/position-dependent data items to be stored in
1089 * a database in an environment opened with the #MDB_FIXEDMAP option.
1090 * Currently the relocation feature is unimplemented and setting
1091 * this function has no effect.
1092 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1093 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1094 * @param[in] rel A #MDB_rel_func function
1095 * @return A non-zero error value on failure and 0 on success. Some possible
1098 * <li>EINVAL - an invalid parameter was specified.
1101 int mdb_set_relfunc(MDB_txn *txn, MDB_dbi dbi, MDB_rel_func *rel);
1103 /** @brief Set a context pointer for a #MDB_FIXEDMAP database's relocation function.
1105 * See #mdb_set_relfunc and #MDB_rel_func for more details.
1106 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1107 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1108 * @param[in] ctx An arbitrary pointer for whatever the application needs.
1109 * It will be passed to the callback function set by #mdb_set_relfunc
1110 * as its \b relctx parameter whenever the callback is invoked.
1111 * @return A non-zero error value on failure and 0 on success. Some possible
1114 * <li>EINVAL - an invalid parameter was specified.
1117 int mdb_set_relctx(MDB_txn *txn, MDB_dbi dbi, void *ctx);
1119 /** @brief Get items from a database.
1121 * This function retrieves key/data pairs from the database. The address
1122 * and length of the data associated with the specified \b key are returned
1123 * in the structure to which \b data refers.
1124 * If the database supports duplicate keys (#MDB_DUPSORT) then the
1125 * first data item for the key will be returned. Retrieval of other
1126 * items requires the use of #mdb_cursor_get().
1128 * @note The memory pointed to by the returned values is owned by the
1129 * database. The caller need not dispose of the memory, and may not
1130 * modify it in any way. For values returned in a read-only transaction
1131 * any modification attempts will cause a SIGSEGV.
1132 * @note Values returned from the database are valid only until a
1133 * subsequent update operation, or the end of the transaction.
1134 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1135 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1136 * @param[in] key The key to search for in the database
1137 * @param[out] data The data corresponding to the key
1138 * @return A non-zero error value on failure and 0 on success. Some possible
1141 * <li>#MDB_NOTFOUND - the key was not in the database.
1142 * <li>EINVAL - an invalid parameter was specified.
1145 int mdb_get(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data);
1147 /** @brief Store items into a database.
1149 * This function stores key/data pairs in the database. The default behavior
1150 * is to enter the new key/data pair, replacing any previously existing key
1151 * if duplicates are disallowed, or adding a duplicate data item if
1152 * duplicates are allowed (#MDB_DUPSORT).
1153 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1154 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1155 * @param[in] key The key to store in the database
1156 * @param[in,out] data The data to store
1157 * @param[in] flags Special options for this operation. This parameter
1158 * must be set to 0 or by bitwise OR'ing together one or more of the
1159 * values described here.
1161 * <li>#MDB_NODUPDATA - enter the new key/data pair only if it does not
1162 * already appear in the database. This flag may only be specified
1163 * if the database was opened with #MDB_DUPSORT. The function will
1164 * return #MDB_KEYEXIST if the key/data pair already appears in the
1166 * <li>#MDB_NOOVERWRITE - enter the new key/data pair only if the key
1167 * does not already appear in the database. The function will return
1168 * #MDB_KEYEXIST if the key already appears in the database, even if
1169 * the database supports duplicates (#MDB_DUPSORT). The \b data
1170 * parameter will be set to point to the existing item.
1171 * <li>#MDB_RESERVE - reserve space for data of the given size, but
1172 * don't copy the given data. Instead, return a pointer to the
1173 * reserved space, which the caller can fill in later - before
1174 * the next update operation or the transaction ends. This saves
1175 * an extra memcpy if the data is being generated later.
1176 * MDB does nothing else with this memory, the caller is expected
1177 * to modify all of the space requested.
1178 * <li>#MDB_APPEND - append the given key/data pair to the end of the
1179 * database. No key comparisons are performed. This option allows
1180 * fast bulk loading when keys are already known to be in the
1181 * correct order. Loading unsorted keys with this flag will cause
1183 * <li>#MDB_APPENDDUP - as above, but for sorted dup data.
1185 * @return A non-zero error value on failure and 0 on success. Some possible
1188 * <li>#MDB_MAP_FULL - the database is full, see #mdb_env_set_mapsize().
1189 * <li>#MDB_TXN_FULL - the transaction has too many dirty pages.
1190 * <li>EACCES - an attempt was made to write in a read-only transaction.
1191 * <li>EINVAL - an invalid parameter was specified.
1194 int mdb_put(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data,
1195 unsigned int flags);
1197 /** @brief Delete items from a database.
1199 * This function removes key/data pairs from the database.
1200 * If the database does not support sorted duplicate data items
1201 * (#MDB_DUPSORT) the data parameter is ignored.
1202 * If the database supports sorted duplicates and the data parameter
1203 * is NULL, all of the duplicate data items for the key will be
1204 * deleted. Otherwise, if the data parameter is non-NULL
1205 * only the matching data item will be deleted.
1206 * This function will return #MDB_NOTFOUND if the specified key/data
1207 * pair is not in the database.
1208 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1209 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1210 * @param[in] key The key to delete from the database
1211 * @param[in] data The data to delete
1212 * @return A non-zero error value on failure and 0 on success. Some possible
1215 * <li>EACCES - an attempt was made to write in a read-only transaction.
1216 * <li>EINVAL - an invalid parameter was specified.
1219 int mdb_del(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data);
1221 /** @brief Create a cursor handle.
1223 * A cursor is associated with a specific transaction and database.
1224 * A cursor cannot be used when its database handle is closed. Nor
1225 * when its transaction has ended, except with #mdb_cursor_renew().
1226 * It can be discarded with #mdb_cursor_close().
1227 * A cursor in a write-transaction can be closed before its transaction
1228 * ends, and will otherwise be closed when its transaction ends.
1229 * A cursor in a read-only transaction must be closed explicitly, before
1230 * or after its transaction ends. It can be reused with
1231 * #mdb_cursor_renew() before finally closing it.
1232 * @note Earlier documentation said that cursors in every transaction
1233 * were closed when the transaction committed or aborted.
1234 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1235 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1236 * @param[out] cursor Address where the new #MDB_cursor handle will be stored
1237 * @return A non-zero error value on failure and 0 on success. Some possible
1240 * <li>EINVAL - an invalid parameter was specified.
1243 int mdb_cursor_open(MDB_txn *txn, MDB_dbi dbi, MDB_cursor **cursor);
1245 /** @brief Close a cursor handle.
1247 * The cursor handle will be freed and must not be used again after this call.
1248 * Its transaction must still be live if it is a write-transaction.
1249 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1251 void mdb_cursor_close(MDB_cursor *cursor);
1253 /** @brief Renew a cursor handle.
1255 * A cursor is associated with a specific transaction and database.
1256 * Cursors that are only used in read-only
1257 * transactions may be re-used, to avoid unnecessary malloc/free overhead.
1258 * The cursor may be associated with a new read-only transaction, and
1259 * referencing the same database handle as it was created with.
1260 * This may be done whether the previous transaction is live or dead.
1261 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1262 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1263 * @return A non-zero error value on failure and 0 on success. Some possible
1266 * <li>EINVAL - an invalid parameter was specified.
1269 int mdb_cursor_renew(MDB_txn *txn, MDB_cursor *cursor);
1271 /** @brief Return the cursor's transaction handle.
1273 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1275 MDB_txn *mdb_cursor_txn(MDB_cursor *cursor);
1277 /** @brief Return the cursor's database handle.
1279 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1281 MDB_dbi mdb_cursor_dbi(MDB_cursor *cursor);
1283 /** @brief Retrieve by cursor.
1285 * This function retrieves key/data pairs from the database. The address and length
1286 * of the key are returned in the object to which \b key refers (except for the
1287 * case of the #MDB_SET option, in which the \b key object is unchanged), and
1288 * the address and length of the data are returned in the object to which \b data
1290 * See #mdb_get() for restrictions on using the output values.
1291 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1292 * @param[in,out] key The key for a retrieved item
1293 * @param[in,out] data The data of a retrieved item
1294 * @param[in] op A cursor operation #MDB_cursor_op
1295 * @return A non-zero error value on failure and 0 on success. Some possible
1298 * <li>#MDB_NOTFOUND - no matching key found.
1299 * <li>EINVAL - an invalid parameter was specified.
1302 int mdb_cursor_get(MDB_cursor *cursor, MDB_val *key, MDB_val *data,
1305 /** @brief Store by cursor.
1307 * This function stores key/data pairs into the database.
1308 * If the function fails for any reason, the state of the cursor will be
1309 * unchanged. If the function succeeds and an item is inserted into the
1310 * database, the cursor is always positioned to refer to the newly inserted item.
1311 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1312 * @param[in] key The key operated on.
1313 * @param[in] data The data operated on.
1314 * @param[in] flags Options for this operation. This parameter
1315 * must be set to 0 or one of the values described here.
1317 * <li>#MDB_CURRENT - overwrite the data of the key/data pair to which
1318 * the cursor refers with the specified data item. The \b key
1319 * parameter is ignored.
1320 * <li>#MDB_NODUPDATA - enter the new key/data pair only if it does not
1321 * already appear in the database. This flag may only be specified
1322 * if the database was opened with #MDB_DUPSORT. The function will
1323 * return #MDB_KEYEXIST if the key/data pair already appears in the
1325 * <li>#MDB_NOOVERWRITE - enter the new key/data pair only if the key
1326 * does not already appear in the database. The function will return
1327 * #MDB_KEYEXIST if the key already appears in the database, even if
1328 * the database supports duplicates (#MDB_DUPSORT).
1329 * <li>#MDB_RESERVE - reserve space for data of the given size, but
1330 * don't copy the given data. Instead, return a pointer to the
1331 * reserved space, which the caller can fill in later. This saves
1332 * an extra memcpy if the data is being generated later.
1333 * <li>#MDB_APPEND - append the given key/data pair to the end of the
1334 * database. No key comparisons are performed. This option allows
1335 * fast bulk loading when keys are already known to be in the
1336 * correct order. Loading unsorted keys with this flag will cause
1338 * <li>#MDB_APPENDDUP - as above, but for sorted dup data.
1339 * <li>#MDB_MULTIPLE - store multiple contiguous data elements in a
1340 * single request. This flag may only be specified if the database
1341 * was opened with #MDB_DUPFIXED. The \b data argument must be an
1342 * array of two MDB_vals. The mv_size of the first MDB_val must be
1343 * the size of a single data element. The mv_data of the first MDB_val
1344 * must point to the beginning of the array of contiguous data elements.
1345 * The mv_size of the second MDB_val must be the count of the number
1346 * of data elements to store. On return this field will be set to
1347 * the count of the number of elements actually written. The mv_data
1348 * of the second MDB_val is unused.
1350 * @return A non-zero error value on failure and 0 on success. Some possible
1353 * <li>#MDB_MAP_FULL - the database is full, see #mdb_env_set_mapsize().
1354 * <li>#MDB_TXN_FULL - the transaction has too many dirty pages.
1355 * <li>EACCES - an attempt was made to modify a read-only database.
1356 * <li>EINVAL - an invalid parameter was specified.
1359 int mdb_cursor_put(MDB_cursor *cursor, MDB_val *key, MDB_val *data,
1360 unsigned int flags);
1362 /** @brief Delete current key/data pair
1364 * This function deletes the key/data pair to which the cursor refers.
1365 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1366 * @param[in] flags Options for this operation. This parameter
1367 * must be set to 0 or one of the values described here.
1369 * <li>#MDB_NODUPDATA - delete all of the data items for the current key.
1370 * This flag may only be specified if the database was opened with #MDB_DUPSORT.
1372 * @return A non-zero error value on failure and 0 on success. Some possible
1375 * <li>EACCES - an attempt was made to modify a read-only database.
1376 * <li>EINVAL - an invalid parameter was specified.
1379 int mdb_cursor_del(MDB_cursor *cursor, unsigned int flags);
1381 /** @brief Return count of duplicates for current key.
1383 * This call is only valid on databases that support sorted duplicate
1384 * data items #MDB_DUPSORT.
1385 * @param[in] cursor A cursor handle returned by #mdb_cursor_open()
1386 * @param[out] countp Address where the count will be stored
1387 * @return A non-zero error value on failure and 0 on success. Some possible
1390 * <li>EINVAL - cursor is not initialized, or an invalid parameter was specified.
1393 int mdb_cursor_count(MDB_cursor *cursor, size_t *countp);
1395 /** @brief Compare two data items according to a particular database.
1397 * This returns a comparison as if the two data items were keys in the
1398 * specified database.
1399 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1400 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1401 * @param[in] a The first item to compare
1402 * @param[in] b The second item to compare
1403 * @return < 0 if a < b, 0 if a == b, > 0 if a > b
1405 int mdb_cmp(MDB_txn *txn, MDB_dbi dbi, const MDB_val *a, const MDB_val *b);
1407 /** @brief Compare two data items according to a particular database.
1409 * This returns a comparison as if the two items were data items of
1410 * the specified database. The database must have the #MDB_DUPSORT flag.
1411 * @param[in] txn A transaction handle returned by #mdb_txn_begin()
1412 * @param[in] dbi A database handle returned by #mdb_dbi_open()
1413 * @param[in] a The first item to compare
1414 * @param[in] b The second item to compare
1415 * @return < 0 if a < b, 0 if a == b, > 0 if a > b
1417 int mdb_dcmp(MDB_txn *txn, MDB_dbi dbi, const MDB_val *a, const MDB_val *b);
1419 /** @brief A callback function used to print a message from the library.
1421 * @param[in] msg The string to be printed.
1422 * @param[in] ctx An arbitrary context pointer for the callback.
1423 * @return < 0 on failure, >= 0 on success.
1425 typedef int (MDB_msg_func)(const char *msg, void *ctx);
1427 /** @brief Dump the entries in the reader lock table.
1429 * @param[in] env An environment handle returned by #mdb_env_create()
1430 * @param[in] func A #MDB_msg_func function
1431 * @param[in] ctx Anything the message function needs
1432 * @return < 0 on failure, >= 0 on success.
1434 int mdb_reader_list(MDB_env *env, MDB_msg_func *func, void *ctx);
1436 /** @brief Check for stale entries in the reader lock table.
1438 * @param[in] env An environment handle returned by #mdb_env_create()
1439 * @param[out] dead Number of stale slots that were cleared
1440 * @return 0 on success, non-zero on failure.
1442 int mdb_reader_check(MDB_env *env, int *dead);
1448 #endif /* _LMDB_H_ */