Add MDB_NOTLS envflag.

vmware
Howard Chu 12 years ago committed by Hallvard Furuseth
parent afe488d8a9
commit 272e4e98ad
  1. 52
      libraries/liblmdb/lmdb.h
  2. 19
      libraries/liblmdb/mdb.c

@ -66,6 +66,7 @@
* *
* - A thread can only use one transaction at a time, plus any child * - A thread can only use one transaction at a time, plus any child
* transactions. Each transaction belongs to one thread. See below. * transactions. Each transaction belongs to one thread. See below.
* The #MDB_NOTLS flag changes this for read-only transactions.
* *
* - Use an MDB_env* in the process which opened it, without fork()ing. * - Use an MDB_env* in the process which opened it, without fork()ing.
* *
@ -249,6 +250,8 @@ typedef void (MDB_rel_func)(MDB_val *item, void *oldptr, void *newptr, void *rel
#define MDB_WRITEMAP 0x80000 #define MDB_WRITEMAP 0x80000
/** use asynchronous msync when MDB_WRITEMAP is used */ /** use asynchronous msync when MDB_WRITEMAP is used */
#define MDB_MAPASYNC 0x100000 #define MDB_MAPASYNC 0x100000
/** tie reader locktable slots to #MDB_txn objects instead of to threads */
#define MDB_NOTLS 0x200000
/** @} */ /** @} */
/** @defgroup mdb_dbi_open Database Flags /** @defgroup mdb_dbi_open Database Flags
@ -392,8 +395,8 @@ typedef struct MDB_envinfo {
size_t me_mapsize; /**< Size of the data memory map */ size_t me_mapsize; /**< Size of the data memory map */
size_t me_last_pgno; /**< ID of the last used page */ size_t me_last_pgno; /**< ID of the last used page */
size_t me_last_txnid; /**< ID of the last committed transaction */ size_t me_last_txnid; /**< ID of the last committed transaction */
unsigned int me_maxreaders; /**< maximum number of threads for the environment */ unsigned int me_maxreaders; /**< max reader slots in the environment */
unsigned int me_numreaders; /**< maximum number of threads used in the environment */ unsigned int me_numreaders; /**< max reader slots used in the environment */
} MDB_envinfo; } MDB_envinfo;
/** @brief Return the mdb library version information. /** @brief Return the mdb library version information.
@ -492,6 +495,15 @@ int mdb_env_create(MDB_env **env);
* database or lose the last transactions. Calling #mdb_env_sync() * database or lose the last transactions. Calling #mdb_env_sync()
* ensures on-disk database integrity until next commit. * ensures on-disk database integrity until next commit.
* This flag may be changed at any time using #mdb_env_set_flags(). * This flag may be changed at any time using #mdb_env_set_flags().
* <li>#MDB_NOTLS
* Don't use Thread-Local Storage. Tie reader locktable slots to
* #MDB_txn objects instead of to threads. I.e. #mdb_txn_reset() keeps
* the slot reseved for the #MDB_txn object. A thread may use parallel
* read-only transactions. A read-only transaction may span threads if
* the user synchronizes its use. Applications that multiplex many
* user threads over individual OS threads need this option. Such an
* application must also serialize the write transactions in an OS
* thread, since MDB's write locking is unaware of the user threads.
* </ul> * </ul>
* @param[in] mode The UNIX permissions to set on created files. This parameter * @param[in] mode The UNIX permissions to set on created files. This parameter
* is ignored on Windows. * is ignored on Windows.
@ -626,13 +638,17 @@ int mdb_env_get_path(MDB_env *env, const char **path);
*/ */
int mdb_env_set_mapsize(MDB_env *env, size_t size); int mdb_env_set_mapsize(MDB_env *env, size_t size);
/** @brief Set the maximum number of threads for the environment. /** @brief Set the maximum number of threads/reader slots for the environment.
* *
* This defines the number of slots in the lock table that is used to track readers in the * This defines the number of slots in the lock table that is used to track readers in the
* the environment. The default is 126. * the environment. The default is 126.
* Starting a read-only transaction normally ties a lock table slot to the
* current thread until the environment closes or the thread exits. If
* MDB_NOTLS is in use, #mdb_txn_begin() instead ties the slot to the
* MDB_txn object until it or the #MDB_env object is destroyed.
* This function may only be called after #mdb_env_create() and before #mdb_env_open(). * This function may only be called after #mdb_env_create() and before #mdb_env_open().
* @param[in] env An environment handle returned by #mdb_env_create() * @param[in] env An environment handle returned by #mdb_env_create()
* @param[in] readers The maximum number of threads * @param[in] readers The maximum number of reader lock table slots
* @return A non-zero error value on failure and 0 on success. Some possible * @return A non-zero error value on failure and 0 on success. Some possible
* errors are: * errors are:
* <ul> * <ul>
@ -641,7 +657,7 @@ int mdb_env_set_mapsize(MDB_env *env, size_t size);
*/ */
int mdb_env_set_maxreaders(MDB_env *env, unsigned int readers); int mdb_env_set_maxreaders(MDB_env *env, unsigned int readers);
/** @brief Get the maximum number of threads for the environment. /** @brief Get the maximum number of threads/reader slots for the environment.
* *
* @param[in] env An environment handle returned by #mdb_env_create() * @param[in] env An environment handle returned by #mdb_env_create()
* @param[out] readers Address of an integer to store the number of readers * @param[out] readers Address of an integer to store the number of readers
@ -672,8 +688,9 @@ int mdb_env_set_maxdbs(MDB_env *env, MDB_dbi dbs);
/** @brief Create a transaction for use with the environment. /** @brief Create a transaction for use with the environment.
* *
* The transaction handle may be discarded using #mdb_txn_abort() or #mdb_txn_commit(). * The transaction handle may be discarded using #mdb_txn_abort() or #mdb_txn_commit().
* @note Transactions may not span threads; a transaction must only be used by a * @note A transaction and its cursors must only be used by a single
* single thread. Also, a thread may only have a single transaction. * thread, and a thread may only have a single transaction at a time.
* If #MDB_NOTLS is in use, this does not apply to read-only transactions.
* @note Cursors may not span transactions; each cursor must be opened and closed * @note Cursors may not span transactions; each cursor must be opened and closed
* within a single transaction. * within a single transaction.
* @param[in] env An environment handle returned by #mdb_env_create() * @param[in] env An environment handle returned by #mdb_env_create()
@ -730,11 +747,13 @@ void mdb_txn_abort(MDB_txn *txn);
/** @brief Reset a read-only transaction. /** @brief Reset a read-only transaction.
* *
* This releases the current reader lock but doesn't free the * Abort the transaction like #mdb_txn_abort(), but keep the transaction
* transaction handle, allowing it to be used again later by #mdb_txn_renew(). * handle. #mdb_txn_renew() may reuse the handle. This saves allocation
* It otherwise has the same effect as #mdb_txn_abort() but saves some memory * overhead if the process will start a new read-only transaction soon,
* allocation/deallocation overhead if a thread is going to start a new * and also locking overhead if #MDB_NOTLS is in use. The reader table
* read-only transaction again soon. * lock is released, but the table slot stays tied to its thread or
* #MDB_txn. Use mdb_txn_abort() to discard a reset handle, and to free
* its lock table slot if MDB_NOTLS is in use.
* All cursors opened within the transaction must be closed before the transaction * All cursors opened within the transaction must be closed before the transaction
* is reset. * is reset.
* Reader locks generally don't interfere with writers, but they keep old * Reader locks generally don't interfere with writers, but they keep old
@ -1045,8 +1064,8 @@ int mdb_del(MDB_txn *txn, MDB_dbi dbi, MDB_val *key, MDB_val *data);
/** @brief Create a cursor handle. /** @brief Create a cursor handle.
* *
* Cursors are associated with a specific transaction and database and * A cursor is associated with a specific transaction and database.
* may not span threads. * It must be closed before its transaction ends.
* @param[in] txn A transaction handle returned by #mdb_txn_begin() * @param[in] txn A transaction handle returned by #mdb_txn_begin()
* @param[in] dbi A database handle returned by #mdb_dbi_open() * @param[in] dbi A database handle returned by #mdb_dbi_open()
* @param[out] cursor Address where the new #MDB_cursor handle will be stored * @param[out] cursor Address where the new #MDB_cursor handle will be stored
@ -1067,8 +1086,9 @@ void mdb_cursor_close(MDB_cursor *cursor);
/** @brief Renew a cursor handle. /** @brief Renew a cursor handle.
* *
* Cursors are associated with a specific transaction and database and * A cursor is associated with a specific transaction and database.
* may not span threads. Cursors that are only used in read-only * It must be closed before its transaction ends.
* Cursors that are only used in read-only
* transactions may be re-used, to avoid unnecessary malloc/free overhead. * transactions may be re-used, to avoid unnecessary malloc/free overhead.
* The cursor may be associated with a new read-only transaction, and * The cursor may be associated with a new read-only transaction, and
* referencing the same database handle as it was created with. * referencing the same database handle as it was created with.

@ -406,6 +406,8 @@ typedef uint16_t indx_t;
* slot's address is saved in thread-specific data so that subsequent read * slot's address is saved in thread-specific data so that subsequent read
* transactions started by the same thread need no further locking to proceed. * transactions started by the same thread need no further locking to proceed.
* *
* If #MDB_NOTLS is set, the slot address is not saved in thread-specific data.
*
* No reader table is used if the database is on a read-only filesystem. * No reader table is used if the database is on a read-only filesystem.
* *
* Since the database uses multi-version concurrency control, readers don't * Since the database uses multi-version concurrency control, readers don't
@ -1792,7 +1794,8 @@ mdb_txn_renew0(MDB_txn *txn)
txn->mt_txnid = env->me_metas[i]->mm_txnid; txn->mt_txnid = env->me_metas[i]->mm_txnid;
txn->mt_u.reader = NULL; txn->mt_u.reader = NULL;
} else { } else {
MDB_reader *r = pthread_getspecific(env->me_txkey); MDB_reader *r = (env->me_flags & MDB_NOTLS) ? txn->mt_u.reader :
pthread_getspecific(env->me_txkey);
if (r) { if (r) {
if (r->mr_pid != env->me_pid || r->mr_txnid != (txnid_t)-1) if (r->mr_pid != env->me_pid || r->mr_txnid != (txnid_t)-1)
return MDB_BAD_RSLOT; return MDB_BAD_RSLOT;
@ -1816,7 +1819,8 @@ mdb_txn_renew0(MDB_txn *txn)
env->me_numreaders = env->me_txns->mti_numreaders; env->me_numreaders = env->me_txns->mti_numreaders;
UNLOCK_MUTEX_R(env); UNLOCK_MUTEX_R(env);
r = &env->me_txns->mti_readers[i]; r = &env->me_txns->mti_readers[i];
if ((rc = pthread_setspecific(env->me_txkey, r)) != 0) { if (!(env->me_flags & MDB_NOTLS) &&
(rc = pthread_setspecific(env->me_txkey, r)) != 0) {
env->me_txns->mti_readers[i].mr_pid = 0; env->me_txns->mti_readers[i].mr_pid = 0;
return rc; return rc;
} }
@ -2006,7 +2010,8 @@ mdb_txn_reset0(MDB_txn *txn)
if (F_ISSET(txn->mt_flags, MDB_TXN_RDONLY)) { if (F_ISSET(txn->mt_flags, MDB_TXN_RDONLY)) {
if (txn->mt_u.reader) { if (txn->mt_u.reader) {
txn->mt_u.reader->mr_txnid = (txnid_t)-1; txn->mt_u.reader->mr_txnid = (txnid_t)-1;
txn->mt_u.reader = NULL; /* do not touch mr_txnid again */ if (!(env->me_flags & MDB_NOTLS))
txn->mt_u.reader = NULL; /* txn does not own reader */
} }
txn->mt_numdbs = 0; /* mark txn as reset, do not close DBs again */ txn->mt_numdbs = 0; /* mark txn as reset, do not close DBs again */
} else { } else {
@ -2091,6 +2096,10 @@ mdb_txn_abort(MDB_txn *txn)
mdb_txn_abort(txn->mt_child); mdb_txn_abort(txn->mt_child);
mdb_txn_reset0(txn); mdb_txn_reset0(txn);
/* Free reader slot tied to this txn (if MDB_NOTLS && writable FS) */
if ((txn->mt_flags & MDB_TXN_RDONLY) && txn->mt_u.reader)
txn->mt_u.reader->mr_pid = 0;
free(txn); free(txn);
} }
@ -3240,7 +3249,7 @@ mdb_env_setup_locks(MDB_env *env, char *lpath, int mode, int *excl)
fcntl(env->me_lfd, F_SETFD, fdflags); fcntl(env->me_lfd, F_SETFD, fdflags);
#endif #endif
{ if (!(env->me_flags & MDB_NOTLS)) {
rc = pthread_key_create(&env->me_txkey, mdb_env_reader_dest); rc = pthread_key_create(&env->me_txkey, mdb_env_reader_dest);
if (rc) if (rc)
goto fail; goto fail;
@ -3418,7 +3427,7 @@ fail:
* environment and re-opening it with the new flags. * environment and re-opening it with the new flags.
*/ */
#define CHANGEABLE (MDB_NOSYNC|MDB_NOMETASYNC|MDB_MAPASYNC) #define CHANGEABLE (MDB_NOSYNC|MDB_NOMETASYNC|MDB_MAPASYNC)
#define CHANGELESS (MDB_FIXEDMAP|MDB_NOSUBDIR|MDB_RDONLY|MDB_WRITEMAP) #define CHANGELESS (MDB_FIXEDMAP|MDB_NOSUBDIR|MDB_RDONLY|MDB_WRITEMAP|MDB_NOTLS)
int int
mdb_env_open(MDB_env *env, const char *path, unsigned int flags, mdb_mode_t mode) mdb_env_open(MDB_env *env, const char *path, unsigned int flags, mdb_mode_t mode)

Loading…
Cancel
Save