1/*
2 * See the file LICENSE for redistribution information.
3 *
4 * Copyright (c) 1996, 2013 Oracle and/or its affiliates.  All rights reserved.
5 *
6 * $Id$
7 *
8 * db.h include file layout:
9 *	General.
10 *	Database Environment.
11 *	Locking subsystem.
12 *	Logging subsystem.
13 *	Shared buffer cache (mpool) subsystem.
14 *	Transaction subsystem.
15 *	Access methods.
16 *	Access method cursors.
17 *	Dbm/Ndbm, Hsearch historic interfaces.
18 */
19
20#ifndef _DB_H_
21#define	_DB_H_
22
23#ifndef	__NO_SYSTEM_INCLUDES
24#include <sys/types.h>
25@inttypes_h_decl@
26@stdint_h_decl@
27@stddef_h_decl@
28#include <stdio.h>
29@unistd_h_decl@
30@thread_h_decl@
31#endif
32
33@platform_header@
34#if defined(__cplusplus)
35extern "C" {
36#endif
37
38@DB_CONST@
39@DB_PROTO1@
40@DB_PROTO2@
41
42/*
43 * Berkeley DB version information.
44 */
45#define	DB_VERSION_FAMILY	@DB_VERSION_FAMILY@
46#define	DB_VERSION_RELEASE	@DB_VERSION_RELEASE@
47#define	DB_VERSION_MAJOR	@DB_VERSION_MAJOR@
48#define	DB_VERSION_MINOR	@DB_VERSION_MINOR@
49#define	DB_VERSION_PATCH	@DB_VERSION_PATCH@
50#define	DB_VERSION_STRING	@DB_VERSION_STRING@
51#define	DB_VERSION_FULL_STRING	@DB_VERSION_FULL_STRING@
52
53/*
54 * !!!
55 * Berkeley DB uses specifically sized types.  If they're not provided by
56 * the system, typedef them here.
57 *
58 * We protect them against multiple inclusion using __BIT_TYPES_DEFINED__,
59 * as does BIND and Kerberos, since we don't know for sure what #include
60 * files the user is using.
61 *
62 * !!!
63 * We also provide the standard u_int, u_long etc., if they're not provided
64 * by the system.
65 */
66#ifndef	__BIT_TYPES_DEFINED__
67#define	__BIT_TYPES_DEFINED__
68@u_int8_decl@
69@int16_decl@
70@u_int16_decl@
71@int32_decl@
72@u_int32_decl@
73@int64_decl@
74@u_int64_decl@
75#endif
76
77@u_char_decl@
78@u_int_decl@
79@u_long_decl@
80@u_short_decl@
81
82/*
83 * Missing ANSI types.
84 *
85 * uintmax_t --
86 * Largest unsigned type, used to align structures in memory.  We don't store
87 * floating point types in structures, so integral types should be sufficient
88 * (and we don't have to worry about systems that store floats in other than
89 * power-of-2 numbers of bytes).  Additionally this fixes compilers that rewrite
90 * structure assignments and ANSI C memcpy calls to be in-line instructions
91 * that happen to require alignment.
92 *
93 * uintptr_t --
94 * Unsigned type that's the same size as a pointer.  There are places where
95 * DB modifies pointers by discarding the bottom bits to guarantee alignment.
96 * We can't use uintmax_t, it may be larger than the pointer, and compilers
97 * get upset about that.  So far we haven't run on any machine where there's
98 * no unsigned type the same size as a pointer -- here's hoping.
99 */
100@uintmax_t_decl@
101@uintptr_t_decl@
102
103@FILE_t_decl@
104@off_t_decl@
105@pid_t_decl@
106@size_t_decl@
107#ifdef HAVE_MIXED_SIZE_ADDRESSING
108typedef u_int32_t db_size_t;
109#else
110typedef size_t db_size_t;
111#endif
112@ssize_t_decl@
113#ifdef HAVE_MIXED_SIZE_ADDRESSING
114typedef int32_t db_ssize_t;
115#else
116typedef ssize_t db_ssize_t;
117#endif
118@time_t_decl@
119
120/*
121 * Sequences are only available on machines with 64-bit integral types.
122 */
123@db_seq_decl@
124
125/* Thread and process identification. */
126@db_threadid_t_decl@
127
128/* Basic types that are exported or quasi-exported. */
129typedef	u_int32_t	db_pgno_t;	/* Page number type. */
130typedef	u_int16_t	db_indx_t;	/* Page offset type. */
131#define	DB_MAX_PAGES	0xffffffff	/* >= # of pages in a file */
132
133typedef	u_int32_t	db_recno_t;	/* Record number type. */
134#define	DB_MAX_RECORDS	0xffffffff	/* >= # of records in a tree */
135
136typedef u_int32_t	db_timeout_t;	/* Type of a timeout. */
137
138/*
139 * Region offsets are the difference between a pointer in a region and the
140 * region's base address.  With private environments, both addresses are the
141 * result of calling malloc, and we can't assume anything about what malloc
142 * will return, so region offsets have to be able to hold differences between
143 * arbitrary pointers.
144 */
145typedef	db_size_t	roff_t;
146
147/*
148 * Forward structure declarations, so we can declare pointers and
149 * applications can get type checking.
150 */
151struct __channel;	typedef struct __channel CHANNEL;
152struct __db;		typedef struct __db DB;
153struct __db_bt_stat;	typedef struct __db_bt_stat DB_BTREE_STAT;
154struct __db_channel;	typedef struct __db_channel DB_CHANNEL;
155struct __db_cipher;	typedef struct __db_cipher DB_CIPHER;
156struct __db_compact;	typedef struct __db_compact DB_COMPACT;
157struct __db_dbt;	typedef struct __db_dbt DBT;
158struct __db_distab;	typedef struct __db_distab DB_DISTAB;
159struct __db_env;	typedef struct __db_env DB_ENV;
160struct __db_h_stat;	typedef struct __db_h_stat DB_HASH_STAT;
161struct __db_heap_rid;	typedef struct __db_heap_rid DB_HEAP_RID;
162struct __db_heap_stat;	typedef struct __db_heap_stat DB_HEAP_STAT;
163struct __db_ilock;	typedef struct __db_ilock DB_LOCK_ILOCK;
164struct __db_lock_hstat;	typedef struct __db_lock_hstat DB_LOCK_HSTAT;
165struct __db_lock_pstat;	typedef struct __db_lock_pstat DB_LOCK_PSTAT;
166struct __db_lock_stat;	typedef struct __db_lock_stat DB_LOCK_STAT;
167struct __db_lock_u;	typedef struct __db_lock_u DB_LOCK;
168struct __db_locker;	typedef struct __db_locker DB_LOCKER;
169struct __db_lockreq;	typedef struct __db_lockreq DB_LOCKREQ;
170struct __db_locktab;	typedef struct __db_locktab DB_LOCKTAB;
171struct __db_log;	typedef struct __db_log DB_LOG;
172struct __db_log_cursor;	typedef struct __db_log_cursor DB_LOGC;
173struct __db_log_stat;	typedef struct __db_log_stat DB_LOG_STAT;
174struct __db_lsn;	typedef struct __db_lsn DB_LSN;
175struct __db_mpool;	typedef struct __db_mpool DB_MPOOL;
176struct __db_mpool_fstat;typedef struct __db_mpool_fstat DB_MPOOL_FSTAT;
177struct __db_mpool_stat;	typedef struct __db_mpool_stat DB_MPOOL_STAT;
178struct __db_mpoolfile;	typedef struct __db_mpoolfile DB_MPOOLFILE;
179struct __db_mutex_stat;	typedef struct __db_mutex_stat DB_MUTEX_STAT;
180struct __db_mutex_t;	typedef struct __db_mutex_t DB_MUTEX;
181struct __db_mutexmgr;	typedef struct __db_mutexmgr DB_MUTEXMGR;
182struct __db_preplist;	typedef struct __db_preplist DB_PREPLIST;
183struct __db_qam_stat;	typedef struct __db_qam_stat DB_QUEUE_STAT;
184struct __db_rep;	typedef struct __db_rep DB_REP;
185struct __db_rep_stat;	typedef struct __db_rep_stat DB_REP_STAT;
186struct __db_repmgr_conn_err;
187	typedef struct __db_repmgr_conn_err DB_REPMGR_CONN_ERR;
188struct __db_repmgr_site;typedef struct __db_repmgr_site DB_REPMGR_SITE;
189struct __db_repmgr_stat;typedef struct __db_repmgr_stat DB_REPMGR_STAT;
190struct __db_seq_record; typedef struct __db_seq_record DB_SEQ_RECORD;
191struct __db_seq_stat;	typedef struct __db_seq_stat DB_SEQUENCE_STAT;
192struct __db_site;	typedef struct __db_site DB_SITE;
193struct __db_sequence;	typedef struct __db_sequence DB_SEQUENCE;
194struct __db_thread_info;typedef struct __db_thread_info DB_THREAD_INFO;
195struct __db_txn;	typedef struct __db_txn DB_TXN;
196struct __db_txn_active;	typedef struct __db_txn_active DB_TXN_ACTIVE;
197struct __db_txn_stat;	typedef struct __db_txn_stat DB_TXN_STAT;
198struct __db_txn_token;	typedef struct __db_txn_token DB_TXN_TOKEN;
199struct __db_txnmgr;	typedef struct __db_txnmgr DB_TXNMGR;
200struct __dbc;		typedef struct __dbc DBC;
201struct __dbc_internal;	typedef struct __dbc_internal DBC_INTERNAL;
202struct __env;		typedef struct __env ENV;
203struct __fh_t;		typedef struct __fh_t DB_FH;
204struct __fname;		typedef struct __fname FNAME;
205struct __key_range;	typedef struct __key_range DB_KEY_RANGE;
206struct __mpoolfile;	typedef struct __mpoolfile MPOOLFILE;
207struct __db_logvrfy_config;
208typedef struct __db_logvrfy_config DB_LOG_VERIFY_CONFIG;
209
210/*
211 * The Berkeley DB API flags are automatically-generated -- the following flag
212 * names are no longer used, but remain for compatibility reasons.
213 */
214#define	DB_DEGREE_2	      DB_READ_COMMITTED
215#define	DB_DIRTY_READ	      DB_READ_UNCOMMITTED
216#define	DB_JOINENV	      0x0
217
218/* Key/data structure -- a Data-Base Thang. */
219struct __db_dbt {
220	void	 *data;			/* Key/data */
221	u_int32_t size;			/* key/data length */
222
223	u_int32_t ulen;			/* RO: length of user buffer. */
224	u_int32_t dlen;			/* RO: get/put record length. */
225	u_int32_t doff;			/* RO: get/put record offset. */
226
227	void *app_data;
228
229#define	DB_DBT_APPMALLOC	0x001	/* Callback allocated memory. */
230#define	DB_DBT_BULK		0x002	/* Internal: Insert if duplicate. */
231#define	DB_DBT_DUPOK		0x004	/* Internal: Insert if duplicate. */
232#define	DB_DBT_ISSET		0x008	/* Lower level calls set value. */
233#define	DB_DBT_MALLOC		0x010	/* Return in malloc'd memory. */
234#define	DB_DBT_MULTIPLE		0x020	/* References multiple records. */
235#define	DB_DBT_PARTIAL		0x040	/* Partial put/get. */
236#define	DB_DBT_REALLOC		0x080	/* Return in realloc'd memory. */
237#define	DB_DBT_READONLY		0x100	/* Readonly, don't update. */
238#define	DB_DBT_STREAMING	0x200	/* Internal: DBT is being streamed. */
239#define	DB_DBT_USERCOPY		0x400	/* Use the user-supplied callback. */
240#define	DB_DBT_USERMEM		0x800	/* Return in user's memory. */
241	u_int32_t flags;
242};
243
244/*******************************************************
245 * Mutexes.
246 *******************************************************/
247/*
248 * When mixed size addressing is supported mutexes need to be the same size
249 * independent of the process address size is.
250 */
251#ifdef HAVE_MIXED_SIZE_ADDRESSING
252typedef db_size_t	db_mutex_t;
253#else
254typedef uintptr_t	db_mutex_t;
255#endif
256
257struct __db_mutex_stat { /* SHARED */
258	/* The following fields are maintained in the region's copy. */
259	u_int32_t st_mutex_align;	/* Mutex alignment */
260	u_int32_t st_mutex_tas_spins;	/* Mutex test-and-set spins */
261	u_int32_t st_mutex_init;	/* Initial mutex count */
262	u_int32_t st_mutex_cnt;		/* Mutex count */
263	u_int32_t st_mutex_max;		/* Mutex max */
264	u_int32_t st_mutex_free;	/* Available mutexes */
265	u_int32_t st_mutex_inuse;	/* Mutexes in use */
266	u_int32_t st_mutex_inuse_max;	/* Maximum mutexes ever in use */
267
268	/* The following fields are filled-in from other places. */
269#ifndef __TEST_DB_NO_STATISTICS
270	uintmax_t st_region_wait;	/* Region lock granted after wait. */
271	uintmax_t st_region_nowait;	/* Region lock granted without wait. */
272	roff_t	  st_regsize;		/* Region size. */
273	roff_t	  st_regmax;		/* Region max. */
274#endif
275};
276
277/* This is the length of the buffer passed to DB_ENV->thread_id_string() */
278#define	DB_THREADID_STRLEN	128
279
280/*******************************************************
281 * Locking.
282 *******************************************************/
283#define	DB_LOCKVERSION	1
284
285#define	DB_FILE_ID_LEN		20	/* Unique file ID length. */
286
287/*
288 * Deadlock detector modes; used in the DB_ENV structure to configure the
289 * locking subsystem.
290 */
291#define	DB_LOCK_NORUN		0
292#define	DB_LOCK_DEFAULT		1	/* Default policy. */
293#define	DB_LOCK_EXPIRE		2	/* Only expire locks, no detection. */
294#define	DB_LOCK_MAXLOCKS	3	/* Select locker with max locks. */
295#define	DB_LOCK_MAXWRITE	4	/* Select locker with max writelocks. */
296#define	DB_LOCK_MINLOCKS	5	/* Select locker with min locks. */
297#define	DB_LOCK_MINWRITE	6	/* Select locker with min writelocks. */
298#define	DB_LOCK_OLDEST		7	/* Select oldest locker. */
299#define	DB_LOCK_RANDOM		8	/* Select random locker. */
300#define	DB_LOCK_YOUNGEST	9	/* Select youngest locker. */
301
302/*
303 * Simple R/W lock modes and for multi-granularity intention locking.
304 *
305 * !!!
306 * These values are NOT random, as they are used as an index into the lock
307 * conflicts arrays, i.e., DB_LOCK_IWRITE must be == 3, and DB_LOCK_IREAD
308 * must be == 4.
309 */
310typedef enum {
311	DB_LOCK_NG=0,			/* Not granted. */
312	DB_LOCK_READ=1,			/* Shared/read. */
313	DB_LOCK_WRITE=2,		/* Exclusive/write. */
314	DB_LOCK_WAIT=3,			/* Wait for event */
315	DB_LOCK_IWRITE=4,		/* Intent exclusive/write. */
316	DB_LOCK_IREAD=5,		/* Intent to share/read. */
317	DB_LOCK_IWR=6,			/* Intent to read and write. */
318	DB_LOCK_READ_UNCOMMITTED=7,	/* Degree 1 isolation. */
319	DB_LOCK_WWRITE=8		/* Was Written. */
320} db_lockmode_t;
321
322/*
323 * Request types.
324 */
325typedef enum {
326	DB_LOCK_DUMP=0,			/* Display held locks. */
327	DB_LOCK_GET=1,			/* Get the lock. */
328	DB_LOCK_GET_TIMEOUT=2,		/* Get lock with a timeout. */
329	DB_LOCK_INHERIT=3,		/* Pass locks to parent. */
330	DB_LOCK_PUT=4,			/* Release the lock. */
331	DB_LOCK_PUT_ALL=5,		/* Release locker's locks. */
332	DB_LOCK_PUT_OBJ=6,		/* Release locker's locks on obj. */
333	DB_LOCK_PUT_READ=7,		/* Release locker's read locks. */
334	DB_LOCK_TIMEOUT=8,		/* Force a txn to timeout. */
335	DB_LOCK_TRADE=9,		/* Trade locker ids on a lock. */
336	DB_LOCK_UPGRADE_WRITE=10	/* Upgrade writes for dirty reads. */
337} db_lockop_t;
338
339/*
340 * Status of a lock.
341 */
342typedef enum  {
343	DB_LSTAT_ABORTED=1,		/* Lock belongs to an aborted txn. */
344	DB_LSTAT_EXPIRED=2,		/* Lock has expired. */
345	DB_LSTAT_FREE=3,		/* Lock is unallocated. */
346	DB_LSTAT_HELD=4,		/* Lock is currently held. */
347	DB_LSTAT_PENDING=5,		/* Lock was waiting and has been
348					 * promoted; waiting for the owner
349					 * to run and upgrade it to held. */
350	DB_LSTAT_WAITING=6		/* Lock is on the wait queue. */
351}db_status_t;
352
353/* Lock statistics structure. */
354struct __db_lock_stat { /* SHARED */
355	u_int32_t st_id;		/* Last allocated locker ID. */
356	u_int32_t st_cur_maxid;		/* Current maximum unused ID. */
357	u_int32_t st_initlocks;		/* Initial number of locks in table. */
358	u_int32_t st_initlockers;	/* Initial num of lockers in table. */
359	u_int32_t st_initobjects;	/* Initial num of objects in table. */
360	u_int32_t st_locks;		/* Current number of locks in table. */
361	u_int32_t st_lockers;		/* Current num of lockers in table. */
362	u_int32_t st_objects;		/* Current num of objects in table. */
363	u_int32_t st_maxlocks;		/* Maximum number of locks in table. */
364	u_int32_t st_maxlockers;	/* Maximum num of lockers in table. */
365	u_int32_t st_maxobjects;	/* Maximum num of objects in table. */
366	u_int32_t st_partitions;	/* number of partitions. */
367	u_int32_t st_tablesize;		/* Size of object hash table. */
368	int32_t   st_nmodes;		/* Number of lock modes. */
369	u_int32_t st_nlockers;		/* Current number of lockers. */
370#ifndef __TEST_DB_NO_STATISTICS
371	u_int32_t st_nlocks;		/* Current number of locks. */
372	u_int32_t st_maxnlocks;		/* Maximum number of locks so far. */
373	u_int32_t st_maxhlocks;		/* Maximum number of locks in any bucket. */
374	uintmax_t st_locksteals;	/* Number of lock steals so far. */
375	uintmax_t st_maxlsteals;	/* Maximum number steals in any partition. */
376	u_int32_t st_maxnlockers;	/* Maximum number of lockers so far. */
377	u_int32_t st_nobjects;		/* Current number of objects. */
378	u_int32_t st_maxnobjects;	/* Maximum number of objects so far. */
379	u_int32_t st_maxhobjects;	/* Maximum number of objectsin any bucket. */
380	uintmax_t st_objectsteals;	/* Number of objects steals so far. */
381	uintmax_t st_maxosteals;	/* Maximum number of steals in any partition. */
382	uintmax_t st_nrequests;		/* Number of lock gets. */
383	uintmax_t st_nreleases;		/* Number of lock puts. */
384	uintmax_t st_nupgrade;		/* Number of lock upgrades. */
385	uintmax_t st_ndowngrade;	/* Number of lock downgrades. */
386	uintmax_t st_lock_wait;		/* Lock conflicts w/ subsequent wait */
387	uintmax_t st_lock_nowait;	/* Lock conflicts w/o subsequent wait */
388	uintmax_t st_ndeadlocks;	/* Number of lock deadlocks. */
389	db_timeout_t st_locktimeout;	/* Lock timeout. */
390	uintmax_t st_nlocktimeouts;	/* Number of lock timeouts. */
391	db_timeout_t st_txntimeout;	/* Transaction timeout. */
392	uintmax_t st_ntxntimeouts;	/* Number of transaction timeouts. */
393	uintmax_t st_part_wait;		/* Partition lock granted after wait. */
394	uintmax_t st_part_nowait;	/* Partition lock granted without wait. */
395	uintmax_t st_part_max_wait;	/* Max partition lock granted after wait. */
396	uintmax_t st_part_max_nowait;	/* Max partition lock granted without wait. */
397	uintmax_t st_objs_wait;	/* 	Object lock granted after wait. */
398	uintmax_t st_objs_nowait;	/* Object lock granted without wait. */
399	uintmax_t st_lockers_wait;	/* Locker lock granted after wait. */
400	uintmax_t st_lockers_nowait;	/* Locker lock granted without wait. */
401	uintmax_t st_region_wait;	/* Region lock granted after wait. */
402	uintmax_t st_region_nowait;	/* Region lock granted without wait. */
403	u_int32_t st_hash_len;		/* Max length of bucket. */
404	roff_t	  st_regsize;		/* Region size. */
405#endif
406};
407
408struct __db_lock_hstat { /* SHARED */
409	uintmax_t st_nrequests;		/* Number of lock gets. */
410	uintmax_t st_nreleases;		/* Number of lock puts. */
411	uintmax_t st_nupgrade;		/* Number of lock upgrades. */
412	uintmax_t st_ndowngrade;	/* Number of lock downgrades. */
413	u_int32_t st_nlocks;		/* Current number of locks. */
414	u_int32_t st_maxnlocks;		/* Maximum number of locks so far. */
415	u_int32_t st_nobjects;		/* Current number of objects. */
416	u_int32_t st_maxnobjects;	/* Maximum number of objects so far. */
417	uintmax_t st_lock_wait;		/* Lock conflicts w/ subsequent wait */
418	uintmax_t st_lock_nowait;	/* Lock conflicts w/o subsequent wait */
419	uintmax_t st_nlocktimeouts;	/* Number of lock timeouts. */
420	uintmax_t st_ntxntimeouts;	/* Number of transaction timeouts. */
421	u_int32_t st_hash_len;		/* Max length of bucket. */
422};
423
424struct __db_lock_pstat { /* SHARED */
425	u_int32_t st_nlocks;		/* Current number of locks. */
426	u_int32_t st_maxnlocks;		/* Maximum number of locks so far. */
427	u_int32_t st_nobjects;		/* Current number of objects. */
428	u_int32_t st_maxnobjects;	/* Maximum number of objects so far. */
429	uintmax_t st_locksteals;	/* Number of lock steals so far. */
430	uintmax_t st_objectsteals;	/* Number of objects steals so far. */
431};
432
433/*
434 * DB_LOCK_ILOCK --
435 *	Internal DB access method lock.
436 */
437struct __db_ilock { /* SHARED */
438	db_pgno_t pgno;			/* Page being locked. */
439	u_int8_t fileid[DB_FILE_ID_LEN];/* File id. */
440#define	DB_HANDLE_LOCK		1
441#define	DB_RECORD_LOCK		2
442#define	DB_PAGE_LOCK		3
443#define	DB_DATABASE_LOCK	4
444	u_int32_t type;			/* Type of lock. */
445};
446
447/*
448 * DB_LOCK --
449 *	The structure is allocated by the caller and filled in during a
450 *	lock_get request (or a lock_vec/DB_LOCK_GET).
451 */
452struct __db_lock_u { /* SHARED */
453	roff_t		off;		/* Offset of the lock in the region */
454	u_int32_t	ndx;		/* Index of the object referenced by
455					 * this lock; used for locking. */
456	u_int32_t	gen;		/* Generation number of this lock. */
457	db_lockmode_t	mode;		/* mode of this lock. */
458};
459
460/* Lock request structure. */
461struct __db_lockreq {
462	db_lockop_t	 op;		/* Operation. */
463	db_lockmode_t	 mode;		/* Requested mode. */
464	db_timeout_t	 timeout;	/* Time to expire lock. */
465	DBT		*obj;		/* Object being locked. */
466	DB_LOCK		 lock;		/* Lock returned. */
467};
468
469/*******************************************************
470 * Logging.
471 *******************************************************/
472#define	DB_LOGVERSION	19		/* Current log version. */
473#define	DB_LOGVERSION_LATCHING 15	/* Log version using latching: db-4.8 */
474#define	DB_LOGCHKSUM	12		/* Check sum headers: db-4.5 */
475#define	DB_LOGOLDVER	8		/* Oldest version supported: db-4.2 */
476#define	DB_LOGMAGIC	0x040988
477
478/*
479 * A DB_LSN has two parts, a fileid which identifies a specific file, and an
480 * offset within that file.  The fileid is an unsigned 4-byte quantity that
481 * uniquely identifies a file within the log directory -- currently a simple
482 * counter inside the log.  The offset is also an unsigned 4-byte value.  The
483 * log manager guarantees the offset is never more than 4 bytes by switching
484 * to a new log file before the maximum length imposed by an unsigned 4-byte
485 * offset is reached.
486 */
487struct __db_lsn { /* SHARED */
488	u_int32_t	file;		/* File ID. */
489	u_int32_t	offset;		/* File offset. */
490};
491
492/*
493 * Application-specified log record types start at DB_user_BEGIN, and must not
494 * equal or exceed DB_debug_FLAG.
495 *
496 * DB_debug_FLAG is the high-bit of the u_int32_t that specifies a log record
497 * type.  If the flag is set, it's a log record that was logged for debugging
498 * purposes only, even if it reflects a database change -- the change was part
499 * of a non-durable transaction.
500 */
501#define	DB_user_BEGIN		10000
502#define	DB_debug_FLAG		0x80000000
503
504/*
505 * DB_LOGC --
506 *	Log cursor.
507 */
508struct __db_log_cursor {
509	ENV	 *env;			/* Environment */
510
511	DB_FH	 *fhp;			/* File handle. */
512	DB_LSN	  lsn;			/* Cursor: LSN */
513	u_int32_t len;			/* Cursor: record length */
514	u_int32_t prev;			/* Cursor: previous record's offset */
515
516	DBT	  dbt;			/* Return DBT. */
517	DB_LSN    p_lsn;		/* Persist LSN. */
518	u_int32_t p_version;		/* Persist version. */
519
520	u_int8_t *bp;			/* Allocated read buffer. */
521	u_int32_t bp_size;		/* Read buffer length in bytes. */
522	u_int32_t bp_rlen;		/* Read buffer valid data length. */
523	DB_LSN	  bp_lsn;		/* Read buffer first byte LSN. */
524
525	u_int32_t bp_maxrec;		/* Max record length in the log file. */
526
527	/* DB_LOGC PUBLIC HANDLE LIST BEGIN */
528	int (*close) __P((DB_LOGC *, u_int32_t));
529	int (*get) __P((DB_LOGC *, DB_LSN *, DBT *, u_int32_t));
530	int (*version) __P((DB_LOGC *, u_int32_t *, u_int32_t));
531	/* DB_LOGC PUBLIC HANDLE LIST END */
532
533#define	DB_LOG_DISK		0x01	/* Log record came from disk. */
534#define	DB_LOG_LOCKED		0x02	/* Log region already locked */
535#define	DB_LOG_SILENT_ERR	0x04	/* Turn-off error messages. */
536	u_int32_t flags;
537};
538
539/* Log statistics structure. */
540struct __db_log_stat { /* SHARED */
541	u_int32_t st_magic;		/* Log file magic number. */
542	u_int32_t st_version;		/* Log file version number. */
543	int32_t   st_mode;		/* Log file permissions mode. */
544	u_int32_t st_lg_bsize;		/* Log buffer size. */
545	u_int32_t st_lg_size;		/* Log file size. */
546	u_int32_t st_wc_bytes;		/* Bytes to log since checkpoint. */
547	u_int32_t st_wc_mbytes;		/* Megabytes to log since checkpoint. */
548	u_int32_t st_fileid_init;	/* Initial allocation for fileids. */
549#ifndef __TEST_DB_NO_STATISTICS
550	u_int32_t st_nfileid;		/* Current number of fileids. */
551	u_int32_t st_maxnfileid;	/* Maximum number of fileids used. */
552	uintmax_t st_record;		/* Records entered into the log. */
553	u_int32_t st_w_bytes;		/* Bytes to log. */
554	u_int32_t st_w_mbytes;		/* Megabytes to log. */
555	uintmax_t st_wcount;		/* Total I/O writes to the log. */
556	uintmax_t st_wcount_fill;	/* Overflow writes to the log. */
557	uintmax_t st_rcount;		/* Total I/O reads from the log. */
558	uintmax_t st_scount;		/* Total syncs to the log. */
559	uintmax_t st_region_wait;	/* Region lock granted after wait. */
560	uintmax_t st_region_nowait;	/* Region lock granted without wait. */
561	u_int32_t st_cur_file;		/* Current log file number. */
562	u_int32_t st_cur_offset;	/* Current log file offset. */
563	u_int32_t st_disk_file;		/* Known on disk log file number. */
564	u_int32_t st_disk_offset;	/* Known on disk log file offset. */
565	u_int32_t st_maxcommitperflush;	/* Max number of commits in a flush. */
566	u_int32_t st_mincommitperflush;	/* Min number of commits in a flush. */
567	roff_t	  st_regsize;		/* Region size. */
568#endif
569};
570
571/*
572 * We need to record the first log record of a transaction.  For user
573 * defined logging this macro returns the place to put that information,
574 * if it is need in rlsnp, otherwise it leaves it unchanged.  We also
575 * need to track the last record of the transaction, this returns the
576 * place to put that info.
577 */
578#define	DB_SET_TXN_LSNP(txn, blsnp, llsnp)		\
579	((txn)->set_txn_lsnp(txn, blsnp, llsnp))
580
581/*
582 * Definition of the structure which specifies marshalling of log records.
583 */
584typedef enum {
585	LOGREC_Done,
586	LOGREC_ARG,
587	LOGREC_HDR,
588	LOGREC_DATA,
589	LOGREC_DB,
590	LOGREC_DBOP,
591	LOGREC_DBT,
592	LOGREC_LOCKS,
593	LOGREC_OP,
594	LOGREC_PGDBT,
595	LOGREC_PGDDBT,
596	LOGREC_PGLIST,
597	LOGREC_POINTER,
598	LOGREC_TIME
599} log_rec_type_t;
600
601typedef const struct __log_rec_spec {
602	log_rec_type_t	type;
603	u_int32_t	offset;
604	const char 	*name;
605	const char	fmt[4];
606} DB_LOG_RECSPEC;
607
608/*
609 * Size of a DBT in a log record.
610 */
611#define	LOG_DBT_SIZE(dbt)						\
612    (sizeof(u_int32_t) + ((dbt) == NULL ? 0 : (dbt)->size))
613
614/*******************************************************
615 * Shared buffer cache (mpool).
616 *******************************************************/
617/* Priority values for DB_MPOOLFILE->{put,set_priority}. */
618typedef enum {
619	DB_PRIORITY_UNCHANGED=0,
620	DB_PRIORITY_VERY_LOW=1,
621	DB_PRIORITY_LOW=2,
622	DB_PRIORITY_DEFAULT=3,
623	DB_PRIORITY_HIGH=4,
624	DB_PRIORITY_VERY_HIGH=5
625} DB_CACHE_PRIORITY;
626
627/* Per-process DB_MPOOLFILE information. */
628struct __db_mpoolfile {
629	DB_FH	  *fhp;			/* Underlying file handle. */
630
631	/*
632	 * !!!
633	 * The ref, pinref and q fields are protected by the region lock.
634	 */
635	u_int32_t  ref;			/* Reference count. */
636
637	u_int32_t pinref;		/* Pinned block reference count. */
638
639	/*
640	 * !!!
641	 * Explicit representations of structures from queue.h.
642	 * TAILQ_ENTRY(__db_mpoolfile) q;
643	 */
644	struct {
645		struct __db_mpoolfile *tqe_next;
646		struct __db_mpoolfile **tqe_prev;
647	} q;				/* Linked list of DB_MPOOLFILE's. */
648
649	/*
650	 * !!!
651	 * The rest of the fields (with the exception of the MP_FLUSH flag)
652	 * are not thread-protected, even when they may be modified at any
653	 * time by the application.  The reason is the DB_MPOOLFILE handle
654	 * is single-threaded from the viewpoint of the application, and so
655	 * the only fields needing to be thread-protected are those accessed
656	 * by checkpoint or sync threads when using DB_MPOOLFILE structures
657	 * to flush buffers from the cache.
658	 */
659	ENV	       *env;		/* Environment */
660	MPOOLFILE      *mfp;		/* Underlying MPOOLFILE. */
661
662	u_int32_t	clear_len;	/* Cleared length on created pages. */
663	u_int8_t			/* Unique file ID. */
664			fileid[DB_FILE_ID_LEN];
665	int		ftype;		/* File type. */
666	int32_t		lsn_offset;	/* LSN offset in page. */
667	u_int32_t	gbytes, bytes;	/* Maximum file size. */
668	DBT	       *pgcookie;	/* Byte-string passed to pgin/pgout. */
669	int32_t		priority;	/* Cache priority. */
670
671	void	       *addr;		/* Address of mmap'd region. */
672	size_t		len;		/* Length of mmap'd region. */
673
674	u_int32_t	config_flags;	/* Flags to DB_MPOOLFILE->set_flags. */
675
676	/* DB_MPOOLFILE PUBLIC HANDLE LIST BEGIN */
677	int (*close) __P((DB_MPOOLFILE *, u_int32_t));
678	int (*get)
679	    __P((DB_MPOOLFILE *, db_pgno_t *, DB_TXN *, u_int32_t, void *));
680	int (*get_clear_len) __P((DB_MPOOLFILE *, u_int32_t *));
681	int (*get_fileid) __P((DB_MPOOLFILE *, u_int8_t *));
682	int (*get_flags) __P((DB_MPOOLFILE *, u_int32_t *));
683	int (*get_ftype) __P((DB_MPOOLFILE *, int *));
684	int (*get_last_pgno) __P((DB_MPOOLFILE *, db_pgno_t *));
685	int (*get_lsn_offset) __P((DB_MPOOLFILE *, int32_t *));
686	int (*get_maxsize) __P((DB_MPOOLFILE *, u_int32_t *, u_int32_t *));
687	int (*get_pgcookie) __P((DB_MPOOLFILE *, DBT *));
688	int (*get_priority) __P((DB_MPOOLFILE *, DB_CACHE_PRIORITY *));
689	int (*open) __P((DB_MPOOLFILE *, const char *, u_int32_t, int, size_t));
690	int (*put) __P((DB_MPOOLFILE *, void *, DB_CACHE_PRIORITY, u_int32_t));
691	int (*set_clear_len) __P((DB_MPOOLFILE *, u_int32_t));
692	int (*set_fileid) __P((DB_MPOOLFILE *, u_int8_t *));
693	int (*set_flags) __P((DB_MPOOLFILE *, u_int32_t, int));
694	int (*set_ftype) __P((DB_MPOOLFILE *, int));
695	int (*set_lsn_offset) __P((DB_MPOOLFILE *, int32_t));
696	int (*set_maxsize) __P((DB_MPOOLFILE *, u_int32_t, u_int32_t));
697	int (*set_pgcookie) __P((DB_MPOOLFILE *, DBT *));
698	int (*set_priority) __P((DB_MPOOLFILE *, DB_CACHE_PRIORITY));
699	int (*sync) __P((DB_MPOOLFILE *));
700	/* DB_MPOOLFILE PUBLIC HANDLE LIST END */
701
702	/*
703	 * MP_FILEID_SET, MP_OPEN_CALLED and MP_READONLY do not need to be
704	 * thread protected because they are initialized before the file is
705	 * linked onto the per-process lists, and never modified.
706	 *
707	 * MP_FLUSH is thread protected because it is potentially read/set by
708	 * multiple threads of control.
709	 */
710#define	MP_FILEID_SET	0x001		/* Application supplied a file ID. */
711#define	MP_FLUSH	0x002		/* Was used to flush a buffer. */
712#define	MP_FOR_FLUSH	0x004		/* Was opened to flush a buffer. */
713#define	MP_MULTIVERSION	0x008		/* Opened for multiversion access. */
714#define	MP_OPEN_CALLED	0x010		/* File opened. */
715#define	MP_READONLY	0x020		/* File is readonly. */
716#define	MP_DUMMY	0x040		/* File is dummy for __memp_fput. */
717	u_int32_t  flags;
718};
719
720/* Mpool statistics structure. */
721struct __db_mpool_stat { /* SHARED */
722	u_int32_t st_gbytes;		/* Total cache size: GB. */
723	u_int32_t st_bytes;		/* Total cache size: B. */
724	u_int32_t st_ncache;		/* Number of cache regions. */
725	u_int32_t st_max_ncache;	/* Maximum number of regions. */
726	db_size_t st_mmapsize;		/* Maximum file size for mmap. */
727	int32_t st_maxopenfd;		/* Maximum number of open fd's. */
728	int32_t st_maxwrite;		/* Maximum buffers to write. */
729	db_timeout_t st_maxwrite_sleep;	/* Sleep after writing max buffers. */
730	u_int32_t st_pages;		/* Total number of pages. */
731#ifndef __TEST_DB_NO_STATISTICS
732	u_int32_t st_map;		/* Pages from mapped files. */
733	uintmax_t st_cache_hit;	/* Pages found in the cache. */
734	uintmax_t st_cache_miss;	/* Pages not found in the cache. */
735	uintmax_t st_page_create;	/* Pages created in the cache. */
736	uintmax_t st_page_in;		/* Pages read in. */
737	uintmax_t st_page_out;		/* Pages written out. */
738	uintmax_t st_ro_evict;		/* Clean pages forced from the cache. */
739	uintmax_t st_rw_evict;		/* Dirty pages forced from the cache. */
740	uintmax_t st_page_trickle;	/* Pages written by memp_trickle. */
741	u_int32_t st_page_clean;	/* Clean pages. */
742	u_int32_t st_page_dirty;	/* Dirty pages. */
743	u_int32_t st_hash_buckets;	/* Number of hash buckets. */
744	u_int32_t st_hash_mutexes;	/* Number of hash bucket mutexes. */
745	u_int32_t st_pagesize;		/* Assumed page size. */
746	u_int32_t st_hash_searches;	/* Total hash chain searches. */
747	u_int32_t st_hash_longest;	/* Longest hash chain searched. */
748	uintmax_t st_hash_examined;	/* Total hash entries searched. */
749	uintmax_t st_hash_nowait;	/* Hash lock granted with nowait. */
750	uintmax_t st_hash_wait;		/* Hash lock granted after wait. */
751	uintmax_t st_hash_max_nowait;	/* Max hash lock granted with nowait. */
752	uintmax_t st_hash_max_wait;	/* Max hash lock granted after wait. */
753	uintmax_t st_region_nowait;	/* Region lock granted with nowait. */
754	uintmax_t st_region_wait;	/* Region lock granted after wait. */
755	uintmax_t st_mvcc_frozen;	/* Buffers frozen. */
756	uintmax_t st_mvcc_thawed;	/* Buffers thawed. */
757	uintmax_t st_mvcc_freed;	/* Frozen buffers freed. */
758	uintmax_t st_alloc;		/* Number of page allocations. */
759	uintmax_t st_alloc_buckets;	/* Buckets checked during allocation. */
760	uintmax_t st_alloc_max_buckets;/* Max checked during allocation. */
761	uintmax_t st_alloc_pages;	/* Pages checked during allocation. */
762	uintmax_t st_alloc_max_pages;	/* Max checked during allocation. */
763	uintmax_t st_io_wait;		/* Thread waited on buffer I/O. */
764	uintmax_t st_sync_interrupted;	/* Number of times sync interrupted. */
765	roff_t	  st_regsize;		/* Region size. */
766	roff_t	  st_regmax;		/* Region max. */
767#endif
768};
769
770/*
771 * Mpool file statistics structure.
772 * The first fields in this structure must mirror the __db_mpool_fstat_int
773 * structure, since content is mem copied between the two.
774 */
775struct __db_mpool_fstat {
776	u_int32_t st_pagesize;		/* Page size. */
777#ifndef __TEST_DB_NO_STATISTICS
778	u_int32_t st_map;		/* Pages from mapped files. */
779	uintmax_t st_cache_hit;	/* Pages found in the cache. */
780	uintmax_t st_cache_miss;	/* Pages not found in the cache. */
781	uintmax_t st_page_create;	/* Pages created in the cache. */
782	uintmax_t st_page_in;		/* Pages read in. */
783	uintmax_t st_page_out;		/* Pages written out. */
784	uintmax_t st_backup_spins;	/* Number of spins during a copy. */
785#endif
786	char *file_name;	/* File name. */
787};
788
789/*******************************************************
790 * Transactions and recovery.
791 *******************************************************/
792#define	DB_TXNVERSION	1
793
794typedef enum {
795	DB_TXN_ABORT=0,			/* Public. */
796	DB_TXN_APPLY=1,			/* Public. */
797	DB_TXN_BACKWARD_ROLL=3,		/* Public. */
798	DB_TXN_FORWARD_ROLL=4,		/* Public. */
799	DB_TXN_OPENFILES=5,		/* Internal. */
800	DB_TXN_POPENFILES=6,		/* Internal. */
801	DB_TXN_PRINT=7,			/* Public. */
802	DB_TXN_LOG_VERIFY=8		/* Internal. */
803} db_recops;
804
805/*
806 * BACKWARD_ALLOC is used during the forward pass to pick up any aborted
807 * allocations for files that were created during the forward pass.
808 * The main difference between _ALLOC and _ROLL is that the entry for
809 * the file not exist during the rollforward pass.
810 */
811#define	DB_UNDO(op)	((op) == DB_TXN_ABORT || (op) == DB_TXN_BACKWARD_ROLL)
812#define	DB_REDO(op)	((op) == DB_TXN_FORWARD_ROLL || (op) == DB_TXN_APPLY)
813
814struct __db_txn {
815	DB_TXNMGR	*mgrp;		/* Pointer to transaction manager. */
816	DB_TXN		*parent;	/* Pointer to transaction's parent. */
817	DB_THREAD_INFO	*thread_info;	/* Pointer to thread information. */
818
819	u_int32_t	txnid;		/* Unique transaction id. */
820	char		*name;		/* Transaction name. */
821	DB_LOCKER	*locker;	/* Locker for this txn. */
822
823	void		*td;		/* Detail structure within region. */
824	db_timeout_t	lock_timeout;	/* Timeout for locks for this txn. */
825	void		*txn_list;	/* Undo information for parent. */
826
827	/*
828	 * !!!
829	 * Explicit representations of structures from queue.h.
830	 * TAILQ_ENTRY(__db_txn) links;
831	 */
832	struct {
833		struct __db_txn *tqe_next;
834		struct __db_txn **tqe_prev;
835	} links;			/* Links transactions off manager. */
836
837	/*
838	 * !!!
839	 * Explicit representations of structures from shqueue.h.
840	 * SH_TAILQ_ENTRY xa_links;
841	 * These links link together transactions that are active in
842	 * the same thread of control.
843	 */
844	struct {
845		db_ssize_t stqe_next;
846		db_ssize_t stqe_prev;
847	} xa_links;			/* Links XA transactions. */
848
849	/*
850	 * !!!
851	 * Explicit representations of structures from queue.h.
852	 * TAILQ_HEAD(__kids, __db_txn) kids;
853	 */
854	struct __kids {
855		struct __db_txn *tqh_first;
856		struct __db_txn **tqh_last;
857	} kids;
858
859	/*
860	 * !!!
861	 * Explicit representations of structures from queue.h.
862	 * TAILQ_HEAD(__events, __txn_event) events;
863	 */
864	struct {
865		struct __txn_event *tqh_first;
866		struct __txn_event **tqh_last;
867	} events;			/* Links deferred events. */
868
869	/*
870	 * !!!
871	 * Explicit representations of structures from queue.h.
872	 * STAILQ_HEAD(__logrec, __txn_logrec) logs;
873	 */
874	struct {
875		struct __txn_logrec *stqh_first;
876		struct __txn_logrec **stqh_last;
877	} logs;				/* Links in memory log records. */
878
879	/*
880	 * !!!
881	 * Explicit representations of structures from queue.h.
882	 * TAILQ_ENTRY(__db_txn) klinks;
883	 */
884	struct {
885		struct __db_txn *tqe_next;
886		struct __db_txn **tqe_prev;
887	} klinks;			/* Links of children in parent. */
888
889	/*
890	 * !!!
891	 * Explicit representations of structures from queue.h.
892	 * TAILQ_HEAD(__my_cursors, __dbc) my_cursors;
893	 */
894	struct __my_cursors {
895		struct __dbc *tqh_first;
896		struct __dbc **tqh_last;
897	} my_cursors;
898
899	/*
900	 * !!!
901	 * Explicit representations of structures from queue.h.
902	 * TAILQ_HEAD(__femfs, MPOOLFILE) femfs;
903	 *
904	 * These are DBs involved in file extension in this transaction.
905	 */
906	struct __femfs {
907		DB *tqh_first;
908		DB **tqh_last;
909	} femfs;
910
911	DB_TXN_TOKEN	*token_buffer;	/* User's commit token buffer. */
912	void	*api_internal;		/* C++ API private. */
913	void	*xml_internal;		/* XML API private. */
914
915	u_int32_t	cursors;	/* Number of cursors open for txn */
916
917	/* DB_TXN PUBLIC HANDLE LIST BEGIN */
918	int	  (*abort) __P((DB_TXN *));
919	int	  (*commit) __P((DB_TXN *, u_int32_t));
920	int	  (*discard) __P((DB_TXN *, u_int32_t));
921	int	  (*get_name) __P((DB_TXN *, const char **));
922	int	  (*get_priority) __P((DB_TXN *, u_int32_t *));
923	u_int32_t (*id) __P((DB_TXN *));
924	int	  (*prepare) __P((DB_TXN *, u_int8_t *));
925	int	  (*set_commit_token) __P((DB_TXN *, DB_TXN_TOKEN *));
926	int	  (*set_name) __P((DB_TXN *, const char *));
927	int	  (*set_priority) __P((DB_TXN *, u_int32_t));
928	int	  (*set_timeout) __P((DB_TXN *, db_timeout_t, u_int32_t));
929	/* DB_TXN PUBLIC HANDLE LIST END */
930
931	/* DB_TXN PRIVATE HANDLE LIST BEGIN */
932	void	  (*set_txn_lsnp) __P((DB_TXN *txn, DB_LSN **, DB_LSN **));
933	/* DB_TXN PRIVATE HANDLE LIST END */
934
935#define	TXN_XA_THREAD_NOTA		0
936#define	TXN_XA_THREAD_ASSOCIATED	1
937#define	TXN_XA_THREAD_SUSPENDED		2
938#define	TXN_XA_THREAD_UNASSOCIATED 	3
939	u_int32_t	xa_thr_status;
940
941#define	TXN_CHILDCOMMIT		0x00001	/* Txn has committed. */
942#define	TXN_COMPENSATE		0x00002	/* Compensating transaction. */
943#define	TXN_DEADLOCK		0x00004	/* Txn has deadlocked. */
944#define	TXN_FAMILY		0x00008	/* Cursors/children are independent. */
945#define	TXN_IGNORE_LEASE	0x00010	/* Skip lease check at commit time. */
946#define	TXN_INFAMILY		0x00020	/* Part of a transaction family. */
947#define	TXN_LOCKTIMEOUT		0x00040	/* Txn has a lock timeout. */
948#define	TXN_MALLOC		0x00080	/* Structure allocated by TXN system. */
949#define	TXN_NOSYNC		0x00100	/* Do not sync on prepare and commit. */
950#define	TXN_NOWAIT		0x00200	/* Do not wait on locks. */
951#define	TXN_PRIVATE		0x00400	/* Txn owned by cursor. */
952#define	TXN_READONLY		0x00800	/* CDS group handle. */
953#define	TXN_READ_COMMITTED	0x01000	/* Txn has degree 2 isolation. */
954#define	TXN_READ_UNCOMMITTED	0x02000	/* Txn has degree 1 isolation. */
955#define	TXN_RESTORED		0x04000	/* Txn has been restored. */
956#define	TXN_SNAPSHOT		0x08000	/* Snapshot Isolation. */
957#define	TXN_SYNC		0x10000	/* Write and sync on prepare/commit. */
958#define	TXN_WRITE_NOSYNC	0x20000	/* Write only on prepare/commit. */
959#define TXN_BULK		0x40000 /* Enable bulk loading optimization. */
960	u_int32_t	flags;
961};
962
963#define	TXN_SYNC_FLAGS (TXN_SYNC | TXN_NOSYNC | TXN_WRITE_NOSYNC)
964
965/*
966 * Structure used for two phase commit interface.
967 * We set the size of our global transaction id (gid) to be 128 in order
968 * to match that defined by the XA X/Open standard.
969 */
970#define	DB_GID_SIZE	128
971struct __db_preplist {
972	DB_TXN	*txn;
973	u_int8_t gid[DB_GID_SIZE];
974};
975
976/* Transaction statistics structure. */
977struct __db_txn_active {
978	u_int32_t txnid;		/* Transaction ID */
979	u_int32_t parentid;		/* Transaction ID of parent */
980	pid_t     pid;			/* Process owning txn ID */
981	db_threadid_t tid;		/* Thread owning txn ID */
982
983	DB_LSN	  lsn;			/* LSN when transaction began */
984
985	DB_LSN	  read_lsn;		/* Read LSN for MVCC */
986	u_int32_t mvcc_ref;		/* MVCC reference count */
987
988	u_int32_t priority;		/* Deadlock resolution priority */
989
990#define	TXN_ABORTED		1
991#define	TXN_COMMITTED		2
992#define	TXN_NEED_ABORT		3
993#define	TXN_PREPARED		4
994#define	TXN_RUNNING		5
995	u_int32_t status;		/* Status of the transaction */
996
997#define	TXN_XA_ACTIVE		1
998#define	TXN_XA_DEADLOCKED	2
999#define	TXN_XA_IDLE		3
1000#define	TXN_XA_PREPARED		4
1001#define	TXN_XA_ROLLEDBACK	5
1002	u_int32_t xa_status;		/* XA status */
1003
1004	u_int8_t  gid[DB_GID_SIZE];	/* Global transaction ID */
1005	char	  name[51];		/* 50 bytes of name, nul termination */
1006};
1007
1008struct __db_txn_stat {
1009	u_int32_t st_nrestores;		/* number of restored transactions
1010					   after recovery. */
1011#ifndef __TEST_DB_NO_STATISTICS
1012	DB_LSN	  st_last_ckp;		/* lsn of the last checkpoint */
1013	time_t	  st_time_ckp;		/* time of last checkpoint */
1014	u_int32_t st_last_txnid;	/* last transaction id given out */
1015	u_int32_t st_inittxns;		/* inital txns allocated */
1016	u_int32_t st_maxtxns;		/* maximum txns possible */
1017	uintmax_t st_naborts;		/* number of aborted transactions */
1018	uintmax_t st_nbegins;		/* number of begun transactions */
1019	uintmax_t st_ncommits;		/* number of committed transactions */
1020	u_int32_t st_nactive;		/* number of active transactions */
1021	u_int32_t st_nsnapshot;		/* number of snapshot transactions */
1022	u_int32_t st_maxnactive;	/* maximum active transactions */
1023	u_int32_t st_maxnsnapshot;	/* maximum snapshot transactions */
1024	uintmax_t st_region_wait;	/* Region lock granted after wait. */
1025	uintmax_t st_region_nowait;	/* Region lock granted without wait. */
1026	roff_t	  st_regsize;		/* Region size. */
1027	DB_TXN_ACTIVE *st_txnarray;	/* array of active transactions */
1028#endif
1029};
1030
1031#define	DB_TXN_TOKEN_SIZE		20
1032struct __db_txn_token {
1033	u_int8_t buf[DB_TXN_TOKEN_SIZE];
1034};
1035
1036/*******************************************************
1037 * Replication.
1038 *******************************************************/
1039/* Special, out-of-band environment IDs. */
1040#define	DB_EID_BROADCAST	-1
1041#define	DB_EID_INVALID		-2
1042#define	DB_EID_MASTER		-3
1043
1044#define	DB_REP_DEFAULT_PRIORITY		100
1045
1046/* Acknowledgement policies; 0 reserved as OOB. */
1047#define	DB_REPMGR_ACKS_ALL		1
1048#define	DB_REPMGR_ACKS_ALL_AVAILABLE	2
1049#define	DB_REPMGR_ACKS_ALL_PEERS	3
1050#define	DB_REPMGR_ACKS_NONE		4
1051#define	DB_REPMGR_ACKS_ONE		5
1052#define	DB_REPMGR_ACKS_ONE_PEER		6
1053#define	DB_REPMGR_ACKS_QUORUM		7
1054
1055/* Replication timeout configuration values. */
1056#define	DB_REP_ACK_TIMEOUT		1	/* RepMgr acknowledgements. */
1057#define	DB_REP_CHECKPOINT_DELAY		2	/* Master checkpoint delay. */
1058#define	DB_REP_CONNECTION_RETRY		3	/* RepMgr connections. */
1059#define	DB_REP_ELECTION_RETRY		4	/* RepMgr elect retries. */
1060#define	DB_REP_ELECTION_TIMEOUT		5	/* Rep normal elections. */
1061#define	DB_REP_FULL_ELECTION_TIMEOUT	6	/* Rep full elections. */
1062#define	DB_REP_HEARTBEAT_MONITOR	7	/* RepMgr client HB monitor. */
1063#define	DB_REP_HEARTBEAT_SEND		8	/* RepMgr master send freq. */
1064#define	DB_REP_LEASE_TIMEOUT		9	/* Master leases. */
1065
1066/*
1067 * Event notification types.  (Tcl testing interface currently assumes there are
1068 * no more than 32 of these.)
1069 */
1070#define	DB_EVENT_PANIC			 0
1071#define	DB_EVENT_REG_ALIVE		 1
1072#define	DB_EVENT_REG_PANIC		 2
1073#define	DB_EVENT_REP_CLIENT		 3
1074#define	DB_EVENT_REP_CONNECT_BROKEN	 4
1075#define	DB_EVENT_REP_CONNECT_ESTD	 5
1076#define	DB_EVENT_REP_CONNECT_TRY_FAILED	 6
1077#define	DB_EVENT_REP_DUPMASTER		 7
1078#define	DB_EVENT_REP_ELECTED		 8
1079#define	DB_EVENT_REP_ELECTION_FAILED	 9
1080#define	DB_EVENT_REP_INIT_DONE		10
1081#define	DB_EVENT_REP_JOIN_FAILURE	11
1082#define	DB_EVENT_REP_LOCAL_SITE_REMOVED	12
1083#define	DB_EVENT_REP_MASTER		13
1084#define	DB_EVENT_REP_MASTER_FAILURE	14
1085#define	DB_EVENT_REP_NEWMASTER		15
1086#define	DB_EVENT_REP_PERM_FAILED	16
1087#define	DB_EVENT_REP_SITE_ADDED		17
1088#define	DB_EVENT_REP_SITE_REMOVED	18
1089#define	DB_EVENT_REP_STARTUPDONE	19
1090#define	DB_EVENT_REP_WOULD_ROLLBACK	20	/* Undocumented; C API only. */
1091#define	DB_EVENT_WRITE_FAILED		21
1092#define	DB_EVENT_NO_SUCH_EVENT		 0xffffffff /* OOB sentinel value */
1093
1094/* Replication Manager site status. */
1095struct __db_repmgr_site {
1096	int eid;
1097	char *host;
1098	u_int port;
1099
1100#define	DB_REPMGR_CONNECTED	1
1101#define	DB_REPMGR_DISCONNECTED	2
1102	u_int32_t status;
1103
1104#define	DB_REPMGR_ISPEER	0x01
1105	u_int32_t flags;
1106};
1107
1108/* Replication statistics. */
1109struct __db_rep_stat { /* SHARED */
1110	/* !!!
1111	 * Many replication statistics fields cannot be protected by a mutex
1112	 * without an unacceptable performance penalty, since most message
1113	 * processing is done without the need to hold a region-wide lock.
1114	 * Fields whose comments end with a '+' may be updated without holding
1115	 * the replication or log mutexes (as appropriate), and thus may be
1116	 * off somewhat (or, on unreasonable architectures under unlucky
1117	 * circumstances, garbaged).
1118	 */
1119	u_int32_t st_startup_complete;	/* Site completed client sync-up. */
1120#ifndef __TEST_DB_NO_STATISTICS
1121	uintmax_t st_log_queued;	/* Log records currently queued.+ */
1122	u_int32_t st_status;		/* Current replication status. */
1123	DB_LSN st_next_lsn;		/* Next LSN to use or expect. */
1124	DB_LSN st_waiting_lsn;		/* LSN we're awaiting, if any. */
1125	DB_LSN st_max_perm_lsn;		/* Maximum permanent LSN. */
1126	db_pgno_t st_next_pg;		/* Next pg we expect. */
1127	db_pgno_t st_waiting_pg;	/* pg we're awaiting, if any. */
1128
1129	u_int32_t st_dupmasters;	/* # of times a duplicate master
1130					   condition was detected.+ */
1131	db_ssize_t st_env_id;		/* Current environment ID. */
1132	u_int32_t st_env_priority;	/* Current environment priority. */
1133	uintmax_t st_bulk_fills;	/* Bulk buffer fills. */
1134	uintmax_t st_bulk_overflows;	/* Bulk buffer overflows. */
1135	uintmax_t st_bulk_records;	/* Bulk records stored. */
1136	uintmax_t st_bulk_transfers;	/* Transfers of bulk buffers. */
1137	uintmax_t st_client_rerequests;/* Number of forced rerequests. */
1138	uintmax_t st_client_svc_req;	/* Number of client service requests
1139					   received by this client. */
1140	uintmax_t st_client_svc_miss;	/* Number of client service requests
1141					   missing on this client. */
1142	u_int32_t st_gen;		/* Current generation number. */
1143	u_int32_t st_egen;		/* Current election gen number. */
1144	uintmax_t st_lease_chk;		/* Lease validity checks. */
1145	uintmax_t st_lease_chk_misses;	/* Lease checks invalid. */
1146	uintmax_t st_lease_chk_refresh;	/* Lease refresh attempts. */
1147	uintmax_t st_lease_sends;	/* Lease messages sent live. */
1148
1149	uintmax_t st_log_duplicated;	/* Log records received multiply.+ */
1150	uintmax_t st_log_queued_max;	/* Max. log records queued at once.+ */
1151	uintmax_t st_log_queued_total;	/* Total # of log recs. ever queued.+ */
1152	uintmax_t st_log_records;	/* Log records received and put.+ */
1153	uintmax_t st_log_requested;	/* Log recs. missed and requested.+ */
1154	db_ssize_t st_master;		/* Env. ID of the current master. */
1155	uintmax_t st_master_changes;	/* # of times we've switched masters. */
1156	uintmax_t st_msgs_badgen;	/* Messages with a bad generation #.+ */
1157	uintmax_t st_msgs_processed;	/* Messages received and processed.+ */
1158	uintmax_t st_msgs_recover;	/* Messages ignored because this site
1159					   was a client in recovery.+ */
1160	uintmax_t st_msgs_send_failures;/* # of failed message sends.+ */
1161	uintmax_t st_msgs_sent;	/* # of successful message sends.+ */
1162	uintmax_t st_newsites;		/* # of NEWSITE msgs. received.+ */
1163	u_int32_t st_nsites;		/* Current number of sites we will
1164					   assume during elections. */
1165	uintmax_t st_nthrottles;	/* # of times we were throttled. */
1166	uintmax_t st_outdated;		/* # of times we detected and returned
1167					   an OUTDATED condition.+ */
1168	uintmax_t st_pg_duplicated;	/* Pages received multiply.+ */
1169	uintmax_t st_pg_records;	/* Pages received and stored.+ */
1170	uintmax_t st_pg_requested;	/* Pages missed and requested.+ */
1171	uintmax_t st_txns_applied;	/* # of transactions applied.+ */
1172	uintmax_t st_startsync_delayed;/* # of STARTSYNC msgs delayed.+ */
1173
1174	/* Elections generally. */
1175	uintmax_t st_elections;	/* # of elections held.+ */
1176	uintmax_t st_elections_won;	/* # of elections won by this site.+ */
1177
1178	/* Statistics about an in-progress election. */
1179	db_ssize_t st_election_cur_winner;	/* Current front-runner. */
1180	u_int32_t st_election_gen;	/* Election generation number. */
1181	u_int32_t st_election_datagen;	/* Election data generation number. */
1182	DB_LSN st_election_lsn;		/* Max. LSN of current winner. */
1183	u_int32_t st_election_nsites;	/* # of "registered voters". */
1184	u_int32_t st_election_nvotes;	/* # of "registered voters" needed. */
1185	u_int32_t st_election_priority;	/* Current election priority. */
1186	int32_t   st_election_status;	/* Current election status. */
1187	u_int32_t st_election_tiebreaker;/* Election tiebreaker value. */
1188	u_int32_t st_election_votes;	/* Votes received in this round. */
1189	u_int32_t st_election_sec;	/* Last election time seconds. */
1190	u_int32_t st_election_usec;	/* Last election time useconds. */
1191	u_int32_t st_max_lease_sec;	/* Maximum lease timestamp seconds. */
1192	u_int32_t st_max_lease_usec;	/* Maximum lease timestamp useconds. */
1193
1194	/* Undocumented statistics only used by the test system. */
1195#ifdef	CONFIG_TEST
1196	u_int32_t st_filefail_cleanups;	/* # of FILE_FAIL cleanups done. */
1197#endif
1198#endif
1199};
1200
1201/* Replication Manager statistics. */
1202struct __db_repmgr_stat { /* SHARED */
1203	uintmax_t st_perm_failed;	/* # of insufficiently ack'ed msgs. */
1204	uintmax_t st_msgs_queued;	/* # msgs queued for network delay. */
1205	uintmax_t st_msgs_dropped;	/* # msgs discarded due to excessive
1206					   queue length. */
1207	uintmax_t st_connection_drop;	/* Existing connections dropped. */
1208	uintmax_t st_connect_fail;	/* Failed new connection attempts. */
1209	uintmax_t st_elect_threads;	/* # of active election threads. */
1210	uintmax_t st_max_elect_threads;	/* Max concurrent e-threads ever. */
1211};
1212
1213/* Replication Manager connection error. */
1214struct __db_repmgr_conn_err {
1215	int		eid;		/* Replication Environment ID. */
1216	int		error;		/* System networking error code. */
1217};
1218
1219/*******************************************************
1220 * Sequences.
1221 *******************************************************/
1222/*
1223 * The storage record for a sequence.
1224 */
1225struct __db_seq_record {
1226	u_int32_t	seq_version;	/* Version size/number. */
1227	u_int32_t	flags;		/* DB_SEQ_XXX Flags. */
1228	db_seq_t	seq_value;	/* Current value. */
1229	db_seq_t	seq_max;	/* Max permitted. */
1230	db_seq_t	seq_min;	/* Min permitted. */
1231};
1232
1233/*
1234 * Handle for a sequence object.
1235 */
1236struct __db_sequence {
1237	DB		*seq_dbp;	/* DB handle for this sequence. */
1238	db_mutex_t	mtx_seq;	/* Mutex if sequence is threaded. */
1239	DB_SEQ_RECORD	*seq_rp;	/* Pointer to current data. */
1240	DB_SEQ_RECORD	seq_record;	/* Data from DB_SEQUENCE. */
1241	int32_t		seq_cache_size; /* Number of values cached. */
1242	db_seq_t	seq_last_value;	/* Last value cached. */
1243	db_seq_t	seq_prev_value;	/* Last value returned. */
1244	DBT		seq_key;	/* DBT pointing to sequence key. */
1245	DBT		seq_data;	/* DBT pointing to seq_record. */
1246
1247	/* API-private structure: used by C++ and Java. */
1248	void		*api_internal;
1249
1250	/* DB_SEQUENCE PUBLIC HANDLE LIST BEGIN */
1251	int		(*close) __P((DB_SEQUENCE *, u_int32_t));
1252	int		(*get) __P((DB_SEQUENCE *,
1253			      DB_TXN *, int32_t, db_seq_t *, u_int32_t));
1254	int		(*get_cachesize) __P((DB_SEQUENCE *, int32_t *));
1255	int		(*get_db) __P((DB_SEQUENCE *, DB **));
1256	int		(*get_flags) __P((DB_SEQUENCE *, u_int32_t *));
1257	int		(*get_key) __P((DB_SEQUENCE *, DBT *));
1258	int		(*get_range) __P((DB_SEQUENCE *,
1259			     db_seq_t *, db_seq_t *));
1260	int		(*initial_value) __P((DB_SEQUENCE *, db_seq_t));
1261	int		(*open) __P((DB_SEQUENCE *,
1262			    DB_TXN *, DBT *, u_int32_t));
1263	int		(*remove) __P((DB_SEQUENCE *, DB_TXN *, u_int32_t));
1264	int		(*set_cachesize) __P((DB_SEQUENCE *, int32_t));
1265	int		(*set_flags) __P((DB_SEQUENCE *, u_int32_t));
1266	int		(*set_range) __P((DB_SEQUENCE *, db_seq_t, db_seq_t));
1267	int		(*stat) __P((DB_SEQUENCE *,
1268			    DB_SEQUENCE_STAT **, u_int32_t));
1269	int		(*stat_print) __P((DB_SEQUENCE *, u_int32_t));
1270	/* DB_SEQUENCE PUBLIC HANDLE LIST END */
1271};
1272
1273struct __db_seq_stat { /* SHARED */
1274	uintmax_t st_wait;		/* Sequence lock granted w/o wait. */
1275	uintmax_t st_nowait;		/* Sequence lock granted after wait. */
1276	db_seq_t  st_current;		/* Current value in db. */
1277	db_seq_t  st_value;		/* Current cached value. */
1278	db_seq_t  st_last_value;	/* Last cached value. */
1279	db_seq_t  st_min;		/* Minimum value. */
1280	db_seq_t  st_max;		/* Maximum value. */
1281	int32_t   st_cache_size;	/* Cache size. */
1282	u_int32_t st_flags;		/* Flag value. */
1283};
1284
1285/*******************************************************
1286 * Access methods.
1287 *******************************************************/
1288/*
1289 * Any new methods need to retain the original numbering.  The type
1290 * is written in a log record so must be maintained.
1291 */
1292typedef enum {
1293	DB_BTREE=1,
1294	DB_HASH=2,
1295	DB_HEAP=6,
1296	DB_RECNO=3,
1297	DB_QUEUE=4,
1298	DB_UNKNOWN=5			/* Figure it out on open. */
1299} DBTYPE;
1300
1301#define	DB_RENAMEMAGIC	0x030800	/* File has been renamed. */
1302
1303#define	DB_BTREEVERSION	9		/* Current btree version. */
1304#define	DB_BTREEOLDVER	8		/* Oldest btree version supported. */
1305#define	DB_BTREEMAGIC	0x053162
1306
1307#define	DB_HASHVERSION	9		/* Current hash version. */
1308#define	DB_HASHOLDVER	7		/* Oldest hash version supported. */
1309#define	DB_HASHMAGIC	0x061561
1310
1311#define	DB_HEAPVERSION	1		/* Current heap version. */
1312#define	DB_HEAPOLDVER	1		/* Oldest heap version supported. */
1313#define	DB_HEAPMAGIC	0x074582
1314
1315#define	DB_QAMVERSION	4		/* Current queue version. */
1316#define	DB_QAMOLDVER	3		/* Oldest queue version supported. */
1317#define	DB_QAMMAGIC	0x042253
1318
1319#define	DB_SEQUENCE_VERSION 2		/* Current sequence version. */
1320#define	DB_SEQUENCE_OLDVER  1		/* Oldest sequence version supported. */
1321
1322/*
1323 * DB access method and cursor operation values.  Each value is an operation
1324 * code to which additional bit flags are added.
1325 */
1326#define	DB_AFTER		 1	/* Dbc.put */
1327#define	DB_APPEND		 2	/* Db.put */
1328#define	DB_BEFORE		 3	/* Dbc.put */
1329#define	DB_CONSUME		 4	/* Db.get */
1330#define	DB_CONSUME_WAIT		 5	/* Db.get */
1331#define	DB_CURRENT		 6	/* Dbc.get, Dbc.put, DbLogc.get */
1332#define	DB_FIRST		 7	/* Dbc.get, DbLogc->get */
1333#define	DB_GET_BOTH		 8	/* Db.get, Dbc.get */
1334#define	DB_GET_BOTHC		 9	/* Dbc.get (internal) */
1335#define	DB_GET_BOTH_RANGE	10	/* Db.get, Dbc.get */
1336#define	DB_GET_RECNO		11	/* Dbc.get */
1337#define	DB_JOIN_ITEM		12	/* Dbc.get; don't do primary lookup */
1338#define	DB_KEYFIRST		13	/* Dbc.put */
1339#define	DB_KEYLAST		14	/* Dbc.put */
1340#define	DB_LAST			15	/* Dbc.get, DbLogc->get */
1341#define	DB_NEXT			16	/* Dbc.get, DbLogc->get */
1342#define	DB_NEXT_DUP		17	/* Dbc.get */
1343#define	DB_NEXT_NODUP		18	/* Dbc.get */
1344#define	DB_NODUPDATA		19	/* Db.put, Dbc.put */
1345#define	DB_NOOVERWRITE		20	/* Db.put */
1346#define	DB_OVERWRITE_DUP	21	/* Dbc.put, Db.put; no DB_KEYEXIST */
1347#define	DB_POSITION		22	/* Dbc.dup */
1348#define	DB_PREV			23	/* Dbc.get, DbLogc->get */
1349#define	DB_PREV_DUP		24	/* Dbc.get */
1350#define	DB_PREV_NODUP		25	/* Dbc.get */
1351#define	DB_SET			26	/* Dbc.get, DbLogc->get */
1352#define	DB_SET_RANGE		27	/* Dbc.get */
1353#define	DB_SET_RECNO		28	/* Db.get, Dbc.get */
1354#define	DB_UPDATE_SECONDARY	29	/* Dbc.get, Dbc.del (internal) */
1355#define	DB_SET_LTE		30	/* Dbc.get (internal) */
1356#define	DB_GET_BOTH_LTE		31	/* Dbc.get (internal) */
1357
1358/* This has to change when the max opcode hits 255. */
1359#define	DB_OPFLAGS_MASK	0x000000ff	/* Mask for operations flags. */
1360
1361/*
1362 * DB (user visible) error return codes.
1363 *
1364 * !!!
1365 * We don't want our error returns to conflict with other packages where
1366 * possible, so pick a base error value that's hopefully not common.  We
1367 * document that we own the error name space from -30,800 to -30,999.
1368 */
1369/* DB (public) error return codes. */
1370#define	DB_BUFFER_SMALL		(-30999)/* User memory too small for return. */
1371#define	DB_DONOTINDEX		(-30998)/* "Null" return from 2ndary callbk. */
1372#define	DB_FOREIGN_CONFLICT	(-30997)/* A foreign db constraint triggered. */
1373#define	DB_HEAP_FULL		(-30996)/* No free space in a heap file. */
1374#define	DB_KEYEMPTY		(-30995)/* Key/data deleted or never created. */
1375#define	DB_KEYEXIST		(-30994)/* The key/data pair already exists. */
1376#define	DB_LOCK_DEADLOCK	(-30993)/* Deadlock. */
1377#define	DB_LOCK_NOTGRANTED	(-30992)/* Lock unavailable. */
1378#define	DB_LOG_BUFFER_FULL	(-30991)/* In-memory log buffer full. */
1379#define	DB_LOG_VERIFY_BAD	(-30990)/* Log verification failed. */
1380#define	DB_NOSERVER		(-30989)/* Server panic return. */
1381#define	DB_NOTFOUND		(-30988)/* Key/data pair not found (EOF). */
1382#define	DB_OLD_VERSION		(-30987)/* Out-of-date version. */
1383#define	DB_PAGE_NOTFOUND	(-30986)/* Requested page not found. */
1384#define	DB_REP_DUPMASTER	(-30985)/* There are two masters. */
1385#define	DB_REP_HANDLE_DEAD	(-30984)/* Rolled back a commit. */
1386#define	DB_REP_HOLDELECTION	(-30983)/* Time to hold an election. */
1387#define	DB_REP_IGNORE		(-30982)/* This msg should be ignored.*/
1388#define	DB_REP_ISPERM		(-30981)/* Cached not written perm written.*/
1389#define	DB_REP_JOIN_FAILURE	(-30980)/* Unable to join replication group. */
1390#define	DB_REP_LEASE_EXPIRED	(-30979)/* Master lease has expired. */
1391#define	DB_REP_LOCKOUT		(-30978)/* API/Replication lockout now. */
1392#define	DB_REP_NEWSITE		(-30977)/* New site entered system. */
1393#define	DB_REP_NOTPERM		(-30976)/* Permanent log record not written. */
1394#define	DB_REP_UNAVAIL		(-30975)/* Site cannot currently be reached. */
1395#define	DB_REP_WOULDROLLBACK	(-30974)/* UNDOC: rollback inhibited by app. */
1396#define	DB_RUNRECOVERY		(-30973)/* Panic return. */
1397#define	DB_SECONDARY_BAD	(-30972)/* Secondary index corrupt. */
1398#define	DB_TIMEOUT		(-30971)/* Timed out on read consistency. */
1399#define	DB_VERIFY_BAD		(-30970)/* Verify failed; bad format. */
1400#define	DB_VERSION_MISMATCH	(-30969)/* Environment version mismatch. */
1401
1402/* DB (private) error return codes. */
1403#define	DB_ALREADY_ABORTED	(-30899)
1404#define	DB_CHKSUM_FAIL		(-30898)/* Checksum failed. */
1405#define	DB_DELETED		(-30897)/* Recovery file marked deleted. */
1406#define	DB_EVENT_NOT_HANDLED	(-30896)/* Forward event to application. */
1407#define	DB_NEEDSPLIT		(-30895)/* Page needs to be split. */
1408#define	DB_REP_BULKOVF		(-30894)/* Rep bulk buffer overflow. */
1409#define	DB_REP_LOGREADY		(-30893)/* Rep log ready for recovery. */
1410#define	DB_REP_NEWMASTER	(-30892)/* We have learned of a new master. */
1411#define	DB_REP_PAGEDONE		(-30891)/* This page was already done. */
1412#define	DB_SURPRISE_KID		(-30890)/* Child commit where parent
1413					   didn't know it was a parent. */
1414#define	DB_SWAPBYTES		(-30889)/* Database needs byte swapping. */
1415#define	DB_TXN_CKP		(-30888)/* Encountered ckp record in log. */
1416#define	DB_VERIFY_FATAL		(-30887)/* DB->verify cannot proceed. */
1417
1418/* Database handle. */
1419struct __db {
1420	/*******************************************************
1421	 * Public: owned by the application.
1422	 *******************************************************/
1423	u_int32_t pgsize;		/* Database logical page size. */
1424	DB_CACHE_PRIORITY priority;	/* Database priority in cache. */
1425
1426					/* Callbacks. */
1427	int (*db_append_recno) __P((DB *, DBT *, db_recno_t));
1428	void (*db_feedback) __P((DB *, int, int));
1429	int (*dup_compare) __P((DB *, const DBT *, const DBT *));
1430
1431	void	*app_private;		/* Application-private handle. */
1432
1433	/*******************************************************
1434	 * Private: owned by DB.
1435	 *******************************************************/
1436	DB_ENV	*dbenv;			/* Backing public environment. */
1437	ENV	*env;			/* Backing private environment. */
1438
1439	DBTYPE	 type;			/* DB access method type. */
1440
1441	DB_MPOOLFILE *mpf;		/* Backing buffer pool. */
1442
1443	db_mutex_t mutex;		/* Synchronization for free threading */
1444
1445	char *fname, *dname;		/* File/database passed to DB->open. */
1446	const char *dirname;		/* Directory of DB file. */
1447	u_int32_t open_flags;		/* Flags passed to DB->open. */
1448
1449	u_int8_t fileid[DB_FILE_ID_LEN];/* File's unique ID for locking. */
1450
1451	u_int32_t adj_fileid;		/* File's unique ID for curs. adj. */
1452
1453#define	DB_LOGFILEID_INVALID	-1
1454	FNAME *log_filename;		/* File's naming info for logging. */
1455
1456	db_pgno_t meta_pgno;		/* Meta page number */
1457	DB_LOCKER *locker;		/* Locker for handle locking. */
1458	DB_LOCKER *cur_locker;		/* Current handle lock holder. */
1459	DB_TXN *cur_txn;		/* Opening transaction. */
1460	DB_LOCKER *associate_locker;	/* Locker for DB->associate call. */
1461	DB_LOCK	 handle_lock;		/* Lock held on this handle. */
1462
1463	time_t	 timestamp;		/* Handle timestamp for replication. */
1464	u_int32_t fid_gen;		/* Rep generation number for fids. */
1465
1466	/*
1467	 * Returned data memory for DB->get() and friends.
1468	 */
1469	DBT	 my_rskey;		/* Secondary key. */
1470	DBT	 my_rkey;		/* [Primary] key. */
1471	DBT	 my_rdata;		/* Data. */
1472
1473	/*
1474	 * !!!
1475	 * Some applications use DB but implement their own locking outside of
1476	 * DB.  If they're using fcntl(2) locking on the underlying database
1477	 * file, and we open and close a file descriptor for that file, we will
1478	 * discard their locks.  The DB_FCNTL_LOCKING flag to DB->open is an
1479	 * undocumented interface to support this usage which leaves any file
1480	 * descriptors we open until DB->close.  This will only work with the
1481	 * DB->open interface and simple caches, e.g., creating a transaction
1482	 * thread may open/close file descriptors this flag doesn't protect.
1483	 * Locking with fcntl(2) on a file that you don't own is a very, very
1484	 * unsafe thing to do.  'Nuff said.
1485	 */
1486	DB_FH	*saved_open_fhp;	/* Saved file handle. */
1487
1488	/*
1489	 * Linked list of DBP's, linked from the ENV, used to keep track
1490	 * of all open db handles for cursor adjustment.
1491	 *
1492	 * !!!
1493	 * Explicit representations of structures from queue.h.
1494	 * TAILQ_ENTRY(__db) dblistlinks;
1495	 */
1496	struct {
1497		struct __db *tqe_next;
1498		struct __db **tqe_prev;
1499	} dblistlinks;
1500
1501	/*
1502	 * Cursor queues.
1503	 *
1504	 * !!!
1505	 * Explicit representations of structures from queue.h.
1506	 * TAILQ_HEAD(__cq_fq, __dbc) free_queue;
1507	 * TAILQ_HEAD(__cq_aq, __dbc) active_queue;
1508	 * TAILQ_HEAD(__cq_jq, __dbc) join_queue;
1509	 */
1510	struct __cq_fq {
1511		struct __dbc *tqh_first;
1512		struct __dbc **tqh_last;
1513	} free_queue;
1514	struct __cq_aq {
1515		struct __dbc *tqh_first;
1516		struct __dbc **tqh_last;
1517	} active_queue;
1518	struct __cq_jq {
1519		struct __dbc *tqh_first;
1520		struct __dbc **tqh_last;
1521	} join_queue;
1522
1523	/*
1524	 * Secondary index support.
1525	 *
1526	 * Linked list of secondary indices -- set in the primary.
1527	 *
1528	 * !!!
1529	 * Explicit representations of structures from queue.h.
1530	 * LIST_HEAD(s_secondaries, __db);
1531	 */
1532	struct {
1533		struct __db *lh_first;
1534	} s_secondaries;
1535
1536	/*
1537	 * List entries for secondaries, and reference count of how many
1538	 * threads are updating this secondary (see Dbc.put).
1539	 *
1540	 * !!!
1541	 * Note that these are synchronized by the primary's mutex, but
1542	 * filled in in the secondaries.
1543	 *
1544	 * !!!
1545	 * Explicit representations of structures from queue.h.
1546	 * LIST_ENTRY(__db) s_links;
1547	 */
1548	struct {
1549		struct __db *le_next;
1550		struct __db **le_prev;
1551	} s_links;
1552	u_int32_t s_refcnt;
1553
1554	/* Secondary callback and free functions -- set in the secondary. */
1555	int	(*s_callback) __P((DB *, const DBT *, const DBT *, DBT *));
1556
1557	/* Reference to primary -- set in the secondary. */
1558	DB	*s_primary;
1559
1560#define	DB_ASSOC_IMMUTABLE_KEY    0x00000001 /* Secondary key is immutable. */
1561#define	DB_ASSOC_CREATE    0x00000002 /* Secondary db populated on open. */
1562
1563	/* Flags passed to associate -- set in the secondary. */
1564	u_int32_t s_assoc_flags;
1565
1566	/*
1567	 * Foreign key support.
1568	 *
1569	 * Linked list of primary dbs -- set in the foreign db
1570	 *
1571	 * !!!
1572	 * Explicit representations of structures from queue.h.
1573	 * LIST_HEAD(f_primaries, __db);
1574	 */
1575	struct {
1576		struct __db_foreign_info *lh_first;
1577	} f_primaries;
1578
1579	/*
1580	 * !!!
1581	 * Explicit representations of structures from queue.h.
1582	 * TAILQ_ENTRY(__db) felink;
1583	 *
1584	 * Links in a list of DBs involved in file extension
1585	 * during a transaction.  These are to be used only while the
1586	 * metadata is locked.
1587	 */
1588	struct {
1589		struct __db *tqe_next;
1590		struct __db **tqe_prev;
1591	} felink;
1592
1593	/* Reference to foreign -- set in the secondary. */
1594	DB      *s_foreign;
1595
1596	/* API-private structure: used by DB 1.85, C++, Java, Perl and Tcl */
1597	void	*api_internal;
1598
1599	/* Subsystem-private structure. */
1600	void	*bt_internal;		/* Btree/Recno access method. */
1601	void	*h_internal;		/* Hash access method. */
1602	void	*heap_internal;		/* Heap access method. */
1603	void	*p_internal;		/* Partition informaiton. */
1604	void	*q_internal;		/* Queue access method. */
1605
1606	/* DB PUBLIC HANDLE LIST BEGIN */
1607	int  (*associate) __P((DB *, DB_TXN *, DB *,
1608		int (*)(DB *, const DBT *, const DBT *, DBT *), u_int32_t));
1609	int  (*associate_foreign) __P((DB *, DB *,
1610		int (*)(DB *, const DBT *, DBT *, const DBT *, int *),
1611		u_int32_t));
1612	int  (*close) __P((DB *, u_int32_t));
1613	int  (*compact) __P((DB *,
1614		DB_TXN *, DBT *, DBT *, DB_COMPACT *, u_int32_t, DBT *));
1615	int  (*cursor) __P((DB *, DB_TXN *, DBC **, u_int32_t));
1616	int  (*del) __P((DB *, DB_TXN *, DBT *, u_int32_t));
1617	void (*err) __P((DB *, int, const char *, ...));
1618	void (*errx) __P((DB *, const char *, ...));
1619	int  (*exists) __P((DB *, DB_TXN *, DBT *, u_int32_t));
1620	int  (*fd) __P((DB *, int *));
1621	int  (*get) __P((DB *, DB_TXN *, DBT *, DBT *, u_int32_t));
1622	int  (*get_alloc) __P((DB *, void *(**)(size_t),
1623		void *(**)(void *, size_t), void (**)(void *)));
1624	int  (*get_append_recno) __P((DB *, int (**)(DB *, DBT *, db_recno_t)));
1625	int  (*get_assoc_flags) __P((DB *, u_int32_t *));
1626	int  (*get_bt_compare)
1627		__P((DB *, int (**)(DB *, const DBT *, const DBT *)));
1628	int  (*get_bt_compress) __P((DB *,
1629		int (**)(DB *,
1630		const DBT *, const DBT *, const DBT *, const DBT *, DBT *),
1631		int (**)(DB *, const DBT *, const DBT *, DBT *, DBT *, DBT *)));
1632	int  (*get_bt_minkey) __P((DB *, u_int32_t *));
1633	int  (*get_bt_prefix)
1634		__P((DB *, size_t (**)(DB *, const DBT *, const DBT *)));
1635	int  (*get_byteswapped) __P((DB *, int *));
1636	int  (*get_cachesize) __P((DB *, u_int32_t *, u_int32_t *, int *));
1637	int  (*get_create_dir) __P((DB *, const char **));
1638	int  (*get_dbname) __P((DB *, const char **, const char **));
1639	int  (*get_dup_compare)
1640		__P((DB *, int (**)(DB *, const DBT *, const DBT *)));
1641	int  (*get_encrypt_flags) __P((DB *, u_int32_t *));
1642	DB_ENV *(*get_env) __P((DB *));
1643	void (*get_errcall) __P((DB *,
1644		void (**)(const DB_ENV *, const char *, const char *)));
1645	void (*get_errfile) __P((DB *, FILE **));
1646	void (*get_errpfx) __P((DB *, const char **));
1647	int  (*get_feedback) __P((DB *, void (**)(DB *, int, int)));
1648	int  (*get_flags) __P((DB *, u_int32_t *));
1649	int  (*get_h_compare)
1650		__P((DB *, int (**)(DB *, const DBT *, const DBT *)));
1651	int  (*get_h_ffactor) __P((DB *, u_int32_t *));
1652	int  (*get_h_hash)
1653		__P((DB *, u_int32_t (**)(DB *, const void *, u_int32_t)));
1654	int  (*get_h_nelem) __P((DB *, u_int32_t *));
1655	int  (*get_heapsize) __P((DB *, u_int32_t *, u_int32_t *));
1656	int  (*get_heap_regionsize) __P((DB *, u_int32_t *));
1657	int  (*get_lk_exclusive) __P((DB *, int *, int *));
1658	int  (*get_lorder) __P((DB *, int *));
1659	DB_MPOOLFILE *(*get_mpf) __P((DB *));
1660	void (*get_msgcall) __P((DB *,
1661	    void (**)(const DB_ENV *, const char *)));
1662	void (*get_msgfile) __P((DB *, FILE **));
1663	int  (*get_multiple) __P((DB *));
1664	int  (*get_open_flags) __P((DB *, u_int32_t *));
1665	int  (*get_pagesize) __P((DB *, u_int32_t *));
1666	int  (*get_partition_callback) __P((DB *,
1667		u_int32_t *, u_int32_t (**)(DB *, DBT *key)));
1668	int  (*get_partition_dirs) __P((DB *, const char ***));
1669	int  (*get_partition_keys) __P((DB *, u_int32_t *, DBT **));
1670	int  (*get_priority) __P((DB *, DB_CACHE_PRIORITY *));
1671	int  (*get_q_extentsize) __P((DB *, u_int32_t *));
1672	int  (*get_re_delim) __P((DB *, int *));
1673	int  (*get_re_len) __P((DB *, u_int32_t *));
1674	int  (*get_re_pad) __P((DB *, int *));
1675	int  (*get_re_source) __P((DB *, const char **));
1676	int  (*get_transactional) __P((DB *));
1677	int  (*get_type) __P((DB *, DBTYPE *));
1678	int  (*join) __P((DB *, DBC **, DBC **, u_int32_t));
1679	int  (*key_range)
1680		__P((DB *, DB_TXN *, DBT *, DB_KEY_RANGE *, u_int32_t));
1681	int  (*open) __P((DB *,
1682		DB_TXN *, const char *, const char *, DBTYPE, u_int32_t, int));
1683	int  (*pget) __P((DB *, DB_TXN *, DBT *, DBT *, DBT *, u_int32_t));
1684	int  (*put) __P((DB *, DB_TXN *, DBT *, DBT *, u_int32_t));
1685	int  (*remove) __P((DB *, const char *, const char *, u_int32_t));
1686	int  (*rename) __P((DB *,
1687		const char *, const char *, const char *, u_int32_t));
1688	int  (*set_alloc) __P((DB *, void *(*)(size_t),
1689		void *(*)(void *, size_t), void (*)(void *)));
1690	int  (*set_append_recno) __P((DB *, int (*)(DB *, DBT *, db_recno_t)));
1691	int  (*set_bt_compare)
1692		__P((DB *, int (*)(DB *, const DBT *, const DBT *)));
1693	int  (*set_bt_compress) __P((DB *,
1694		int (*)(DB *, const DBT *, const DBT *, const DBT *, const DBT *, DBT *),
1695		int (*)(DB *, const DBT *, const DBT *, DBT *, DBT *, DBT *)));
1696	int  (*set_bt_minkey) __P((DB *, u_int32_t));
1697	int  (*set_bt_prefix)
1698		__P((DB *, size_t (*)(DB *, const DBT *, const DBT *)));
1699	int  (*set_cachesize) __P((DB *, u_int32_t, u_int32_t, int));
1700	int  (*set_create_dir) __P((DB *, const char *));
1701	int  (*set_dup_compare)
1702		__P((DB *, int (*)(DB *, const DBT *, const DBT *)));
1703	int  (*set_encrypt) __P((DB *, const char *, u_int32_t));
1704	void (*set_errcall) __P((DB *,
1705		void (*)(const DB_ENV *, const char *, const char *)));
1706	void (*set_errfile) __P((DB *, FILE *));
1707	void (*set_errpfx) __P((DB *, const char *));
1708	int  (*set_feedback) __P((DB *, void (*)(DB *, int, int)));
1709	int  (*set_flags) __P((DB *, u_int32_t));
1710	int  (*set_h_compare)
1711		__P((DB *, int (*)(DB *, const DBT *, const DBT *)));
1712	int  (*set_h_ffactor) __P((DB *, u_int32_t));
1713	int  (*set_h_hash)
1714		__P((DB *, u_int32_t (*)(DB *, const void *, u_int32_t)));
1715	int  (*set_h_nelem) __P((DB *, u_int32_t));
1716	int  (*set_heapsize) __P((DB *, u_int32_t, u_int32_t, u_int32_t));
1717	int  (*set_heap_regionsize) __P((DB *, u_int32_t));
1718	int  (*set_lk_exclusive) __P((DB *, int));
1719	int  (*set_lorder) __P((DB *, int));
1720	void (*set_msgcall) __P((DB *, void (*)(const DB_ENV *, const char *)));
1721	void (*set_msgfile) __P((DB *, FILE *));
1722	int  (*set_pagesize) __P((DB *, u_int32_t));
1723	int  (*set_paniccall) __P((DB *, void (*)(DB_ENV *, int)));
1724	int  (*set_partition) __P((DB *,
1725		u_int32_t, DBT *, u_int32_t (*)(DB *, DBT *key)));
1726	int  (*set_partition_dirs) __P((DB *, const char **));
1727	int  (*set_priority) __P((DB *, DB_CACHE_PRIORITY));
1728	int  (*set_q_extentsize) __P((DB *, u_int32_t));
1729	int  (*set_re_delim) __P((DB *, int));
1730	int  (*set_re_len) __P((DB *, u_int32_t));
1731	int  (*set_re_pad) __P((DB *, int));
1732	int  (*set_re_source) __P((DB *, const char *));
1733	int  (*sort_multiple) __P((DB *, DBT *, DBT *, u_int32_t));
1734	int  (*stat) __P((DB *, DB_TXN *, void *, u_int32_t));
1735	int  (*stat_print) __P((DB *, u_int32_t));
1736	int  (*sync) __P((DB *, u_int32_t));
1737	int  (*truncate) __P((DB *, DB_TXN *, u_int32_t *, u_int32_t));
1738	int  (*upgrade) __P((DB *, const char *, u_int32_t));
1739	int  (*verify)
1740		__P((DB *, const char *, const char *, FILE *, u_int32_t));
1741	/* DB PUBLIC HANDLE LIST END */
1742
1743	/* DB PRIVATE HANDLE LIST BEGIN */
1744	int  (*dump) __P((DB *, const char *,
1745		int (*)(void *, const void *), void *, int, int));
1746	int  (*db_am_remove) __P((DB *, DB_THREAD_INFO *,
1747		DB_TXN *, const char *, const char *, u_int32_t));
1748	int  (*db_am_rename) __P((DB *, DB_THREAD_INFO *,
1749		DB_TXN *, const char *, const char *, const char *));
1750	/* DB PRIVATE HANDLE LIST END */
1751
1752	/*
1753	 * Never called; these are a place to save function pointers
1754	 * so that we can undo an associate.
1755	 */
1756	int  (*stored_get) __P((DB *, DB_TXN *, DBT *, DBT *, u_int32_t));
1757	int  (*stored_close) __P((DB *, u_int32_t));
1758
1759	/* Alternative handle close function, used by C++ API. */
1760	int  (*alt_close) __P((DB *, u_int32_t));
1761
1762#define	DB_OK_BTREE	0x01
1763#define	DB_OK_HASH	0x02
1764#define	DB_OK_HEAP	0x04
1765#define	DB_OK_QUEUE	0x08
1766#define	DB_OK_RECNO	0x10
1767	u_int32_t	am_ok;		/* Legal AM choices. */
1768
1769	/*
1770	 * This field really ought to be an AM_FLAG, but we have
1771	 * have run out of bits.  If/when we decide to split up
1772	 * the flags, we can incorporate it.
1773	 */
1774	int	 preserve_fid;		/* Do not free fileid on close. */
1775
1776#define	DB_AM_CHKSUM		0x00000001 /* Checksumming */
1777#define	DB_AM_COMPENSATE	0x00000002 /* Created by compensating txn */
1778#define	DB_AM_COMPRESS		0x00000004 /* Compressed BTree */
1779#define	DB_AM_CREATED		0x00000008 /* Database was created upon open */
1780#define	DB_AM_CREATED_MSTR	0x00000010 /* Encompassing file was created */
1781#define	DB_AM_DBM_ERROR		0x00000020 /* Error in DBM/NDBM database */
1782#define	DB_AM_DELIMITER		0x00000040 /* Variable length delimiter set */
1783#define	DB_AM_DISCARD		0x00000080 /* Discard any cached pages */
1784#define	DB_AM_DUP		0x00000100 /* DB_DUP */
1785#define	DB_AM_DUPSORT		0x00000200 /* DB_DUPSORT */
1786#define	DB_AM_ENCRYPT		0x00000400 /* Encryption */
1787#define	DB_AM_FIXEDLEN		0x00000800 /* Fixed-length records */
1788#define	DB_AM_INMEM		0x00001000 /* In-memory; no sync on close */
1789#define	DB_AM_INORDER		0x00002000 /* DB_INORDER */
1790#define	DB_AM_IN_RENAME		0x00004000 /* File is being renamed */
1791#define	DB_AM_NOT_DURABLE	0x00008000 /* Do not log changes */
1792#define	DB_AM_OPEN_CALLED	0x00010000 /* DB->open called */
1793#define	DB_AM_PAD		0x00020000 /* Fixed-length record pad */
1794#define	DB_AM_PARTDB		0x00040000 /* Handle for a database partition */
1795#define	DB_AM_PGDEF		0x00080000 /* Page size was defaulted */
1796#define	DB_AM_RDONLY		0x00100000 /* Database is readonly */
1797#define	DB_AM_READ_UNCOMMITTED	0x00200000 /* Support degree 1 isolation */
1798#define	DB_AM_RECNUM		0x00400000 /* DB_RECNUM */
1799#define	DB_AM_RECOVER		0x00800000 /* DB opened by recovery routine */
1800#define	DB_AM_RENUMBER		0x01000000 /* DB_RENUMBER */
1801#define	DB_AM_REVSPLITOFF	0x02000000 /* DB_REVSPLITOFF */
1802#define	DB_AM_SECONDARY		0x04000000 /* Database is a secondary index */
1803#define	DB_AM_SNAPSHOT		0x08000000 /* DB_SNAPSHOT */
1804#define	DB_AM_SUBDB		0x10000000 /* Subdatabases supported */
1805#define	DB_AM_SWAP		0x20000000 /* Pages need to be byte-swapped */
1806#define	DB_AM_TXN		0x40000000 /* Opened in a transaction */
1807#define	DB_AM_VERIFYING		0x80000000 /* DB handle is in the verifier */
1808	u_int32_t orig_flags;		   /* Flags at  open, for refresh */
1809	u_int32_t flags;
1810
1811#define DB2_AM_EXCL		0x00000001 /* Exclusively lock the handle */
1812#define DB2_AM_INTEXCL		0x00000002 /* Internal exclusive lock. */
1813#define DB2_AM_NOWAIT		0x00000004 /* Do not wait for handle lock */
1814	u_int32_t orig_flags2;		   /* Second flags word; for refresh */
1815	u_int32_t flags2;		   /* Second flags word */
1816};
1817
1818/*
1819 * Macros for bulk operations.  These are only intended for the C API.
1820 * For C++, use DbMultiple*Iterator or DbMultiple*Builder.
1821 *
1822 * Bulk operations store multiple entries into a single DBT structure. The
1823 * following macros assist with creating and reading these Multiple DBTs.
1824 *
1825 * The basic layout for single data items is:
1826 *
1827 * -------------------------------------------------------------------------
1828 * | data1 | ... | dataN | ..... |-1 | dNLen | dNOff | ... | d1Len | d1Off |
1829 * -------------------------------------------------------------------------
1830 *
1831 * For the DB_MULTIPLE_KEY* macros, the items are in key/data pairs, so data1
1832 * would be a key, and data2 its corresponding value (N is always even).
1833 *
1834 * For the DB_MULTIPLE_RECNO* macros, the record number is stored along with
1835 * the len/off pair in the "header" section, and the list is zero terminated
1836 * (since -1 is a valid record number):
1837 *
1838 * --------------------------------------------------------------------------
1839 * | d1 |..| dN |..| 0 | dNLen | dNOff | recnoN |..| d1Len | d1Off | recno1 |
1840 * --------------------------------------------------------------------------
1841 */
1842#define	DB_MULTIPLE_INIT(pointer, dbt)					\
1843	(pointer = (u_int8_t *)(dbt)->data +				\
1844	    (dbt)->ulen - sizeof(u_int32_t))
1845
1846#define	DB_MULTIPLE_NEXT(pointer, dbt, retdata, retdlen)		\
1847	do {								\
1848		u_int32_t *__p = (u_int32_t *)(pointer);		\
1849		if (*__p == (u_int32_t)-1) {				\
1850			retdata = NULL;					\
1851			pointer = NULL;					\
1852			break;						\
1853		}							\
1854		retdata = (u_int8_t *)(dbt)->data + *__p--;		\
1855		retdlen = *__p--;					\
1856		pointer = __p;						\
1857		if (retdlen == 0 && retdata == (u_int8_t *)(dbt)->data)	\
1858			retdata = NULL;					\
1859	} while (0)
1860
1861#define	DB_MULTIPLE_KEY_NEXT(pointer, dbt, retkey, retklen, retdata, retdlen) \
1862	do {								\
1863		u_int32_t *__p = (u_int32_t *)(pointer);		\
1864		if (*__p == (u_int32_t)-1) {				\
1865			retdata = NULL;					\
1866			retkey = NULL;					\
1867			pointer = NULL;					\
1868			break;						\
1869		}							\
1870		retkey = (u_int8_t *)(dbt)->data + *__p--;		\
1871		retklen = *__p--;					\
1872		retdata = (u_int8_t *)(dbt)->data + *__p--;		\
1873		retdlen = *__p--;					\
1874		pointer = __p;						\
1875	} while (0)
1876
1877#define	DB_MULTIPLE_RECNO_NEXT(pointer, dbt, recno, retdata, retdlen)   \
1878	do {								\
1879		u_int32_t *__p = (u_int32_t *)(pointer);		\
1880		if (*__p == (u_int32_t)0) {				\
1881			recno = 0;					\
1882			retdata = NULL;					\
1883			pointer = NULL;					\
1884			break;						\
1885		}							\
1886		recno = *__p--;						\
1887		retdata = (u_int8_t *)(dbt)->data + *__p--;		\
1888		retdlen = *__p--;					\
1889		pointer = __p;						\
1890	} while (0)
1891
1892#define DB_MULTIPLE_WRITE_INIT(pointer, dbt)				\
1893	do {								\
1894		(dbt)->flags |= DB_DBT_BULK;				\
1895		pointer = (u_int8_t *)(dbt)->data +			\
1896		    (dbt)->ulen - sizeof(u_int32_t);			\
1897		*(u_int32_t *)(pointer) = (u_int32_t)-1;		\
1898	} while (0)
1899
1900#define DB_MULTIPLE_RESERVE_NEXT(pointer, dbt, writedata, writedlen)	\
1901	do {								\
1902		u_int32_t *__p = (u_int32_t *)(pointer);		\
1903		u_int32_t __off = ((pointer) ==	(u_int8_t *)(dbt)->data +\
1904		    (dbt)->ulen - sizeof(u_int32_t)) ?  0 : __p[1] + __p[2];\
1905		if ((u_int8_t *)(dbt)->data + __off + (writedlen) >	\
1906		    (u_int8_t *)(__p - 2))				\
1907			writedata = NULL;				\
1908		else {							\
1909			writedata = (u_int8_t *)(dbt)->data + __off;	\
1910			__p[0] = __off;					\
1911			__p[-1] = (u_int32_t)(writedlen);		\
1912			__p[-2] = (u_int32_t)-1;			\
1913			pointer = __p - 2;				\
1914		}							\
1915	} while (0)
1916
1917#define DB_MULTIPLE_WRITE_NEXT(pointer, dbt, writedata, writedlen)	\
1918	do {								\
1919		void *__destd;						\
1920		DB_MULTIPLE_RESERVE_NEXT((pointer), (dbt),		\
1921		    __destd, (writedlen));				\
1922		if (__destd == NULL)					\
1923			pointer = NULL;					\
1924		else							\
1925			memcpy(__destd, (writedata), (writedlen));	\
1926	} while (0)
1927
1928#define DB_MULTIPLE_KEY_RESERVE_NEXT(pointer, dbt, writekey, writeklen, writedata, writedlen) \
1929	do {								\
1930		u_int32_t *__p = (u_int32_t *)(pointer);		\
1931		u_int32_t __off = ((pointer) == (u_int8_t *)(dbt)->data +\
1932		    (dbt)->ulen - sizeof(u_int32_t)) ?  0 : __p[1] + __p[2];\
1933		if ((u_int8_t *)(dbt)->data + __off + (writeklen) +	\
1934		    (writedlen) > (u_int8_t *)(__p - 4)) {		\
1935			writekey = NULL;				\
1936			writedata = NULL;				\
1937		} else {						\
1938			writekey = (u_int8_t *)(dbt)->data + __off;	\
1939			__p[0] = __off;					\
1940			__p[-1] = (u_int32_t)(writeklen);		\
1941			__p -= 2;					\
1942			__off += (u_int32_t)(writeklen);		\
1943			writedata = (u_int8_t *)(dbt)->data + __off;	\
1944			__p[0] = __off;					\
1945			__p[-1] = (u_int32_t)(writedlen);		\
1946			__p[-2] = (u_int32_t)-1;			\
1947			pointer = __p - 2;				\
1948		}							\
1949	} while (0)
1950
1951#define DB_MULTIPLE_KEY_WRITE_NEXT(pointer, dbt, writekey, writeklen, writedata, writedlen) \
1952	do {								\
1953		void *__destk, *__destd;				\
1954		DB_MULTIPLE_KEY_RESERVE_NEXT((pointer), (dbt),		\
1955		    __destk, (writeklen), __destd, (writedlen));	\
1956		if (__destk == NULL)					\
1957			pointer = NULL;					\
1958		else {							\
1959			memcpy(__destk, (writekey), (writeklen));	\
1960			if (__destd != NULL)				\
1961				memcpy(__destd, (writedata), (writedlen));\
1962		}							\
1963	} while (0)
1964
1965#define DB_MULTIPLE_RECNO_WRITE_INIT(pointer, dbt)			\
1966	do {								\
1967		(dbt)->flags |= DB_DBT_BULK;				\
1968		pointer = (u_int8_t *)(dbt)->data +			\
1969		    (dbt)->ulen - sizeof(u_int32_t);			\
1970		*(u_int32_t *)(pointer) = 0;				\
1971	} while (0)
1972
1973#define DB_MULTIPLE_RECNO_RESERVE_NEXT(pointer, dbt, recno, writedata, writedlen) \
1974	do {								\
1975		u_int32_t *__p = (u_int32_t *)(pointer);		\
1976		u_int32_t __off = ((pointer) == (u_int8_t *)(dbt)->data +\
1977		    (dbt)->ulen - sizeof(u_int32_t)) ? 0 : __p[1] + __p[2]; \
1978		if (((u_int8_t *)(dbt)->data + __off) + (writedlen) >	\
1979		    (u_int8_t *)(__p - 3))				\
1980			writedata = NULL;				\
1981		else {							\
1982			writedata = (u_int8_t *)(dbt)->data + __off;	\
1983			__p[0] = (u_int32_t)(recno);			\
1984			__p[-1] = __off;				\
1985			__p[-2] = (u_int32_t)(writedlen);		\
1986			__p[-3] = 0;					\
1987			pointer = __p - 3;				\
1988		}							\
1989	} while (0)
1990
1991#define DB_MULTIPLE_RECNO_WRITE_NEXT(pointer, dbt, recno, writedata, writedlen)\
1992	do {								\
1993		void *__destd;						\
1994		DB_MULTIPLE_RECNO_RESERVE_NEXT((pointer), (dbt),	\
1995		    (recno), __destd, (writedlen));			\
1996		if (__destd == NULL)					\
1997			pointer = NULL;					\
1998		else if ((writedlen) != 0)				\
1999			memcpy(__destd, (writedata), (writedlen));	\
2000	} while (0)
2001
2002struct __db_heap_rid {
2003	db_pgno_t pgno;			/* Page number. */
2004	db_indx_t indx;			/* Index in the offset table. */
2005};
2006#define DB_HEAP_RID_SZ	(sizeof(db_pgno_t) + sizeof(db_indx_t))
2007
2008/*******************************************************
2009 * Access method cursors.
2010 *******************************************************/
2011struct __dbc {
2012	DB *dbp;			/* Backing database */
2013	DB_ENV *dbenv;			/* Backing environment */
2014	ENV *env;			/* Backing environment */
2015
2016	DB_THREAD_INFO *thread_info;	/* Thread that owns this cursor. */
2017	DB_TXN	 *txn;			/* Associated transaction. */
2018	DB_CACHE_PRIORITY priority;	/* Priority in cache. */
2019
2020	/*
2021	 * Active/free cursor queues.
2022	 *
2023	 * !!!
2024	 * Explicit representations of structures from queue.h.
2025	 * TAILQ_ENTRY(__dbc) links;
2026	 */
2027	struct {
2028		DBC *tqe_next;
2029		DBC **tqe_prev;
2030	} links;
2031
2032	/*
2033	 * Cursor queue of the owning transaction.
2034	 *
2035	 * !!!
2036	 * Explicit representations of structures from queue.h.
2037	 * TAILQ_ENTRY(__dbc) txn_cursors;
2038	 */
2039	struct {
2040		DBC *tqe_next;	/* next element */
2041		DBC **tqe_prev;	/* address of previous next element */
2042	} txn_cursors;
2043
2044	/*
2045	 * The DBT *'s below are used by the cursor routines to return
2046	 * data to the user when DBT flags indicate that DB should manage
2047	 * the returned memory.  They point at a DBT containing the buffer
2048	 * and length that will be used, and "belonging" to the handle that
2049	 * should "own" this memory.  This may be a "my_*" field of this
2050	 * cursor--the default--or it may be the corresponding field of
2051	 * another cursor, a DB handle, a join cursor, etc.  In general, it
2052	 * will be whatever handle the user originally used for the current
2053	 * DB interface call.
2054	 */
2055	DBT	 *rskey;		/* Returned secondary key. */
2056	DBT	 *rkey;			/* Returned [primary] key. */
2057	DBT	 *rdata;		/* Returned data. */
2058
2059	DBT	  my_rskey;		/* Space for returned secondary key. */
2060	DBT	  my_rkey;		/* Space for returned [primary] key. */
2061	DBT	  my_rdata;		/* Space for returned data. */
2062
2063	DB_LOCKER *lref;		/* Reference to default locker. */
2064	DB_LOCKER *locker;		/* Locker for this operation. */
2065	DBT	  lock_dbt;		/* DBT referencing lock. */
2066	DB_LOCK_ILOCK lock;		/* Object to be locked. */
2067	DB_LOCK	  mylock;		/* CDB lock held on this cursor. */
2068
2069	DBTYPE	  dbtype;		/* Cursor type. */
2070
2071	DBC_INTERNAL *internal;		/* Access method private. */
2072
2073	/* DBC PUBLIC HANDLE LIST BEGIN */
2074	int (*close) __P((DBC *));
2075	int (*cmp) __P((DBC *, DBC *, int *, u_int32_t));
2076	int (*count) __P((DBC *, db_recno_t *, u_int32_t));
2077	int (*del) __P((DBC *, u_int32_t));
2078	int (*dup) __P((DBC *, DBC **, u_int32_t));
2079	int (*get) __P((DBC *, DBT *, DBT *, u_int32_t));
2080	int (*get_priority) __P((DBC *, DB_CACHE_PRIORITY *));
2081	int (*pget) __P((DBC *, DBT *, DBT *, DBT *, u_int32_t));
2082	int (*put) __P((DBC *, DBT *, DBT *, u_int32_t));
2083	int (*set_priority) __P((DBC *, DB_CACHE_PRIORITY));
2084	/* DBC PUBLIC HANDLE LIST END */
2085
2086	/* The following are the method names deprecated in the 4.6 release. */
2087	int (*c_close) __P((DBC *));
2088	int (*c_count) __P((DBC *, db_recno_t *, u_int32_t));
2089	int (*c_del) __P((DBC *, u_int32_t));
2090	int (*c_dup) __P((DBC *, DBC **, u_int32_t));
2091	int (*c_get) __P((DBC *, DBT *, DBT *, u_int32_t));
2092	int (*c_pget) __P((DBC *, DBT *, DBT *, DBT *, u_int32_t));
2093	int (*c_put) __P((DBC *, DBT *, DBT *, u_int32_t));
2094
2095	/* DBC PRIVATE HANDLE LIST BEGIN */
2096	int (*am_bulk) __P((DBC *, DBT *, u_int32_t));
2097	int (*am_close) __P((DBC *, db_pgno_t, int *));
2098	int (*am_del) __P((DBC *, u_int32_t));
2099	int (*am_destroy) __P((DBC *));
2100	int (*am_get) __P((DBC *, DBT *, DBT *, u_int32_t, db_pgno_t *));
2101	int (*am_put) __P((DBC *, DBT *, DBT *, u_int32_t, db_pgno_t *));
2102	int (*am_writelock) __P((DBC *));
2103	/* DBC PRIVATE HANDLE LIST END */
2104
2105/*
2106 * DBC_DONTLOCK and DBC_RECOVER are used during recovery and transaction
2107 * abort.  If a transaction is being aborted or recovered then DBC_RECOVER
2108 * will be set and locking and logging will be disabled on this cursor.  If
2109 * we are performing a compensating transaction (e.g. free page processing)
2110 * then DB_DONTLOCK will be set to inhibit locking, but logging will still
2111 * be required. DB_DONTLOCK is also used if the whole database is locked.
2112 */
2113#define	DBC_ACTIVE		0x00001	/* Cursor in use. */
2114#define	DBC_BULK		0x00002	/* Bulk update cursor. */
2115#define	DBC_DONTLOCK		0x00004	/* Don't lock on this cursor. */
2116#define	DBC_DOWNREV		0x00008	/* Down rev replication master. */
2117#define	DBC_DUPLICATE		0x00010	/* Create a duplicate cursor. */
2118#define	DBC_ERROR		0x00020	/* Error in this request. */
2119#define	DBC_FAMILY		0x00040 /* Part of a locker family. */
2120#define	DBC_FROM_DB_GET		0x00080 /* Called from the DB->get() method. */
2121#define	DBC_MULTIPLE		0x00100	/* Return Multiple data. */
2122#define	DBC_MULTIPLE_KEY	0x00200	/* Return Multiple keys and data. */
2123#define	DBC_OPD			0x00400	/* Cursor references off-page dups. */
2124#define	DBC_OWN_LID		0x00800	/* Free lock id on destroy. */
2125#define	DBC_PARTITIONED		0x01000	/* Cursor for a partitioned db. */
2126#define	DBC_READ_COMMITTED	0x02000	/* Cursor has degree 2 isolation. */
2127#define	DBC_READ_UNCOMMITTED	0x04000	/* Cursor has degree 1 isolation. */
2128#define	DBC_RECOVER		0x08000	/* Recovery cursor; don't log/lock. */
2129#define	DBC_RMW			0x10000	/* Acquire write flag in read op. */
2130#define	DBC_TRANSIENT		0x20000	/* Cursor is transient. */
2131#define	DBC_WAS_READ_COMMITTED	0x40000	/* Cursor holds a read commited lock. */
2132#define	DBC_WRITECURSOR		0x80000	/* Cursor may be used to write (CDB). */
2133#define	DBC_WRITER	       0x100000	/* Cursor immediately writing (CDB). */
2134	u_int32_t flags;
2135};
2136
2137/* Key range statistics structure */
2138struct __key_range {
2139	double less;
2140	double equal;
2141	double greater;
2142};
2143
2144/* Btree/Recno statistics structure. */
2145struct __db_bt_stat { /* SHARED */
2146	u_int32_t bt_magic;		/* Magic number. */
2147	u_int32_t bt_version;		/* Version number. */
2148	u_int32_t bt_metaflags;		/* Metadata flags. */
2149	u_int32_t bt_nkeys;		/* Number of unique keys. */
2150	u_int32_t bt_ndata;		/* Number of data items. */
2151	u_int32_t bt_pagecnt;		/* Page count. */
2152	u_int32_t bt_pagesize;		/* Page size. */
2153	u_int32_t bt_minkey;		/* Minkey value. */
2154	u_int32_t bt_re_len;		/* Fixed-length record length. */
2155	u_int32_t bt_re_pad;		/* Fixed-length record pad. */
2156	u_int32_t bt_levels;		/* Tree levels. */
2157	u_int32_t bt_int_pg;		/* Internal pages. */
2158	u_int32_t bt_leaf_pg;		/* Leaf pages. */
2159	u_int32_t bt_dup_pg;		/* Duplicate pages. */
2160	u_int32_t bt_over_pg;		/* Overflow pages. */
2161	u_int32_t bt_empty_pg;		/* Empty pages. */
2162	u_int32_t bt_free;		/* Pages on the free list. */
2163	uintmax_t bt_int_pgfree;	/* Bytes free in internal pages. */
2164	uintmax_t bt_leaf_pgfree;	/* Bytes free in leaf pages. */
2165	uintmax_t bt_dup_pgfree;	/* Bytes free in duplicate pages. */
2166	uintmax_t bt_over_pgfree;	/* Bytes free in overflow pages. */
2167};
2168
2169struct __db_compact {
2170	/* Input Parameters. */
2171	u_int32_t	compact_fillpercent;	/* Desired fillfactor: 1-100 */
2172	db_timeout_t	compact_timeout;	/* Lock timeout. */
2173	u_int32_t	compact_pages;		/* Max pages to process. */
2174	/* Output Stats. */
2175	u_int32_t	compact_empty_buckets;	/* Empty hash buckets found. */
2176	u_int32_t	compact_pages_free;	/* Number of pages freed. */
2177	u_int32_t	compact_pages_examine;	/* Number of pages examine. */
2178	u_int32_t	compact_levels;		/* Number of levels removed. */
2179	u_int32_t	compact_deadlock;	/* Number of deadlocks. */
2180	db_pgno_t	compact_pages_truncated; /* Pages truncated to OS. */
2181	/* Internal. */
2182	db_pgno_t	compact_truncate;	/* Exchange pages above here. */
2183};
2184
2185/* Hash statistics structure. */
2186struct __db_h_stat { /* SHARED */
2187	u_int32_t hash_magic;		/* Magic number. */
2188	u_int32_t hash_version;		/* Version number. */
2189	u_int32_t hash_metaflags;	/* Metadata flags. */
2190	u_int32_t hash_nkeys;		/* Number of unique keys. */
2191	u_int32_t hash_ndata;		/* Number of data items. */
2192	u_int32_t hash_pagecnt;		/* Page count. */
2193	u_int32_t hash_pagesize;	/* Page size. */
2194	u_int32_t hash_ffactor;		/* Fill factor specified at create. */
2195	u_int32_t hash_buckets;		/* Number of hash buckets. */
2196	u_int32_t hash_free;		/* Pages on the free list. */
2197	uintmax_t hash_bfree;		/* Bytes free on bucket pages. */
2198	u_int32_t hash_bigpages;	/* Number of big key/data pages. */
2199	uintmax_t hash_big_bfree;	/* Bytes free on big item pages. */
2200	u_int32_t hash_overflows;	/* Number of overflow pages. */
2201	uintmax_t hash_ovfl_free;	/* Bytes free on ovfl pages. */
2202	u_int32_t hash_dup;		/* Number of dup pages. */
2203	uintmax_t hash_dup_free;	/* Bytes free on duplicate pages. */
2204};
2205
2206/* Heap statistics structure. */
2207struct __db_heap_stat { /* SHARED */
2208	u_int32_t heap_magic;		/* Magic number. */
2209	u_int32_t heap_version;		/* Version number. */
2210	u_int32_t heap_metaflags;	/* Metadata flags. */
2211	u_int32_t heap_nrecs;		/* Number of records. */
2212	u_int32_t heap_pagecnt;		/* Page count. */
2213	u_int32_t heap_pagesize;	/* Page size. */
2214	u_int32_t heap_nregions;	/* Number of regions. */
2215	u_int32_t heap_regionsize;	/* Number of pages in a region. */
2216};
2217
2218/* Queue statistics structure. */
2219struct __db_qam_stat { /* SHARED */
2220	u_int32_t qs_magic;		/* Magic number. */
2221	u_int32_t qs_version;		/* Version number. */
2222	u_int32_t qs_metaflags;		/* Metadata flags. */
2223	u_int32_t qs_nkeys;		/* Number of unique keys. */
2224	u_int32_t qs_ndata;		/* Number of data items. */
2225	u_int32_t qs_pagesize;		/* Page size. */
2226	u_int32_t qs_extentsize;	/* Pages per extent. */
2227	u_int32_t qs_pages;		/* Data pages. */
2228	u_int32_t qs_re_len;		/* Fixed-length record length. */
2229	u_int32_t qs_re_pad;		/* Fixed-length record pad. */
2230	u_int32_t qs_pgfree;		/* Bytes free in data pages. */
2231	u_int32_t qs_first_recno;	/* First not deleted record. */
2232	u_int32_t qs_cur_recno;		/* Next available record number. */
2233};
2234
2235/*******************************************************
2236 * Environment.
2237 *******************************************************/
2238#define	DB_REGION_MAGIC	0x120897	/* Environment magic number. */
2239
2240/*
2241 * Database environment structure.
2242 *
2243 * This is the public database environment handle.  The private environment
2244 * handle is the ENV structure.   The user owns this structure, the library
2245 * owns the ENV structure.  The reason there are two structures is because
2246 * the user's configuration outlives any particular DB_ENV->open call, and
2247 * separate structures allows us to easily discard internal information without
2248 * discarding the user's configuration.
2249 *
2250 * Fields in the DB_ENV structure should normally be set only by application
2251 * DB_ENV handle methods.
2252 */
2253
2254/*
2255 * Memory configuration types.
2256 */
2257typedef enum {
2258	DB_MEM_LOCK=1,
2259	DB_MEM_LOCKOBJECT=2,
2260	DB_MEM_LOCKER=3,
2261	DB_MEM_LOGID=4,
2262	DB_MEM_TRANSACTION=5,
2263	DB_MEM_THREAD=6
2264} DB_MEM_CONFIG;
2265
2266/*
2267 * Backup configuration types.
2268 */
2269typedef enum {
2270	DB_BACKUP_READ_COUNT = 1,
2271	DB_BACKUP_READ_SLEEP = 2,
2272	DB_BACKUP_SIZE = 3,
2273	DB_BACKUP_WRITE_DIRECT = 4
2274} DB_BACKUP_CONFIG;
2275
2276struct __db_env {
2277	ENV *env;			/* Linked ENV structure */
2278
2279	/*
2280	 * The DB_ENV structure can be used concurrently, so field access is
2281	 * protected.
2282	 */
2283	db_mutex_t mtx_db_env;		/* DB_ENV structure mutex */
2284
2285					/* Error message callback */
2286	void (*db_errcall) __P((const DB_ENV *, const char *, const char *));
2287	FILE		*db_errfile;	/* Error message file stream */
2288	const char	*db_errpfx;	/* Error message prefix */
2289
2290					/* Other message callback */
2291	void (*db_msgcall) __P((const DB_ENV *, const char *));
2292	FILE		*db_msgfile;	/* Other message file stream */
2293
2294	/* Other application callback functions */
2295	int   (*app_dispatch) __P((DB_ENV *, DBT *, DB_LSN *, db_recops));
2296	void  (*db_event_func) __P((DB_ENV *, u_int32_t, void *));
2297	void  (*db_feedback) __P((DB_ENV *, int, int));
2298	void  (*db_free) __P((void *));
2299	void  (*db_paniccall) __P((DB_ENV *, int));
2300	void *(*db_malloc) __P((size_t));
2301	void *(*db_realloc) __P((void *, size_t));
2302	int   (*is_alive) __P((DB_ENV *, pid_t, db_threadid_t, u_int32_t));
2303	void  (*thread_id) __P((DB_ENV *, pid_t *, db_threadid_t *));
2304	char *(*thread_id_string) __P((DB_ENV *, pid_t, db_threadid_t, char *));
2305
2306	/* Application specified paths */
2307	char	*db_log_dir;		/* Database log file directory */
2308	char	*db_md_dir;		/* Persistent metadata directory */
2309	char	*db_tmp_dir;		/* Database tmp file directory */
2310
2311	char    *db_create_dir;		/* Create directory for data files */
2312	char   **db_data_dir;		/* Database data file directories */
2313	int	 data_cnt;		/* Database data file slots */
2314	int	 data_next;		/* Next database data file slot */
2315
2316	char	*intermediate_dir_mode;	/* Intermediate directory perms */
2317
2318	long	 shm_key;		/* shmget key */
2319
2320	char	*passwd;		/* Cryptography support */
2321	size_t	 passwd_len;
2322
2323	/* Private handle references */
2324	void	*app_private;		/* Application-private handle */
2325	void	*api1_internal;		/* C++, Perl API private */
2326	void	*api2_internal;		/* Java API private */
2327
2328	u_int32_t	verbose;	/* DB_VERB_XXX flags */
2329
2330	/* Mutex configuration */
2331	u_int32_t	mutex_align;	/* Mutex alignment */
2332	u_int32_t	mutex_cnt;	/* Number of mutexes to configure */
2333	u_int32_t	mutex_inc;	/* Number of mutexes to add */
2334	u_int32_t	mutex_max;	/* Max number of mutexes */
2335	u_int32_t	mutex_tas_spins;/* Test-and-set spin count */
2336
2337	/* Locking configuration */
2338	u_int8_t       *lk_conflicts;	/* Two dimensional conflict matrix */
2339	int		lk_modes;	/* Number of lock modes in table */
2340	u_int32_t	lk_detect;	/* Deadlock detect on all conflicts */
2341	u_int32_t	lk_max;	/* Maximum number of locks */
2342	u_int32_t	lk_max_lockers;/* Maximum number of lockers */
2343	u_int32_t	lk_max_objects;/* Maximum number of locked objects */
2344	u_int32_t	lk_init;	/* Initial number of locks */
2345	u_int32_t	lk_init_lockers;/* Initial number of lockers */
2346	u_int32_t	lk_init_objects;/* Initial number of locked objects */
2347	u_int32_t	lk_partitions ;/* Number of object partitions */
2348	db_timeout_t	lk_timeout;	/* Lock timeout period */
2349	/* Used during initialization */
2350	u_int32_t	locker_t_size;	/* Locker hash table size. */
2351	u_int32_t	object_t_size;	/* Object hash table size. */
2352
2353	/* Logging configuration */
2354	u_int32_t	lg_bsize;	/* Buffer size */
2355	u_int32_t	lg_fileid_init;	/* Initial allocation for fname structs */
2356	int		lg_filemode;	/* Log file permission mode */
2357	u_int32_t	lg_regionmax;	/* Region size */
2358	u_int32_t	lg_size;	/* Log file size */
2359	u_int32_t	lg_flags;	/* Log configuration */
2360
2361	/* Memory pool configuration */
2362	u_int32_t	mp_gbytes;	/* Cache size: GB */
2363	u_int32_t	mp_bytes;	/* Cache size: bytes */
2364	u_int32_t	mp_max_gbytes;	/* Maximum cache size: GB */
2365	u_int32_t	mp_max_bytes;	/* Maximum cache size: bytes */
2366	size_t		mp_mmapsize;	/* Maximum file size for mmap */
2367	int		mp_maxopenfd;	/* Maximum open file descriptors */
2368	int		mp_maxwrite;	/* Maximum buffers to write */
2369	u_int		mp_ncache;	/* Initial number of cache regions */
2370	u_int32_t	mp_pagesize;	/* Average page size */
2371	u_int32_t	mp_tablesize;	/* Approximate hash table size */
2372	u_int32_t	mp_mtxcount;	/* Number of mutexs */
2373					/* Sleep after writing max buffers */
2374	db_timeout_t	mp_maxwrite_sleep;
2375
2376	/* Transaction configuration */
2377	u_int32_t	tx_init;	/* Initial number of transactions */
2378	u_int32_t	tx_max;		/* Maximum number of transactions */
2379	time_t		tx_timestamp;	/* Recover to specific timestamp */
2380	db_timeout_t	tx_timeout;	/* Timeout for transactions */
2381
2382	/* Thread tracking configuration */
2383	u_int32_t	thr_init;	/* Thread count */
2384	u_int32_t	thr_max;	/* Thread max */
2385	roff_t		memory_max;	/* Maximum region memory */
2386
2387	/*
2388	 * The following fields are not strictly user-owned, but they outlive
2389	 * the ENV structure, and so are stored here.
2390	 */
2391	DB_FH		*registry;	/* DB_REGISTER file handle */
2392	u_int32_t	registry_off;	/*
2393					 * Offset of our slot.  We can't use
2394					 * off_t because its size depends on
2395					 * build settings.
2396					 */
2397        db_timeout_t	envreg_timeout; /* DB_REGISTER wait timeout */
2398
2399#define	DB_ENV_AUTO_COMMIT	0x00000001 /* DB_AUTO_COMMIT */
2400#define	DB_ENV_CDB_ALLDB	0x00000002 /* CDB environment wide locking */
2401#define	DB_ENV_FAILCHK		0x00000004 /* Failchk is running */
2402#define	DB_ENV_DIRECT_DB	0x00000008 /* DB_DIRECT_DB set */
2403#define	DB_ENV_DSYNC_DB		0x00000010 /* DB_DSYNC_DB set */
2404#define	DB_ENV_DATABASE_LOCKING	0x00000020 /* Try database-level locking */
2405#define	DB_ENV_MULTIVERSION	0x00000040 /* DB_MULTIVERSION set */
2406#define	DB_ENV_NOLOCKING	0x00000080 /* DB_NOLOCKING set */
2407#define	DB_ENV_NOMMAP		0x00000100 /* DB_NOMMAP set */
2408#define	DB_ENV_NOPANIC		0x00000200 /* Okay if panic set */
2409#define	DB_ENV_OVERWRITE	0x00000400 /* DB_OVERWRITE set */
2410#define	DB_ENV_REGION_INIT	0x00000800 /* DB_REGION_INIT set */
2411#define	DB_ENV_TIME_NOTGRANTED	0x00001000 /* DB_TIME_NOTGRANTED set */
2412#define	DB_ENV_TXN_NOSYNC	0x00002000 /* DB_TXN_NOSYNC set */
2413#define	DB_ENV_TXN_NOWAIT	0x00004000 /* DB_TXN_NOWAIT set */
2414#define	DB_ENV_TXN_SNAPSHOT	0x00008000 /* DB_TXN_SNAPSHOT set */
2415#define	DB_ENV_TXN_WRITE_NOSYNC	0x00010000 /* DB_TXN_WRITE_NOSYNC set */
2416#define	DB_ENV_YIELDCPU		0x00020000 /* DB_YIELDCPU set */
2417#define DB_ENV_HOTBACKUP	0x00040000 /* DB_HOTBACKUP_IN_PROGRESS set */
2418#define DB_ENV_NOFLUSH		0x00080000 /* DB_NOFLUSH set */
2419	u_int32_t flags;
2420
2421	/* DB_ENV PUBLIC HANDLE LIST BEGIN */
2422	int  (*add_data_dir) __P((DB_ENV *, const char *));
2423	int  (*backup)	__P((DB_ENV *, const char *, u_int32_t));
2424	int  (*cdsgroup_begin) __P((DB_ENV *, DB_TXN **));
2425	int  (*close) __P((DB_ENV *, u_int32_t));
2426	int  (*dbbackup) __P((DB_ENV *, const char *, const char *, u_int32_t));
2427	int  (*dbremove) __P((DB_ENV *,
2428		DB_TXN *, const char *, const char *, u_int32_t));
2429	int  (*dbrename) __P((DB_ENV *,
2430		DB_TXN *, const char *, const char *, const char *, u_int32_t));
2431	void (*err) __P((const DB_ENV *, int, const char *, ...));
2432	void (*errx) __P((const DB_ENV *, const char *, ...));
2433	int  (*failchk) __P((DB_ENV *, u_int32_t));
2434	int  (*fileid_reset) __P((DB_ENV *, const char *, u_int32_t));
2435	int  (*get_alloc) __P((DB_ENV *, void *(**)(size_t),
2436		void *(**)(void *, size_t), void (**)(void *)));
2437	int  (*get_app_dispatch)
2438		__P((DB_ENV *, int (**)(DB_ENV *, DBT *, DB_LSN *, db_recops)));
2439	int  (*get_cache_max) __P((DB_ENV *, u_int32_t *, u_int32_t *));
2440	int  (*get_cachesize) __P((DB_ENV *, u_int32_t *, u_int32_t *, int *));
2441	int  (*get_create_dir) __P((DB_ENV *, const char **));
2442	int  (*get_data_dirs) __P((DB_ENV *, const char ***));
2443	int  (*get_data_len) __P((DB_ENV *, u_int32_t *));
2444	int  (*get_backup_callbacks) __P((DB_ENV *,
2445		int (**)(DB_ENV *, const char *, const char *, void **),
2446		int (**)(DB_ENV *, u_int32_t, u_int32_t, u_int32_t, u_int8_t *, void *),
2447		int (**)(DB_ENV *, const char *, void *)));
2448	int  (*get_backup_config) __P((DB_ENV *, DB_BACKUP_CONFIG, u_int32_t *));
2449	int  (*get_encrypt_flags) __P((DB_ENV *, u_int32_t *));
2450	void (*get_errcall) __P((DB_ENV *,
2451		void (**)(const DB_ENV *, const char *, const char *)));
2452	void (*get_errfile) __P((DB_ENV *, FILE **));
2453	void (*get_errpfx) __P((DB_ENV *, const char **));
2454	int  (*get_flags) __P((DB_ENV *, u_int32_t *));
2455	int  (*get_feedback) __P((DB_ENV *, void (**)(DB_ENV *, int, int)));
2456	int  (*get_home) __P((DB_ENV *, const char **));
2457	int  (*get_intermediate_dir_mode) __P((DB_ENV *, const char **));
2458	int  (*get_isalive) __P((DB_ENV *,
2459		int (**)(DB_ENV *, pid_t, db_threadid_t, u_int32_t)));
2460	int  (*get_lg_bsize) __P((DB_ENV *, u_int32_t *));
2461	int  (*get_lg_dir) __P((DB_ENV *, const char **));
2462	int  (*get_lg_filemode) __P((DB_ENV *, int *));
2463	int  (*get_lg_max) __P((DB_ENV *, u_int32_t *));
2464	int  (*get_lg_regionmax) __P((DB_ENV *, u_int32_t *));
2465	int  (*get_lk_conflicts) __P((DB_ENV *, const u_int8_t **, int *));
2466	int  (*get_lk_detect) __P((DB_ENV *, u_int32_t *));
2467	int  (*get_lk_max_lockers) __P((DB_ENV *, u_int32_t *));
2468	int  (*get_lk_max_locks) __P((DB_ENV *, u_int32_t *));
2469	int  (*get_lk_max_objects) __P((DB_ENV *, u_int32_t *));
2470	int  (*get_lk_partitions) __P((DB_ENV *, u_int32_t *));
2471	int  (*get_lk_priority) __P((DB_ENV *, u_int32_t, u_int32_t *));
2472	int  (*get_lk_tablesize) __P((DB_ENV *, u_int32_t *));
2473	int  (*get_memory_init) __P((DB_ENV *, DB_MEM_CONFIG, u_int32_t *));
2474	int  (*get_memory_max) __P((DB_ENV *, u_int32_t *, u_int32_t *));
2475	int  (*get_metadata_dir) __P((DB_ENV *, const char **));
2476	int  (*get_mp_max_openfd) __P((DB_ENV *, int *));
2477	int  (*get_mp_max_write) __P((DB_ENV *, int *, db_timeout_t *));
2478	int  (*get_mp_mmapsize) __P((DB_ENV *, size_t *));
2479	int  (*get_mp_mtxcount) __P((DB_ENV *, u_int32_t *));
2480	int  (*get_mp_pagesize) __P((DB_ENV *, u_int32_t *));
2481	int  (*get_mp_tablesize) __P((DB_ENV *, u_int32_t *));
2482	void (*get_msgcall)
2483		__P((DB_ENV *, void (**)(const DB_ENV *, const char *)));
2484	void (*get_msgfile) __P((DB_ENV *, FILE **));
2485	int  (*get_open_flags) __P((DB_ENV *, u_int32_t *));
2486	int  (*get_shm_key) __P((DB_ENV *, long *));
2487	int  (*get_thread_count) __P((DB_ENV *, u_int32_t *));
2488	int  (*get_thread_id_fn)
2489		__P((DB_ENV *, void (**)(DB_ENV *, pid_t *, db_threadid_t *)));
2490	int  (*get_thread_id_string_fn) __P((DB_ENV *,
2491		char *(**)(DB_ENV *, pid_t, db_threadid_t, char *)));
2492	int  (*get_timeout) __P((DB_ENV *, db_timeout_t *, u_int32_t));
2493	int  (*get_tmp_dir) __P((DB_ENV *, const char **));
2494	int  (*get_tx_max) __P((DB_ENV *, u_int32_t *));
2495	int  (*get_tx_timestamp) __P((DB_ENV *, time_t *));
2496	int  (*get_verbose) __P((DB_ENV *, u_int32_t, int *));
2497	int  (*is_bigendian) __P((void));
2498	int  (*lock_detect) __P((DB_ENV *, u_int32_t, u_int32_t, int *));
2499	int  (*lock_get) __P((DB_ENV *,
2500		u_int32_t, u_int32_t, DBT *, db_lockmode_t, DB_LOCK *));
2501	int  (*lock_id) __P((DB_ENV *, u_int32_t *));
2502	int  (*lock_id_free) __P((DB_ENV *, u_int32_t));
2503	int  (*lock_put) __P((DB_ENV *, DB_LOCK *));
2504	int  (*lock_stat) __P((DB_ENV *, DB_LOCK_STAT **, u_int32_t));
2505	int  (*lock_stat_print) __P((DB_ENV *, u_int32_t));
2506	int  (*lock_vec) __P((DB_ENV *,
2507		u_int32_t, u_int32_t, DB_LOCKREQ *, int, DB_LOCKREQ **));
2508	int  (*log_archive) __P((DB_ENV *, char **[], u_int32_t));
2509	int  (*log_cursor) __P((DB_ENV *, DB_LOGC **, u_int32_t));
2510	int  (*log_file) __P((DB_ENV *, const DB_LSN *, char *, size_t));
2511	int  (*log_flush) __P((DB_ENV *, const DB_LSN *));
2512	int  (*log_get_config) __P((DB_ENV *, u_int32_t, int *));
2513	int  (*log_printf) __P((DB_ENV *, DB_TXN *, const char *, ...));
2514	int  (*log_put) __P((DB_ENV *, DB_LSN *, const DBT *, u_int32_t));
2515	int  (*log_put_record) __P((DB_ENV *, DB *, DB_TXN *, DB_LSN *,
2516		u_int32_t, u_int32_t, u_int32_t, u_int32_t,
2517		DB_LOG_RECSPEC *, ...));
2518	int  (*log_read_record) __P((DB_ENV *, DB **,
2519		void *, void *, DB_LOG_RECSPEC *, u_int32_t, void **));
2520	int  (*log_set_config) __P((DB_ENV *, u_int32_t, int));
2521	int  (*log_stat) __P((DB_ENV *, DB_LOG_STAT **, u_int32_t));
2522	int  (*log_stat_print) __P((DB_ENV *, u_int32_t));
2523	int  (*log_verify) __P((DB_ENV *, const DB_LOG_VERIFY_CONFIG *));
2524	int  (*lsn_reset) __P((DB_ENV *, const char *, u_int32_t));
2525	int  (*memp_fcreate) __P((DB_ENV *, DB_MPOOLFILE **, u_int32_t));
2526	int  (*memp_register) __P((DB_ENV *, int, int (*)(DB_ENV *, db_pgno_t,
2527		void *, DBT *), int (*)(DB_ENV *, db_pgno_t, void *, DBT *)));
2528	int  (*memp_stat) __P((DB_ENV *,
2529		DB_MPOOL_STAT **, DB_MPOOL_FSTAT ***, u_int32_t));
2530	int  (*memp_stat_print) __P((DB_ENV *, u_int32_t));
2531	int  (*memp_sync) __P((DB_ENV *, DB_LSN *));
2532	int  (*memp_trickle) __P((DB_ENV *, int, int *));
2533	int  (*mutex_alloc) __P((DB_ENV *, u_int32_t, db_mutex_t *));
2534	int  (*mutex_free) __P((DB_ENV *, db_mutex_t));
2535	int  (*mutex_get_align) __P((DB_ENV *, u_int32_t *));
2536	int  (*mutex_get_increment) __P((DB_ENV *, u_int32_t *));
2537	int  (*mutex_get_init) __P((DB_ENV *, u_int32_t *));
2538	int  (*mutex_get_max) __P((DB_ENV *, u_int32_t *));
2539	int  (*mutex_get_tas_spins) __P((DB_ENV *, u_int32_t *));
2540	int  (*mutex_lock) __P((DB_ENV *, db_mutex_t));
2541	int  (*mutex_set_align) __P((DB_ENV *, u_int32_t));
2542	int  (*mutex_set_increment) __P((DB_ENV *, u_int32_t));
2543	int  (*mutex_set_init) __P((DB_ENV *, u_int32_t));
2544	int  (*mutex_set_max) __P((DB_ENV *, u_int32_t));
2545	int  (*mutex_set_tas_spins) __P((DB_ENV *, u_int32_t));
2546	int  (*mutex_stat) __P((DB_ENV *, DB_MUTEX_STAT **, u_int32_t));
2547	int  (*mutex_stat_print) __P((DB_ENV *, u_int32_t));
2548	int  (*mutex_unlock) __P((DB_ENV *, db_mutex_t));
2549	int  (*open) __P((DB_ENV *, const char *, u_int32_t, int));
2550	int  (*remove) __P((DB_ENV *, const char *, u_int32_t));
2551	int  (*rep_elect) __P((DB_ENV *, u_int32_t, u_int32_t, u_int32_t));
2552	int  (*rep_flush) __P((DB_ENV *));
2553	int  (*rep_get_clockskew) __P((DB_ENV *, u_int32_t *, u_int32_t *));
2554	int  (*rep_get_config) __P((DB_ENV *, u_int32_t, int *));
2555	int  (*rep_get_limit) __P((DB_ENV *, u_int32_t *, u_int32_t *));
2556	int  (*rep_get_nsites) __P((DB_ENV *, u_int32_t *));
2557	int  (*rep_get_priority) __P((DB_ENV *, u_int32_t *));
2558	int  (*rep_get_request) __P((DB_ENV *, u_int32_t *, u_int32_t *));
2559	int  (*rep_get_timeout) __P((DB_ENV *, int, u_int32_t *));
2560	int  (*rep_process_message)
2561		__P((DB_ENV *, DBT *, DBT *, int, DB_LSN *));
2562	int  (*rep_set_clockskew) __P((DB_ENV *, u_int32_t, u_int32_t));
2563	int  (*rep_set_config) __P((DB_ENV *, u_int32_t, int));
2564	int  (*rep_set_limit) __P((DB_ENV *, u_int32_t, u_int32_t));
2565	int  (*rep_set_nsites) __P((DB_ENV *, u_int32_t));
2566	int  (*rep_set_priority) __P((DB_ENV *, u_int32_t));
2567	int  (*rep_set_request) __P((DB_ENV *, u_int32_t, u_int32_t));
2568	int  (*rep_set_timeout) __P((DB_ENV *, int, db_timeout_t));
2569	int  (*rep_set_transport) __P((DB_ENV *, int, int (*)(DB_ENV *,
2570		const DBT *, const DBT *, const DB_LSN *, int, u_int32_t)));
2571	int  (*rep_start) __P((DB_ENV *, DBT *, u_int32_t));
2572	int  (*rep_stat) __P((DB_ENV *, DB_REP_STAT **, u_int32_t));
2573	int  (*rep_stat_print) __P((DB_ENV *, u_int32_t));
2574	int  (*rep_sync) __P((DB_ENV *, u_int32_t));
2575	int  (*repmgr_channel) __P((DB_ENV *, int, DB_CHANNEL **, u_int32_t));
2576	int  (*repmgr_get_ack_policy) __P((DB_ENV *, int *));
2577	int  (*repmgr_local_site) __P((DB_ENV *, DB_SITE **));
2578	int  (*repmgr_msg_dispatch) __P((DB_ENV *,
2579		void (*)(DB_ENV *, DB_CHANNEL *, DBT *, u_int32_t, u_int32_t),
2580		u_int32_t));
2581	int  (*repmgr_set_ack_policy) __P((DB_ENV *, int));
2582	int  (*repmgr_site)
2583		__P((DB_ENV *, const char *, u_int, DB_SITE**, u_int32_t));
2584	int  (*repmgr_site_by_eid) __P((DB_ENV *, int, DB_SITE**));
2585	int  (*repmgr_site_list) __P((DB_ENV *, u_int *, DB_REPMGR_SITE **));
2586	int  (*repmgr_start) __P((DB_ENV *, int, u_int32_t));
2587	int  (*repmgr_stat) __P((DB_ENV *, DB_REPMGR_STAT **, u_int32_t));
2588	int  (*repmgr_stat_print) __P((DB_ENV *, u_int32_t));
2589	int  (*set_alloc) __P((DB_ENV *, void *(*)(size_t),
2590		void *(*)(void *, size_t), void (*)(void *)));
2591	int  (*set_app_dispatch)
2592		__P((DB_ENV *, int (*)(DB_ENV *, DBT *, DB_LSN *, db_recops)));
2593	int  (*set_cache_max) __P((DB_ENV *, u_int32_t, u_int32_t));
2594	int  (*set_cachesize) __P((DB_ENV *, u_int32_t, u_int32_t, int));
2595	int  (*set_create_dir) __P((DB_ENV *, const char *));
2596	int  (*set_data_dir) __P((DB_ENV *, const char *));
2597	int  (*set_data_len) __P((DB_ENV *, u_int32_t));
2598	int  (*set_backup_callbacks) __P((DB_ENV *,
2599		int (*)(DB_ENV *, const char *, const char *, void **),
2600		int (*)(DB_ENV *, u_int32_t,
2601		    u_int32_t, u_int32_t, u_int8_t *, void *),
2602		int (*)(DB_ENV *, const char *, void *)));
2603	int  (*set_backup_config) __P((DB_ENV *, DB_BACKUP_CONFIG, u_int32_t));
2604	int  (*set_encrypt) __P((DB_ENV *, const char *, u_int32_t));
2605	void (*set_errcall) __P((DB_ENV *,
2606		void (*)(const DB_ENV *, const char *, const char *)));
2607	void (*set_errfile) __P((DB_ENV *, FILE *));
2608	void (*set_errpfx) __P((DB_ENV *, const char *));
2609	int  (*set_event_notify)
2610		__P((DB_ENV *, void (*)(DB_ENV *, u_int32_t, void *)));
2611	int  (*set_feedback) __P((DB_ENV *, void (*)(DB_ENV *, int, int)));
2612	int  (*set_flags) __P((DB_ENV *, u_int32_t, int));
2613	int  (*set_intermediate_dir_mode) __P((DB_ENV *, const char *));
2614	int  (*set_isalive) __P((DB_ENV *,
2615		int (*)(DB_ENV *, pid_t, db_threadid_t, u_int32_t)));
2616	int  (*set_lg_bsize) __P((DB_ENV *, u_int32_t));
2617	int  (*set_lg_dir) __P((DB_ENV *, const char *));
2618	int  (*set_lg_filemode) __P((DB_ENV *, int));
2619	int  (*set_lg_max) __P((DB_ENV *, u_int32_t));
2620	int  (*set_lg_regionmax) __P((DB_ENV *, u_int32_t));
2621	int  (*set_lk_conflicts) __P((DB_ENV *, u_int8_t *, int));
2622	int  (*set_lk_detect) __P((DB_ENV *, u_int32_t));
2623	int  (*set_lk_max_lockers) __P((DB_ENV *, u_int32_t));
2624	int  (*set_lk_max_locks) __P((DB_ENV *, u_int32_t));
2625	int  (*set_lk_max_objects) __P((DB_ENV *, u_int32_t));
2626	int  (*set_lk_partitions) __P((DB_ENV *, u_int32_t));
2627	int  (*set_lk_priority) __P((DB_ENV *, u_int32_t, u_int32_t));
2628	int  (*set_lk_tablesize) __P((DB_ENV *, u_int32_t));
2629	int  (*set_memory_init) __P((DB_ENV *, DB_MEM_CONFIG, u_int32_t));
2630	int  (*set_memory_max) __P((DB_ENV *, u_int32_t, u_int32_t));
2631	int  (*set_metadata_dir) __P((DB_ENV *, const char *));
2632	int  (*set_mp_max_openfd) __P((DB_ENV *, int));
2633	int  (*set_mp_max_write) __P((DB_ENV *, int, db_timeout_t));
2634	int  (*set_mp_mmapsize) __P((DB_ENV *, size_t));
2635	int  (*set_mp_mtxcount) __P((DB_ENV *, u_int32_t));
2636	int  (*set_mp_pagesize) __P((DB_ENV *, u_int32_t));
2637	int  (*set_mp_tablesize) __P((DB_ENV *, u_int32_t));
2638	void (*set_msgcall)
2639		__P((DB_ENV *, void (*)(const DB_ENV *, const char *)));
2640	void (*set_msgfile) __P((DB_ENV *, FILE *));
2641	int  (*set_paniccall) __P((DB_ENV *, void (*)(DB_ENV *, int)));
2642	int  (*set_shm_key) __P((DB_ENV *, long));
2643	int  (*set_thread_count) __P((DB_ENV *, u_int32_t));
2644	int  (*set_thread_id)
2645		__P((DB_ENV *, void (*)(DB_ENV *, pid_t *, db_threadid_t *)));
2646	int  (*set_thread_id_string) __P((DB_ENV *,
2647		char *(*)(DB_ENV *, pid_t, db_threadid_t, char *)));
2648	int  (*set_timeout) __P((DB_ENV *, db_timeout_t, u_int32_t));
2649	int  (*set_tmp_dir) __P((DB_ENV *, const char *));
2650	int  (*set_tx_max) __P((DB_ENV *, u_int32_t));
2651	int  (*set_tx_timestamp) __P((DB_ENV *, time_t *));
2652	int  (*set_verbose) __P((DB_ENV *, u_int32_t, int));
2653	int  (*txn_applied) __P((DB_ENV *,
2654		DB_TXN_TOKEN *, db_timeout_t, u_int32_t));
2655	int  (*stat_print) __P((DB_ENV *, u_int32_t));
2656	int  (*txn_begin) __P((DB_ENV *, DB_TXN *, DB_TXN **, u_int32_t));
2657	int  (*txn_checkpoint) __P((DB_ENV *, u_int32_t, u_int32_t, u_int32_t));
2658	int  (*txn_recover) __P((DB_ENV *,
2659		DB_PREPLIST *, long, long *, u_int32_t));
2660	int  (*txn_stat) __P((DB_ENV *, DB_TXN_STAT **, u_int32_t));
2661	int  (*txn_stat_print) __P((DB_ENV *, u_int32_t));
2662	/* DB_ENV PUBLIC HANDLE LIST END */
2663
2664	/* DB_ENV PRIVATE HANDLE LIST BEGIN */
2665	int  (*prdbt) __P((DBT *, int,
2666		const char *, void *, int (*)(void *, const void *), int, int));
2667	/* DB_ENV PRIVATE HANDLE LIST END */
2668};
2669
2670/*
2671 * Dispatch structure for recovery, log verification and print routines. Since
2672 * internal and external routines take different arguments (ENV versus DB_ENV),
2673 * we need something more elaborate than a single pointer and size.
2674 */
2675struct __db_distab {
2676	int   (**int_dispatch) __P((ENV *, DBT *, DB_LSN *, db_recops, void *));
2677	size_t	int_size;
2678	int   (**ext_dispatch) __P((DB_ENV *, DBT *, DB_LSN *, db_recops));
2679	size_t	ext_size;
2680};
2681
2682/*
2683 * Log verification configuration structure.
2684 */
2685struct __db_logvrfy_config {
2686	int continue_after_fail, verbose;
2687	u_int32_t cachesize;
2688	const char *temp_envhome;
2689	const char *dbfile, *dbname;
2690	DB_LSN start_lsn, end_lsn;
2691	time_t start_time, end_time;
2692};
2693
2694struct __db_channel {
2695	CHANNEL *channel;	/* Pointer to internal state details. */
2696	int eid;		/* Env. ID passed in constructor. */
2697	db_timeout_t timeout;
2698
2699	/* DB_CHANNEL PUBLIC HANDLE LIST BEGIN */
2700	int (*close) __P((DB_CHANNEL *, u_int32_t));
2701	int (*send_msg) __P((DB_CHANNEL *, DBT *, u_int32_t, u_int32_t));
2702	int (*send_request) __P((DB_CHANNEL *,
2703		DBT *, u_int32_t, DBT *, db_timeout_t, u_int32_t));
2704	int  (*set_timeout) __P((DB_CHANNEL *, db_timeout_t));
2705	/* DB_CHANNEL PUBLIC HANDLE LIST END */
2706};
2707
2708struct __db_site {
2709	ENV *env;
2710	int eid;
2711	const char *host;
2712	u_int port;
2713	u_int32_t flags;
2714
2715	/* DB_SITE PUBLIC HANDLE LIST BEGIN */
2716	int (*get_address) __P((DB_SITE *, const char **, u_int *));
2717	int (*get_config) __P((DB_SITE *, u_int32_t, u_int32_t *));
2718	int (*get_eid) __P((DB_SITE *, int *));
2719	int (*set_config) __P((DB_SITE *, u_int32_t, u_int32_t));
2720	int (*remove) __P((DB_SITE *));
2721	int (*close) __P((DB_SITE *));
2722	/* DB_SITE PUBLIC HANDLE LIST END */
2723};
2724
2725#if DB_DBM_HSEARCH != 0
2726/*******************************************************
2727 * Dbm/Ndbm historic interfaces.
2728 *******************************************************/
2729typedef struct __db DBM;
2730
2731#define	DBM_INSERT	0		/* Flags to dbm_store(). */
2732#define	DBM_REPLACE	1
2733
2734/*
2735 * The DB support for ndbm(3) always appends this suffix to the
2736 * file name to avoid overwriting the user's original database.
2737 */
2738#define	DBM_SUFFIX	".db"
2739
2740#if defined(_XPG4_2)
2741typedef struct {
2742	char *dptr;
2743	size_t dsize;
2744} datum;
2745#else
2746typedef struct {
2747	char *dptr;
2748	int dsize;
2749} datum;
2750#endif
2751
2752/*
2753 * Translate NDBM calls into DB calls so that DB doesn't step on the
2754 * application's name space.
2755 */
2756#define	dbm_clearerr(a)		__db_ndbm_clearerr@DB_VERSION_UNIQUE_NAME@(a)
2757#define	dbm_close(a)		__db_ndbm_close@DB_VERSION_UNIQUE_NAME@(a)
2758#define	dbm_delete(a, b)	__db_ndbm_delete@DB_VERSION_UNIQUE_NAME@(a, b)
2759#define	dbm_dirfno(a)		__db_ndbm_dirfno@DB_VERSION_UNIQUE_NAME@(a)
2760#define	dbm_error(a)		__db_ndbm_error@DB_VERSION_UNIQUE_NAME@(a)
2761#define	dbm_fetch(a, b)		__db_ndbm_fetch@DB_VERSION_UNIQUE_NAME@(a, b)
2762#define	dbm_firstkey(a)		__db_ndbm_firstkey@DB_VERSION_UNIQUE_NAME@(a)
2763#define	dbm_nextkey(a)		__db_ndbm_nextkey@DB_VERSION_UNIQUE_NAME@(a)
2764#define	dbm_open(a, b, c)	__db_ndbm_open@DB_VERSION_UNIQUE_NAME@(a, b, c)
2765#define	dbm_pagfno(a)		__db_ndbm_pagfno@DB_VERSION_UNIQUE_NAME@(a)
2766#define	dbm_rdonly(a)		__db_ndbm_rdonly@DB_VERSION_UNIQUE_NAME@(a)
2767#define	dbm_store(a, b, c, d) \
2768	__db_ndbm_store@DB_VERSION_UNIQUE_NAME@(a, b, c, d)
2769
2770/*
2771 * Translate DBM calls into DB calls so that DB doesn't step on the
2772 * application's name space.
2773 *
2774 * The global variables dbrdonly, dirf and pagf were not retained when 4BSD
2775 * replaced the dbm interface with ndbm, and are not supported here.
2776 */
2777#define	dbminit(a)	__db_dbm_init@DB_VERSION_UNIQUE_NAME@(a)
2778#define	dbmclose	__db_dbm_close@DB_VERSION_UNIQUE_NAME@
2779#if !defined(__cplusplus)
2780#define	delete(a)	__db_dbm_delete@DB_VERSION_UNIQUE_NAME@(a)
2781#define	store(a, b)	__db_dbm_store@DB_VERSION_UNIQUE_NAME@(a, b)
2782#endif
2783#define	fetch(a)	__db_dbm_fetch@DB_VERSION_UNIQUE_NAME@(a)
2784#define	firstkey	__db_dbm_firstkey@DB_VERSION_UNIQUE_NAME@
2785#define	nextkey(a)	__db_dbm_nextkey@DB_VERSION_UNIQUE_NAME@(a)
2786
2787/*******************************************************
2788 * Hsearch historic interface.
2789 *******************************************************/
2790typedef enum {
2791	FIND, ENTER
2792} ACTION;
2793
2794typedef struct entry {
2795	char *key;
2796	char *data;
2797} ENTRY;
2798
2799#define	hcreate(a)	__db_hcreate@DB_VERSION_UNIQUE_NAME@(a)
2800#define	hdestroy	__db_hdestroy@DB_VERSION_UNIQUE_NAME@
2801#define	hsearch(a, b)	__db_hsearch@DB_VERSION_UNIQUE_NAME@(a, b)
2802
2803#endif /* DB_DBM_HSEARCH */
2804
2805#if defined(__cplusplus)
2806}
2807#endif
2808
2809@platform_footer@
2810#endif /* !_DB_H_ */
2811