2 * @brief memory-mapped database library
4 * A Btree-based database management library modeled loosely on the
5 * BerkeleyDB API, but much simplified.
8 * Copyright 2011 Howard Chu, Symas Corp.
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted only as authorized by the OpenLDAP
15 * A copy of this license is available in the file LICENSE in the
16 * top-level directory of the distribution or, alternatively, at
17 * <http://www.OpenLDAP.org/license.html>.
19 * This code is derived from btree.c written by Martin Hedenfalk.
21 * Copyright (c) 2009, 2010 Martin Hedenfalk <martin@bzero.se>
23 * Permission to use, copy, modify, and distribute this software for any
24 * purpose with or without fee is hereby granted, provided that the above
25 * copyright notice and this permission notice appear in all copies.
27 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
28 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
29 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
30 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
31 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
32 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
33 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
35 #include <sys/types.h>
37 #include <sys/param.h>
43 #ifdef HAVE_SYS_FILE_H
66 /** @defgroup internal MDB Internals
69 /** @defgroup compat Windows Compatibility Macros
70 * A bunch of macros to minimize the amount of platform-specific ifdefs
71 * needed throughout the rest of the code. When the features this library
72 * needs are similar enough to POSIX to be hidden in a one-or-two line
73 * replacement, this macro approach is used.
77 #define pthread_t DWORD
78 #define pthread_mutex_t HANDLE
79 #define pthread_key_t DWORD
80 #define pthread_self() GetCurrentThreadId()
81 #define pthread_key_create(x,y) *(x) = TlsAlloc()
82 #define pthread_key_delete(x) TlsFree(x)
83 #define pthread_getspecific(x) TlsGetValue(x)
84 #define pthread_setspecific(x,y) TlsSetValue(x,y)
85 #define pthread_mutex_unlock(x) ReleaseMutex(x)
86 #define pthread_mutex_lock(x) WaitForSingleObject(x, INFINITE)
87 #define LOCK_MUTEX_R(env) pthread_mutex_lock(env->me_rmutex)
88 #define UNLOCK_MUTEX_R(env) pthread_mutex_unlock(env->me_rmutex)
89 #define LOCK_MUTEX_W(env) pthread_mutex_lock(env->me_wmutex)
90 #define UNLOCK_MUTEX_W(env) pthread_mutex_unlock(env->me_wmutex)
91 #define getpid() GetCurrentProcessId()
92 #define fdatasync(fd) !FlushFileBuffers(fd)
93 #define ErrCode() GetLastError()
94 #define GetPageSize(x) {SYSTEM_INFO si; GetSystemInfo(&si); (x) = si.dwPageSize;}
95 #define close(fd) CloseHandle(fd)
96 #define munmap(ptr,len) UnmapViewOfFile(ptr)
98 /** Lock the reader mutex.
100 #define LOCK_MUTEX_R(env) pthread_mutex_lock(&env->me_txns->mti_mutex)
101 /** Unlock the reader mutex.
103 #define UNLOCK_MUTEX_R(env) pthread_mutex_unlock(&env->me_txns->mti_mutex)
105 /** Lock the writer mutex.
106 * Only a single write transaction is allowed at a time. Other writers
107 * will block waiting for this mutex.
109 #define LOCK_MUTEX_W(env) pthread_mutex_lock(&env->me_txns->mti_wmutex)
110 /** Unlock the writer mutex.
112 #define UNLOCK_MUTEX_W(env) pthread_mutex_unlock(&env->me_txns->mti_wmutex)
114 /** Get the error code for the last failed system function.
116 #define ErrCode() errno
118 /** An abstraction for a file handle.
119 * On POSIX systems file handles are small integers. On Windows
120 * they're opaque pointers.
124 /** A value for an invalid file handle.
125 * Mainly used to initialize file variables and signify that they are
128 #define INVALID_HANDLE_VALUE -1
130 /** Get the size of a memory page for the system.
131 * This is the basic size that the platform's memory manager uses, and is
132 * fundamental to the use of memory-mapped files.
134 #define GetPageSize(x) (x) = sysconf(_SC_PAGE_SIZE)
140 /** A flag for opening a file and requesting synchronous data writes.
141 * This is only used when writing a meta page. It's not strictly needed;
142 * we could just do a normal write and then immediately perform a flush.
143 * But if this flag is available it saves us an extra system call.
145 * @note If O_DSYNC is undefined but exists in /usr/include,
146 * preferably set some compiler flag to get the definition.
147 * Otherwise compile with the less efficient -DMDB_DSYNC=O_SYNC.
150 # define MDB_DSYNC O_DSYNC
154 /** A page number in the database.
155 * Note that 64 bit page numbers are overkill, since pages themselves
156 * already represent 12-13 bits of addressable memory, and the OS will
157 * always limit applications to a maximum of 63 bits of address space.
159 * @note In the #MDB_node structure, we only store 48 bits of this value,
160 * which thus limits us to only 60 bits of addressable data.
162 typedef ULONG pgno_t;
164 /** @defgroup debug Debug Macros
168 /** Enable debug output.
169 * Set this to 1 for copious tracing. Set to 2 to add dumps of all IDLs
170 * read from and written to the database (used for free space management).
175 #if !(__STDC_VERSION__ >= 199901L || defined(__GNUC__))
176 # define DPRINTF (void) /* Vararg macros may be unsupported */
178 /** Print a debug message with printf formatting. */
179 # define DPRINTF(fmt, ...) /**< Requires 2 or more args */ \
180 fprintf(stderr, "%s:%d:(%p) " fmt "\n", __func__, __LINE__, pthread_self(), __VA_ARGS__)
182 # define DPRINTF(fmt, ...) ((void) 0)
184 /** Print a debug string.
185 * The string is printed literally, with no format processing.
187 #define DPUTS(arg) DPRINTF("%s", arg)
190 /** A default memory page size.
191 * The actual size is platform-dependent, but we use this for
192 * boot-strapping. We probably should not be using this any more.
193 * The #GetPageSize() macro is used to get the actual size.
195 * Note that we don't currently support Huge pages. On Linux,
196 * regular data files cannot use Huge pages, and in general
197 * Huge pages aren't actually pageable. We rely on the OS
198 * demand-pager to read our data and page it out when memory
199 * pressure from other processes is high. So until OSs have
200 * actual paging support for Huge pages, they're not viable.
202 #define PAGESIZE 4096
204 /** The minimum number of keys required in a database page.
205 * Setting this to a larger value will place a smaller bound on the
206 * maximum size of a data item. Data items larger than this size will
207 * be pushed into overflow pages instead of being stored directly in
208 * the B-tree node. This value used to default to 4. With a page size
209 * of 4096 bytes that meant that any item larger than 1024 bytes would
210 * go into an overflow page. That also meant that on average 2-3KB of
211 * each overflow page was wasted space. The value cannot be lower than
212 * 2 because then there would no longer be a tree structure. With this
213 * value, items larger than 2KB will go into overflow pages, and on
214 * average only 1KB will be wasted.
216 #define MDB_MINKEYS 2
218 /** A stamp that identifies a file as an MDB file.
219 * There's nothing special about this value other than that it is easily
220 * recognizable, and it will reflect any byte order mismatches.
222 #define MDB_MAGIC 0xBEEFC0DE
224 /** The version number for a database's file format. */
225 #define MDB_VERSION 1
227 /** The maximum size of a key in the database.
228 * While data items have essentially unbounded size, we require that
229 * keys all fit onto a regular page. This limit could be raised a bit
230 * further if needed; to something just under #PAGESIZE / #MDB_MINKEYS.
232 #define MAXKEYSIZE 511
237 * This is used for printing a hex dump of a key's contents.
239 #define DKBUF char kbuf[(MAXKEYSIZE*2+1)]
240 /** Display a key in hex.
242 * Invoke a function to display a key in hex.
244 #define DKEY(x) mdb_dkey(x, kbuf)
250 /** @defgroup lazylock Lazy Locking
251 * Macros for locks that are't actually needed.
252 * The DB view is always consistent because all writes are wrapped in
253 * the wmutex. Finer-grained locks aren't necessary.
257 /** Use lazy locking. I.e., don't lock these accesses at all. */
261 /** Grab the reader lock */
262 #define LAZY_MUTEX_LOCK(x)
263 /** Release the reader lock */
264 #define LAZY_MUTEX_UNLOCK(x)
265 /** Release the DB table reader/writer lock */
266 #define LAZY_RWLOCK_UNLOCK(x)
267 /** Grab the DB table write lock */
268 #define LAZY_RWLOCK_WRLOCK(x)
269 /** Grab the DB table read lock */
270 #define LAZY_RWLOCK_RDLOCK(x)
271 /** Declare the DB table rwlock */
272 #define LAZY_RWLOCK_DEF(x)
273 /** Initialize the DB table rwlock */
274 #define LAZY_RWLOCK_INIT(x,y)
275 /** Destroy the DB table rwlock */
276 #define LAZY_RWLOCK_DESTROY(x)
278 #define LAZY_MUTEX_LOCK(x) pthread_mutex_lock(x)
279 #define LAZY_MUTEX_UNLOCK(x) pthread_mutex_unlock(x)
280 #define LAZY_RWLOCK_UNLOCK(x) pthread_rwlock_unlock(x)
281 #define LAZY_RWLOCK_WRLOCK(x) pthread_rwlock_wrlock(x)
282 #define LAZY_RWLOCK_RDLOCK(x) pthread_rwlock_rdlock(x)
283 #define LAZY_RWLOCK_DEF(x) pthread_rwlock_t x
284 #define LAZY_RWLOCK_INIT(x,y) pthread_rwlock_init(x,y)
285 #define LAZY_RWLOCK_DESTROY(x) pthread_rwlock_destroy(x)
289 /** An invalid page number.
290 * Mainly used to denote an empty tree.
292 #define P_INVALID (~0UL)
294 /** Test if a flag \b f is set in a flag word \b w. */
295 #define F_ISSET(w, f) (((w) & (f)) == (f))
297 /** Used for offsets within a single page.
298 * Since memory pages are typically 4 or 8KB in size, 12-13 bits,
301 typedef uint16_t indx_t;
303 /** Default size of memory map.
304 * This is certainly too small for any actual applications. Apps should always set
305 * the size explicitly using #mdb_env_set_mapsize().
307 #define DEFAULT_MAPSIZE 1048576
309 /** @defgroup readers Reader Lock Table
310 * Readers don't acquire any locks for their data access. Instead, they
311 * simply record their transaction ID in the reader table. The reader
312 * mutex is needed just to find an empty slot in the reader table. The
313 * slot's address is saved in thread-specific data so that subsequent read
314 * transactions started by the same thread need no further locking to proceed.
316 * Since the database uses multi-version concurrency control, readers don't
317 * actually need any locking. This table is used to keep track of which
318 * readers are using data from which old transactions, so that we'll know
319 * when a particular old transaction is no longer in use, Old transactions
320 * that have freed any data pages can then have their freed pages reclaimed
321 * for use by a later write transaction.
323 * The lock table is constructed such that reader slots are aligned with the
324 * processor's cache line size. Any slot is only ever used by one thread.
325 * This alignment guarantees that there will be no contention or cache
326 * thrashing as threads update their own slot info, and also eliminates
327 * any need for locking when accessing a slot.
329 * A writer thread will scan every slot in the table to determine the oldest
330 * outstanding reader transaction. Any freed pages older than this will be
331 * reclaimed by the writer. The writer doesn't use any locks when scanning
332 * this table. This means that there's no guarantee that the writer will
333 * see the most up-to-date reader info, but that's not required for correct
334 * operation - all we need is to know the upper bound on the oldest reader,
335 * we don't care at all about the newest reader. So the only consequence of
336 * reading stale information here is that old pages might hang around a
337 * while longer before being reclaimed. That's actually good anyway, because
338 * the longer we delay reclaiming old pages, the more likely it is that a
339 * string of contiguous pages can be found after coalescing old pages from
340 * many old transactions together.
342 * @todo We don't actually do such coalescing yet, we grab pages from one
343 * old transaction at a time.
346 /** Number of slots in the reader table.
347 * This value was chosen somewhat arbitrarily. 126 readers plus a
348 * couple mutexes fit exactly into 8KB on my development machine.
349 * Applications should set the table size using #mdb_env_set_maxreaders().
351 #define DEFAULT_READERS 126
353 /** The size of a CPU cache line in bytes. We want our lock structures
354 * aligned to this size to avoid false cache line sharing in the
356 * This value works for most CPUs. For Itanium this should be 128.
362 /** The information we store in a single slot of the reader table.
363 * In addition to a transaction ID, we also record the process and
364 * thread ID that owns a slot, so that we can detect stale information,
365 * e.g. threads or processes that went away without cleaning up.
366 * @note We currently don't check for stale records. We simply re-init
367 * the table when we know that we're the only process opening the
370 typedef struct MDB_rxbody {
371 /** The current Transaction ID when this transaction began.
372 * Multiple readers that start at the same time will probably have the
373 * same ID here. Again, it's not important to exclude them from
374 * anything; all we need to know is which version of the DB they
375 * started from so we can avoid overwriting any data used in that
376 * particular version.
379 /** The process ID of the process owning this reader txn. */
381 /** The thread ID of the thread owning this txn. */
385 /** The actual reader record, with cacheline padding. */
386 typedef struct MDB_reader {
389 /** shorthand for mrb_txnid */
390 #define mr_txnid mru.mrx.mrb_txnid
391 #define mr_pid mru.mrx.mrb_pid
392 #define mr_tid mru.mrx.mrb_tid
393 /** cache line alignment */
394 char pad[(sizeof(MDB_rxbody)+CACHELINE-1) & ~(CACHELINE-1)];
398 /** The header for the reader table.
399 * The table resides in a memory-mapped file. (This is a different file
400 * than is used for the main database.)
402 * For POSIX the actual mutexes reside in the shared memory of this
403 * mapped file. On Windows, mutexes are named objects allocated by the
404 * kernel; we store the mutex names in this mapped file so that other
405 * processes can grab them. This same approach will also be used on
406 * MacOSX/Darwin (using named semaphores) since MacOSX doesn't support
407 * process-shared POSIX mutexes.
409 typedef struct MDB_txbody {
410 /** Stamp identifying this as an MDB lock file. It must be set
413 /** Version number of this lock file. Must be set to #MDB_VERSION. */
414 uint32_t mtb_version;
418 /** Mutex protecting access to this table.
419 * This is the reader lock that #LOCK_MUTEX_R acquires.
421 pthread_mutex_t mtb_mutex;
423 /** The ID of the last transaction committed to the database.
424 * This is recorded here only for convenience; the value can always
425 * be determined by reading the main database meta pages.
428 /** The number of slots that have been used in the reader table.
429 * This always records the maximum count, it is not decremented
430 * when readers release their slots.
432 uint32_t mtb_numreaders;
433 /** The ID of the most recent meta page in the database.
434 * This is recorded here only for convenience; the value can always
435 * be determined by reading the main database meta pages.
437 uint32_t mtb_me_toggle;
440 /** The actual reader table definition. */
441 typedef struct MDB_txninfo {
444 #define mti_magic mt1.mtb.mtb_magic
445 #define mti_version mt1.mtb.mtb_version
446 #define mti_mutex mt1.mtb.mtb_mutex
447 #define mti_rmname mt1.mtb.mtb_rmname
448 #define mti_txnid mt1.mtb.mtb_txnid
449 #define mti_numreaders mt1.mtb.mtb_numreaders
450 #define mti_me_toggle mt1.mtb.mtb_me_toggle
451 char pad[(sizeof(MDB_txbody)+CACHELINE-1) & ~(CACHELINE-1)];
456 #define mti_wmname mt2.mt2_wmname
458 pthread_mutex_t mt2_wmutex;
459 #define mti_wmutex mt2.mt2_wmutex
461 char pad[(sizeof(pthread_mutex_t)+CACHELINE-1) & ~(CACHELINE-1)];
463 MDB_reader mti_readers[1];
467 /** Common header for all page types.
468 * Overflow pages occupy a number of contiguous pages with no
469 * headers on any page after the first.
471 typedef struct MDB_page {
473 pgno_t mp_pgno; /**< page number */
474 void * mp_next; /**< for in-memory list of freed structs */
476 #define P_BRANCH 0x01 /**< branch page */
477 #define P_LEAF 0x02 /**< leaf page */
478 #define P_OVERFLOW 0x04 /**< overflow page */
479 #define P_META 0x08 /**< meta page */
480 #define P_DIRTY 0x10 /**< dirty page */
481 #define P_LEAF2 0x20 /**< for #MDB_DUPFIXED records */
485 indx_t mp_lower; /**< lower bound of free space */
486 indx_t mp_upper; /**< upper bound of free space */
488 uint32_t mp_pages; /**< number of overflow pages */
490 indx_t mp_ptrs[1]; /**< dynamic size */
493 /** Size of the page header, excluding dynamic data at the end */
494 #define PAGEHDRSZ ((unsigned) offsetof(MDB_page, mp_ptrs))
496 /** Address of first usable data byte in a page, after the header */
497 #define METADATA(p) ((void *)((char *)(p) + PAGEHDRSZ))
499 /** Number of nodes on a page */
500 #define NUMKEYS(p) (((p)->mp_lower - PAGEHDRSZ) >> 1)
502 /** The amount of space remaining in the page */
503 #define SIZELEFT(p) (indx_t)((p)->mp_upper - (p)->mp_lower)
505 /** The percentage of space used in the page, in tenths of a percent. */
506 #define PAGEFILL(env, p) (1000L * ((env)->me_psize - PAGEHDRSZ - SIZELEFT(p)) / \
507 ((env)->me_psize - PAGEHDRSZ))
508 /** The minimum page fill factor, in tenths of a percent.
509 * Pages emptier than this are candidates for merging.
511 #define FILL_THRESHOLD 250
513 /** Test if a page is a leaf page */
514 #define IS_LEAF(p) F_ISSET((p)->mp_flags, P_LEAF)
515 /** Test if a page is a LEAF2 page */
516 #define IS_LEAF2(p) F_ISSET((p)->mp_flags, P_LEAF2)
517 /** Test if a page is a branch page */
518 #define IS_BRANCH(p) F_ISSET((p)->mp_flags, P_BRANCH)
519 /** Test if a page is an overflow page */
520 #define IS_OVERFLOW(p) F_ISSET((p)->mp_flags, P_OVERFLOW)
522 /** The number of overflow pages needed to store the given size. */
523 #define OVPAGES(size, psize) ((PAGEHDRSZ-1 + (size)) / (psize) + 1)
525 /** Header for a single key/data pair within a page.
526 * We guarantee 2-byte alignment for nodes.
528 typedef struct MDB_node {
529 /** lo and hi are used for data size on leaf nodes and for
530 * child pgno on branch nodes. On 64 bit platforms, flags
531 * is also used for pgno. (branch nodes ignore flags)
533 unsigned short mn_lo;
534 unsigned short mn_hi; /**< part of dsize or pgno */
535 unsigned short mn_flags; /**< flags for special node types */
536 #define F_BIGDATA 0x01 /**< data put on overflow page */
537 #define F_SUBDATA 0x02 /**< data is a sub-database */
538 #define F_DUPDATA 0x04 /**< data has duplicates */
539 unsigned short mn_ksize; /**< key size */
540 char mn_data[1]; /**< key and data are appended here */
543 /** Size of the node header, excluding dynamic data at the end */
544 #define NODESIZE offsetof(MDB_node, mn_data)
546 /** Size of a node in a branch page.
547 * This is just the node header plus the key, there is no data.
549 #define INDXSIZE(k) (NODESIZE + ((k) == NULL ? 0 : (k)->mv_size))
551 /** Size of a node in a leaf page.
552 * This is node header plus key plus data size.
554 #define LEAFSIZE(k, d) (NODESIZE + (k)->mv_size + (d)->mv_size)
556 /** Address of node \i in page \p */
557 #define NODEPTR(p, i) ((MDB_node *)((char *)(p) + (p)->mp_ptrs[i]))
559 /** Address of the key for the node */
560 #define NODEKEY(node) (void *)((node)->mn_data)
562 /** Address of the data for a node */
563 #define NODEDATA(node) (void *)((char *)(node)->mn_data + (node)->mn_ksize)
565 /** Get the page number pointed to by a branch node */
566 #if LONG_MAX == 0x7fffffff
567 #define NODEPGNO(node) ((node)->mn_lo | ((node)->mn_hi << 16))
568 /** Set the page number in a branch node */
569 #define SETPGNO(node,pgno) do { \
570 (node)->mn_lo = (pgno) & 0xffff; (node)->mn_hi = (pgno) >> 16;} while(0)
572 #define NODEPGNO(node) ((node)->mn_lo | ((node)->mn_hi << 16) | ((unsigned long)(node)->mn_flags << 32))
573 /** Set the page number in a branch node */
574 #define SETPGNO(node,pgno) do { \
575 (node)->mn_lo = (pgno) & 0xffff; (node)->mn_hi = (pgno) >> 16; \
576 (node)->mn_flags = (pgno) >> 32; } while(0)
579 /** Get the size of the data in a leaf node */
580 #define NODEDSZ(node) ((node)->mn_lo | ((unsigned)(node)->mn_hi << 16))
581 /** Set the size of the data for a leaf node */
582 #define SETDSZ(node,size) do { \
583 (node)->mn_lo = (size) & 0xffff; (node)->mn_hi = (size) >> 16;} while(0)
584 /** The size of a key in a node */
585 #define NODEKSZ(node) ((node)->mn_ksize)
587 /** The address of a key in a LEAF2 page.
588 * LEAF2 pages are used for #MDB_DUPFIXED sorted-duplicate sub-DBs.
589 * There are no node headers, keys are stored contiguously.
591 #define LEAF2KEY(p, i, ks) ((char *)(p) + PAGEHDRSZ + ((i)*(ks)))
593 /** Set the \b node's key into \b key, if requested. */
594 #define MDB_SET_KEY(node, key) if (key!=NULL) {(key)->mv_size = NODEKSZ(node); (key)->mv_data = NODEKEY(node);}
596 /** Information about a single database in the environment. */
597 typedef struct MDB_db {
598 uint32_t md_pad; /**< also ksize for LEAF2 pages */
599 uint16_t md_flags; /**< @ref mdb_open */
600 uint16_t md_depth; /**< depth of this tree */
601 ULONG md_branch_pages; /**< number of internal pages */
602 ULONG md_leaf_pages; /**< number of leaf pages */
603 ULONG md_overflow_pages; /**< number of overflow pages */
604 ULONG md_entries; /**< number of data items */
605 pgno_t md_root; /**< the root page of this tree */
608 /** Handle for the DB used to track free pages. */
610 /** Handle for the default DB. */
613 /** Meta page content. */
614 typedef struct MDB_meta {
615 /** Stamp identifying this as an MDB data file. It must be set
618 /** Version number of this lock file. Must be set to #MDB_VERSION. */
620 void *mm_address; /**< address for fixed mapping */
621 size_t mm_mapsize; /**< size of mmap region */
622 MDB_db mm_dbs[2]; /**< first is free space, 2nd is main db */
623 /** The size of pages used in this DB */
624 #define mm_psize mm_dbs[0].md_pad
625 /** Any persistent environment flags. @ref mdb_env */
626 #define mm_flags mm_dbs[0].md_flags
627 pgno_t mm_last_pg; /**< last used page in file */
628 ULONG mm_txnid; /**< txnid that committed this page */
631 /** Auxiliary DB info.
632 * The information here is mostly static/read-only. There is
633 * only a single copy of this record in the environment.
634 * The \b md_dirty flag is not read-only, but only a write
635 * transaction can ever update it, and only write transactions
636 * need to worry about it.
638 typedef struct MDB_dbx {
639 MDB_val md_name; /**< name of the database */
640 MDB_cmp_func *md_cmp; /**< function for comparing keys */
641 MDB_cmp_func *md_dcmp; /**< function for comparing data items */
642 MDB_rel_func *md_rel; /**< user relocate function */
643 MDB_dbi md_parent; /**< parent DB of a sub-DB */
644 unsigned int md_dirty; /**< TRUE if DB was written in this txn */
647 /** A database transaction.
648 * Every operation requires a transaction handle.
651 pgno_t mt_next_pgno; /**< next unallocated page */
652 /** The ID of this transaction. IDs are integers incrementing from 1.
653 * Only committed write transactions increment the ID. If a transaction
654 * aborts, the ID may be re-used by the next writer.
657 MDB_env *mt_env; /**< the DB environment */
658 /** The list of pages that became unused during this transaction.
663 ID2L dirty_list; /**< modified pages */
664 MDB_reader *reader; /**< this thread's slot in the reader table */
666 /** Array of records for each DB known in the environment. */
668 /** Array of MDB_db records for each known DB */
670 /** Number of DB records in use. This number only ever increments;
671 * we don't decrement it when individual DB handles are closed.
673 unsigned int mt_numdbs;
675 #define MDB_TXN_RDONLY 0x01 /**< read-only transaction */
676 #define MDB_TXN_ERROR 0x02 /**< an error has occurred */
677 unsigned int mt_flags;
678 /** Tracks which of the two meta pages was used at the start
679 * of this transaction.
681 unsigned int mt_toggle;
684 /** Enough space for 2^32 nodes with minimum of 2 keys per node. I.e., plenty.
685 * At 4 keys per node, enough for 2^64 nodes, so there's probably no need to
686 * raise this on a 64 bit machine.
688 #define CURSOR_STACK 32
692 /** Cursors are used for all DB operations */
694 /** Context used for databases with #MDB_DUPSORT, otherwise NULL */
695 struct MDB_xcursor *mc_xcursor;
696 /** The transaction that owns this cursor */
698 /** The database handle this cursor operates on */
700 unsigned short mc_snum; /**< number of pushed pages */
701 unsigned short mc_top; /**< index of top page, mc_snum-1 */
702 unsigned int mc_flags;
703 #define C_INITIALIZED 0x01 /**< cursor has been initialized and is valid */
704 #define C_EOF 0x02 /**< No more data */
705 #define C_XDIRTY 0x04 /**< @deprecated mc_xcursor needs to be flushed */
706 MDB_page *mc_pg[CURSOR_STACK]; /**< stack of pushed pages */
707 indx_t mc_ki[CURSOR_STACK]; /**< stack of page indices */
710 /** Context for sorted-dup records.
711 * We could have gone to a fully recursive design, with arbitrarily
712 * deep nesting of sub-databases. But for now we only handle these
713 * levels - main DB, optional sub-DB, sorted-duplicate DB.
715 typedef struct MDB_xcursor {
716 /** A sub-cursor for traversing the Dup DB */
717 MDB_cursor mx_cursor;
718 /** A fake transaction struct for pointing to our own table
722 /** Our private DB information tables. Slots 0 and 1 are always
723 * copies of the corresponding slots in the main transaction. These
724 * hold the FREEDB and MAINDB, respectively. If the main cursor is
725 * on a sub-database, that will be copied to slot 2, and the duplicate
726 * database info will be in slot 3. If the main cursor is on the MAINDB
727 * then the duplicate DB info will be in slot 2 and slot 3 will be unused.
734 /** A set of pages freed by an earlier transaction. */
735 typedef struct MDB_oldpages {
736 /** Usually we only read one record from the FREEDB at a time, but
737 * in case we read more, this will chain them together.
739 struct MDB_oldpages *mo_next;
740 /** The ID of the transaction in which these pages were freed. */
742 /** An #IDL of the pages */
743 pgno_t mo_pages[1]; /* dynamic */
746 /** The database environment. */
748 HANDLE me_fd; /**< The main data file */
749 HANDLE me_lfd; /**< The lock file */
750 HANDLE me_mfd; /**< just for writing the meta pages */
751 #define MDB_FATAL_ERROR 0x80000000U
753 uint32_t me_extrapad; /**< unused for now */
754 unsigned int me_maxreaders; /**< size of the reader table */
755 unsigned int me_numdbs; /**< number of DBs opened */
756 unsigned int me_maxdbs; /**< size of the DB table */
757 char *me_path; /**< path to the DB files */
758 char *me_map; /**< the memory map of the data file */
759 MDB_txninfo *me_txns; /**< the memory map of the lock file */
760 MDB_meta *me_metas[2]; /**< pointers to the two meta pages */
761 MDB_txn *me_txn; /**< current write transaction */
762 size_t me_mapsize; /**< size of the data memory map */
763 off_t me_size; /**< current file size */
764 pgno_t me_maxpg; /**< me_mapsize / me_psize */
765 unsigned int me_psize; /**< size of a page, from #GetPageSize */
766 unsigned int me_db_toggle; /**< which DB table is current */
767 MDB_dbx *me_dbxs; /**< array of static DB info */
768 MDB_db *me_dbs[2]; /**< two arrays of MDB_db info */
769 MDB_oldpages *me_pghead; /**< list of old page records */
770 pthread_key_t me_txkey; /**< thread-key for readers */
771 MDB_page *me_dpages; /**< list of malloc'd blocks for re-use */
772 /** IDL of pages that became unused in a write txn */
773 pgno_t me_free_pgs[MDB_IDL_UM_SIZE];
774 /** ID2L of pages that were written during a write txn */
775 ID2 me_dirty_list[MDB_IDL_UM_SIZE];
776 /** rwlock for the DB tables, if #LAZY_LOCKS is false */
777 LAZY_RWLOCK_DEF(me_dblock);
779 HANDLE me_rmutex; /* Windows mutexes don't reside in shared mem */
783 /** max number of pages to commit in one writev() call */
784 #define MDB_COMMIT_PAGES 64
786 static MDB_page *mdb_alloc_page(MDB_cursor *mc, int num);
787 static int mdb_touch(MDB_cursor *mc);
789 static int mdb_search_page_root(MDB_cursor *mc,
790 MDB_val *key, int modify);
791 static int mdb_search_page(MDB_cursor *mc,
792 MDB_val *key, int modify);
794 static int mdb_env_read_header(MDB_env *env, MDB_meta *meta);
795 static int mdb_env_read_meta(MDB_env *env, int *which);
796 static int mdb_env_write_meta(MDB_txn *txn);
797 static int mdb_get_page(MDB_txn *txn, pgno_t pgno, MDB_page **mp);
799 static MDB_node *mdb_search_node(MDB_cursor *mc, MDB_val *key, int *exactp);
800 static int mdb_add_node(MDB_cursor *mc, indx_t indx,
801 MDB_val *key, MDB_val *data, pgno_t pgno, uint8_t flags);
802 static void mdb_del_node(MDB_page *mp, indx_t indx, int ksize);
803 static int mdb_del0(MDB_cursor *mc, MDB_node *leaf);
804 static int mdb_read_data(MDB_txn *txn, MDB_node *leaf, MDB_val *data);
806 static int mdb_rebalance(MDB_cursor *mc);
807 static int mdb_update_key(MDB_page *mp, indx_t indx, MDB_val *key);
808 static int mdb_move_node(MDB_cursor *csrcrc, MDB_cursor *cdstst);
809 static int mdb_merge(MDB_cursor *csrcrc, MDB_cursor *cdstst);
810 static int mdb_split(MDB_cursor *mc, MDB_val *newkey, MDB_val *newdata,
812 static MDB_page *mdb_new_page(MDB_cursor *mc, uint32_t flags, int num);
814 static void cursor_pop_page(MDB_cursor *mc);
815 static int cursor_push_page(MDB_cursor *mc, MDB_page *mp);
817 static int mdb_sibling(MDB_cursor *mc, int move_right);
818 static int mdb_cursor_next(MDB_cursor *mc, MDB_val *key, MDB_val *data, MDB_cursor_op op);
819 static int mdb_cursor_prev(MDB_cursor *mc, MDB_val *key, MDB_val *data, MDB_cursor_op op);
820 static int mdb_cursor_set(MDB_cursor *mc, MDB_val *key, MDB_val *data, MDB_cursor_op op,
822 static int mdb_cursor_first(MDB_cursor *mc, MDB_val *key, MDB_val *data);
823 static int mdb_cursor_last(MDB_cursor *mc, MDB_val *key, MDB_val *data);
825 static void mdb_xcursor_init0(MDB_cursor *mc);
826 static void mdb_xcursor_init1(MDB_cursor *mc, MDB_node *node);
827 static void mdb_xcursor_init2(MDB_cursor *mc);
828 static void mdb_xcursor_fini(MDB_cursor *mc);
830 static size_t mdb_leaf_size(MDB_env *env, MDB_val *key, MDB_val *data);
831 static size_t mdb_branch_size(MDB_env *env, MDB_val *key);
833 static void mdb_default_cmp(MDB_txn *txn, MDB_dbi dbi);
836 static MDB_cmp_func memncmp, memnrcmp, intcmp, cintcmp;
840 static SECURITY_DESCRIPTOR mdb_null_sd;
841 static SECURITY_ATTRIBUTES mdb_all_sa;
842 static int mdb_sec_inited;
846 mdb_version(int *major, int *minor, int *patch)
848 if (major) *major = MDB_VERSION_MAJOR;
849 if (minor) *minor = MDB_VERSION_MINOR;
850 if (patch) *patch = MDB_VERSION_PATCH;
851 return MDB_VERSION_STRING;
854 /** Table of descriptions for MDB @ref error codes */
855 static char *const mdb_errstr[] = {
856 "MDB_KEYEXIST: Key/data pair already exists",
857 "MDB_NOTFOUND: No matching key/data pair found",
858 "MDB_PAGE_NOTFOUND: Requested page not found",
859 "MDB_CORRUPTED: Located page was wrong type",
860 "MDB_PANIC: Update of meta page failed",
861 "MDB_VERSION_MISMATCH: Database environment version mismatch"
865 mdb_strerror(int err)
868 return ("Successful return: 0");
870 if (err >= MDB_KEYEXIST && err <= MDB_VERSION_MISMATCH)
871 return mdb_errstr[err - MDB_KEYEXIST];
873 return strerror(err);
878 mdb_dkey(MDB_val *key, char *buf)
881 unsigned char *c = key->mv_data;
883 if (key->mv_size > MAXKEYSIZE)
886 for (i=0; i<key->mv_size; i++)
887 ptr += sprintf(ptr, "%02x", *c++);
889 sprintf(buf, "%.*s", key->mv_size, key->mv_data);
896 mdb_cmp(MDB_txn *txn, MDB_dbi dbi, const MDB_val *a, const MDB_val *b)
898 return txn->mt_dbxs[dbi].md_cmp(a, b);
902 mdb_dcmp(MDB_txn *txn, MDB_dbi dbi, const MDB_val *a, const MDB_val *b)
904 if (txn->mt_dbxs[dbi].md_dcmp)
905 return txn->mt_dbxs[dbi].md_dcmp(a, b);
907 return EINVAL; /* too bad you can't distinguish this from a valid result */
910 /* Allocate new page(s) for writing */
912 mdb_alloc_page(MDB_cursor *mc, int num)
914 MDB_txn *txn = mc->mc_txn;
916 pgno_t pgno = P_INVALID;
919 if (txn->mt_txnid > 2) {
921 if (!txn->mt_env->me_pghead && mc->mc_dbi != FREE_DBI &&
922 txn->mt_dbs[FREE_DBI].md_root != P_INVALID) {
923 /* See if there's anything in the free DB */
929 m2.mc_dbi = FREE_DBI;
932 mdb_search_page(&m2, NULL, 0);
933 leaf = NODEPTR(m2.mc_pg[m2.mc_top], 0);
934 kptr = (ULONG *)NODEKEY(leaf);
938 oldest = txn->mt_txnid - 1;
939 for (i=0; i<txn->mt_env->me_txns->mti_numreaders; i++) {
940 ULONG mr = txn->mt_env->me_txns->mti_readers[i].mr_txnid;
941 if (mr && mr < oldest)
946 if (oldest > *kptr) {
947 /* It's usable, grab it.
953 mdb_read_data(txn, leaf, &data);
954 idl = (ULONG *)data.mv_data;
955 mop = malloc(sizeof(MDB_oldpages) + MDB_IDL_SIZEOF(idl) - sizeof(pgno_t));
956 mop->mo_next = txn->mt_env->me_pghead;
957 mop->mo_txnid = *kptr;
958 txn->mt_env->me_pghead = mop;
959 memcpy(mop->mo_pages, idl, MDB_IDL_SIZEOF(idl));
964 DPRINTF("IDL read txn %lu root %lu num %lu",
965 mop->mo_txnid, txn->mt_dbs[FREE_DBI].md_root, idl[0]);
966 for (i=0; i<idl[0]; i++) {
967 DPRINTF("IDL %lu", idl[i+1]);
971 /* drop this IDL from the DB */
972 m2.mc_ki[m2.mc_top] = 0;
973 m2.mc_flags = C_INITIALIZED;
974 mdb_cursor_del(&m2, 0);
977 if (txn->mt_env->me_pghead) {
978 MDB_oldpages *mop = txn->mt_env->me_pghead;
980 /* FIXME: For now, always use fresh pages. We
981 * really ought to search the free list for a
986 /* peel pages off tail, so we only have to truncate the list */
987 pgno = MDB_IDL_LAST(mop->mo_pages);
988 if (MDB_IDL_IS_RANGE(mop->mo_pages)) {
990 if (mop->mo_pages[2] > mop->mo_pages[1])
991 mop->mo_pages[0] = 0;
995 if (MDB_IDL_IS_ZERO(mop->mo_pages)) {
996 txn->mt_env->me_pghead = mop->mo_next;
1003 if (pgno == P_INVALID) {
1004 /* DB size is maxed out */
1005 if (txn->mt_next_pgno + num >= txn->mt_env->me_maxpg)
1008 if (txn->mt_env->me_dpages && num == 1) {
1009 np = txn->mt_env->me_dpages;
1010 txn->mt_env->me_dpages = np->mp_next;
1012 if ((np = malloc(txn->mt_env->me_psize * num )) == NULL)
1015 if (pgno == P_INVALID) {
1016 np->mp_pgno = txn->mt_next_pgno;
1017 txn->mt_next_pgno += num;
1021 mid.mid = np->mp_pgno;
1023 mdb_mid2l_insert(txn->mt_u.dirty_list, &mid);
1028 /* Touch a page: make it dirty and re-insert into tree with updated pgno.
1031 mdb_touch(MDB_cursor *mc)
1033 MDB_page *mp = mc->mc_pg[mc->mc_top];
1036 if (!F_ISSET(mp->mp_flags, P_DIRTY)) {
1038 if ((np = mdb_alloc_page(mc, 1)) == NULL)
1040 DPRINTF("touched db %u page %lu -> %lu", mc->mc_dbi, mp->mp_pgno, np->mp_pgno);
1041 assert(mp->mp_pgno != np->mp_pgno);
1042 mdb_midl_append(mc->mc_txn->mt_free_pgs, mp->mp_pgno);
1044 memcpy(np, mp, mc->mc_txn->mt_env->me_psize);
1047 mp->mp_flags |= P_DIRTY;
1049 mc->mc_pg[mc->mc_top] = mp;
1050 /* Update the page number to new touched page. */
1052 SETPGNO(NODEPTR(mc->mc_pg[mc->mc_top-1], mc->mc_ki[mc->mc_top-1]), mp->mp_pgno);
1058 mdb_env_sync(MDB_env *env, int force)
1061 if (force || !F_ISSET(env->me_flags, MDB_NOSYNC)) {
1062 if (fdatasync(env->me_fd))
1069 mdb_txn_reset0(MDB_txn *txn);
1072 mdb_txn_renew0(MDB_txn *txn)
1074 MDB_env *env = txn->mt_env;
1076 if (txn->mt_flags & MDB_TXN_RDONLY) {
1077 MDB_reader *r = pthread_getspecific(env->me_txkey);
1080 pid_t pid = getpid();
1081 pthread_t tid = pthread_self();
1084 for (i=0; i<env->me_txns->mti_numreaders; i++)
1085 if (env->me_txns->mti_readers[i].mr_pid == 0)
1087 if (i == env->me_maxreaders) {
1088 UNLOCK_MUTEX_R(env);
1091 env->me_txns->mti_readers[i].mr_pid = pid;
1092 env->me_txns->mti_readers[i].mr_tid = tid;
1093 if (i >= env->me_txns->mti_numreaders)
1094 env->me_txns->mti_numreaders = i+1;
1095 UNLOCK_MUTEX_R(env);
1096 r = &env->me_txns->mti_readers[i];
1097 pthread_setspecific(env->me_txkey, r);
1099 txn->mt_txnid = env->me_txns->mti_txnid;
1100 txn->mt_toggle = env->me_txns->mti_me_toggle;
1101 r->mr_txnid = txn->mt_txnid;
1102 txn->mt_u.reader = r;
1106 txn->mt_txnid = env->me_txns->mti_txnid+1;
1107 txn->mt_toggle = env->me_txns->mti_me_toggle;
1108 txn->mt_u.dirty_list = env->me_dirty_list;
1109 txn->mt_u.dirty_list[0].mid = 0;
1110 txn->mt_free_pgs = env->me_free_pgs;
1111 txn->mt_free_pgs[0] = 0;
1112 txn->mt_next_pgno = env->me_metas[txn->mt_toggle]->mm_last_pg+1;
1116 /* Copy the DB arrays */
1117 LAZY_RWLOCK_RDLOCK(&env->me_dblock);
1118 txn->mt_numdbs = env->me_numdbs;
1119 txn->mt_dbxs = env->me_dbxs; /* mostly static anyway */
1120 memcpy(txn->mt_dbs, env->me_metas[txn->mt_toggle]->mm_dbs, 2 * sizeof(MDB_db));
1121 if (txn->mt_numdbs > 2)
1122 memcpy(txn->mt_dbs+2, env->me_dbs[env->me_db_toggle]+2,
1123 (txn->mt_numdbs - 2) * sizeof(MDB_db));
1124 LAZY_RWLOCK_UNLOCK(&env->me_dblock);
1130 mdb_txn_renew(MDB_txn *txn)
1137 if (txn->mt_env->me_flags & MDB_FATAL_ERROR) {
1138 DPUTS("environment had fatal error, must shutdown!");
1142 rc = mdb_txn_renew0(txn);
1143 if (rc == MDB_SUCCESS) {
1144 DPRINTF("renew txn %lu%c %p on mdbenv %p, root page %lu",
1145 txn->mt_txnid, (txn->mt_flags & MDB_TXN_RDONLY) ? 'r' : 'w', txn,
1146 (void *)txn->mt_env, txn->mt_dbs[MAIN_DBI].md_root);
1152 mdb_txn_begin(MDB_env *env, unsigned int flags, MDB_txn **ret)
1157 if (env->me_flags & MDB_FATAL_ERROR) {
1158 DPUTS("environment had fatal error, must shutdown!");
1161 if ((txn = calloc(1, sizeof(MDB_txn) + env->me_maxdbs * sizeof(MDB_db))) == NULL) {
1162 DPRINTF("calloc: %s", strerror(ErrCode()));
1165 txn->mt_dbs = (MDB_db *)(txn+1);
1166 if (flags & MDB_RDONLY) {
1167 txn->mt_flags |= MDB_TXN_RDONLY;
1171 rc = mdb_txn_renew0(txn);
1176 DPRINTF("begin txn %lu%c %p on mdbenv %p, root page %lu",
1177 txn->mt_txnid, (txn->mt_flags & MDB_TXN_RDONLY) ? 'r' : 'w', txn,
1178 (void *) env, txn->mt_dbs[MAIN_DBI].md_root);
1185 mdb_txn_reset0(MDB_txn *txn)
1187 MDB_env *env = txn->mt_env;
1189 if (F_ISSET(txn->mt_flags, MDB_TXN_RDONLY)) {
1190 txn->mt_u.reader->mr_txnid = 0;
1196 /* return all dirty pages to dpage list */
1197 for (i=1; i<=txn->mt_u.dirty_list[0].mid; i++) {
1198 dp = txn->mt_u.dirty_list[i].mptr;
1199 if (!IS_OVERFLOW(dp) || dp->mp_pages == 1) {
1200 dp->mp_next = txn->mt_env->me_dpages;
1201 txn->mt_env->me_dpages = dp;
1203 /* large pages just get freed directly */
1208 while ((mop = txn->mt_env->me_pghead)) {
1209 txn->mt_env->me_pghead = mop->mo_next;
1214 for (i=2; i<env->me_numdbs; i++)
1215 env->me_dbxs[i].md_dirty = 0;
1216 UNLOCK_MUTEX_W(env);
1221 mdb_txn_reset(MDB_txn *txn)
1226 DPRINTF("reset txn %lu%c %p on mdbenv %p, root page %lu",
1227 txn->mt_txnid, (txn->mt_flags & MDB_TXN_RDONLY) ? 'r' : 'w', txn,
1228 (void *)txn->mt_env, txn->mt_dbs[MAIN_DBI].md_root);
1230 mdb_txn_reset0(txn);
1234 mdb_txn_abort(MDB_txn *txn)
1239 DPRINTF("abort txn %lu%c %p on mdbenv %p, root page %lu",
1240 txn->mt_txnid, (txn->mt_flags & MDB_TXN_RDONLY) ? 'r' : 'w', txn,
1241 (void *)txn->mt_env, txn->mt_dbs[MAIN_DBI].md_root);
1243 mdb_txn_reset0(txn);
1248 mdb_txn_commit(MDB_txn *txn)
1259 assert(txn != NULL);
1260 assert(txn->mt_env != NULL);
1264 if (F_ISSET(txn->mt_flags, MDB_TXN_RDONLY)) {
1269 if (txn != env->me_txn) {
1270 DPUTS("attempt to commit unknown transaction");
1275 if (F_ISSET(txn->mt_flags, MDB_TXN_ERROR)) {
1276 DPUTS("error flag is set, can't commit");
1281 if (!txn->mt_u.dirty_list[0].mid)
1284 DPRINTF("committing txn %lu %p on mdbenv %p, root page %lu",
1285 txn->mt_txnid, txn, (void *)env, txn->mt_dbs[MAIN_DBI].md_root);
1288 mc.mc_dbi = FREE_DBI;
1291 /* should only be one record now */
1292 if (env->me_pghead) {
1293 /* make sure first page of freeDB is touched and on freelist */
1294 mdb_search_page(&mc, NULL, 1);
1296 /* save to free list */
1297 if (!MDB_IDL_IS_ZERO(txn->mt_free_pgs)) {
1301 /* make sure last page of freeDB is touched and on freelist */
1302 key.mv_size = MAXKEYSIZE+1;
1304 mdb_search_page(&mc, &key, 1);
1306 mdb_midl_sort(txn->mt_free_pgs);
1310 ULONG *idl = txn->mt_free_pgs;
1311 DPRINTF("IDL write txn %lu root %lu num %lu",
1312 txn->mt_txnid, txn->mt_dbs[FREE_DBI].md_root, idl[0]);
1313 for (i=0; i<idl[0]; i++) {
1314 DPRINTF("IDL %lu", idl[i+1]);
1318 /* write to last page of freeDB */
1319 key.mv_size = sizeof(pgno_t);
1320 key.mv_data = (char *)&txn->mt_txnid;
1321 data.mv_data = txn->mt_free_pgs;
1322 /* The free list can still grow during this call,
1323 * despite the pre-emptive touches above. So check
1324 * and make sure the entire thing got written.
1327 i = txn->mt_free_pgs[0];
1328 data.mv_size = MDB_IDL_SIZEOF(txn->mt_free_pgs);
1329 rc = mdb_cursor_put(&mc, &key, &data, 0);
1334 } while (i != txn->mt_free_pgs[0]);
1336 /* should only be one record now */
1337 if (env->me_pghead) {
1341 mop = env->me_pghead;
1342 key.mv_size = sizeof(pgno_t);
1343 key.mv_data = (char *)&mop->mo_txnid;
1344 data.mv_size = MDB_IDL_SIZEOF(mop->mo_pages);
1345 data.mv_data = mop->mo_pages;
1346 mdb_cursor_put(&mc, &key, &data, 0);
1347 free(env->me_pghead);
1348 env->me_pghead = NULL;
1351 /* Update DB root pointers. Their pages have already been
1352 * touched so this is all in-place and cannot fail.
1356 data.mv_size = sizeof(MDB_db);
1358 mc.mc_dbi = MAIN_DBI;
1360 for (i = 2; i < txn->mt_numdbs; i++) {
1361 if (txn->mt_dbxs[i].md_dirty) {
1362 data.mv_data = &txn->mt_dbs[i];
1363 mdb_cursor_put(&mc, &txn->mt_dbxs[i].md_name, &data, 0);
1368 /* Commit up to MDB_COMMIT_PAGES dirty pages to disk until done.
1374 /* Windows actually supports scatter/gather I/O, but only on
1375 * unbuffered file handles. Since we're relying on the OS page
1376 * cache for all our data, that's self-defeating. So we just
1377 * write pages one at a time. We use the ov structure to set
1378 * the write offset, to at least save the overhead of a Seek
1382 memset(&ov, 0, sizeof(ov));
1383 for (; i<=txn->mt_u.dirty_list[0].mid; i++) {
1385 dp = txn->mt_u.dirty_list[i].mptr;
1386 DPRINTF("committing page %lu", dp->mp_pgno);
1387 size = dp->mp_pgno * env->me_psize;
1388 ov.Offset = size & 0xffffffff;
1389 ov.OffsetHigh = size >> 16;
1390 ov.OffsetHigh >>= 16;
1391 /* clear dirty flag */
1392 dp->mp_flags &= ~P_DIRTY;
1393 wsize = env->me_psize;
1394 if (IS_OVERFLOW(dp)) wsize *= dp->mp_pages;
1395 rc = WriteFile(env->me_fd, dp, wsize, NULL, &ov);
1398 DPRINTF("WriteFile: %d", n);
1405 struct iovec iov[MDB_COMMIT_PAGES];
1409 for (; i<=txn->mt_u.dirty_list[0].mid; i++) {
1410 dp = txn->mt_u.dirty_list[i].mptr;
1411 if (dp->mp_pgno != next) {
1413 DPRINTF("committing %u dirty pages", n);
1414 rc = writev(env->me_fd, iov, n);
1418 DPUTS("short write, filesystem full?");
1420 DPRINTF("writev: %s", strerror(n));
1427 lseek(env->me_fd, dp->mp_pgno * env->me_psize, SEEK_SET);
1430 DPRINTF("committing page %lu", dp->mp_pgno);
1431 iov[n].iov_len = env->me_psize;
1432 if (IS_OVERFLOW(dp)) iov[n].iov_len *= dp->mp_pages;
1433 iov[n].iov_base = dp;
1434 size += iov[n].iov_len;
1435 next = dp->mp_pgno + (IS_OVERFLOW(dp) ? dp->mp_pages : 1);
1436 /* clear dirty flag */
1437 dp->mp_flags &= ~P_DIRTY;
1438 if (++n >= MDB_COMMIT_PAGES) {
1448 DPRINTF("committing %u dirty pages", n);
1449 rc = writev(env->me_fd, iov, n);
1453 DPUTS("short write, filesystem full?");
1455 DPRINTF("writev: %s", strerror(n));
1462 /* Drop the dirty pages.
1464 for (i=1; i<=txn->mt_u.dirty_list[0].mid; i++) {
1465 dp = txn->mt_u.dirty_list[i].mptr;
1466 if (!IS_OVERFLOW(dp) || dp->mp_pages == 1) {
1467 dp->mp_next = txn->mt_env->me_dpages;
1468 txn->mt_env->me_dpages = dp;
1472 txn->mt_u.dirty_list[i].mid = 0;
1474 txn->mt_u.dirty_list[0].mid = 0;
1476 if ((n = mdb_env_sync(env, 0)) != 0 ||
1477 (n = mdb_env_write_meta(txn)) != MDB_SUCCESS) {
1484 /* update the DB tables */
1486 int toggle = !env->me_db_toggle;
1489 ip = &env->me_dbs[toggle][2];
1490 jp = &txn->mt_dbs[2];
1491 LAZY_RWLOCK_WRLOCK(&env->me_dblock);
1492 for (i = 2; i < txn->mt_numdbs; i++) {
1493 if (ip->md_root != jp->md_root)
1498 for (i = 2; i < txn->mt_numdbs; i++) {
1499 if (txn->mt_dbxs[i].md_dirty)
1500 txn->mt_dbxs[i].md_dirty = 0;
1502 env->me_db_toggle = toggle;
1503 env->me_numdbs = txn->mt_numdbs;
1504 LAZY_RWLOCK_UNLOCK(&env->me_dblock);
1507 UNLOCK_MUTEX_W(env);
1514 mdb_env_read_header(MDB_env *env, MDB_meta *meta)
1516 char page[PAGESIZE];
1521 /* We don't know the page size yet, so use a minimum value.
1525 if (!ReadFile(env->me_fd, page, PAGESIZE, (DWORD *)&rc, NULL) || rc == 0)
1527 if ((rc = read(env->me_fd, page, PAGESIZE)) == 0)
1532 else if (rc != PAGESIZE) {
1536 DPRINTF("read: %s", strerror(err));
1540 p = (MDB_page *)page;
1542 if (!F_ISSET(p->mp_flags, P_META)) {
1543 DPRINTF("page %lu not a meta page", p->mp_pgno);
1548 if (m->mm_magic != MDB_MAGIC) {
1549 DPUTS("meta has invalid magic");
1553 if (m->mm_version != MDB_VERSION) {
1554 DPRINTF("database is version %u, expected version %u",
1555 m->mm_version, MDB_VERSION);
1556 return MDB_VERSION_MISMATCH;
1559 memcpy(meta, m, sizeof(*m));
1564 mdb_env_init_meta(MDB_env *env, MDB_meta *meta)
1571 DPUTS("writing new meta page");
1575 meta->mm_magic = MDB_MAGIC;
1576 meta->mm_version = MDB_VERSION;
1577 meta->mm_psize = psize;
1578 meta->mm_last_pg = 1;
1579 meta->mm_flags = env->me_flags & 0xffff;
1580 meta->mm_flags |= MDB_INTEGERKEY;
1581 meta->mm_dbs[0].md_root = P_INVALID;
1582 meta->mm_dbs[1].md_root = P_INVALID;
1584 p = calloc(2, psize);
1586 p->mp_flags = P_META;
1589 memcpy(m, meta, sizeof(*meta));
1591 q = (MDB_page *)((char *)p + psize);
1594 q->mp_flags = P_META;
1597 memcpy(m, meta, sizeof(*meta));
1602 rc = WriteFile(env->me_fd, p, psize * 2, &len, NULL);
1603 rc = (len == psize * 2) ? MDB_SUCCESS : ErrCode();
1606 rc = write(env->me_fd, p, psize * 2);
1607 rc = (rc == (int)psize * 2) ? MDB_SUCCESS : ErrCode();
1614 mdb_env_write_meta(MDB_txn *txn)
1617 MDB_meta meta, metab;
1619 int rc, len, toggle;
1625 assert(txn != NULL);
1626 assert(txn->mt_env != NULL);
1628 toggle = !txn->mt_toggle;
1629 DPRINTF("writing meta page %d for root page %lu",
1630 toggle, txn->mt_dbs[MAIN_DBI].md_root);
1634 metab.mm_txnid = env->me_metas[toggle]->mm_txnid;
1635 metab.mm_last_pg = env->me_metas[toggle]->mm_last_pg;
1637 ptr = (char *)&meta;
1638 off = offsetof(MDB_meta, mm_dbs[0].md_depth);
1639 len = sizeof(MDB_meta) - off;
1642 meta.mm_dbs[0] = txn->mt_dbs[0];
1643 meta.mm_dbs[1] = txn->mt_dbs[1];
1644 meta.mm_last_pg = txn->mt_next_pgno - 1;
1645 meta.mm_txnid = txn->mt_txnid;
1648 off += env->me_psize;
1651 /* Write to the SYNC fd */
1654 memset(&ov, 0, sizeof(ov));
1656 WriteFile(env->me_mfd, ptr, len, (DWORD *)&rc, &ov);
1659 rc = pwrite(env->me_mfd, ptr, len, off);
1664 DPUTS("write failed, disk error?");
1665 /* On a failure, the pagecache still contains the new data.
1666 * Write some old data back, to prevent it from being used.
1667 * Use the non-SYNC fd; we know it will fail anyway.
1669 meta.mm_last_pg = metab.mm_last_pg;
1670 meta.mm_txnid = metab.mm_txnid;
1672 WriteFile(env->me_fd, ptr, len, NULL, &ov);
1674 r2 = pwrite(env->me_fd, ptr, len, off);
1676 env->me_flags |= MDB_FATAL_ERROR;
1679 /* Memory ordering issues are irrelevant; since the entire writer
1680 * is wrapped by wmutex, all of these changes will become visible
1681 * after the wmutex is unlocked. Since the DB is multi-version,
1682 * readers will get consistent data regardless of how fresh or
1683 * how stale their view of these values is.
1685 LAZY_MUTEX_LOCK(&env->me_txns->mti_mutex);
1686 txn->mt_env->me_txns->mti_me_toggle = toggle;
1687 txn->mt_env->me_txns->mti_txnid = txn->mt_txnid;
1688 LAZY_MUTEX_UNLOCK(&env->me_txns->mti_mutex);
1694 mdb_env_read_meta(MDB_env *env, int *which)
1698 assert(env != NULL);
1700 if (env->me_metas[0]->mm_txnid < env->me_metas[1]->mm_txnid)
1703 DPRINTF("Using meta page %d", toggle);
1710 mdb_env_create(MDB_env **env)
1714 e = calloc(1, sizeof(MDB_env));
1715 if (!e) return ENOMEM;
1717 e->me_maxreaders = DEFAULT_READERS;
1719 e->me_fd = INVALID_HANDLE_VALUE;
1720 e->me_lfd = INVALID_HANDLE_VALUE;
1721 e->me_mfd = INVALID_HANDLE_VALUE;
1727 mdb_env_set_mapsize(MDB_env *env, size_t size)
1731 env->me_mapsize = size;
1736 mdb_env_set_maxdbs(MDB_env *env, int dbs)
1740 env->me_maxdbs = dbs;
1745 mdb_env_set_maxreaders(MDB_env *env, int readers)
1749 env->me_maxreaders = readers;
1754 mdb_env_get_maxreaders(MDB_env *env, int *readers)
1756 if (!env || !readers)
1758 *readers = env->me_maxreaders;
1763 mdb_env_open2(MDB_env *env, unsigned int flags)
1765 int i, newenv = 0, toggle;
1769 env->me_flags = flags;
1771 memset(&meta, 0, sizeof(meta));
1773 if ((i = mdb_env_read_header(env, &meta)) != 0) {
1776 DPUTS("new mdbenv");
1780 if (!env->me_mapsize) {
1781 env->me_mapsize = newenv ? DEFAULT_MAPSIZE : meta.mm_mapsize;
1787 LONG sizelo, sizehi;
1788 sizelo = env->me_mapsize & 0xffffffff;
1789 sizehi = env->me_mapsize >> 16; /* pointless on WIN32, only needed on W64 */
1791 /* Windows won't create mappings for zero length files.
1792 * Just allocate the maxsize right now.
1795 SetFilePointer(env->me_fd, sizelo, sizehi ? &sizehi : NULL, 0);
1796 if (!SetEndOfFile(env->me_fd))
1798 SetFilePointer(env->me_fd, 0, NULL, 0);
1800 mh = CreateFileMapping(env->me_fd, NULL, PAGE_READONLY,
1801 sizehi, sizelo, NULL);
1804 env->me_map = MapViewOfFileEx(mh, FILE_MAP_READ, 0, 0, env->me_mapsize,
1812 if (meta.mm_address && (flags & MDB_FIXEDMAP))
1814 env->me_map = mmap(meta.mm_address, env->me_mapsize, PROT_READ, i,
1816 if (env->me_map == MAP_FAILED)
1821 meta.mm_mapsize = env->me_mapsize;
1822 if (flags & MDB_FIXEDMAP)
1823 meta.mm_address = env->me_map;
1824 i = mdb_env_init_meta(env, &meta);
1825 if (i != MDB_SUCCESS) {
1826 munmap(env->me_map, env->me_mapsize);
1830 env->me_psize = meta.mm_psize;
1832 env->me_maxpg = env->me_mapsize / env->me_psize;
1834 p = (MDB_page *)env->me_map;
1835 env->me_metas[0] = METADATA(p);
1836 env->me_metas[1] = (MDB_meta *)((char *)env->me_metas[0] + meta.mm_psize);
1838 if ((i = mdb_env_read_meta(env, &toggle)) != 0)
1841 DPRINTF("opened database version %u, pagesize %u",
1842 env->me_metas[toggle]->mm_version, env->me_psize);
1843 DPRINTF("depth: %u", env->me_metas[toggle]->mm_dbs[MAIN_DBI].md_depth);
1844 DPRINTF("entries: %lu", env->me_metas[toggle]->mm_dbs[MAIN_DBI].md_entries);
1845 DPRINTF("branch pages: %lu", env->me_metas[toggle]->mm_dbs[MAIN_DBI].md_branch_pages);
1846 DPRINTF("leaf pages: %lu", env->me_metas[toggle]->mm_dbs[MAIN_DBI].md_leaf_pages);
1847 DPRINTF("overflow pages: %lu", env->me_metas[toggle]->mm_dbs[MAIN_DBI].md_overflow_pages);
1848 DPRINTF("root: %lu", env->me_metas[toggle]->mm_dbs[MAIN_DBI].md_root);
1854 /* Windows doesn't support destructor callbacks for thread-specific storage */
1856 mdb_env_reader_dest(void *ptr)
1858 MDB_reader *reader = ptr;
1860 reader->mr_txnid = 0;
1866 /* downgrade the exclusive lock on the region back to shared */
1868 mdb_env_share_locks(MDB_env *env)
1872 if (env->me_metas[0]->mm_txnid < env->me_metas[1]->mm_txnid)
1874 env->me_txns->mti_me_toggle = toggle;
1875 env->me_txns->mti_txnid = env->me_metas[toggle]->mm_txnid;
1880 /* First acquire a shared lock. The Unlock will
1881 * then release the existing exclusive lock.
1883 memset(&ov, 0, sizeof(ov));
1884 LockFileEx(env->me_lfd, 0, 0, 1, 0, &ov);
1885 UnlockFile(env->me_lfd, 0, 0, 1, 0);
1889 struct flock lock_info;
1890 /* The shared lock replaces the existing lock */
1891 memset((void *)&lock_info, 0, sizeof(lock_info));
1892 lock_info.l_type = F_RDLCK;
1893 lock_info.l_whence = SEEK_SET;
1894 lock_info.l_start = 0;
1895 lock_info.l_len = 1;
1896 fcntl(env->me_lfd, F_SETLK, &lock_info);
1902 mdb_env_setup_locks(MDB_env *env, char *lpath, int mode, int *excl)
1910 if ((env->me_lfd = CreateFile(lpath, GENERIC_READ|GENERIC_WRITE,
1911 FILE_SHARE_READ|FILE_SHARE_WRITE, NULL, OPEN_ALWAYS,
1912 FILE_ATTRIBUTE_NORMAL, NULL)) == INVALID_HANDLE_VALUE) {
1916 /* Try to get exclusive lock. If we succeed, then
1917 * nobody is using the lock region and we should initialize it.
1920 if (LockFile(env->me_lfd, 0, 0, 1, 0)) {
1924 memset(&ov, 0, sizeof(ov));
1925 if (!LockFileEx(env->me_lfd, 0, 0, 1, 0, &ov)) {
1931 size = GetFileSize(env->me_lfd, NULL);
1933 if ((env->me_lfd = open(lpath, O_RDWR|O_CREAT, mode)) == -1) {
1937 /* Try to get exclusive lock. If we succeed, then
1938 * nobody is using the lock region and we should initialize it.
1941 struct flock lock_info;
1942 memset((void *)&lock_info, 0, sizeof(lock_info));
1943 lock_info.l_type = F_WRLCK;
1944 lock_info.l_whence = SEEK_SET;
1945 lock_info.l_start = 0;
1946 lock_info.l_len = 1;
1947 rc = fcntl(env->me_lfd, F_SETLK, &lock_info);
1951 lock_info.l_type = F_RDLCK;
1952 rc = fcntl(env->me_lfd, F_SETLKW, &lock_info);
1959 size = lseek(env->me_lfd, 0, SEEK_END);
1961 rsize = (env->me_maxreaders-1) * sizeof(MDB_reader) + sizeof(MDB_txninfo);
1962 if (size < rsize && *excl) {
1964 SetFilePointer(env->me_lfd, rsize, NULL, 0);
1965 if (!SetEndOfFile(env->me_lfd)) {
1970 if (ftruncate(env->me_lfd, rsize) != 0) {
1977 size = rsize - sizeof(MDB_txninfo);
1978 env->me_maxreaders = size/sizeof(MDB_reader) + 1;
1983 mh = CreateFileMapping(env->me_lfd, NULL, PAGE_READWRITE,
1989 env->me_txns = MapViewOfFileEx(mh, FILE_MAP_WRITE, 0, 0, rsize, NULL);
1991 if (!env->me_txns) {
1997 env->me_txns = mmap(0, rsize, PROT_READ|PROT_WRITE, MAP_SHARED,
1999 if (env->me_txns == MAP_FAILED) {
2007 if (!mdb_sec_inited) {
2008 InitializeSecurityDescriptor(&mdb_null_sd,
2009 SECURITY_DESCRIPTOR_REVISION);
2010 SetSecurityDescriptorDacl(&mdb_null_sd, TRUE, 0, FALSE);
2011 mdb_all_sa.nLength = sizeof(SECURITY_ATTRIBUTES);
2012 mdb_all_sa.bInheritHandle = FALSE;
2013 mdb_all_sa.lpSecurityDescriptor = &mdb_null_sd;
2016 /* FIXME: only using up to 20 characters of the env path here,
2017 * probably not enough to assure uniqueness...
2019 sprintf(env->me_txns->mti_rmname, "Global\\MDBr%.20s", lpath);
2020 ptr = env->me_txns->mti_rmname + sizeof("Global\\MDBr");
2021 while ((ptr = strchr(ptr, '\\')))
2023 env->me_rmutex = CreateMutex(&mdb_all_sa, FALSE, env->me_txns->mti_rmname);
2024 if (!env->me_rmutex) {
2028 sprintf(env->me_txns->mti_rmname, "Global\\MDBw%.20s", lpath);
2029 ptr = env->me_txns->mti_rmname + sizeof("Global\\MDBw");
2030 while ((ptr = strchr(ptr, '\\')))
2032 env->me_wmutex = CreateMutex(&mdb_all_sa, FALSE, env->me_txns->mti_rmname);
2033 if (!env->me_wmutex) {
2038 pthread_mutexattr_t mattr;
2040 pthread_mutexattr_init(&mattr);
2041 rc = pthread_mutexattr_setpshared(&mattr, PTHREAD_PROCESS_SHARED);
2045 pthread_mutex_init(&env->me_txns->mti_mutex, &mattr);
2046 pthread_mutex_init(&env->me_txns->mti_wmutex, &mattr);
2048 env->me_txns->mti_version = MDB_VERSION;
2049 env->me_txns->mti_magic = MDB_MAGIC;
2050 env->me_txns->mti_txnid = 0;
2051 env->me_txns->mti_numreaders = 0;
2052 env->me_txns->mti_me_toggle = 0;
2055 if (env->me_txns->mti_magic != MDB_MAGIC) {
2056 DPUTS("lock region has invalid magic");
2060 if (env->me_txns->mti_version != MDB_VERSION) {
2061 DPRINTF("lock region is version %u, expected version %u",
2062 env->me_txns->mti_version, MDB_VERSION);
2063 rc = MDB_VERSION_MISMATCH;
2067 if (rc != EACCES && rc != EAGAIN) {
2071 env->me_rmutex = OpenMutex(SYNCHRONIZE, FALSE, env->me_txns->mti_rmname);
2072 if (!env->me_rmutex) {
2076 env->me_wmutex = OpenMutex(SYNCHRONIZE, FALSE, env->me_txns->mti_wmname);
2077 if (!env->me_wmutex) {
2087 env->me_lfd = INVALID_HANDLE_VALUE;
2092 /** The name of the lock file in the DB environment */
2093 #define LOCKNAME "/lock.mdb"
2094 /** The name of the data file in the DB environment */
2095 #define DATANAME "/data.mdb"
2097 mdb_env_open(MDB_env *env, const char *path, unsigned int flags, mode_t mode)
2099 int oflags, rc, len, excl;
2100 char *lpath, *dpath;
2103 lpath = malloc(len + sizeof(LOCKNAME) + len + sizeof(DATANAME));
2106 dpath = lpath + len + sizeof(LOCKNAME);
2107 sprintf(lpath, "%s" LOCKNAME, path);
2108 sprintf(dpath, "%s" DATANAME, path);
2110 rc = mdb_env_setup_locks(env, lpath, mode, &excl);
2115 if (F_ISSET(flags, MDB_RDONLY)) {
2116 oflags = GENERIC_READ;
2117 len = OPEN_EXISTING;
2119 oflags = GENERIC_READ|GENERIC_WRITE;
2122 mode = FILE_ATTRIBUTE_NORMAL;
2123 if ((env->me_fd = CreateFile(dpath, oflags, FILE_SHARE_READ|FILE_SHARE_WRITE,
2124 NULL, len, mode, NULL)) == INVALID_HANDLE_VALUE) {
2129 if (F_ISSET(flags, MDB_RDONLY))
2132 oflags = O_RDWR | O_CREAT;
2134 if ((env->me_fd = open(dpath, oflags, mode)) == -1) {
2140 if ((rc = mdb_env_open2(env, flags)) == MDB_SUCCESS) {
2141 /* synchronous fd for meta writes */
2143 if (!(flags & (MDB_RDONLY|MDB_NOSYNC)))
2144 mode |= FILE_FLAG_WRITE_THROUGH;
2145 if ((env->me_mfd = CreateFile(dpath, oflags, FILE_SHARE_READ|FILE_SHARE_WRITE,
2146 NULL, len, mode, NULL)) == INVALID_HANDLE_VALUE) {
2151 if (!(flags & (MDB_RDONLY|MDB_NOSYNC)))
2152 oflags |= MDB_DSYNC;
2153 if ((env->me_mfd = open(dpath, oflags, mode)) == -1) {
2158 env->me_path = strdup(path);
2159 DPRINTF("opened dbenv %p", (void *) env);
2160 pthread_key_create(&env->me_txkey, mdb_env_reader_dest);
2161 LAZY_RWLOCK_INIT(&env->me_dblock, NULL);
2163 mdb_env_share_locks(env);
2164 env->me_dbxs = calloc(env->me_maxdbs, sizeof(MDB_dbx));
2165 env->me_dbs[0] = calloc(env->me_maxdbs, sizeof(MDB_db));
2166 env->me_dbs[1] = calloc(env->me_maxdbs, sizeof(MDB_db));
2172 if (env->me_fd != INVALID_HANDLE_VALUE) {
2174 env->me_fd = INVALID_HANDLE_VALUE;
2176 if (env->me_lfd != INVALID_HANDLE_VALUE) {
2178 env->me_lfd = INVALID_HANDLE_VALUE;
2186 mdb_env_close(MDB_env *env)
2193 while (env->me_dpages) {
2194 dp = env->me_dpages;
2195 env->me_dpages = dp->mp_next;
2199 free(env->me_dbs[1]);
2200 free(env->me_dbs[0]);
2204 LAZY_RWLOCK_DESTROY(&env->me_dblock);
2205 pthread_key_delete(env->me_txkey);
2208 munmap(env->me_map, env->me_mapsize);
2213 pid_t pid = getpid();
2215 for (i=0; i<env->me_txns->mti_numreaders; i++)
2216 if (env->me_txns->mti_readers[i].mr_pid == pid)
2217 env->me_txns->mti_readers[i].mr_pid = 0;
2218 munmap(env->me_txns, (env->me_maxreaders-1)*sizeof(MDB_reader)+sizeof(MDB_txninfo));
2224 /* only for aligned ints */
2226 intcmp(const MDB_val *a, const MDB_val *b)
2228 if (a->mv_size == sizeof(long))
2230 unsigned long *la, *lb;
2235 unsigned int *ia, *ib;
2242 /* ints must always be the same size */
2244 cintcmp(const MDB_val *a, const MDB_val *b)
2246 #if __BYTE_ORDER == __LITTLE_ENDIAN
2247 unsigned short *u, *c;
2250 u = a->mv_data + a->mv_size;
2251 c = b->mv_data + a->mv_size;
2254 } while(!x && u > (unsigned short *)a->mv_data);
2257 return memcmp(a->mv_data, b->mv_data, a->mv_size);
2262 memncmp(const MDB_val *a, const MDB_val *b)
2268 len_diff = a->mv_size - b->mv_size;
2271 diff = memcmp(a->mv_data, b->mv_data, len);
2272 return diff ? diff : len_diff;
2276 memnrcmp(const MDB_val *a, const MDB_val *b)
2278 const unsigned char *p1, *p2, *p1_lim;
2281 if (b->mv_size == 0)
2282 return a->mv_size != 0;
2283 if (a->mv_size == 0)
2286 p1 = (const unsigned char *)a->mv_data + a->mv_size - 1;
2287 p2 = (const unsigned char *)b->mv_data + b->mv_size - 1;
2289 len_diff = a->mv_size - b->mv_size;
2291 p1_lim = p1 - a->mv_size;
2293 p1_lim = p1 - b->mv_size;
2295 while (p1 > p1_lim) {
2305 /* Search for key within a leaf page, using binary search.
2306 * Returns the smallest entry larger or equal to the key.
2307 * If exactp is non-null, stores whether the found entry was an exact match
2308 * in *exactp (1 or 0).
2309 * If kip is non-null, stores the index of the found entry in *kip.
2310 * If no entry larger or equal to the key is found, returns NULL.
2313 mdb_search_node(MDB_cursor *mc, MDB_val *key, int *exactp)
2315 unsigned int i = 0, nkeys;
2318 MDB_page *mp = mc->mc_pg[mc->mc_top];
2319 MDB_node *node = NULL;
2324 nkeys = NUMKEYS(mp);
2326 DPRINTF("searching %u keys in %s page %lu",
2327 nkeys, IS_LEAF(mp) ? "leaf" : "branch",
2332 low = IS_LEAF(mp) ? 0 : 1;
2334 cmp = mc->mc_txn->mt_dbxs[mc->mc_dbi].md_cmp;
2336 nodekey.mv_size = mc->mc_txn->mt_dbs[mc->mc_dbi].md_pad;
2337 node = NODEPTR(mp, 0); /* fake */
2339 while (low <= high) {
2340 i = (low + high) >> 1;
2343 nodekey.mv_data = LEAF2KEY(mp, i, nodekey.mv_size);
2345 node = NODEPTR(mp, i);
2347 nodekey.mv_size = node->mn_ksize;
2348 nodekey.mv_data = NODEKEY(node);
2351 rc = cmp(key, &nodekey);
2355 DPRINTF("found leaf index %u [%s], rc = %i",
2356 i, DKEY(&nodekey), rc);
2358 DPRINTF("found branch index %u [%s -> %lu], rc = %i",
2359 i, DKEY(&nodekey), NODEPGNO(node), rc);
2370 if (rc > 0) { /* Found entry is less than the key. */
2371 i++; /* Skip to get the smallest entry larger than key. */
2373 node = NODEPTR(mp, i);
2376 *exactp = (rc == 0);
2377 /* store the key index */
2378 mc->mc_ki[mc->mc_top] = i;
2380 /* There is no entry larger or equal to the key. */
2383 /* nodeptr is fake for LEAF2 */
2388 cursor_pop_page(MDB_cursor *mc)
2393 top = mc->mc_pg[mc->mc_top];
2398 DPRINTF("popped page %lu off db %u cursor %p", top->mp_pgno,
2399 mc->mc_dbi, (void *) mc);
2404 cursor_push_page(MDB_cursor *mc, MDB_page *mp)
2406 DPRINTF("pushing page %lu on db %u cursor %p", mp->mp_pgno,
2407 mc->mc_dbi, (void *) mc);
2409 if (mc->mc_snum >= CURSOR_STACK)
2412 mc->mc_top = mc->mc_snum++;
2413 mc->mc_pg[mc->mc_top] = mp;
2414 mc->mc_ki[mc->mc_top] = 0;
2420 mdb_get_page(MDB_txn *txn, pgno_t pgno, MDB_page **ret)
2424 if (!F_ISSET(txn->mt_flags, MDB_TXN_RDONLY) && txn->mt_u.dirty_list[0].mid) {
2426 x = mdb_mid2l_search(txn->mt_u.dirty_list, pgno);
2427 if (x <= txn->mt_u.dirty_list[0].mid && txn->mt_u.dirty_list[x].mid == pgno) {
2428 p = txn->mt_u.dirty_list[x].mptr;
2432 if (pgno <= txn->mt_env->me_metas[txn->mt_toggle]->mm_last_pg)
2433 p = (MDB_page *)(txn->mt_env->me_map + txn->mt_env->me_psize * pgno);
2437 DPRINTF("page %lu not found", pgno);
2440 return (p != NULL) ? MDB_SUCCESS : MDB_PAGE_NOTFOUND;
2444 mdb_search_page_root(MDB_cursor *mc, MDB_val *key, int modify)
2446 MDB_page *mp = mc->mc_pg[mc->mc_top];
2451 while (IS_BRANCH(mp)) {
2454 DPRINTF("branch page %lu has %u keys", mp->mp_pgno, NUMKEYS(mp));
2455 assert(NUMKEYS(mp) > 1);
2456 DPRINTF("found index 0 to page %lu", NODEPGNO(NODEPTR(mp, 0)));
2458 if (key == NULL) /* Initialize cursor to first page. */
2459 mc->mc_ki[mc->mc_top] = 0;
2460 else if (key->mv_size > MAXKEYSIZE && key->mv_data == NULL) {
2461 /* cursor to last page */
2462 mc->mc_ki[mc->mc_top] = NUMKEYS(mp)-1;
2465 node = mdb_search_node(mc, key, &exact);
2467 mc->mc_ki[mc->mc_top] = NUMKEYS(mp) - 1;
2469 assert(mc->mc_ki[mc->mc_top] > 0);
2470 mc->mc_ki[mc->mc_top]--;
2475 DPRINTF("following index %u for key [%s]",
2476 mc->mc_ki[mc->mc_top], DKEY(key));
2477 assert(mc->mc_ki[mc->mc_top] < NUMKEYS(mp));
2478 node = NODEPTR(mp, mc->mc_ki[mc->mc_top]);
2480 if ((rc = mdb_get_page(mc->mc_txn, NODEPGNO(node), &mp)))
2483 if ((rc = cursor_push_page(mc, mp)))
2487 if ((rc = mdb_touch(mc)) != 0)
2489 mp = mc->mc_pg[mc->mc_top];
2494 DPRINTF("internal error, index points to a %02X page!?",
2496 return MDB_CORRUPTED;
2499 DPRINTF("found leaf page %lu for key [%s]", mp->mp_pgno,
2500 key ? DKEY(key) : NULL);
2505 /* Search for the page a given key should be in.
2506 * Pushes parent pages on the cursor stack.
2507 * If key is NULL, search for the lowest page (used by mdb_cursor_first).
2508 * If modify is true, visited pages are updated with new page numbers.
2511 mdb_search_page(MDB_cursor *mc, MDB_val *key, int modify)
2516 /* Make sure the txn is still viable, then find the root from
2517 * the txn's db table.
2519 if (F_ISSET(mc->mc_txn->mt_flags, MDB_TXN_ERROR)) {
2520 DPUTS("transaction has failed, must abort");
2523 root = mc->mc_txn->mt_dbs[mc->mc_dbi].md_root;
2525 if (root == P_INVALID) { /* Tree is empty. */
2526 DPUTS("tree is empty");
2527 return MDB_NOTFOUND;
2530 if ((rc = mdb_get_page(mc->mc_txn, root, &mc->mc_pg[0])))
2536 DPRINTF("db %u root page %lu has flags 0x%X",
2537 mc->mc_dbi, root, mc->mc_pg[0]->mp_flags);
2540 /* For sub-databases, update main root first */
2541 if (mc->mc_dbi > MAIN_DBI && !mc->mc_txn->mt_dbxs[mc->mc_dbi].md_dirty) {
2543 mc2.mc_txn = mc->mc_txn;
2544 mc2.mc_dbi = MAIN_DBI;
2545 rc = mdb_search_page(&mc2, &mc->mc_txn->mt_dbxs[mc->mc_dbi].md_name, 1);
2548 mc->mc_txn->mt_dbxs[mc->mc_dbi].md_dirty = 1;
2550 if (!F_ISSET(mc->mc_pg[0]->mp_flags, P_DIRTY)) {
2551 if ((rc = mdb_touch(mc)))
2553 mc->mc_txn->mt_dbs[mc->mc_dbi].md_root = mc->mc_pg[0]->mp_pgno;
2557 return mdb_search_page_root(mc, key, modify);
2561 mdb_read_data(MDB_txn *txn, MDB_node *leaf, MDB_val *data)
2563 MDB_page *omp; /* overflow mpage */
2567 if (!F_ISSET(leaf->mn_flags, F_BIGDATA)) {
2568 data->mv_size = NODEDSZ(leaf);
2569 data->mv_data = NODEDATA(leaf);
2573 /* Read overflow data.
2575 data->mv_size = NODEDSZ(leaf);
2576 memcpy(&pgno, NODEDATA(leaf), sizeof(pgno));
2577 if ((rc = mdb_get_page(txn, pgno, &omp))) {
2578 DPRINTF("read overflow page %lu failed", pgno);
2581 data->mv_data = METADATA(omp);
2587 mdb_get(MDB_txn *txn, MDB_dbi dbi,
2588 MDB_val *key, MDB_val *data)
2597 DPRINTF("===> get db %u key [%s]", dbi, DKEY(key));
2599 if (txn == NULL || !dbi || dbi >= txn->mt_numdbs)
2602 if (key->mv_size == 0 || key->mv_size > MAXKEYSIZE) {
2609 if (txn->mt_dbs[dbi].md_flags & MDB_DUPSORT) {
2610 mc.mc_xcursor = &mx;
2611 mdb_xcursor_init0(&mc);
2613 mc.mc_xcursor = NULL;
2615 return mdb_cursor_set(&mc, key, data, MDB_SET, &exact);
2619 mdb_sibling(MDB_cursor *mc, int move_right)
2626 if (mc->mc_snum < 2) {
2627 return MDB_NOTFOUND; /* root has no siblings */
2629 ptop = mc->mc_top-1;
2631 DPRINTF("parent page is page %lu, index %u",
2632 mc->mc_pg[ptop]->mp_pgno, mc->mc_ki[ptop]);
2634 cursor_pop_page(mc);
2635 if (move_right ? (mc->mc_ki[ptop] + 1u >= NUMKEYS(mc->mc_pg[ptop]))
2636 : (mc->mc_ki[ptop] == 0)) {
2637 DPRINTF("no more keys left, moving to %s sibling",
2638 move_right ? "right" : "left");
2639 if ((rc = mdb_sibling(mc, move_right)) != MDB_SUCCESS)
2646 DPRINTF("just moving to %s index key %u",
2647 move_right ? "right" : "left", mc->mc_ki[ptop]);
2649 assert(IS_BRANCH(mc->mc_pg[ptop]));
2651 indx = NODEPTR(mc->mc_pg[ptop], mc->mc_ki[ptop]);
2652 if ((rc = mdb_get_page(mc->mc_txn, NODEPGNO(indx), &mp)))
2655 cursor_push_page(mc, mp);
2661 mdb_cursor_next(MDB_cursor *mc, MDB_val *key, MDB_val *data, MDB_cursor_op op)
2667 if (mc->mc_flags & C_EOF) {
2668 return MDB_NOTFOUND;
2671 assert(mc->mc_flags & C_INITIALIZED);
2673 mp = mc->mc_pg[mc->mc_top];
2675 if (mc->mc_txn->mt_dbs[mc->mc_dbi].md_flags & MDB_DUPSORT) {
2676 leaf = NODEPTR(mp, mc->mc_ki[mc->mc_top]);
2677 if (F_ISSET(leaf->mn_flags, F_DUPDATA)) {
2678 if (op == MDB_NEXT || op == MDB_NEXT_DUP) {
2679 rc = mdb_cursor_next(&mc->mc_xcursor->mx_cursor, data, NULL, MDB_NEXT);
2680 if (op != MDB_NEXT || rc == MDB_SUCCESS)
2684 mc->mc_xcursor->mx_cursor.mc_flags = 0;
2685 if (op == MDB_NEXT_DUP)
2686 return MDB_NOTFOUND;
2690 DPRINTF("cursor_next: top page is %lu in cursor %p", mp->mp_pgno, (void *) mc);
2692 if (mc->mc_ki[mc->mc_top] + 1u >= NUMKEYS(mp)) {
2693 DPUTS("=====> move to next sibling page");
2694 if (mdb_sibling(mc, 1) != MDB_SUCCESS) {
2695 mc->mc_flags |= C_EOF;
2696 return MDB_NOTFOUND;
2698 mp = mc->mc_pg[mc->mc_top];
2699 DPRINTF("next page is %lu, key index %u", mp->mp_pgno, mc->mc_ki[mc->mc_top]);
2701 mc->mc_ki[mc->mc_top]++;
2703 DPRINTF("==> cursor points to page %lu with %u keys, key index %u",
2704 mp->mp_pgno, NUMKEYS(mp), mc->mc_ki[mc->mc_top]);
2707 key->mv_size = mc->mc_txn->mt_dbs[mc->mc_dbi].md_pad;
2708 key->mv_data = LEAF2KEY(mp, mc->mc_ki[mc->mc_top], key->mv_size);
2712 assert(IS_LEAF(mp));
2713 leaf = NODEPTR(mp, mc->mc_ki[mc->mc_top]);
2715 if (F_ISSET(leaf->mn_flags, F_DUPDATA)) {
2716 mdb_xcursor_init1(mc, leaf);
2719 if ((rc = mdb_read_data(mc->mc_txn, leaf, data) != MDB_SUCCESS))
2722 if (F_ISSET(leaf->mn_flags, F_DUPDATA)) {
2723 rc = mdb_cursor_first(&mc->mc_xcursor->mx_cursor, data, NULL);
2724 if (rc != MDB_SUCCESS)
2729 MDB_SET_KEY(leaf, key);
2734 mdb_cursor_prev(MDB_cursor *mc, MDB_val *key, MDB_val *data, MDB_cursor_op op)
2740 assert(mc->mc_flags & C_INITIALIZED);
2742 mp = mc->mc_pg[mc->mc_top];
2744 if (mc->mc_txn->mt_dbs[mc->mc_dbi].md_flags & MDB_DUPSORT) {
2745 leaf = NODEPTR(mp, mc->mc_ki[mc->mc_top]);
2746 if (op == MDB_PREV || op == MDB_PREV_DUP) {
2747 if (F_ISSET(leaf->mn_flags, F_DUPDATA)) {
2748 rc = mdb_cursor_prev(&mc->mc_xcursor->mx_cursor, data, NULL, MDB_PREV);
2749 if (op != MDB_PREV || rc == MDB_SUCCESS)
2752 mc->mc_xcursor->mx_cursor.mc_flags = 0;
2753 if (op == MDB_PREV_DUP)
2754 return MDB_NOTFOUND;
2759 DPRINTF("cursor_prev: top page is %lu in cursor %p", mp->mp_pgno, (void *) mc);
2761 if (mc->mc_ki[mc->mc_top] == 0) {
2762 DPUTS("=====> move to prev sibling page");
2763 if (mdb_sibling(mc, 0) != MDB_SUCCESS) {
2764 mc->mc_flags &= ~C_INITIALIZED;
2765 return MDB_NOTFOUND;
2767 mp = mc->mc_pg[mc->mc_top];
2768 mc->mc_ki[mc->mc_top] = NUMKEYS(mp) - 1;
2769 DPRINTF("prev page is %lu, key index %u", mp->mp_pgno, mc->mc_ki[mc->mc_top]);
2771 mc->mc_ki[mc->mc_top]--;
2773 mc->mc_flags &= ~C_EOF;
2775 DPRINTF("==> cursor points to page %lu with %u keys, key index %u",
2776 mp->mp_pgno, NUMKEYS(mp), mc->mc_ki[mc->mc_top]);
2779 key->mv_size = mc->mc_txn->mt_dbs[mc->mc_dbi].md_pad;
2780 key->mv_data = LEAF2KEY(mp, mc->mc_ki[mc->mc_top], key->mv_size);
2784 assert(IS_LEAF(mp));
2785 leaf = NODEPTR(mp, mc->mc_ki[mc->mc_top]);
2787 if (F_ISSET(leaf->mn_flags, F_DUPDATA)) {
2788 mdb_xcursor_init1(mc, leaf);
2791 if ((rc = mdb_read_data(mc->mc_txn, leaf, data) != MDB_SUCCESS))
2794 if (F_ISSET(leaf->mn_flags, F_DUPDATA)) {
2795 rc = mdb_cursor_last(&mc->mc_xcursor->mx_cursor, data, NULL);
2796 if (rc != MDB_SUCCESS)
2801 MDB_SET_KEY(leaf, key);
2806 mdb_cursor_set(MDB_cursor *mc, MDB_val *key, MDB_val *data,
2807 MDB_cursor_op op, int *exactp)
2815 assert(key->mv_size > 0);
2817 /* See if we're already on the right page */
2818 if (mc->mc_flags & C_INITIALIZED) {
2821 if (mc->mc_pg[mc->mc_top]->mp_flags & P_LEAF2) {
2822 nodekey.mv_size = mc->mc_txn->mt_dbs[mc->mc_dbi].md_pad;
2823 nodekey.mv_data = LEAF2KEY(mc->mc_pg[mc->mc_top], 0, nodekey.mv_size);
2825 leaf = NODEPTR(mc->mc_pg[mc->mc_top], 0);
2826 MDB_SET_KEY(leaf, &nodekey);
2828 rc = mc->mc_txn->mt_dbxs[mc->mc_dbi].md_cmp(key, &nodekey);
2830 /* Probably happens rarely, but first node on the page
2831 * was the one we wanted.
2833 mc->mc_ki[mc->mc_top] = 0;
2838 leaf = NODEPTR(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
2843 if (NUMKEYS(mc->mc_pg[mc->mc_top]) > 1) {
2844 if (mc->mc_pg[mc->mc_top]->mp_flags & P_LEAF2) {
2845 nodekey.mv_data = LEAF2KEY(mc->mc_pg[mc->mc_top],
2846 NUMKEYS(mc->mc_pg[mc->mc_top])-1, nodekey.mv_size);
2848 leaf = NODEPTR(mc->mc_pg[mc->mc_top], NUMKEYS(mc->mc_pg[mc->mc_top])-1);
2849 MDB_SET_KEY(leaf, &nodekey);
2851 rc = mc->mc_txn->mt_dbxs[mc->mc_dbi].md_cmp(key, &nodekey);
2853 /* last node was the one we wanted */
2854 mc->mc_ki[mc->mc_top] = NUMKEYS(mc->mc_pg[mc->mc_top])-1;
2858 /* This is definitely the right page, skip search_page */
2863 /* If any parents have right-sibs, search.
2864 * Otherwise, there's nothing further.
2866 for (i=0; i<mc->mc_top; i++)
2868 NUMKEYS(mc->mc_pg[i])-1)
2870 if (i == mc->mc_top) {
2871 /* There are no other pages */
2872 mc->mc_ki[mc->mc_top] = NUMKEYS(mc->mc_pg[mc->mc_top]);
2873 return MDB_NOTFOUND;
2878 rc = mdb_search_page(mc, key, 0);
2879 if (rc != MDB_SUCCESS)
2882 assert(IS_LEAF(mc->mc_pg[mc->mc_top]));
2885 leaf = mdb_search_node(mc, key, exactp);
2886 if (exactp != NULL && !*exactp) {
2887 /* MDB_SET specified and not an exact match. */
2888 return MDB_NOTFOUND;
2892 DPUTS("===> inexact leaf not found, goto sibling");
2893 if ((rc = mdb_sibling(mc, 1)) != MDB_SUCCESS)
2894 return rc; /* no entries matched */
2895 mc->mc_ki[mc->mc_top] = 0;
2896 assert(IS_LEAF(mc->mc_pg[mc->mc_top]));
2897 leaf = NODEPTR(mc->mc_pg[mc->mc_top], 0);
2901 mc->mc_flags |= C_INITIALIZED;
2902 mc->mc_flags &= ~C_EOF;
2904 if (IS_LEAF2(mc->mc_pg[mc->mc_top])) {
2905 key->mv_size = mc->mc_txn->mt_dbs[mc->mc_dbi].md_pad;
2906 key->mv_data = LEAF2KEY(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top], key->mv_size);
2910 if (F_ISSET(leaf->mn_flags, F_DUPDATA)) {
2911 mdb_xcursor_init1(mc, leaf);
2914 if (F_ISSET(leaf->mn_flags, F_DUPDATA)) {
2915 if (op == MDB_SET || op == MDB_SET_RANGE) {
2916 rc = mdb_cursor_first(&mc->mc_xcursor->mx_cursor, data, NULL);
2919 if (op == MDB_GET_BOTH) {
2925 rc = mdb_cursor_set(&mc->mc_xcursor->mx_cursor, data, NULL, MDB_SET_RANGE, ex2p);
2926 if (rc != MDB_SUCCESS)
2929 } else if (op == MDB_GET_BOTH || op == MDB_GET_BOTH_RANGE) {
2931 if ((rc = mdb_read_data(mc->mc_txn, leaf, &d2)) != MDB_SUCCESS)
2933 rc = mc->mc_txn->mt_dbxs[mc->mc_dbi].md_dcmp(data, &d2);
2935 if (op == MDB_GET_BOTH || rc > 0)
2936 return MDB_NOTFOUND;
2940 if ((rc = mdb_read_data(mc->mc_txn, leaf, data)) != MDB_SUCCESS)
2945 /* The key already matches in all other cases */
2946 if (op == MDB_SET_RANGE)
2947 MDB_SET_KEY(leaf, key);
2948 DPRINTF("==> cursor placed on key [%s]", DKEY(key));
2954 mdb_cursor_first(MDB_cursor *mc, MDB_val *key, MDB_val *data)
2959 rc = mdb_search_page(mc, NULL, 0);
2960 if (rc != MDB_SUCCESS)
2962 assert(IS_LEAF(mc->mc_pg[mc->mc_top]));
2964 leaf = NODEPTR(mc->mc_pg[mc->mc_top], 0);
2965 mc->mc_flags |= C_INITIALIZED;
2966 mc->mc_flags &= ~C_EOF;
2968 if (IS_LEAF2(mc->mc_pg[mc->mc_top])) {
2969 key->mv_size = mc->mc_txn->mt_dbs[mc->mc_dbi].md_pad;
2970 key->mv_data = LEAF2KEY(mc->mc_pg[mc->mc_top], 0, key->mv_size);
2975 if (F_ISSET(leaf->mn_flags, F_DUPDATA)) {
2976 mdb_xcursor_init1(mc, leaf);
2977 rc = mdb_cursor_first(&mc->mc_xcursor->mx_cursor, data, NULL);
2982 mc->mc_xcursor->mx_cursor.mc_flags = 0;
2983 if ((rc = mdb_read_data(mc->mc_txn, leaf, data)) != MDB_SUCCESS)
2987 MDB_SET_KEY(leaf, key);
2992 mdb_cursor_last(MDB_cursor *mc, MDB_val *key, MDB_val *data)
2998 lkey.mv_size = MAXKEYSIZE+1;
2999 lkey.mv_data = NULL;
3001 rc = mdb_search_page(mc, &lkey, 0);
3002 if (rc != MDB_SUCCESS)
3004 assert(IS_LEAF(mc->mc_pg[mc->mc_top]));
3006 leaf = NODEPTR(mc->mc_pg[mc->mc_top], NUMKEYS(mc->mc_pg[mc->mc_top])-1);
3007 mc->mc_flags |= C_INITIALIZED;
3008 mc->mc_flags &= ~C_EOF;
3010 mc->mc_ki[mc->mc_top] = NUMKEYS(mc->mc_pg[mc->mc_top]) - 1;
3012 if (IS_LEAF2(mc->mc_pg[mc->mc_top])) {
3013 key->mv_size = mc->mc_txn->mt_dbs[mc->mc_dbi].md_pad;
3014 key->mv_data = LEAF2KEY(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top], key->mv_size);
3019 if (F_ISSET(leaf->mn_flags, F_DUPDATA)) {
3020 mdb_xcursor_init1(mc, leaf);
3021 rc = mdb_cursor_last(&mc->mc_xcursor->mx_cursor, data, NULL);
3025 if ((rc = mdb_read_data(mc->mc_txn, leaf, data)) != MDB_SUCCESS)
3030 MDB_SET_KEY(leaf, key);
3035 mdb_cursor_get(MDB_cursor *mc, MDB_val *key, MDB_val *data,
3045 case MDB_GET_BOTH_RANGE:
3046 if (data == NULL || mc->mc_xcursor == NULL) {
3053 if (key == NULL || key->mv_size == 0 || key->mv_size > MAXKEYSIZE) {
3055 } else if (op == MDB_SET_RANGE)
3056 rc = mdb_cursor_set(mc, key, data, op, NULL);
3058 rc = mdb_cursor_set(mc, key, data, op, &exact);
3060 case MDB_GET_MULTIPLE:
3062 !(mc->mc_txn->mt_dbs[mc->mc_dbi].md_flags & MDB_DUPFIXED) ||
3063 !(mc->mc_flags & C_INITIALIZED)) {
3068 if (!(mc->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED) ||
3069 (mc->mc_xcursor->mx_cursor.mc_flags & C_EOF))
3072 case MDB_NEXT_MULTIPLE:
3074 !(mc->mc_txn->mt_dbs[mc->mc_dbi].md_flags & MDB_DUPFIXED)) {
3078 if (!(mc->mc_flags & C_INITIALIZED))
3079 rc = mdb_cursor_first(mc, key, data);
3081 rc = mdb_cursor_next(mc, key, data, MDB_NEXT_DUP);
3082 if (rc == MDB_SUCCESS) {
3083 if (mc->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED) {
3086 mx = &mc->mc_xcursor->mx_cursor;
3087 data->mv_size = NUMKEYS(mx->mc_pg[mx->mc_top]) *
3088 mx->mc_txn->mt_dbs[mx->mc_dbi].md_pad;
3089 data->mv_data = METADATA(mx->mc_pg[mx->mc_top]);
3090 mx->mc_ki[mx->mc_top] = NUMKEYS(mx->mc_pg[mx->mc_top])-1;
3098 case MDB_NEXT_NODUP:
3099 if (!(mc->mc_flags & C_INITIALIZED))
3100 rc = mdb_cursor_first(mc, key, data);
3102 rc = mdb_cursor_next(mc, key, data, op);
3106 case MDB_PREV_NODUP:
3107 if (!(mc->mc_flags & C_INITIALIZED) || (mc->mc_flags & C_EOF))
3108 rc = mdb_cursor_last(mc, key, data);
3110 rc = mdb_cursor_prev(mc, key, data, op);
3113 rc = mdb_cursor_first(mc, key, data);
3117 !(mc->mc_txn->mt_dbs[mc->mc_dbi].md_flags & MDB_DUPSORT) ||
3118 !(mc->mc_flags & C_INITIALIZED) ||
3119 !(mc->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED)) {
3123 rc = mdb_cursor_first(&mc->mc_xcursor->mx_cursor, data, NULL);
3126 rc = mdb_cursor_last(mc, key, data);
3130 !(mc->mc_txn->mt_dbs[mc->mc_dbi].md_flags & MDB_DUPSORT) ||
3131 !(mc->mc_flags & C_INITIALIZED) ||
3132 !(mc->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED)) {
3136 rc = mdb_cursor_last(&mc->mc_xcursor->mx_cursor, data, NULL);
3139 DPRINTF("unhandled/unimplemented cursor operation %u", op);
3148 mdb_cursor_touch(MDB_cursor *mc)
3152 if (mc->mc_dbi > MAIN_DBI && !mc->mc_txn->mt_dbxs[mc->mc_dbi].md_dirty) {
3154 mc2.mc_txn = mc->mc_txn;
3155 mc2.mc_dbi = MAIN_DBI;
3156 rc = mdb_search_page(&mc2, &mc->mc_txn->mt_dbxs[mc->mc_dbi].md_name, 1);
3158 mc->mc_txn->mt_dbxs[mc->mc_dbi].md_dirty = 1;
3160 for (mc->mc_top = 0; mc->mc_top < mc->mc_snum; mc->mc_top++) {
3161 if (!F_ISSET(mc->mc_pg[mc->mc_top]->mp_flags, P_DIRTY)) {
3165 mc->mc_txn->mt_dbs[mc->mc_dbi].md_root =
3166 mc->mc_pg[mc->mc_top]->mp_pgno;
3170 mc->mc_top = mc->mc_snum-1;
3175 mdb_cursor_put(MDB_cursor *mc, MDB_val *key, MDB_val *data,
3179 MDB_val xdata, *rdata, dkey;
3181 char dbuf[PAGESIZE];
3187 if (F_ISSET(mc->mc_txn->mt_flags, MDB_TXN_RDONLY))
3190 DPRINTF("==> put db %u key [%s], size %zu, data size %zu",
3191 mc->mc_dbi, DKEY(key), key->mv_size, data->mv_size);
3195 if (flags == MDB_CURRENT) {
3196 if (!(mc->mc_flags & C_INITIALIZED))
3199 } else if (mc->mc_txn->mt_dbs[mc->mc_dbi].md_root == P_INVALID) {
3201 /* new database, write a root leaf page */
3202 DPUTS("allocating new root leaf page");
3203 if ((np = mdb_new_page(mc, P_LEAF, 1)) == NULL) {
3207 cursor_push_page(mc, np);
3208 mc->mc_txn->mt_dbs[mc->mc_dbi].md_root = np->mp_pgno;
3209 mc->mc_txn->mt_dbs[mc->mc_dbi].md_depth++;
3210 mc->mc_txn->mt_dbxs[mc->mc_dbi].md_dirty = 1;
3211 if ((mc->mc_txn->mt_dbs[mc->mc_dbi].md_flags & (MDB_DUPSORT|MDB_DUPFIXED))
3213 np->mp_flags |= P_LEAF2;
3214 mc->mc_flags |= C_INITIALIZED;
3219 rc = mdb_cursor_set(mc, key, NULL, MDB_SET, &exact);
3220 if (flags == MDB_NOOVERWRITE && rc == 0) {
3221 DPRINTF("duplicate key [%s]", DKEY(key));
3222 return MDB_KEYEXIST;
3224 if (rc && rc != MDB_NOTFOUND)
3228 /* Cursor is positioned, now make sure all pages are writable */
3229 rc2 = mdb_cursor_touch(mc);
3230 if (rc2) return rc2;
3233 /* The key already exists */
3234 if (rc == MDB_SUCCESS) {
3235 /* there's only a key anyway, so this is a no-op */
3236 if (IS_LEAF2(mc->mc_pg[mc->mc_top])) {
3237 unsigned int ksize = mc->mc_txn->mt_dbs[mc->mc_dbi].md_pad;
3238 if (key->mv_size != ksize)
3240 if (flags == MDB_CURRENT) {
3241 char *ptr = LEAF2KEY(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top], ksize);
3242 memcpy(ptr, key->mv_data, ksize);
3247 leaf = NODEPTR(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
3250 if (F_ISSET(mc->mc_txn->mt_dbs[mc->mc_dbi].md_flags, MDB_DUPSORT)) {
3251 /* Was a single item before, must convert now */
3252 if (!F_ISSET(leaf->mn_flags, F_DUPDATA)) {
3253 dkey.mv_size = NODEDSZ(leaf);
3254 dkey.mv_data = dbuf;
3255 memcpy(dbuf, NODEDATA(leaf), dkey.mv_size);
3256 /* data matches, ignore it */
3257 if (!mdb_dcmp(mc->mc_txn, mc->mc_dbi, data, &dkey))
3258 return (flags == MDB_NODUPDATA) ? MDB_KEYEXIST : MDB_SUCCESS;
3259 memset(&dummy, 0, sizeof(dummy));
3260 if (mc->mc_txn->mt_dbs[mc->mc_dbi].md_flags & MDB_DUPFIXED) {
3261 dummy.md_pad = data->mv_size;
3262 dummy.md_flags = MDB_DUPFIXED;
3263 if (mc->mc_txn->mt_dbs[mc->mc_dbi].md_flags & MDB_INTEGERDUP)
3264 dummy.md_flags |= MDB_INTEGERKEY;
3266 dummy.md_root = P_INVALID;
3267 if (dkey.mv_size == sizeof(MDB_db)) {
3268 memcpy(NODEDATA(leaf), &dummy, sizeof(dummy));
3271 mdb_del_node(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top], 0);
3274 xdata.mv_size = sizeof(MDB_db);
3275 xdata.mv_data = &dummy;
3280 /* same size, just replace it */
3281 if (!F_ISSET(leaf->mn_flags, F_BIGDATA) &&
3282 NODEDSZ(leaf) == data->mv_size) {
3283 memcpy(NODEDATA(leaf), data->mv_data, data->mv_size);
3286 mdb_del_node(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top], 0);
3288 DPRINTF("inserting key at index %i", mc->mc_ki[mc->mc_top]);
3294 nsize = IS_LEAF2(mc->mc_pg[mc->mc_top]) ? key->mv_size : mdb_leaf_size(mc->mc_txn->mt_env, key, rdata);
3295 if (SIZELEFT(mc->mc_pg[mc->mc_top]) < nsize) {
3296 rc = mdb_split(mc, key, rdata, P_INVALID);
3298 /* There is room already in this leaf page. */
3299 rc = mdb_add_node(mc, mc->mc_ki[mc->mc_top], key, rdata, 0, 0);
3302 if (rc != MDB_SUCCESS)
3303 mc->mc_txn->mt_flags |= MDB_TXN_ERROR;
3305 /* Remember if we just added a subdatabase */
3306 if (flags & F_SUBDATA) {
3307 leaf = NODEPTR(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
3308 leaf->mn_flags |= F_SUBDATA;
3311 /* Now store the actual data in the child DB. Note that we're
3312 * storing the user data in the keys field, so there are strict
3313 * size limits on dupdata. The actual data fields of the child
3314 * DB are all zero size.
3317 leaf = NODEPTR(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
3319 if (flags == MDB_CURRENT)
3320 mdb_xcursor_init2(mc);
3322 mdb_xcursor_init1(mc, leaf);
3325 if (flags == MDB_NODUPDATA)
3326 flags = MDB_NOOVERWRITE;
3327 /* converted, write the original data first */
3329 rc = mdb_cursor_put(&mc->mc_xcursor->mx_cursor, &dkey, &xdata, flags);
3331 leaf->mn_flags |= F_DUPDATA;
3333 rc = mdb_cursor_put(&mc->mc_xcursor->mx_cursor, data, &xdata, flags);
3334 mdb_xcursor_fini(mc);
3335 memcpy(NODEDATA(leaf),
3336 &mc->mc_xcursor->mx_txn.mt_dbs[mc->mc_xcursor->mx_cursor.mc_dbi],
3339 mc->mc_txn->mt_dbs[mc->mc_dbi].md_entries++;
3346 mdb_cursor_del(MDB_cursor *mc, unsigned int flags)
3351 if (F_ISSET(mc->mc_txn->mt_flags, MDB_TXN_RDONLY))
3354 if (!mc->mc_flags & C_INITIALIZED)
3357 rc = mdb_cursor_touch(mc);
3360 leaf = NODEPTR(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
3362 if (!IS_LEAF2(mc->mc_pg[mc->mc_top]) && F_ISSET(leaf->mn_flags, F_DUPDATA)) {
3363 if (flags != MDB_NODUPDATA) {
3364 mdb_xcursor_init2(mc);
3365 rc = mdb_cursor_del(&mc->mc_xcursor->mx_cursor, 0);
3366 mdb_xcursor_fini(mc);
3367 /* If sub-DB still has entries, we're done */
3368 if (mc->mc_xcursor->mx_txn.mt_dbs[mc->mc_xcursor->mx_cursor.mc_dbi].md_root
3370 memcpy(NODEDATA(leaf),
3371 &mc->mc_xcursor->mx_txn.mt_dbs[mc->mc_xcursor->mx_cursor.mc_dbi],
3373 mc->mc_txn->mt_dbs[mc->mc_dbi].md_entries--;
3376 /* otherwise fall thru and delete the sub-DB */
3379 /* add all the child DB's pages to the free list */
3380 rc = mdb_search_page(&mc->mc_xcursor->mx_cursor, NULL, 0);
3381 if (rc == MDB_SUCCESS) {
3386 mx = &mc->mc_xcursor->mx_cursor;
3387 mc->mc_txn->mt_dbs[mc->mc_dbi].md_entries -=
3388 mx->mc_txn->mt_dbs[mx->mc_dbi].md_entries;
3390 cursor_pop_page(mx);
3392 while (mx->mc_snum > 1) {
3393 for (i=0; i<NUMKEYS(mx->mc_pg[mx->mc_top]); i++) {
3395 ni = NODEPTR(mx->mc_pg[mx->mc_top], i);
3398 mdb_midl_append(mc->mc_txn->mt_free_pgs, pg);
3400 rc = mdb_sibling(mx, 1);
3405 mdb_midl_append(mc->mc_txn->mt_free_pgs,
3406 mx->mc_txn->mt_dbs[mx->mc_dbi].md_root);
3410 return mdb_del0(mc, leaf);
3413 /* Allocate a page and initialize it
3416 mdb_new_page(MDB_cursor *mc, uint32_t flags, int num)
3420 if ((np = mdb_alloc_page(mc, num)) == NULL)
3422 DPRINTF("allocated new mpage %lu, page size %u",
3423 np->mp_pgno, mc->mc_txn->mt_env->me_psize);
3424 np->mp_flags = flags | P_DIRTY;
3425 np->mp_lower = PAGEHDRSZ;
3426 np->mp_upper = mc->mc_txn->mt_env->me_psize;
3429 mc->mc_txn->mt_dbs[mc->mc_dbi].md_branch_pages++;
3430 else if (IS_LEAF(np))
3431 mc->mc_txn->mt_dbs[mc->mc_dbi].md_leaf_pages++;
3432 else if (IS_OVERFLOW(np)) {
3433 mc->mc_txn->mt_dbs[mc->mc_dbi].md_overflow_pages += num;
3441 mdb_leaf_size(MDB_env *env, MDB_val *key, MDB_val *data)
3445 sz = LEAFSIZE(key, data);
3446 if (data->mv_size >= env->me_psize / MDB_MINKEYS) {
3447 /* put on overflow page */
3448 sz -= data->mv_size - sizeof(pgno_t);
3452 return sz + sizeof(indx_t);
3456 mdb_branch_size(MDB_env *env, MDB_val *key)
3461 if (sz >= env->me_psize / MDB_MINKEYS) {
3462 /* put on overflow page */
3463 /* not implemented */
3464 /* sz -= key->size - sizeof(pgno_t); */
3467 return sz + sizeof(indx_t);
3471 mdb_add_node(MDB_cursor *mc, indx_t indx,
3472 MDB_val *key, MDB_val *data, pgno_t pgno, uint8_t flags)
3475 size_t node_size = NODESIZE;
3478 MDB_page *mp = mc->mc_pg[mc->mc_top];
3479 MDB_page *ofp = NULL; /* overflow page */
3482 assert(mp->mp_upper >= mp->mp_lower);
3484 DPRINTF("add to %s page %lu index %i, data size %zu key size %zu [%s]",
3485 IS_LEAF(mp) ? "leaf" : "branch",
3486 mp->mp_pgno, indx, data ? data->mv_size : 0,
3487 key ? key->mv_size : 0, key ? DKEY(key) : NULL);
3490 /* Move higher keys up one slot. */
3491 int ksize = mc->mc_txn->mt_dbs[mc->mc_dbi].md_pad, dif;
3492 char *ptr = LEAF2KEY(mp, indx, ksize);
3493 dif = NUMKEYS(mp) - indx;
3495 memmove(ptr+ksize, ptr, dif*ksize);
3496 /* insert new key */
3497 memcpy(ptr, key->mv_data, ksize);
3499 /* Just using these for counting */
3500 mp->mp_lower += sizeof(indx_t);
3501 mp->mp_upper -= ksize - sizeof(indx_t);
3506 node_size += key->mv_size;
3510 if (F_ISSET(flags, F_BIGDATA)) {
3511 /* Data already on overflow page. */
3512 node_size += sizeof(pgno_t);
3513 } else if (data->mv_size >= mc->mc_txn->mt_env->me_psize / MDB_MINKEYS) {
3514 int ovpages = OVPAGES(data->mv_size, mc->mc_txn->mt_env->me_psize);
3515 /* Put data on overflow page. */
3516 DPRINTF("data size is %zu, put on overflow page",
3518 node_size += sizeof(pgno_t);
3519 if ((ofp = mdb_new_page(mc, P_OVERFLOW, ovpages)) == NULL)
3521 DPRINTF("allocated overflow page %lu", ofp->mp_pgno);
3524 node_size += data->mv_size;
3527 node_size += node_size & 1;
3529 if (node_size + sizeof(indx_t) > SIZELEFT(mp)) {
3530 DPRINTF("not enough room in page %lu, got %u ptrs",
3531 mp->mp_pgno, NUMKEYS(mp));
3532 DPRINTF("upper - lower = %u - %u = %u", mp->mp_upper, mp->mp_lower,
3533 mp->mp_upper - mp->mp_lower);
3534 DPRINTF("node size = %zu", node_size);
3538 /* Move higher pointers up one slot. */
3539 for (i = NUMKEYS(mp); i > indx; i--)
3540 mp->mp_ptrs[i] = mp->mp_ptrs[i - 1];
3542 /* Adjust free space offsets. */
3543 ofs = mp->mp_upper - node_size;
3544 assert(ofs >= mp->mp_lower + sizeof(indx_t));
3545 mp->mp_ptrs[indx] = ofs;
3547 mp->mp_lower += sizeof(indx_t);
3549 /* Write the node data. */
3550 node = NODEPTR(mp, indx);
3551 node->mn_ksize = (key == NULL) ? 0 : key->mv_size;
3552 node->mn_flags = flags;
3554 SETDSZ(node,data->mv_size);
3559 memcpy(NODEKEY(node), key->mv_data, key->mv_size);
3564 if (F_ISSET(flags, F_BIGDATA))
3565 memcpy(node->mn_data + key->mv_size, data->mv_data,
3568 memcpy(node->mn_data + key->mv_size, data->mv_data,
3571 memcpy(node->mn_data + key->mv_size, &ofp->mp_pgno,
3573 memcpy(METADATA(ofp), data->mv_data, data->mv_size);
3581 mdb_del_node(MDB_page *mp, indx_t indx, int ksize)
3584 indx_t i, j, numkeys, ptr;
3588 DPRINTF("delete node %u on %s page %lu", indx,
3589 IS_LEAF(mp) ? "leaf" : "branch", mp->mp_pgno);
3590 assert(indx < NUMKEYS(mp));
3593 int x = NUMKEYS(mp) - 1 - indx;
3594 base = LEAF2KEY(mp, indx, ksize);
3596 memmove(base, base + ksize, x * ksize);
3597 mp->mp_lower -= sizeof(indx_t);
3598 mp->mp_upper += ksize - sizeof(indx_t);
3602 node = NODEPTR(mp, indx);
3603 sz = NODESIZE + node->mn_ksize;
3605 if (F_ISSET(node->mn_flags, F_BIGDATA))
3606 sz += sizeof(pgno_t);
3608 sz += NODEDSZ(node);
3612 ptr = mp->mp_ptrs[indx];
3613 numkeys = NUMKEYS(mp);
3614 for (i = j = 0; i < numkeys; i++) {
3616 mp->mp_ptrs[j] = mp->mp_ptrs[i];
3617 if (mp->mp_ptrs[i] < ptr)
3618 mp->mp_ptrs[j] += sz;
3623 base = (char *)mp + mp->mp_upper;
3624 memmove(base + sz, base, ptr - mp->mp_upper);
3626 mp->mp_lower -= sizeof(indx_t);
3631 mdb_xcursor_init0(MDB_cursor *mc)
3633 MDB_xcursor *mx = mc->mc_xcursor;
3636 mx->mx_txn = *mc->mc_txn;
3637 mx->mx_txn.mt_dbxs = mx->mx_dbxs;
3638 mx->mx_txn.mt_dbs = mx->mx_dbs;
3639 mx->mx_dbxs[0] = mc->mc_txn->mt_dbxs[0];
3640 mx->mx_dbxs[1] = mc->mc_txn->mt_dbxs[1];
3641 if (mc->mc_dbi > 1) {
3642 mx->mx_dbxs[2] = mc->mc_txn->mt_dbxs[mc->mc_dbi];
3647 mx->mx_dbxs[dbn+1].md_parent = dbn;
3648 mx->mx_dbxs[dbn+1].md_cmp = mx->mx_dbxs[dbn].md_dcmp;
3649 mx->mx_dbxs[dbn+1].md_rel = mx->mx_dbxs[dbn].md_rel;
3650 mx->mx_dbxs[dbn+1].md_dirty = 0;
3651 mx->mx_txn.mt_numdbs = dbn+2;
3652 mx->mx_txn.mt_u = mc->mc_txn->mt_u;
3654 mx->mx_cursor.mc_xcursor = NULL;
3655 mx->mx_cursor.mc_txn = &mx->mx_txn;
3656 mx->mx_cursor.mc_dbi = dbn+1;
3660 mdb_xcursor_init1(MDB_cursor *mc, MDB_node *node)
3662 MDB_db *db = NODEDATA(node);
3663 MDB_xcursor *mx = mc->mc_xcursor;
3665 mx->mx_dbs[0] = mc->mc_txn->mt_dbs[0];
3666 mx->mx_dbs[1] = mc->mc_txn->mt_dbs[1];
3667 if (mc->mc_dbi > 1) {
3668 mx->mx_dbs[2] = mc->mc_txn->mt_dbs[mc->mc_dbi];
3669 mx->mx_dbxs[2].md_dirty = mc->mc_txn->mt_dbxs[mc->mc_dbi].md_dirty;
3674 DPRINTF("Sub-db %u for db %u root page %lu", dbn, mc->mc_dbi, db->md_root);
3675 mx->mx_dbs[dbn] = *db;
3676 if (F_ISSET(mc->mc_pg[mc->mc_top]->mp_flags, P_DIRTY))
3677 mx->mx_dbxs[dbn].md_dirty = 1;
3678 mx->mx_dbxs[dbn].md_name.mv_data = NODEKEY(node);
3679 mx->mx_dbxs[dbn].md_name.mv_size = node->mn_ksize;
3680 mx->mx_txn.mt_next_pgno = mc->mc_txn->mt_next_pgno;
3681 mx->mx_cursor.mc_snum = 0;
3682 mx->mx_cursor.mc_flags = 0;
3686 mdb_xcursor_init2(MDB_cursor *mc)
3688 MDB_xcursor *mx = mc->mc_xcursor;
3690 mx->mx_dbs[0] = mc->mc_txn->mt_dbs[0];
3691 mx->mx_dbs[1] = mc->mc_txn->mt_dbs[1];
3692 if (mc->mc_dbi > 1) {
3693 mx->mx_dbs[2] = mc->mc_txn->mt_dbs[mc->mc_dbi];
3694 mx->mx_dbxs[2].md_dirty = mc->mc_txn->mt_dbxs[mc->mc_dbi].md_dirty;
3699 DPRINTF("Sub-db %u for db %u root page %lu", dbn, mc->mc_dbi,
3700 mx->mx_dbs[dbn].md_root);
3701 mx->mx_txn.mt_next_pgno = mc->mc_txn->mt_next_pgno;
3705 mdb_xcursor_fini(MDB_cursor *mc)
3707 MDB_xcursor *mx = mc->mc_xcursor;
3708 mc->mc_txn->mt_next_pgno = mx->mx_txn.mt_next_pgno;
3709 mc->mc_txn->mt_dbs[0] = mx->mx_dbs[0];
3710 mc->mc_txn->mt_dbs[1] = mx->mx_dbs[1];
3711 if (mc->mc_dbi > 1) {
3712 mc->mc_txn->mt_dbs[mc->mc_dbi] = mx->mx_dbs[2];
3713 mc->mc_txn->mt_dbxs[mc->mc_dbi].md_dirty = mx->mx_dbxs[2].md_dirty;
3718 mdb_cursor_open(MDB_txn *txn, MDB_dbi dbi, MDB_cursor **ret)
3721 size_t size = sizeof(MDB_cursor);
3723 if (txn == NULL || ret == NULL || !dbi || dbi >= txn->mt_numdbs)
3726 if (txn->mt_dbs[dbi].md_flags & MDB_DUPSORT)
3727 size += sizeof(MDB_xcursor);
3729 if ((mc = calloc(1, size)) != NULL) {
3732 if (txn->mt_dbs[dbi].md_flags & MDB_DUPSORT) {
3733 MDB_xcursor *mx = (MDB_xcursor *)(mc + 1);
3734 mc->mc_xcursor = mx;
3735 mdb_xcursor_init0(mc);
3746 /* Return the count of duplicate data items for the current key */
3748 mdb_cursor_count(MDB_cursor *mc, unsigned long *countp)
3752 if (mc == NULL || countp == NULL)
3755 if (!(mc->mc_txn->mt_dbs[mc->mc_dbi].md_flags & MDB_DUPSORT))
3758 leaf = NODEPTR(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top]);
3759 if (!F_ISSET(leaf->mn_flags, F_DUPDATA)) {
3762 if (!(mc->mc_xcursor->mx_cursor.mc_flags & C_INITIALIZED))
3765 *countp = mc->mc_xcursor->mx_txn.mt_dbs[mc->mc_xcursor->mx_cursor.mc_dbi].md_entries;
3771 mdb_cursor_close(MDB_cursor *mc)
3779 mdb_update_key(MDB_page *mp, indx_t indx, MDB_val *key)
3781 indx_t ptr, i, numkeys;
3788 node = NODEPTR(mp, indx);
3789 ptr = mp->mp_ptrs[indx];
3790 DPRINTF("update key %u (ofs %u) [%.*s] to [%s] on page %lu",
3792 (int)node->mn_ksize, (char *)NODEKEY(node),
3796 delta = key->mv_size - node->mn_ksize;
3798 if (delta > 0 && SIZELEFT(mp) < delta) {
3799 DPRINTF("OUCH! Not enough room, delta = %d", delta);
3803 numkeys = NUMKEYS(mp);
3804 for (i = 0; i < numkeys; i++) {
3805 if (mp->mp_ptrs[i] <= ptr)
3806 mp->mp_ptrs[i] -= delta;
3809 base = (char *)mp + mp->mp_upper;
3810 len = ptr - mp->mp_upper + NODESIZE;
3811 memmove(base - delta, base, len);
3812 mp->mp_upper -= delta;
3814 node = NODEPTR(mp, indx);
3815 node->mn_ksize = key->mv_size;
3818 memcpy(NODEKEY(node), key->mv_data, key->mv_size);
3823 /* Move a node from csrc to cdst.
3826 mdb_move_node(MDB_cursor *csrc, MDB_cursor *cdst)
3833 /* Mark src and dst as dirty. */
3834 if ((rc = mdb_touch(csrc)) ||
3835 (rc = mdb_touch(cdst)))
3838 if (IS_LEAF2(csrc->mc_pg[csrc->mc_top])) {
3839 srcnode = NODEPTR(csrc->mc_pg[csrc->mc_top], 0); /* fake */
3840 key.mv_size = csrc->mc_txn->mt_dbs[csrc->mc_dbi].md_pad;
3841 key.mv_data = LEAF2KEY(csrc->mc_pg[csrc->mc_top], csrc->mc_ki[csrc->mc_top], key.mv_size);
3843 data.mv_data = NULL;
3845 if (csrc->mc_ki[csrc->mc_top] == 0 && IS_BRANCH(csrc->mc_pg[csrc->mc_top])) {
3846 unsigned int snum = csrc->mc_snum;
3847 /* must find the lowest key below src */
3848 mdb_search_page_root(csrc, NULL, 0);
3849 srcnode = NODEPTR(csrc->mc_pg[csrc->mc_top], 0);
3850 csrc->mc_snum = snum--;
3851 csrc->mc_top = snum;
3853 srcnode = NODEPTR(csrc->mc_pg[csrc->mc_top], csrc->mc_ki[csrc->mc_top]);
3855 key.mv_size = NODEKSZ(srcnode);
3856 key.mv_data = NODEKEY(srcnode);
3857 data.mv_size = NODEDSZ(srcnode);
3858 data.mv_data = NODEDATA(srcnode);
3860 DPRINTF("moving %s node %u [%s] on page %lu to node %u on page %lu",
3861 IS_LEAF(csrc->mc_pg[csrc->mc_top]) ? "leaf" : "branch",
3862 csrc->mc_ki[csrc->mc_top],
3864 csrc->mc_pg[csrc->mc_top]->mp_pgno,
3865 cdst->mc_ki[cdst->mc_top], cdst->mc_pg[cdst->mc_top]->mp_pgno);
3867 /* Add the node to the destination page.
3869 rc = mdb_add_node(cdst, cdst->mc_ki[cdst->mc_top], &key, &data, NODEPGNO(srcnode),
3871 if (rc != MDB_SUCCESS)
3874 /* Delete the node from the source page.
3876 mdb_del_node(csrc->mc_pg[csrc->mc_top], csrc->mc_ki[csrc->mc_top], key.mv_size);
3878 /* Update the parent separators.
3880 if (csrc->mc_ki[csrc->mc_top] == 0) {
3881 if (csrc->mc_ki[csrc->mc_top-1] != 0) {
3882 if (IS_LEAF2(csrc->mc_pg[csrc->mc_top])) {
3883 key.mv_data = LEAF2KEY(csrc->mc_pg[csrc->mc_top], csrc->mc_ki[csrc->mc_top], key.mv_size);
3885 srcnode = NODEPTR(csrc->mc_pg[csrc->mc_top], csrc->mc_ki[csrc->mc_top]);
3886 key.mv_size = NODEKSZ(srcnode);
3887 key.mv_data = NODEKEY(srcnode);
3889 DPRINTF("update separator for source page %lu to [%s]",
3890 csrc->mc_pg[csrc->mc_top]->mp_pgno, DKEY(&key));
3891 if ((rc = mdb_update_key(csrc->mc_pg[csrc->mc_top-1], csrc->mc_ki[csrc->mc_top-1],
3892 &key)) != MDB_SUCCESS)
3895 if (IS_BRANCH(csrc->mc_pg[csrc->mc_top])) {
3897 nullkey.mv_size = 0;
3898 assert(mdb_update_key(csrc->mc_pg[csrc->mc_top], 0, &nullkey) == MDB_SUCCESS);
3902 if (cdst->mc_ki[cdst->mc_top] == 0) {
3903 if (cdst->mc_ki[cdst->mc_top-1] != 0) {
3904 if (IS_LEAF2(csrc->mc_pg[csrc->mc_top])) {
3905 key.mv_data = LEAF2KEY(cdst->mc_pg[cdst->mc_top], 0, key.mv_size);
3907 srcnode = NODEPTR(cdst->mc_pg[cdst->mc_top], 0);
3908 key.mv_size = NODEKSZ(srcnode);
3909 key.mv_data = NODEKEY(srcnode);
3911 DPRINTF("update separator for destination page %lu to [%s]",
3912 cdst->mc_pg[cdst->mc_top]->mp_pgno, DKEY(&key));
3913 if ((rc = mdb_update_key(cdst->mc_pg[cdst->mc_top-1], cdst->mc_ki[cdst->mc_top-1],
3914 &key)) != MDB_SUCCESS)
3917 if (IS_BRANCH(cdst->mc_pg[cdst->mc_top])) {
3919 nullkey.mv_size = 0;
3920 assert(mdb_update_key(cdst->mc_pg[cdst->mc_top], 0, &nullkey) == MDB_SUCCESS);
3928 mdb_merge(MDB_cursor *csrc, MDB_cursor *cdst)
3935 DPRINTF("merging page %lu into %lu", csrc->mc_pg[csrc->mc_top]->mp_pgno, cdst->mc_pg[cdst->mc_top]->mp_pgno);
3937 assert(csrc->mc_snum > 1); /* can't merge root page */
3938 assert(cdst->mc_snum > 1);
3940 /* Mark dst as dirty. */
3941 if ((rc = mdb_touch(cdst)))
3944 /* Move all nodes from src to dst.
3946 j = NUMKEYS(cdst->mc_pg[cdst->mc_top]);
3947 if (IS_LEAF2(csrc->mc_pg[csrc->mc_top])) {
3948 key.mv_size = csrc->mc_txn->mt_dbs[csrc->mc_dbi].md_pad;
3949 key.mv_data = METADATA(csrc->mc_pg[csrc->mc_top]);
3950 for (i = 0; i < NUMKEYS(csrc->mc_pg[csrc->mc_top]); i++, j++) {
3951 rc = mdb_add_node(cdst, j, &key, NULL, 0, 0);
3952 if (rc != MDB_SUCCESS)
3954 key.mv_data = (char *)key.mv_data + key.mv_size;
3957 for (i = 0; i < NUMKEYS(csrc->mc_pg[csrc->mc_top]); i++, j++) {
3958 srcnode = NODEPTR(csrc->mc_pg[csrc->mc_top], i);
3960 key.mv_size = srcnode->mn_ksize;
3961 key.mv_data = NODEKEY(srcnode);
3962 data.mv_size = NODEDSZ(srcnode);
3963 data.mv_data = NODEDATA(srcnode);
3964 rc = mdb_add_node(cdst, j, &key, &data, NODEPGNO(srcnode), srcnode->mn_flags);
3965 if (rc != MDB_SUCCESS)
3970 DPRINTF("dst page %lu now has %u keys (%.1f%% filled)",
3971 cdst->mc_pg[cdst->mc_top]->mp_pgno, NUMKEYS(cdst->mc_pg[cdst->mc_top]), (float)PAGEFILL(cdst->mc_txn->mt_env, cdst->mc_pg[cdst->mc_top]) / 10);
3973 /* Unlink the src page from parent and add to free list.
3975 mdb_del_node(csrc->mc_pg[csrc->mc_top-1], csrc->mc_ki[csrc->mc_top-1], 0);
3976 if (csrc->mc_ki[csrc->mc_top-1] == 0) {
3978 if ((rc = mdb_update_key(csrc->mc_pg[csrc->mc_top-1], 0, &key)) != MDB_SUCCESS)
3982 mdb_midl_append(csrc->mc_txn->mt_free_pgs, csrc->mc_pg[csrc->mc_top]->mp_pgno);
3983 if (IS_LEAF(csrc->mc_pg[csrc->mc_top]))
3984 csrc->mc_txn->mt_dbs[csrc->mc_dbi].md_leaf_pages--;
3986 csrc->mc_txn->mt_dbs[csrc->mc_dbi].md_branch_pages--;
3987 cursor_pop_page(csrc);
3989 return mdb_rebalance(csrc);
3993 mdb_cursor_copy(const MDB_cursor *csrc, MDB_cursor *cdst)
3997 cdst->mc_txn = csrc->mc_txn;
3998 cdst->mc_dbi = csrc->mc_dbi;
3999 cdst->mc_snum = csrc->mc_snum;
4000 cdst->mc_top = csrc->mc_top;
4001 cdst->mc_flags = csrc->mc_flags;
4003 for (i=0; i<csrc->mc_snum; i++) {
4004 cdst->mc_pg[i] = csrc->mc_pg[i];
4005 cdst->mc_ki[i] = csrc->mc_ki[i];
4010 mdb_rebalance(MDB_cursor *mc)
4018 DPRINTF("rebalancing %s page %lu (has %u keys, %.1f%% full)",
4019 IS_LEAF(mc->mc_pg[mc->mc_top]) ? "leaf" : "branch",
4020 mc->mc_pg[mc->mc_top]->mp_pgno, NUMKEYS(mc->mc_pg[mc->mc_top]), (float)PAGEFILL(mc->mc_txn->mt_env, mc->mc_pg[mc->mc_top]) / 10);
4022 if (PAGEFILL(mc->mc_txn->mt_env, mc->mc_pg[mc->mc_top]) >= FILL_THRESHOLD) {
4023 DPRINTF("no need to rebalance page %lu, above fill threshold",
4024 mc->mc_pg[mc->mc_top]->mp_pgno);
4028 if (mc->mc_snum < 2) {
4029 if (NUMKEYS(mc->mc_pg[mc->mc_top]) == 0) {
4030 DPUTS("tree is completely empty");
4031 mc->mc_txn->mt_dbs[mc->mc_dbi].md_root = P_INVALID;
4032 mc->mc_txn->mt_dbs[mc->mc_dbi].md_depth = 0;
4033 mc->mc_txn->mt_dbs[mc->mc_dbi].md_leaf_pages = 0;
4034 mdb_midl_append(mc->mc_txn->mt_free_pgs, mc->mc_pg[mc->mc_top]->mp_pgno);
4035 } else if (IS_BRANCH(mc->mc_pg[mc->mc_top]) && NUMKEYS(mc->mc_pg[mc->mc_top]) == 1) {
4036 DPUTS("collapsing root page!");
4037 mdb_midl_append(mc->mc_txn->mt_free_pgs, mc->mc_pg[mc->mc_top]->mp_pgno);
4038 mc->mc_txn->mt_dbs[mc->mc_dbi].md_root = NODEPGNO(NODEPTR(mc->mc_pg[mc->mc_top], 0));
4039 if ((rc = mdb_get_page(mc->mc_txn, mc->mc_txn->mt_dbs[mc->mc_dbi].md_root, &root)))
4041 mc->mc_txn->mt_dbs[mc->mc_dbi].md_depth--;
4042 mc->mc_txn->mt_dbs[mc->mc_dbi].md_branch_pages--;
4044 DPUTS("root page doesn't need rebalancing");
4048 /* The parent (branch page) must have at least 2 pointers,
4049 * otherwise the tree is invalid.
4051 ptop = mc->mc_top-1;
4052 assert(NUMKEYS(mc->mc_pg[ptop]) > 1);
4054 /* Leaf page fill factor is below the threshold.
4055 * Try to move keys from left or right neighbor, or
4056 * merge with a neighbor page.
4061 mdb_cursor_copy(mc, &mn);
4062 mn.mc_xcursor = NULL;
4064 if (mc->mc_ki[ptop] == 0) {
4065 /* We're the leftmost leaf in our parent.
4067 DPUTS("reading right neighbor");
4069 node = NODEPTR(mc->mc_pg[ptop], mn.mc_ki[ptop]);
4070 if ((rc = mdb_get_page(mc->mc_txn, NODEPGNO(node), &mn.mc_pg[mn.mc_top])))
4072 mn.mc_ki[mn.mc_top] = 0;
4073 mc->mc_ki[mc->mc_top] = NUMKEYS(mc->mc_pg[mc->mc_top]);
4075 /* There is at least one neighbor to the left.
4077 DPUTS("reading left neighbor");
4079 node = NODEPTR(mc->mc_pg[ptop], mn.mc_ki[ptop]);
4080 if ((rc = mdb_get_page(mc->mc_txn, NODEPGNO(node), &mn.mc_pg[mn.mc_top])))
4082 mn.mc_ki[mn.mc_top] = NUMKEYS(mn.mc_pg[mn.mc_top]) - 1;
4083 mc->mc_ki[mc->mc_top] = 0;
4086 DPRINTF("found neighbor page %lu (%u keys, %.1f%% full)",
4087 mn.mc_pg[mn.mc_top]->mp_pgno, NUMKEYS(mn.mc_pg[mn.mc_top]), (float)PAGEFILL(mc->mc_txn->mt_env, mn.mc_pg[mn.mc_top]) / 10);
4089 /* If the neighbor page is above threshold and has at least two
4090 * keys, move one key from it.
4092 * Otherwise we should try to merge them.
4094 if (PAGEFILL(mc->mc_txn->mt_env, mn.mc_pg[mn.mc_top]) >= FILL_THRESHOLD && NUMKEYS(mn.mc_pg[mn.mc_top]) >= 2)
4095 return mdb_move_node(&mn, mc);
4096 else { /* FIXME: if (has_enough_room()) */
4097 if (mc->mc_ki[ptop] == 0)
4098 return mdb_merge(&mn, mc);
4100 return mdb_merge(mc, &mn);
4105 mdb_del0(MDB_cursor *mc, MDB_node *leaf)
4109 /* add overflow pages to free list */
4110 if (!IS_LEAF2(mc->mc_pg[mc->mc_top]) && F_ISSET(leaf->mn_flags, F_BIGDATA)) {
4114 memcpy(&pg, NODEDATA(leaf), sizeof(pg));
4115 ovpages = OVPAGES(NODEDSZ(leaf), mc->mc_txn->mt_env->me_psize);
4116 for (i=0; i<ovpages; i++) {
4117 DPRINTF("freed ov page %lu", pg);
4118 mdb_midl_append(mc->mc_txn->mt_free_pgs, pg);
4122 mdb_del_node(mc->mc_pg[mc->mc_top], mc->mc_ki[mc->mc_top], mc->mc_txn->mt_dbs[mc->mc_dbi].md_pad);
4123 mc->mc_txn->mt_dbs[mc->mc_dbi].md_entries--;
4124 rc = mdb_rebalance(mc);
4125 if (rc != MDB_SUCCESS)
4126 mc->mc_txn->mt_flags |= MDB_TXN_ERROR;
4132 mdb_del(MDB_txn *txn, MDB_dbi dbi,
4133 MDB_val *key, MDB_val *data)
4138 MDB_val rdata, *xdata;
4142 assert(key != NULL);
4144 DPRINTF("====> delete db %u key [%s]", dbi, DKEY(key));
4146 if (txn == NULL || !dbi || dbi >= txn->mt_numdbs)
4149 if (F_ISSET(txn->mt_flags, MDB_TXN_RDONLY)) {
4153 if (key->mv_size == 0 || key->mv_size > MAXKEYSIZE) {
4160 if (txn->mt_dbs[dbi].md_flags & MDB_DUPSORT) {
4161 mc.mc_xcursor = &mx;
4162 mdb_xcursor_init0(&mc);
4164 mc.mc_xcursor = NULL;
4176 rc = mdb_cursor_set(&mc, key, xdata, op, &exact);
4178 rc = mdb_cursor_del(&mc, data ? 0 : MDB_NODUPDATA);
4182 /* Split page <mc->top>, and insert <key,(data|newpgno)> in either left or
4183 * right sibling, at index <mc->ki> (as if unsplit). Updates mc->top and
4184 * mc->ki with the actual values after split, ie if mc->top and mc->ki
4185 * refer to a node in the new right sibling page.
4188 mdb_split(MDB_cursor *mc, MDB_val *newkey, MDB_val *newdata, pgno_t newpgno)
4191 int rc = MDB_SUCCESS, ins_new = 0;
4194 unsigned int i, j, split_indx, nkeys, pmax;
4196 MDB_val sepkey, rkey, rdata;
4198 MDB_page *mp, *rp, *pp;
4203 mp = mc->mc_pg[mc->mc_top];
4204 newindx = mc->mc_ki[mc->mc_top];
4206 DPRINTF("-----> splitting %s page %lu and adding [%s] at index %i",
4207 IS_LEAF(mp) ? "leaf" : "branch", mp->mp_pgno,
4208 DKEY(newkey), mc->mc_ki[mc->mc_top]);
4210 if (mc->mc_snum < 2) {
4211 if ((pp = mdb_new_page(mc, P_BRANCH, 1)) == NULL)
4213 /* shift current top to make room for new parent */
4214 mc->mc_pg[1] = mc->mc_pg[0];
4215 mc->mc_ki[1] = mc->mc_ki[0];
4218 mc->mc_txn->mt_dbs[mc->mc_dbi].md_root = pp->mp_pgno;
4219 DPRINTF("root split! new root = %lu", pp->mp_pgno);
4220 mc->mc_txn->mt_dbs[mc->mc_dbi].md_depth++;
4222 /* Add left (implicit) pointer. */
4223 if ((rc = mdb_add_node(mc, 0, NULL, NULL, mp->mp_pgno, 0)) != MDB_SUCCESS) {
4224 /* undo the pre-push */
4225 mc->mc_pg[0] = mc->mc_pg[1];
4226 mc->mc_ki[0] = mc->mc_ki[1];
4227 mc->mc_txn->mt_dbs[mc->mc_dbi].md_root = mp->mp_pgno;
4228 mc->mc_txn->mt_dbs[mc->mc_dbi].md_depth--;
4235 ptop = mc->mc_top-1;
4236 DPRINTF("parent branch page is %lu", mc->mc_pg[ptop]->mp_pgno);
4239 /* Create a right sibling. */
4240 if ((rp = mdb_new_page(mc, mp->mp_flags, 1)) == NULL)
4242 mdb_cursor_copy(mc, &mn);
4243 mn.mc_pg[mn.mc_top] = rp;
4244 mn.mc_ki[ptop] = mc->mc_ki[ptop]+1;
4245 DPRINTF("new right sibling: page %lu", rp->mp_pgno);
4247 nkeys = NUMKEYS(mp);
4248 split_indx = nkeys / 2 + 1;
4253 unsigned int lsize, rsize, ksize;
4254 /* Move half of the keys to the right sibling */
4256 x = mc->mc_ki[mc->mc_top] - split_indx;
4257 ksize = mc->mc_txn->mt_dbs[mc->mc_dbi].md_pad;
4258 split = LEAF2KEY(mp, split_indx, ksize);
4259 rsize = (nkeys - split_indx) * ksize;
4260 lsize = (nkeys - split_indx) * sizeof(indx_t);
4261 mp->mp_lower -= lsize;
4262 rp->mp_lower += lsize;
4263 mp->mp_upper += rsize - lsize;
4264 rp->mp_upper -= rsize - lsize;
4265 sepkey.mv_size = ksize;
4266 if (newindx == split_indx) {
4267 sepkey.mv_data = newkey->mv_data;
4269 sepkey.mv_data = split;
4272 ins = LEAF2KEY(mp, mc->mc_ki[mc->mc_top], ksize);
4273 memcpy(rp->mp_ptrs, split, rsize);
4274 sepkey.mv_data = rp->mp_ptrs;
4275 memmove(ins+ksize, ins, (split_indx - mc->mc_ki[mc->mc_top]) * ksize);
4276 memcpy(ins, newkey->mv_data, ksize);
4277 mp->mp_lower += sizeof(indx_t);
4278 mp->mp_upper -= ksize - sizeof(indx_t);
4281 memcpy(rp->mp_ptrs, split, x * ksize);
4282 ins = LEAF2KEY(rp, x, ksize);
4283 memcpy(ins, newkey->mv_data, ksize);
4284 memcpy(ins+ksize, split + x * ksize, rsize - x * ksize);
4285 rp->mp_lower += sizeof(indx_t);
4286 rp->mp_upper -= ksize - sizeof(indx_t);
4287 mc->mc_ki[mc->mc_top] = x;
4288 mc->mc_pg[mc->mc_top] = rp;
4293 /* For leaf pages, check the split point based on what
4294 * fits where, since otherwise add_node can fail.
4297 unsigned int psize, nsize;
4298 /* Maximum free space in an empty page */
4299 pmax = mc->mc_txn->mt_env->me_psize - PAGEHDRSZ;
4300 nsize = mdb_leaf_size(mc->mc_txn->mt_env, newkey, newdata);
4301 if (newindx < split_indx) {
4303 for (i=0; i<split_indx; i++) {
4304 node = NODEPTR(mp, i);
4305 psize += NODESIZE + NODEKSZ(node) + sizeof(indx_t);
4306 if (F_ISSET(node->mn_flags, F_BIGDATA))
4307 psize += sizeof(pgno_t);
4309 psize += NODEDSZ(node);
4318 for (i=nkeys-1; i>=split_indx; i--) {
4319 node = NODEPTR(mp, i);
4320 psize += NODESIZE + NODEKSZ(node) + sizeof(indx_t);
4321 if (F_ISSET(node->mn_flags, F_BIGDATA))
4322 psize += sizeof(pgno_t);
4324 psize += NODEDSZ(node);
4334 /* First find the separating key between the split pages.
4336 if (newindx == split_indx) {
4337 sepkey.mv_size = newkey->mv_size;
4338 sepkey.mv_data = newkey->mv_data;
4340 node = NODEPTR(mp, split_indx);
4341 sepkey.mv_size = node->mn_ksize;
4342 sepkey.mv_data = NODEKEY(node);
4346 DPRINTF("separator is [%s]", DKEY(&sepkey));
4348 /* Copy separator key to the parent.
4350 if (SIZELEFT(mn.mc_pg[ptop]) < mdb_branch_size(mc->mc_txn->mt_env, &sepkey)) {
4353 rc = mdb_split(&mn, &sepkey, NULL, rp->mp_pgno);
4355 /* Right page might now have changed parent.
4356 * Check if left page also changed parent.
4358 if (mn.mc_pg[ptop] != mc->mc_pg[ptop] &&
4359 mc->mc_ki[ptop] >= NUMKEYS(mc->mc_pg[ptop])) {
4360 mc->mc_pg[ptop] = mn.mc_pg[ptop];
4361 mc->mc_ki[ptop] = mn.mc_ki[ptop] - 1;
4365 rc = mdb_add_node(&mn, mn.mc_ki[ptop], &sepkey, NULL, rp->mp_pgno, 0);
4371 if (rc != MDB_SUCCESS) {
4375 /* Move half of the keys to the right sibling. */
4377 /* grab a page to hold a temporary copy */
4378 if (mc->mc_txn->mt_env->me_dpages) {
4379 copy = mc->mc_txn->mt_env->me_dpages;
4380 mc->mc_txn->mt_env->me_dpages = copy->mp_next;
4382 if ((copy = malloc(mc->mc_txn->mt_env->me_psize)) == NULL)
4386 copy->mp_pgno = mp->mp_pgno;
4387 copy->mp_flags = mp->mp_flags;
4388 copy->mp_lower = PAGEHDRSZ;
4389 copy->mp_upper = mc->mc_txn->mt_env->me_psize;
4390 mc->mc_pg[mc->mc_top] = copy;
4391 for (i = j = 0; i <= nkeys; j++) {
4392 if (i == split_indx) {
4393 /* Insert in right sibling. */
4394 /* Reset insert index for right sibling. */
4395 j = (i == newindx && ins_new);
4396 mc->mc_pg[mc->mc_top] = rp;
4399 if (i == newindx && !ins_new) {
4400 /* Insert the original entry that caused the split. */
4401 rkey.mv_data = newkey->mv_data;
4402 rkey.mv_size = newkey->mv_size;
4404 rdata.mv_data = newdata->mv_data;
4405 rdata.mv_size = newdata->mv_size;
4412 /* Update page and index for the new key. */
4413 mc->mc_ki[mc->mc_top] = j;
4414 } else if (i == nkeys) {
4417 node = NODEPTR(mp, i);
4418 rkey.mv_data = NODEKEY(node);
4419 rkey.mv_size = node->mn_ksize;
4421 rdata.mv_data = NODEDATA(node);
4422 rdata.mv_size = NODEDSZ(node);
4424 pgno = NODEPGNO(node);
4425 flags = node->mn_flags;
4430 if (!IS_LEAF(mp) && j == 0) {
4431 /* First branch index doesn't need key data. */
4435 rc = mdb_add_node(mc, j, &rkey, &rdata, pgno, flags);
4438 /* reset back to original page */
4439 if (newindx < split_indx)
4440 mc->mc_pg[mc->mc_top] = mp;
4442 nkeys = NUMKEYS(copy);
4443 for (i=0; i<nkeys; i++)
4444 mp->mp_ptrs[i] = copy->mp_ptrs[i];
4445 mp->mp_lower = copy->mp_lower;
4446 mp->mp_upper = copy->mp_upper;
4447 memcpy(NODEPTR(mp, nkeys-1), NODEPTR(copy, nkeys-1),
4448 mc->mc_txn->mt_env->me_psize - copy->mp_upper);
4450 /* return tmp page to freelist */
4451 copy->mp_next = mc->mc_txn->mt_env->me_dpages;
4452 mc->mc_txn->mt_env->me_dpages = copy;
4457 mdb_put(MDB_txn *txn, MDB_dbi dbi,
4458 MDB_val *key, MDB_val *data, unsigned int flags)
4463 assert(key != NULL);
4464 assert(data != NULL);
4466 if (txn == NULL || !dbi || dbi >= txn->mt_numdbs)
4469 if (F_ISSET(txn->mt_flags, MDB_TXN_RDONLY)) {
4473 if (key->mv_size == 0 || key->mv_size > MAXKEYSIZE) {
4477 if ((flags & (MDB_NOOVERWRITE|MDB_NODUPDATA)) != flags)
4484 if (txn->mt_dbs[dbi].md_flags & MDB_DUPSORT) {
4485 mc.mc_xcursor = &mx;
4486 mdb_xcursor_init0(&mc);
4488 mc.mc_xcursor = NULL;
4490 return mdb_cursor_put(&mc, key, data, flags);
4494 mdb_env_set_flags(MDB_env *env, unsigned int flag, int onoff)
4496 #define CHANGEABLE (MDB_NOSYNC)
4497 if ((flag & CHANGEABLE) != flag)
4500 env->me_flags |= flag;
4502 env->me_flags &= ~flag;
4507 mdb_env_get_flags(MDB_env *env, unsigned int *arg)
4512 *arg = env->me_flags;
4517 mdb_env_get_path(MDB_env *env, const char **arg)
4522 *arg = env->me_path;
4527 mdb_stat0(MDB_env *env, MDB_db *db, MDB_stat *arg)
4529 arg->ms_psize = env->me_psize;
4530 arg->ms_depth = db->md_depth;
4531 arg->ms_branch_pages = db->md_branch_pages;
4532 arg->ms_leaf_pages = db->md_leaf_pages;
4533 arg->ms_overflow_pages = db->md_overflow_pages;
4534 arg->ms_entries = db->md_entries;
4539 mdb_env_stat(MDB_env *env, MDB_stat *arg)
4543 if (env == NULL || arg == NULL)
4546 mdb_env_read_meta(env, &toggle);
4548 return mdb_stat0(env, &env->me_metas[toggle]->mm_dbs[MAIN_DBI], arg);
4552 mdb_default_cmp(MDB_txn *txn, MDB_dbi dbi)
4554 if (txn->mt_dbs[dbi].md_flags & MDB_REVERSEKEY)
4555 txn->mt_dbxs[dbi].md_cmp = memnrcmp;
4556 else if (txn->mt_dbs[dbi].md_flags & MDB_INTEGERKEY)
4557 txn->mt_dbxs[dbi].md_cmp = cintcmp;
4559 txn->mt_dbxs[dbi].md_cmp = memncmp;
4561 if (txn->mt_dbs[dbi].md_flags & MDB_DUPSORT) {
4562 if (txn->mt_dbs[dbi].md_flags & MDB_INTEGERDUP) {
4563 if (txn->mt_dbs[dbi].md_flags & MDB_DUPFIXED)
4564 txn->mt_dbxs[dbi].md_dcmp = intcmp;
4566 txn->mt_dbxs[dbi].md_dcmp = cintcmp;
4567 } else if (txn->mt_dbs[dbi].md_flags & MDB_REVERSEDUP) {
4568 txn->mt_dbxs[dbi].md_dcmp = memnrcmp;
4570 txn->mt_dbxs[dbi].md_dcmp = memncmp;
4573 txn->mt_dbxs[dbi].md_dcmp = NULL;
4577 int mdb_open(MDB_txn *txn, const char *name, unsigned int flags, MDB_dbi *dbi)
4584 if (txn->mt_dbxs[FREE_DBI].md_cmp == NULL) {
4585 mdb_default_cmp(txn, FREE_DBI);
4591 if (flags & (MDB_DUPSORT|MDB_REVERSEKEY|MDB_INTEGERKEY))
4592 txn->mt_dbs[MAIN_DBI].md_flags |= (flags & (MDB_DUPSORT|MDB_REVERSEKEY|MDB_INTEGERKEY));
4593 mdb_default_cmp(txn, MAIN_DBI);
4597 if (txn->mt_dbxs[MAIN_DBI].md_cmp == NULL) {
4598 mdb_default_cmp(txn, MAIN_DBI);
4601 /* Is the DB already open? */
4603 for (i=2; i<txn->mt_numdbs; i++) {
4604 if (len == txn->mt_dbxs[i].md_name.mv_size &&
4605 !strncmp(name, txn->mt_dbxs[i].md_name.mv_data, len)) {
4611 if (txn->mt_numdbs >= txn->mt_env->me_maxdbs - 1)
4614 /* Find the DB info */
4616 key.mv_data = (void *)name;
4617 rc = mdb_get(txn, MAIN_DBI, &key, &data);
4619 /* Create if requested */
4620 if (rc == MDB_NOTFOUND && (flags & MDB_CREATE)) {
4623 data.mv_size = sizeof(MDB_db);
4624 data.mv_data = &dummy;
4625 memset(&dummy, 0, sizeof(dummy));
4626 dummy.md_root = P_INVALID;
4627 dummy.md_flags = flags & 0xffff;
4629 mc.mc_dbi = MAIN_DBI;
4631 rc = mdb_cursor_put(&mc, &key, &data, F_SUBDATA);
4635 /* OK, got info, add to table */
4636 if (rc == MDB_SUCCESS) {
4637 txn->mt_dbxs[txn->mt_numdbs].md_name.mv_data = strdup(name);
4638 txn->mt_dbxs[txn->mt_numdbs].md_name.mv_size = len;
4639 txn->mt_dbxs[txn->mt_numdbs].md_rel = NULL;
4640 txn->mt_dbxs[txn->mt_numdbs].md_parent = MAIN_DBI;
4641 txn->mt_dbxs[txn->mt_numdbs].md_dirty = dirty;
4642 memcpy(&txn->mt_dbs[txn->mt_numdbs], data.mv_data, sizeof(MDB_db));
4643 *dbi = txn->mt_numdbs;
4644 txn->mt_env->me_dbs[0][txn->mt_numdbs] = txn->mt_dbs[txn->mt_numdbs];
4645 txn->mt_env->me_dbs[1][txn->mt_numdbs] = txn->mt_dbs[txn->mt_numdbs];
4646 mdb_default_cmp(txn, txn->mt_numdbs);
4653 int mdb_stat(MDB_txn *txn, MDB_dbi dbi, MDB_stat *arg)
4655 if (txn == NULL || arg == NULL || dbi >= txn->mt_numdbs)
4658 return mdb_stat0(txn->mt_env, &txn->mt_dbs[dbi], arg);
4661 void mdb_close(MDB_txn *txn, MDB_dbi dbi)
4664 if (dbi <= MAIN_DBI || dbi >= txn->mt_numdbs)
4666 ptr = txn->mt_dbxs[dbi].md_name.mv_data;
4667 txn->mt_dbxs[dbi].md_name.mv_data = NULL;
4668 txn->mt_dbxs[dbi].md_name.mv_size = 0;
4672 int mdb_set_compare(MDB_txn *txn, MDB_dbi dbi, MDB_cmp_func *cmp)
4674 if (txn == NULL || !dbi || dbi >= txn->mt_numdbs)
4677 txn->mt_dbxs[dbi].md_cmp = cmp;
4681 int mdb_set_dupsort(MDB_txn *txn, MDB_dbi dbi, MDB_cmp_func *cmp)
4683 if (txn == NULL || !dbi || dbi >= txn->mt_numdbs)
4686 txn->mt_dbxs[dbi].md_dcmp = cmp;
4690 int mdb_set_relfunc(MDB_txn *txn, MDB_dbi dbi, MDB_rel_func *rel)
4692 if (txn == NULL || !dbi || dbi >= txn->mt_numdbs)
4695 txn->mt_dbxs[dbi].md_rel = rel;