/* back-bdb.h - bdb back-end header file */
/* $OpenLDAP$ */
-/*
- * Copyright 2000-2002 The OpenLDAP Foundation, All Rights Reserved.
- * COPYING RESTRICTIONS APPLY, see COPYRIGHT file
+/* This work is part of OpenLDAP Software <http://www.openldap.org/>.
+ *
+ * Copyright 2000-2005 The OpenLDAP Foundation.
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted only as authorized by the OpenLDAP
+ * Public License.
+ *
+ * A copy of this license is available in the file LICENSE in the
+ * top-level directory of the distribution or, alternatively, at
+ * <http://www.OpenLDAP.org/license.html>.
*/
#ifndef _BACK_BDB_H_
#define _BACK_BDB_H_
#include <portable.h>
-#include <db.h>
-
#include "slap.h"
+#include <db.h>
+#include "alock.h"
LDAP_BEGIN_DECL
-#define BDB_FILTER_INDICES 1
-#define BDB_IDL_MULTI 1
-/* #define BDB_HIER 1 */
+#define DB_VERSION_FULL ((DB_VERSION_MAJOR << 24) | (DB_VERSION_MINOR << 16) | DB_VERSION_PATCH)
+
+#define BDB_SUBENTRIES 1
#define DN_BASE_PREFIX SLAP_INDEX_EQUALITY_PREFIX
#define DN_ONE_PREFIX '%'
#define bv2DBT(bv,t) ((t)->data = (bv)->bv_val, \
(t)->size = (bv)->bv_len )
-#define DEFAULT_MODE 0600
+#define BDB_TXN_RETRIES 16
-#define BDB_TXN_RETRIES 16
-
-#ifdef BDB_HIER
-#define BDB_DBENV_HOME LDAP_RUNDIR LDAP_DIRSEP "openldap-hdb"
-#else
-#define BDB_DBENV_HOME LDAP_RUNDIR LDAP_DIRSEP "openldap-bdb"
-#endif
-
-#ifdef BDB_SUBDIRS
-#define BDB_TMP_SUBDIR LDAP_DIRSEP "tmp"
-#define BDB_LG_SUBDIR LDAP_DIRSEP "log"
-#define BDB_DATA_SUBDIR LDAP_DIRSEP "data"
-#endif
+#define BDB_MAX_ADD_LOOP 30
#define BDB_SUFFIX ".bdb"
#define BDB_ID2ENTRY 0
-#ifdef BDB_HIER
-#define BDB_ID2PARENT 1
-#else
#define BDB_DN2ID 1
-#endif
#define BDB_NDB 2
/* The bdb on-disk entry format is pretty space-inefficient. Average
* sized user entries are 3-4K each. You need at least two entries to
* fit into a single database page, more is better. 64K is BDB's
- * upper bound. The same issues arise with IDLs in the index databases,
- * but it's nearly impossible to avoid overflows there.
- *
- * When using BDB_IDL_MULTI, the IDL size is no longer an issue. Smaller
- * pages are better for concurrency.
+ * upper bound. Smaller pages are better for concurrency.
*/
#ifndef BDB_ID2ENTRY_PAGESIZE
#define BDB_ID2ENTRY_PAGESIZE 16384
#endif
#ifndef BDB_PAGESIZE
-#ifdef BDB_IDL_MULTI
#define BDB_PAGESIZE 4096 /* BDB's original default */
-#else
-#define BDB_PAGESIZE 16384
#endif
+
+#define DEFAULT_CACHE_SIZE 1000
+
+/* The default search IDL stack cache depth */
+#define DEFAULT_SEARCH_STACK_DEPTH 16
+
+/* The minimum we can function with */
+#define MINIMUM_SEARCH_STACK_DEPTH 8
+
+typedef struct bdb_idl_cache_entry_s {
+ struct berval kstr;
+ ldap_pvt_thread_rdwr_t idl_entry_rwlock;
+ ID *idl;
+ DB *db;
+ struct bdb_idl_cache_entry_s* idl_lru_prev;
+ struct bdb_idl_cache_entry_s* idl_lru_next;
+} bdb_idl_cache_entry_t;
+
+/* BDB backend specific entry info */
+typedef struct bdb_entry_info {
+ struct bdb_entry_info *bei_parent;
+ ID bei_id;
+
+ /* we use the bei_id as a lockobj, but we need to make the size != 4
+ * to avoid conflicting with BDB's internal locks. So add a byte here
+ * that is always zero.
+ */
+ char bei_lockpad;
+
+ short bei_state;
+#define CACHE_ENTRY_DELETED 1
+#define CACHE_ENTRY_NO_KIDS 2
+#define CACHE_ENTRY_NOT_LINKED 4
+#define CACHE_ENTRY_NO_GRANDKIDS 8
+#define CACHE_ENTRY_LOADING 0x10
+
+ /*
+ * remaining fields require backend cache lock to access
+ */
+ struct berval bei_nrdn;
+#ifdef BDB_HIER
+ struct berval bei_rdn;
+ int bei_modrdns; /* track renames */
+ int bei_ckids; /* number of kids cached */
+ int bei_dkids; /* number of kids on-disk, plus 1 */
+#endif
+ Entry *bei_e;
+ Avlnode *bei_kids;
+#ifdef SLAP_ZONE_ALLOC
+ struct bdb_info *bei_bdb;
+ int bei_zseq;
#endif
+ ldap_pvt_thread_mutex_t bei_kids_mutex;
+
+ struct bdb_entry_info *bei_lrunext; /* for cache lru list */
+ struct bdb_entry_info *bei_lruprev;
+} EntryInfo;
+#undef BEI
+#define BEI(e) ((EntryInfo *) ((e)->e_private))
+/* for the in-core cache of entries */
+typedef struct bdb_cache {
+ int c_maxsize;
+ int c_cursize;
+ int c_eiused; /* EntryInfo's in use */
+ int c_leaves; /* EntryInfo leaf nodes */
+ EntryInfo c_dntree;
+ EntryInfo *c_eifree; /* free list */
+ Avlnode *c_idtree;
+ EntryInfo *c_lruhead; /* lru - add accessed entries here */
+ EntryInfo *c_lrutail; /* lru - rem lru entries from here */
+ ldap_pvt_thread_rdwr_t c_rwlock;
+ ldap_pvt_thread_mutex_t lru_mutex;
+ u_int32_t c_locker; /* used by lru cleaner */
+#ifdef SLAP_ZONE_ALLOC
+ void *c_zctx;
+#endif
+} Cache;
+
+#define CACHE_READ_LOCK 0
+#define CACHE_WRITE_LOCK 1
+
#define BDB_INDICES 128
struct bdb_db_info {
int bi_db_opflags; /* db-specific flags */
slap_mask_t bi_defaultmask;
+ Cache bi_cache;
Avlnode *bi_attrs;
-#ifdef BDB_HIER
- Avlnode *bi_tree;
- ldap_pvt_thread_rdwr_t bi_tree_rdwr;
- void *bi_troot;
- int bi_nrdns;
-#endif
+ void *bi_search_stack;
+ int bi_search_stack_depth;
+ int bi_linear_index;
- int bi_txn;
int bi_txn_cp;
u_int32_t bi_txn_cp_min;
u_int32_t bi_txn_cp_kbyte;
+ void *bi_txn_cp_task;
+ void *bi_index_task;
-#ifndef NO_THREADS
int bi_lock_detect;
- int bi_lock_detect_seconds;
- ldap_pvt_thread_t bi_lock_detect_tid;
-#endif
+ long bi_shm_key;
ID bi_lastid;
ldap_pvt_thread_mutex_t bi_lastid_mutex;
+ int bi_idl_cache_max_size;
+ int bi_idl_cache_size;
+ Avlnode *bi_idl_tree;
+ bdb_idl_cache_entry_t *bi_idl_lru_head;
+ bdb_idl_cache_entry_t *bi_idl_lru_tail;
+ ldap_pvt_thread_rdwr_t bi_idl_tree_rwlock;
+ ldap_pvt_thread_mutex_t bi_idl_tree_lrulock;
+ alock_info_t bi_alock_info;
+ char *bi_db_config_path;
+ BerVarray bi_db_config;
+ int bi_flags;
+#define BDB_IS_OPEN 0x01
+#define BDB_HAS_CONFIG 0x02
+#define BDB_UPD_CONFIG 0x04
+#define BDB_DEL_INDEX 0x08
+#define BDB_RE_OPEN 0x10
+#ifdef BDB_HIER
+ int bi_modrdns; /* number of modrdns completed */
+ ldap_pvt_thread_mutex_t bi_modrdns_mutex;
+#endif
};
#define bi_id2entry bi_databases[BDB_ID2ENTRY]
-#ifdef BDB_HIER
-#define bi_id2parent bi_databases[BDB_ID2PARENT]
-#else
#define bi_dn2id bi_databases[BDB_DN2ID]
-#endif
+
+struct bdb_lock_info {
+ struct bdb_lock_info *bli_next;
+ ID bli_id;
+ DB_LOCK bli_lock;
+};
struct bdb_op_info {
BackendDB* boi_bdb;
DB_TXN* boi_txn;
- int boi_err;
+ u_int32_t boi_err;
+ u_int32_t boi_locker;
+ int boi_acl_cache;
+ struct bdb_lock_info *boi_locks; /* used when no txn */
};
+#define DB_OPEN(db, file, name, type, flags, mode) \
+ (db)->open(db, file, name, type, flags, mode)
+
#if DB_VERSION_MAJOR < 4
-#define TXN_CHECKPOINT(env, k, m, f) txn_checkpoint(env, k, m, f)
-#define TXN_ID(txn) txn_id(txn)
-#define LOCK_DETECT(env, f, t, a) lock_detect(env, f, t, a)
-#define LOCK_GET(env, i, f, o, m, l) lock_get(env, i, f, o, m, l)
+#define LOCK_DETECT(env,f,t,a) lock_detect(env, f, t, a)
+#define LOCK_GET(env,i,f,o,m,l) lock_get(env, i, f, o, m, l)
+#define LOCK_PUT(env,l) lock_put(env, l)
+#define TXN_CHECKPOINT(env,k,m,f) txn_checkpoint(env, k, m, f)
+#define TXN_BEGIN(env,p,t,f) txn_begin((env), p, t, f)
+#define TXN_PREPARE(txn,gid) txn_prepare((txn), (gid))
+#define TXN_COMMIT(txn,f) txn_commit((txn), (f))
+#define TXN_ABORT(txn) txn_abort((txn))
+#define TXN_ID(txn) txn_id(txn)
+#define XLOCK_ID(env, locker) lock_id(env, locker)
+#define XLOCK_ID_FREE(env, locker) lock_id_free(env, locker)
#else
-#define TXN_CHECKPOINT(env, k, m, f) (env)->txn_checkpoint(env, k, m, f)
-#define TXN_ID(txn) (txn)->id(txn)
-#define LOCK_DETECT(env, f, t, a) (env)->lock_detect(env, f, t, a)
-#define LOCK_GET(env, i, f, o, m, l) (env)->lock_get(env, i, f, o, m, l)
+#define LOCK_DETECT(env,f,t,a) (env)->lock_detect(env, f, t, a)
+#define LOCK_GET(env,i,f,o,m,l) (env)->lock_get(env, i, f, o, m, l)
+#define LOCK_PUT(env,l) (env)->lock_put(env, l)
+#define TXN_CHECKPOINT(env,k,m,f) (env)->txn_checkpoint(env, k, m, f)
+#define TXN_BEGIN(env,p,t,f) (env)->txn_begin((env), p, t, f)
+#define TXN_PREPARE(txn,g) (txn)->prepare((txn), (g))
+#define TXN_COMMIT(txn,f) (txn)->commit((txn), (f))
+#define TXN_ABORT(txn) (txn)->abort((txn))
+#define TXN_ID(txn) (txn)->id(txn)
+#define XLOCK_ID(env, locker) (env)->lock_id(env, locker)
+#define XLOCK_ID_FREE(env, locker) (env)->lock_id_free(env, locker)
+
+/* BDB 4.1.17 adds txn arg to db->open */
+#if DB_VERSION_FULL >= 0x04010011
+#undef DB_OPEN
+#define DB_OPEN(db, file, name, type, flags, mode) \
+ (db)->open(db, NULL, file, name, type, flags, mode)
#endif
+#endif
+
+#define BDB_REUSE_LOCKERS
+
+#define BDB_CSN_COMMIT 0
+#define BDB_CSN_ABORT 1
+#define BDB_CSN_RETRY 2
+
+/* Copy an ID "src" to pointer "dst" in big-endian byte order */
+#define BDB_ID2DISK( src, dst ) \
+ do { int i0; ID tmp; unsigned char *_p; \
+ tmp = (src); _p = (unsigned char *)(dst); \
+ for ( i0=sizeof(ID)-1; i0>=0; i0-- ) { \
+ _p[i0] = tmp & 0xff; tmp >>= 8; \
+ } \
+ } while(0)
+
+/* Copy a pointer "src" to a pointer "dst" from big-endian to native order */
+#define BDB_DISK2ID( src, dst ) \
+ do { int i0; ID tmp = 0; unsigned char *_p; \
+ _p = (unsigned char *)(src); \
+ for ( i0=0; i0<sizeof(ID); i0++ ) { \
+ tmp <<= 8; tmp |= *_p++; \
+ } *(dst) = tmp; \
+ } while (0)
+
LDAP_END_DECL
+/* for the cache of attribute information (which are indexed, etc.) */
+typedef struct bdb_attrinfo {
+ AttributeDescription *ai_desc; /* attribute description cn;lang-en */
+ slap_mask_t ai_indexmask; /* how the attr is indexed */
+ slap_mask_t ai_newmask; /* new settings to replace old mask */
+#ifdef LDAP_COMP_MATCH
+ ComponentReference* ai_cr; /*component indexing*/
+#endif
+} AttrInfo;
+
+/* These flags must not clash with SLAP_INDEX flags or ops in slap.h! */
+#define BDB_INDEX_DELETING 0x8000U /* index is being modified */
+#define BDB_INDEX_UPDATE_OP 0x03 /* performing an index update */
+
#include "proto-bdb.h"
#endif /* _BACK_BDB_H_ */