DBD-SQLcipher
view release on metacpan or search on metacpan
# $dbh->{AutoCommit} stays intact;
$dbh->{AutoCommit} = 1; # ends the transactional mode
This "AutoCommit" mode is independent from the autocommit mode of the
internal SQLcipher library, which always begins by a "BEGIN" statement, and
ends by a "COMMIT" or a <ROLLBACK>.
Transaction and Database Locking
Transaction by "AutoCommit" or "begin_work" is nice and handy, but
sometimes you may get an annoying "database is locked" error. This
typically happens when someone begins a transaction, and tries to write
to a database while other person is reading from the database (in
another transaction). You might be surprised but SQLcipher doesn't lock a
database when you just begin a normal (deferred) transaction to maximize
concurrency. It reserves a lock when you issue a statement to write, but
until you actually try to write with a "commit" statement, it allows
other people to read from the database. However, reading from the
database also requires "shared lock", and that prevents to give you the
"exclusive lock" you reserved, thus you get the "database is locked"
error, and other people will get the same error if they try to write
afterwards, as you still have a "pending" lock. "busy_timeout" doesn't
help in this case.
To avoid this, set a transaction type explicitly. You can issue a "begin
immediate transaction" (or "begin exclusive transaction") for each
transaction, or set "sqlite_use_immediate_transaction" database handle
attribute to true (since 1.30_02) to always use an immediate transaction
(even when you simply use "begin_work" or turn off the "AutoCommit".).
lib/DBD/SQLcipher.pm view on Meta::CPAN
See L<http://www.sqlite.org/c3ref/open.html> for details.
=head2 DBD::SQLcipher And File::Temp
When you use L<File::Temp> to create a temporary file/directory for
SQLcipher databases, you need to remember:
=over 4
=item tempfile may be locked exclusively
You may want to use C<tempfile()> to create a temporary database
filename for DBD::SQLcipher, but as noted in L<File::Temp>'s POD,
this file may have an exclusive lock under some operating systems
(notably Mac OSX), and result in a "database is locked" error.
To avoid this, set EXLOCK option to false when you call tempfile().
($fh, $filename) = tempfile($template, EXLOCK => 0);
=item CLEANUP may not work unless a database is disconnected
When you set CLEANUP option to true when you create a temporary
directory with C<tempdir()> or C<newdir()>, you may have to
disconnect databases explicitly before the temporary directory
is gone (notably under MS Windows).
** New error codes may be added in future versions of SQLite.
**
** See also: [extended result code definitions]
*/
#define SQLITE_OK 0 /* Successful result */
/* beginning-of-error-codes */
#define SQLITE_ERROR 1 /* SQL error or missing database */
#define SQLITE_INTERNAL 2 /* Internal logic error in SQLite */
#define SQLITE_PERM 3 /* Access permission denied */
#define SQLITE_ABORT 4 /* Callback routine requested an abort */
#define SQLITE_BUSY 5 /* The database file is locked */
#define SQLITE_LOCKED 6 /* A table in the database is locked */
#define SQLITE_NOMEM 7 /* A malloc() failed */
#define SQLITE_READONLY 8 /* Attempt to write a readonly database */
#define SQLITE_INTERRUPT 9 /* Operation terminated by sqlite3_interrupt()*/
#define SQLITE_IOERR 10 /* Some kind of disk I/O error occurred */
#define SQLITE_CORRUPT 11 /* The database disk image is malformed */
#define SQLITE_NOTFOUND 12 /* Unknown opcode in sqlite3_file_control() */
#define SQLITE_FULL 13 /* Insertion failed because database is full */
#define SQLITE_CANTOPEN 14 /* Unable to open the database file */
#define SQLITE_PROTOCOL 15 /* Database lock protocol error */
#define SQLITE_EMPTY 16 /* Database is empty */
** this file-control is NULL. However, if the database file is being synced
** as part of a multi-database commit, the argument points to a nul-terminated
** string containing the transactions master-journal file name. VFSes that
** do not need this signal should silently ignore this opcode. Applications
** should not call [sqlite3_file_control()] with this opcode as doing so may
** disrupt the operation of the specialized VFSes that do require it.
**
** <li>[[SQLITE_FCNTL_COMMIT_PHASETWO]]
** The [SQLITE_FCNTL_COMMIT_PHASETWO] opcode is generated internally by SQLite
** and sent to the VFS after a transaction has been committed immediately
** but before the database is unlocked. VFSes that do not need this signal
** should silently ignore this opcode. Applications should not call
** [sqlite3_file_control()] with this opcode as doing so may disrupt the
** operation of the specialized VFSes that do require it.
**
** <li>[[SQLITE_FCNTL_WIN32_AV_RETRY]]
** ^The [SQLITE_FCNTL_WIN32_AV_RETRY] opcode is used to configure automatic
** retry counts and intervals for certain disk I/O operations for the
** windows [VFS] in order to provide robustness in the presence of
** anti-virus programs. By default, the windows VFS will retry file read,
** file write, and file delete operations up to 10 times, with a delay
** <ul>
** <li> SQLITE_SHM_LOCK | SQLITE_SHM_SHARED
** <li> SQLITE_SHM_LOCK | SQLITE_SHM_EXCLUSIVE
** <li> SQLITE_SHM_UNLOCK | SQLITE_SHM_SHARED
** <li> SQLITE_SHM_UNLOCK | SQLITE_SHM_EXCLUSIVE
** </ul>
**
** When unlocking, the same SHARED or EXCLUSIVE flag must be supplied as
** was given on the corresponding lock.
**
** The xShmLock method can transition between unlocked and SHARED or
** between unlocked and EXCLUSIVE. It cannot transition between SHARED
** and EXCLUSIVE.
*/
#define SQLITE_SHM_UNLOCK 1
#define SQLITE_SHM_LOCK 2
#define SQLITE_SHM_SHARED 4
#define SQLITE_SHM_EXCLUSIVE 8
/*
** CAPI3REF: Maximum xShmLock index
**
SQLITE_API int sqlite3_complete16(const void *sql);
/*
** CAPI3REF: Register A Callback To Handle SQLITE_BUSY Errors
** KEYWORDS: {busy-handler callback} {busy handler}
**
** ^The sqlite3_busy_handler(D,X,P) routine sets a callback function X
** that might be invoked with argument P whenever
** an attempt is made to access a database table associated with
** [database connection] D when another thread
** or process has the table locked.
** The sqlite3_busy_handler() interface is used to implement
** [sqlite3_busy_timeout()] and [PRAGMA busy_timeout].
**
** ^If the busy callback is NULL, then [SQLITE_BUSY]
** is returned immediately upon encountering the lock. ^If the busy callback
** is not NULL, then the callback might be invoked with two arguments.
**
** ^The first argument to the busy handler is a copy of the void* pointer which
** is the third argument to sqlite3_busy_handler(). ^The second argument to
** the busy handler callback is the number of times that the busy handler has
**
** The presence of a busy handler does not guarantee that it will be invoked
** when there is lock contention. ^If SQLite determines that invoking the busy
** handler could result in a deadlock, it will go ahead and return [SQLITE_BUSY]
** to the application instead of invoking the
** busy handler.
** Consider a scenario where one process is holding a read lock that
** it is trying to promote to a reserved lock and
** a second process is holding a reserved lock that it is trying
** to promote to an exclusive lock. The first process cannot proceed
** because it is blocked by the second and the second process cannot
** proceed because it is blocked by the first. If both processes
** invoke the busy handlers, neither will make any progress. Therefore,
** SQLite returns [SQLITE_BUSY] for the first process, hoping that this
** will induce the first process to release its read lock and allow
** the second process to proceed.
**
** ^The default busy callback is NULL.
**
** ^(There can only be a single busy handler defined for each
** [database connection]. Setting a new busy handler clears any
** previously set handler.)^ ^Note that calling [sqlite3_busy_timeout()]
**
** A busy handler must not close the database connection
** or [prepared statement] that invoked the busy handler.
*/
SQLITE_API int sqlite3_busy_handler(sqlite3*, int(*)(void*,int), void*);
/*
** CAPI3REF: Set A Busy Timeout
**
** ^This routine sets a [sqlite3_busy_handler | busy handler] that sleeps
** for a specified amount of time when a table is locked. ^The handler
** will sleep multiple times until at least "ms" milliseconds of sleeping
** have accumulated. ^After at least "ms" milliseconds of sleeping,
** the handler returns 0 which causes [sqlite3_step()] to return
** [SQLITE_BUSY].
**
** ^Calling this routine with an argument less than or equal to zero
** turns off all busy handlers.
**
** ^(There can only be a single busy handler for a particular
** [database connection] at any given moment. If another busy handler
** CAPI3REF: Online Backup API.
**
** The backup API copies the content of one database into another.
** It is useful either for creating backups of databases or
** for copying in-memory databases to or from persistent files.
**
** See Also: [Using the SQLite Online Backup API]
**
** ^SQLite holds a write transaction open on the destination database file
** for the duration of the backup operation.
** ^The source database is read-locked only while it is being read;
** it is not locked continuously for the entire backup operation.
** ^Thus, the backup may be performed on a live source database without
** preventing other database connections from
** reading or writing to the source database while the backup is underway.
**
** ^(To perform a backup operation:
** <ol>
** <li><b>sqlite3_backup_init()</b> is called once to initialize the
** backup,
** <li><b>sqlite3_backup_step()</b> is called one or more times to transfer
** the data between the two databases, and finally
** errors are considered fatal.)^ The application must accept
** that the backup operation has failed and pass the backup operation handle
** to the sqlite3_backup_finish() to release associated resources.
**
** ^The first call to sqlite3_backup_step() obtains an exclusive lock
** on the destination file. ^The exclusive lock is not released until either
** sqlite3_backup_finish() is called or the backup operation is complete
** and sqlite3_backup_step() returns [SQLITE_DONE]. ^Every call to
** sqlite3_backup_step() obtains a [shared lock] on the source database that
** lasts for the duration of the sqlite3_backup_step() call.
** ^Because the source database is not locked between calls to
** sqlite3_backup_step(), the source database may be modified mid-way
** through the backup process. ^If the source database is modified by an
** external process or via a database connection other than the one being
** used by the backup operation, then the backup will be automatically
** restarted by the next call to sqlite3_backup_step(). ^If the source
** database is modified by the using the same database connection as is used
** by the backup operation, then the backup database is automatically
** updated at the same time.
**
** [[sqlite3_backup_finish()]] <b>sqlite3_backup_finish()</b>
** ^This API may be used to register a callback that SQLite will invoke
** when the connection currently holding the required lock relinquishes it.
** ^This API is only available if the library was compiled with the
** [SQLITE_ENABLE_UNLOCK_NOTIFY] C-preprocessor symbol defined.
**
** See Also: [Using the SQLite Unlock Notification Feature].
**
** ^Shared-cache locks are released when a database connection concludes
** its current transaction, either by committing it or rolling it back.
**
** ^When a connection (known as the blocked connection) fails to obtain a
** shared-cache lock and SQLITE_LOCKED is returned to the caller, the
** identity of the database connection (the blocking connection) that
** has locked the required resource is stored internally. ^After an
** application receives an SQLITE_LOCKED error, it may call the
** sqlite3_unlock_notify() method with the blocked connection handle as
** the first argument to register for a callback that will be invoked
** when the blocking connections current transaction is concluded. ^The
** callback is invoked from within the [sqlite3_step] or [sqlite3_close]
** call that concludes the blocking connections transaction.
**
** ^(If sqlite3_unlock_notify() is called in a multi-threaded application,
** there is a chance that the blocking connection will have already
** concluded its transaction by the time sqlite3_unlock_notify() is invoked.
** If this happens, then the specified callback is invoked immediately,
** from within the call to sqlite3_unlock_notify().)^
**
** ^If the blocked connection is attempting to obtain a write-lock on a
** shared-cache table, and more than one other connection currently holds
** a read-lock on the same table, then SQLite arbitrarily selects one of
** the other connections to use as the blocking connection.
**
** ^(There may be at most one unlock-notify callback registered by a
** blocked connection. If sqlite3_unlock_notify() is called when the
** blocked connection already has a registered unlock-notify callback,
** then the new callback replaces the old.)^ ^If sqlite3_unlock_notify() is
** called with a NULL pointer as its second argument, then any existing
** unlock-notify callback is canceled. ^The blocked connections
** unlock-notify callback may also be canceled by closing the blocked
** connection using [sqlite3_close()].
**
** The unlock-notify callback is not reentrant. If an application invokes
** any sqlite3_xxx API functions from within an unlock-notify callback, a
** crash or deadlock may be the result.
**
** ^Unless deadlock is detected (see below), sqlite3_unlock_notify() always
** returns SQLITE_OK.
**
** <b>Callback Invocation Details</b>
**
** When an unlock-notify callback is registered, the application provides a
** single void* pointer that is passed to the callback when it is invoked.
** However, the signature of the callback function allows SQLite to pass
** it an array of void* context pointers. The first argument passed to
** an unlock-notify callback is a pointer to an array of void* pointers,
** and the second is the number of entries in the array.
**
** When a blocking connections transaction is concluded, there may be
** more than one blocked connection that has registered for an unlock-notify
** callback. ^If two or more such blocked connections have specified the
** same callback function, then instead of invoking the callback function
** multiple times, it is invoked once with the set of void* context pointers
** specified by the blocked connections bundled together into an array.
** This gives the application an opportunity to prioritize any actions
** related to the set of unblocked database connections.
**
** <b>Deadlock Detection</b>
**
** Assuming that after registering for an unlock-notify callback a
** database waits for the callback to be issued before taking any further
** action (a reasonable assumption), then using this API may cause the
** application to deadlock. For example, if connection X is waiting for
** connection Y's transaction to be concluded, and similarly connection
** Y is waiting on connection X's transaction, then neither connection
** will proceed and the system may remain deadlocked indefinitely.
**
** To avoid this scenario, the sqlite3_unlock_notify() performs deadlock
** detection. ^If a given call to sqlite3_unlock_notify() would put the
** system in a deadlocked state, then SQLITE_LOCKED is returned and no
** unlock-notify callback is registered. The system is said to be in
** a deadlocked state if connection A has registered for an unlock-notify
** callback on the conclusion of connection B's transaction, and connection
** B has itself registered for an unlock-notify callback when connection
** A's transaction is concluded. ^Indirect deadlock is also detected, so
** the system is also considered to be deadlocked if connection B has
** registered for an unlock-notify callback on the conclusion of connection
** C's transaction, where connection C is waiting on connection A. ^Any
** number of levels of indirection are allowed.
**
** <b>The "DROP TABLE" Exception</b>
**
** When a call to [sqlite3_step()] returns SQLITE_LOCKED, it is almost
** always appropriate to call sqlite3_unlock_notify(). There is however,
** one exception. When executing a "DROP TABLE" or "DROP INDEX" statement,
** SQLite checks if there are any currently executing SELECT statements
** invoked immediately. If the application then re-attempts the "DROP TABLE"
** or "DROP INDEX" query, an infinite loop might be the result.
**
** One way around this problem is to check the extended error code returned
** by an sqlite3_step() call. ^(If there is a blocking connection, then the
** extended error code is set to SQLITE_LOCKED_SHAREDCACHE. Otherwise, in
** the special "DROP TABLE/INDEX" case, the extended error code is just
** SQLITE_LOCKED.)^
*/
SQLITE_API int sqlite3_unlock_notify(
sqlite3 *pBlocked, /* Waiting connection */
void (*xNotify)(void **apArg, int nArg), /* Callback function to invoke */
void *pNotifyArg /* Argument to pass to xNotify */
);
/*
** CAPI3REF: String Comparison
**
** ^The [sqlite3_stricmp()] and [sqlite3_strnicmp()] APIs allow applications
** and extensions to compare the contents of two buffers containing UTF-8
** unlock so that it can proceed.
**
** When X.pBlockingConnection==Y, that means that something that X tried
** tried to do recently failed with an SQLITE_LOCKED error due to locks
** held by Y.
*/
sqlite3 *pBlockingConnection; /* Connection that caused SQLITE_LOCKED */
sqlite3 *pUnlockConnection; /* Connection to watch for unlock */
void *pUnlockArg; /* Argument to xUnlockNotify */
void (*xUnlockNotify)(void **, int); /* Unlock notify callback */
sqlite3 *pNextBlocked; /* Next in list of all blocked connections */
#endif
#ifdef SQLITE_USER_AUTHENTICATION
sqlite3_userauth auth; /* User authentication information */
#endif
};
/*
** A macro to discover the encoding of a database.
*/
#define SCHEMA_ENC(db) ((db)->aDb[0].pSchema->enc)
#else
#define sqlite3ExprSetHeight(x,y)
#define sqlite3SelectExprHeight(x) 0
#define sqlite3ExprCheckHeight(x,y)
#endif
SQLITE_PRIVATE u32 sqlite3Get4byte(const u8*);
SQLITE_PRIVATE void sqlite3Put4byte(u8*, u32);
#ifdef SQLITE_ENABLE_UNLOCK_NOTIFY
SQLITE_PRIVATE void sqlite3ConnectionBlocked(sqlite3 *, sqlite3 *);
SQLITE_PRIVATE void sqlite3ConnectionUnlocked(sqlite3 *db);
SQLITE_PRIVATE void sqlite3ConnectionClosed(sqlite3 *db);
#else
#define sqlite3ConnectionBlocked(x,y)
#define sqlite3ConnectionUnlocked(x)
#define sqlite3ConnectionClosed(x)
#endif
#ifdef SQLITE_DEBUG
SQLITE_PRIVATE void sqlite3ParserTrace(FILE*, char *);
#endif
/*
** If the SQLITE_ENABLE IOTRACE exists then the global variable
** sqlite3IoTrace is a pointer to a printf-like routine used to
** The pBt pointer itself may not be changed while there exists cursors
** in the referenced BtShared that point back to this Btree since those
** cursors have to go through this Btree to find their BtShared and
** they often do so without holding sqlite3.mutex.
*/
struct Btree {
sqlite3 *db; /* The database connection holding this btree */
BtShared *pBt; /* Sharable content of this btree */
u8 inTrans; /* TRANS_NONE, TRANS_READ or TRANS_WRITE */
u8 sharable; /* True if we can share pBt with another db */
u8 locked; /* True if db currently has pBt locked */
int wantToLock; /* Number of nested calls to sqlite3BtreeEnter() */
int nBackup; /* Number of backup operations reading this btree */
u32 iDataVersion; /* Combines with pBt->pPager->iDataVersion */
Btree *pNext; /* List of other sharable Btrees from the same db */
Btree *pPrev; /* Back pointer of the same list */
#ifndef SQLITE_OMIT_SHARED_CACHE
BtLock lock; /* Object used to lock page 1 */
#endif
};
sqlite3_vfs *pVfs,
const char *zPath,
sqlite3_file *pFile,
int flags,
int *pFlagsOut
){
int rc;
DO_OS_MALLOC_TEST(0);
/* 0x87f7f is a mask of SQLITE_OPEN_ flags that are valid to be passed
** down into the VFS layer. Some SQLITE_OPEN_ flags (for example,
** SQLITE_OPEN_FULLMUTEX or SQLITE_OPEN_SHAREDCACHE) are blocked before
** reaching the VFS. */
rc = pVfs->xOpen(pVfs, zPath, pFile, flags & 0x87f7f, pFlagsOut);
assert( rc==SQLITE_OK || pFile->pMethods==0 );
return rc;
}
SQLITE_PRIVATE int sqlite3OsDelete(sqlite3_vfs *pVfs, const char *zPath, int dirSync){
DO_OS_MALLOC_TEST(0);
assert( dirSync==0 || dirSync==1 );
return pVfs->xDelete(pVfs, zPath, dirSync);
}
SQLITE3_MUTEX_INITIALIZER,
SQLITE3_MUTEX_INITIALIZER,
SQLITE3_MUTEX_INITIALIZER
};
static int winMutex_isInit = 0;
static int winMutex_isNt = -1; /* <0 means "need to query" */
/* As the winMutexInit() and winMutexEnd() functions are called as part
** of the sqlite3_initialize() and sqlite3_shutdown() processing, the
** "interlocked" magic used here is probably not strictly necessary.
*/
static LONG SQLITE_WIN32_VOLATILE winMutex_lock = 0;
SQLITE_API int sqlite3_win32_is_nt(void); /* os_win.c */
SQLITE_API void sqlite3_win32_sleep(DWORD milliseconds); /* os_win.c */
static int winMutexInit(void){
/* The first to increment to 1 does actual initialization */
if( InterlockedCompareExchange(&winMutex_lock, 1, 0)==0 ){
int i;
for(i=0; i<ArraySize(winMutex_staticMutexes); i++){
#if SQLITE_OS_WINRT
InitializeCriticalSectionEx(&winMutex_staticMutexes[i].mutex, 0, 0);
#else
InitializeCriticalSection(&winMutex_staticMutexes[i].mutex);
#endif
}
winMutex_isInit = 1;
}else{
while( !winMutex_isInit ){
sqlite3_win32_sleep(1);
}
}
return SQLITE_OK;
}
static int winMutexEnd(void){
/* The first to decrement to 0 does actual shutdown
** (which should be the last to shutdown.) */
if( InterlockedCompareExchange(&winMutex_lock, 0, 1)==1 ){
if( winMutex_isInit==1 ){
int i;
for(i=0; i<ArraySize(winMutex_staticMutexes); i++){
DeleteCriticalSection(&winMutex_staticMutexes[i].mutex);
}
winMutex_isInit = 0;
}
}
return SQLITE_OK;
}
** canonical filename and implemented in the previous division.)
**
** The sqlite3_file structure for POSIX is no longer just an integer file
** descriptor. It is now a structure that holds the integer file
** descriptor and a pointer to a structure that describes the internal
** locks on the corresponding inode. There is one locking structure
** per inode, so if the same inode is opened twice, both unixFile structures
** point to the same locking structure. The locking structure keeps
** a reference count (so we will know when to delete it) and a "cnt"
** field that tells us its internal lock status. cnt==0 means the
** file is unlocked. cnt==-1 means the file has an exclusive lock.
** cnt>0 means there are cnt shared locks on the file.
**
** Any attempt to lock or unlock a file first checks the locking
** structure. The fcntl() system call is only invoked to set a
** POSIX lock if the internal lock structure transitions between
** a locked and an unlocked state.
**
** But wait: there are yet more problems with POSIX advisory locks.
**
** If you close a file descriptor that points to a file that has locks,
** all locks on that file that are owned by the current process are
** released. To work around this problem, each unixInodeInfo object
** maintains a count of the number of pending locks on tha inode.
** When an attempt is made to close an unixFile, if there are
** other unixFile open on the same inode that are holding locks, the call
** to close() the file descriptor is deferred until all of the locks clear.
** confusion with SQLite lock names). The algorithms are complicated
** slightly in order to be compatible with windows systems simultaneously
** accessing the same database file, in case that is ever required.
**
** Symbols defined in os.h indentify the 'pending byte' and the 'reserved
** byte', each single bytes at well known offsets, and the 'shared byte
** range', a range of 510 bytes at a well known offset.
**
** To obtain a SHARED lock, a read-lock is obtained on the 'pending
** byte'. If this is successful, a random byte from the 'shared byte
** range' is read-locked and the lock on the 'pending byte' released.
**
** A process may only obtain a RESERVED lock after it has a SHARED lock.
** A RESERVED lock is implemented by grabbing a write-lock on the
** 'reserved byte'.
**
** A process may only obtain a PENDING lock after it has obtained a
** SHARED lock. A PENDING lock is implemented by obtaining a write-lock
** on the 'pending byte'. This ensures that no new SHARED locks can be
** obtained, but existing SHARED locks are allowed to persist. A process
** does not have to obtain a RESERVED lock on the way to a PENDING lock.
** unixFile, do nothing. Don't use the end_lock: exit path, as
** unixEnterMutex() hasn't been called yet.
*/
if( pFile->eFileLock>=eFileLock ){
OSTRACE(("LOCK %d %s ok (already held) (unix)\n", pFile->h,
azFileLock(eFileLock)));
return SQLITE_OK;
}
/* Make sure the locking sequence is correct.
** (1) We never move from unlocked to anything higher than shared lock.
** (2) SQLite never explicitly requests a pendig lock.
** (3) A shared lock is always held when a reserve lock is requested.
*/
assert( pFile->eFileLock!=NO_LOCK || eFileLock==SHARED_LOCK );
assert( eFileLock!=PENDING_LOCK );
assert( eFileLock!=RESERVED_LOCK || pFile->eFileLock==SHARED_LOCK );
/* This mutex is needed because pFile->pInode is shared across threads
*/
unixEnterMutex();
}
/*
** Lower the locking level on file descriptor pFile to eFileLock. eFileLock
** must be either NO_LOCK or SHARED_LOCK.
**
** If the locking level of the file descriptor is already at or below
** the requested locking level, this routine is a no-op.
**
** If handleNFSUnlock is true, then on downgrading an EXCLUSIVE_LOCK to SHARED
** the byte range is divided into 2 parts and the first part is unlocked then
** set to a read lock, then the other part is simply unlocked. This works
** around a bug in BSD NFS lockd (also seen on MacOSX 10.3+) that fails to
** remove the write lock on a region when a read lock is set.
*/
static int posixUnlock(sqlite3_file *id, int eFileLock, int handleNFSUnlock){
unixFile *pFile = (unixFile*)id;
unixInodeInfo *pInode;
struct flock lock;
int rc = SQLITE_OK;
assert( pFile );
#define DOTLOCK_SUFFIX ".lock"
/*
** This routine checks if there is a RESERVED lock held on the specified
** file by this or any other process. If such a lock is held, set *pResOut
** to a non-zero value otherwise *pResOut is set to zero. The return value
** is set to SQLITE_OK unless an I/O error occurs during lock checking.
**
** In dotfile locking, either a lock exists or it does not. So in this
** variation of CheckReservedLock(), *pResOut is set to true if any lock
** is held on the file and false if the file is unlocked.
*/
static int dotlockCheckReservedLock(sqlite3_file *id, int *pResOut) {
int rc = SQLITE_OK;
int reserved = 0;
unixFile *pFile = (unixFile*)id;
SimulateIOError( return SQLITE_IOERR_CHECKRESERVEDLOCK; );
assert( pFile );
int rc = SQLITE_OK;
/* if we already have a lock, it is exclusive.
** Just adjust level and punt on outta here. */
if (pFile->eFileLock > NO_LOCK) {
pFile->eFileLock = eFileLock;
rc = SQLITE_OK;
goto sem_end_lock;
}
/* lock semaphore now but bail out when already locked. */
if( sem_trywait(pSem)==-1 ){
rc = SQLITE_BUSY;
goto sem_end_lock;
}
/* got it, set the type and return ok */
pFile->eFileLock = eFileLock;
sem_end_lock:
return rc;
typedef struct afpLockingContext afpLockingContext;
struct afpLockingContext {
int reserved;
const char *dbPath; /* Name of the open file */
};
struct ByteRangeLockPB2
{
unsigned long long offset; /* offset to first byte to lock */
unsigned long long length; /* nbr of bytes to lock */
unsigned long long retRangeStart; /* nbr of 1st byte locked if successful */
unsigned char unLockFlag; /* 1 = unlock, 0 = lock */
unsigned char startEndFlag; /* 1=rel to end of fork, 0=rel to start */
int fd; /* file desc to assoc this lock with */
};
#define afpfsByteRangeLock2FSCTL _IOWR('z', 23, struct ByteRangeLockPB2)
/*
** This is a utility for setting or clearing a bit-range lock on an
** AFP filesystem.
**
** Return SQLITE_OK on success, SQLITE_BUSY on failure.
*/
static int afpSetLock(
const char *path, /* Name of the file to be locked or unlocked */
unixFile *pFile, /* Open file descriptor on path */
unsigned long long offset, /* First byte to be locked */
unsigned long long length, /* Number of bytes to lock */
int setLockFlag /* True to set lock. False to clear lock */
){
struct ByteRangeLockPB2 pb;
int err;
pb.unLockFlag = setLockFlag ? 0 : 1;
pb.startEndFlag = 0;
pb.offset = offset;
pb.length = length;
** unixFile, do nothing. Don't use the afp_end_lock: exit path, as
** unixEnterMutex() hasn't been called yet.
*/
if( pFile->eFileLock>=eFileLock ){
OSTRACE(("LOCK %d %s ok (already held) (afp)\n", pFile->h,
azFileLock(eFileLock)));
return SQLITE_OK;
}
/* Make sure the locking sequence is correct
** (1) We never move from unlocked to anything higher than shared lock.
** (2) SQLite never explicitly requests a pendig lock.
** (3) A shared lock is always held when a reserve lock is requested.
*/
assert( pFile->eFileLock!=NO_LOCK || eFileLock==SHARED_LOCK );
assert( eFileLock!=PENDING_LOCK );
assert( eFileLock!=RESERVED_LOCK || pFile->eFileLock==SHARED_LOCK );
/* This mutex is needed because pFile->pInode is shared across threads
*/
unixEnterMutex();
}
if( pShmNode->isReadonly && rc==SQLITE_OK ) rc = SQLITE_READONLY;
sqlite3_mutex_leave(pShmNode->mutex);
return rc;
}
/*
** Change the lock state for a shared-memory segment.
**
** Note that the relationship between SHAREd and EXCLUSIVE locks is a little
** different here than in posix. In xShmLock(), one can go from unlocked
** to shared and back or from unlocked to exclusive and back. But one may
** not go from shared to exclusive or from exclusive to shared.
*/
static int unixShmLock(
sqlite3_file *fd, /* Database file holding the shared memory */
int ofst, /* First lock to acquire or release */
int n, /* Number of locks to acquire or release */
int flags /* What to do with the lock */
){
unixFile *pDbFd = (unixFile*)fd; /* Connection holding shared memory */
unixShm *p = pDbFd->pShm; /* The shared memory being locked */
unixShm *pX; /* For looping over all siblings */
unixShmNode *pShmNode = p->pShmNode; /* The underlying file iNode */
int rc = SQLITE_OK; /* Result code */
u16 mask; /* Mask of locks to take or release */
assert( pShmNode==pDbFd->pInode->pShmNode );
assert( pShmNode->pInode==pDbFd->pInode );
assert( ofst>=0 && ofst+n<=SQLITE_SHM_NLOCK );
assert( n>=1 );
assert( flags==(SQLITE_SHM_LOCK | SQLITE_SHM_SHARED)
** lock. If any do, return SQLITE_BUSY right away.
*/
for(pX=pShmNode->pFirst; pX; pX=pX->pNext){
if( (pX->exclMask & mask)!=0 || (pX->sharedMask & mask)!=0 ){
rc = SQLITE_BUSY;
break;
}
}
/* Get the exclusive locks at the system level. Then if successful
** also mark the local connection as being locked.
*/
if( rc==SQLITE_OK ){
rc = unixShmSystemLock(pShmNode, F_WRLCK, ofst+UNIX_SHM_BASE, n);
if( rc==SQLITE_OK ){
assert( (p->sharedMask & mask)==0 );
p->exclMask |= mask;
}
}
}
sqlite3_mutex_leave(pShmNode->mutex);
{ "smbfs", &afpIoMethods },
{ "webdav", &nolockIoMethods },
{ 0, 0 }
};
int i;
struct statfs fsInfo;
struct flock lockInfo;
if( !filePath ){
/* If filePath==NULL that means we are dealing with a transient file
** that does not need to be locked. */
return &nolockIoMethods;
}
if( statfs(filePath, &fsInfo) != -1 ){
if( fsInfo.f_flags & MNT_RDONLY ){
return &nolockIoMethods;
}
for(i=0; aMap[i].zFilesystem; i++){
if( strcmp(fsInfo.f_fstypename, aMap[i].zFilesystem)==0 ){
return aMap[i].pMethods;
}
** This is for VXWorks only.
*/
static const sqlite3_io_methods *autolockIoFinderImpl(
const char *filePath, /* name of the database file */
unixFile *pNew /* the open file object */
){
struct flock lockInfo;
if( !filePath ){
/* If filePath==NULL that means we are dealing with a transient file
** that does not need to be locked. */
return &nolockIoMethods;
}
/* Test if fcntl() is supported and use POSIX style locks.
** Otherwise fall back to the named semaphore method.
*/
lockInfo.l_len = 1;
lockInfo.l_start = 0;
lockInfo.l_whence = SEEK_SET;
lockInfo.l_type = F_RDLCK;
#if SQLITE_OS_WINRT && (!defined(SQLITE_OMIT_WAL) || SQLITE_MAX_MMAP_SIZE>0)
{ "CreateFileMappingFromApp", (SYSCALL)CreateFileMappingFromApp, 0 },
#else
{ "CreateFileMappingFromApp", (SYSCALL)0, 0 },
#endif
#define osCreateFileMappingFromApp ((HANDLE(WINAPI*)(HANDLE, \
LPSECURITY_ATTRIBUTES,ULONG,ULONG64,LPCWSTR))aSyscall[75].pCurrent)
/*
** NOTE: On some sub-platforms, the InterlockedCompareExchange "function"
** is really just a macro that uses a compiler intrinsic (e.g. x64).
** So do not try to make this is into a redefinable interface.
*/
#if defined(InterlockedCompareExchange)
{ "InterlockedCompareExchange", (SYSCALL)0, 0 },
#define osInterlockedCompareExchange InterlockedCompareExchange
#else
{ "InterlockedCompareExchange", (SYSCALL)InterlockedCompareExchange, 0 },
#define osInterlockedCompareExchange ((LONG(WINAPI*)(LONG \
SQLITE_WIN32_VOLATILE*, LONG,LONG))aSyscall[76].pCurrent)
#endif /* defined(InterlockedCompareExchange) */
}; /* End of the overrideable system calls */
/*
** This is the xSetSystemCall() method of sqlite3_vfs for all of the
** "win32" VFSes. Return SQLITE_OK opon successfully updating the
** system call pointer, or SQLITE_NOTFOUND if there is no configurable
** system call named zName.
*/
static int winSetSystemCall(
** based on the NT kernel.
*/
SQLITE_API int sqlite3_win32_is_nt(void){
#if SQLITE_OS_WINRT
/*
** NOTE: The WinRT sub-platform is always assumed to be based on the NT
** kernel.
*/
return 1;
#elif defined(SQLITE_WIN32_GETVERSIONEX) && SQLITE_WIN32_GETVERSIONEX
if( osInterlockedCompareExchange(&sqlite3_os_type, 0, 0)==0 ){
#if defined(SQLITE_WIN32_HAS_ANSI)
OSVERSIONINFOA sInfo;
sInfo.dwOSVersionInfoSize = sizeof(sInfo);
osGetVersionExA(&sInfo);
osInterlockedCompareExchange(&sqlite3_os_type,
(sInfo.dwPlatformId == VER_PLATFORM_WIN32_NT) ? 2 : 1, 0);
#elif defined(SQLITE_WIN32_HAS_WIDE)
OSVERSIONINFOW sInfo;
sInfo.dwOSVersionInfoSize = sizeof(sInfo);
osGetVersionExW(&sInfo);
osInterlockedCompareExchange(&sqlite3_os_type,
(sInfo.dwPlatformId == VER_PLATFORM_WIN32_NT) ? 2 : 1, 0);
#endif
}
return osInterlockedCompareExchange(&sqlite3_os_type, 2, 2)==2;
#elif SQLITE_TEST
return osInterlockedCompareExchange(&sqlite3_os_type, 2, 2)==2;
#else
/*
** NOTE: All sub-platforms where the GetVersionEx[AW] functions are
** deprecated are always assumed to be based on the NT kernel.
*/
return 1;
#endif
}
#ifdef SQLITE_WIN32_MALLOC
/*
** Apply advisory locks for all n bytes beginning at ofst.
*/
#define _SHM_UNLCK 1
#define _SHM_RDLCK 2
#define _SHM_WRLCK 3
static int winShmSystemLock(
winShmNode *pFile, /* Apply locks to this open shared-memory segment */
int lockType, /* _SHM_UNLCK, _SHM_RDLCK, or _SHM_WRLCK */
int ofst, /* Offset to first byte to be locked/unlocked */
int nByte /* Number of bytes to lock or unlock */
){
int rc = 0; /* Result code form Lock/UnlockFileEx() */
/* Access to the winShmNode object is serialized by the caller */
assert( sqlite3_mutex_held(pFile->mutex) || pFile->nRef==0 );
OSTRACE(("SHM-LOCK file=%p, lock=%d, offset=%d, size=%d\n",
pFile->hFile.h, lockType, ofst, nByte));
/*
** Change the lock state for a shared-memory segment.
*/
static int winShmLock(
sqlite3_file *fd, /* Database file holding the shared memory */
int ofst, /* First lock to acquire or release */
int n, /* Number of locks to acquire or release */
int flags /* What to do with the lock */
){
winFile *pDbFd = (winFile*)fd; /* Connection holding shared memory */
winShm *p = pDbFd->pShm; /* The shared memory being locked */
winShm *pX; /* For looping over all siblings */
winShmNode *pShmNode = p->pShmNode;
int rc = SQLITE_OK; /* Result code */
u16 mask; /* Mask of locks to take or release */
assert( ofst>=0 && ofst+n<=SQLITE_SHM_NLOCK );
assert( n>=1 );
assert( flags==(SQLITE_SHM_LOCK | SQLITE_SHM_SHARED)
|| flags==(SQLITE_SHM_LOCK | SQLITE_SHM_EXCLUSIVE)
|| flags==(SQLITE_SHM_UNLOCK | SQLITE_SHM_SHARED)
** lock. If any do, return SQLITE_BUSY right away.
*/
for(pX=pShmNode->pFirst; pX; pX=pX->pNext){
if( (pX->exclMask & mask)!=0 || (pX->sharedMask & mask)!=0 ){
rc = SQLITE_BUSY;
break;
}
}
/* Get the exclusive locks at the system level. Then if successful
** also mark the local connection as being locked.
*/
if( rc==SQLITE_OK ){
rc = winShmSystemLock(pShmNode, _SHM_WRLCK, ofst+WIN_SHM_BASE, n);
if( rc==SQLITE_OK ){
assert( (p->sharedMask & mask)==0 );
p->exclMask |= mask;
}
}
}
sqlite3_mutex_leave(pShmNode->mutex);
** WRITER_DBMOD -> WRITER_FINISHED [sqlite3PagerCommitPhaseOne]
** WRITER_*** -> READER [pager_end_transaction]
**
** WRITER_*** -> ERROR [pager_error]
** ERROR -> OPEN [pager_unlock]
**
**
** OPEN:
**
** The pager starts up in this state. Nothing is guaranteed in this
** state - the file may or may not be locked and the database size is
** unknown. The database may not be read or written.
**
** * No read or write transaction is active.
** * Any lock, or no lock at all, may be held on the database file.
** * The dbSize, dbOrigSize and dbFileSize variables may not be trusted.
**
** READER:
**
** In this state all the requirements for reading the database in
** rollback (non-WAL) mode are met. Unless the pager is (or recently
** pagerUnlockDb() take a conservative approach - eLock is always updated
** when unlocking the file, and only updated when locking the file if the
** VFS call is successful. This way, the Pager.eLock variable may be set
** to a less exclusive (lower) value than the lock that is actually held
** at the system level, but it is never set to a more exclusive value.
**
** This is usually safe. If an xUnlock fails or appears to fail, there may
** be a few redundant xLock() calls or a lock may be held for longer than
** required, but nothing really goes wrong.
**
** The exception is when the database file is unlocked as the pager moves
** from ERROR to OPEN state. At this point there may be a hot-journal file
** in the file-system that needs to be rolled back (as part of an OPEN->SHARED
** transition, by the same pager or any other). If the call to xUnlock()
** fails at this point and the pager is left holding an EXCLUSIVE lock, this
** can confuse the call to xCheckReservedLock() call made later as part
** of hot-journal detection.
**
** xCheckReservedLock() is defined as returning true "if there is a RESERVED
** lock held by this process or any others". So xCheckReservedLock may
** return true because the caller itself is holding an EXCLUSIVE lock (but
}
return rc;
}
/*
** This function is a no-op if the pager is in exclusive mode and not
** in the ERROR state. Otherwise, it switches the pager to PAGER_OPEN
** state.
**
** If the pager is not in exclusive-access mode, the database file is
** completely unlocked. If the file is unlocked and the file-system does
** not exhibit the UNDELETABLE_WHEN_OPEN property, the journal file is
** closed (if it is open).
**
** If the pager is in ERROR state when this function is called, the
** contents of the pager cache are discarded before switching back to
** the OPEN state. Regardless of whether the pager is in exclusive-mode
** or not, any journal file left in the file-system will be treated
** as a hot-journal and rolled back the next time a read-transaction
** is opened (by this or by any other connection).
*/
** Ticket #1171: The statement journal might contain page content that is
** different from the page content at the start of the transaction.
** This occurs when a page is changed prior to the start of a statement
** then changed again within the statement. When rolling back such a
** statement we must not write to the original database unless we know
** for certain that original page contents are synced into the main rollback
** journal. Otherwise, a power loss might leave modified data in the
** database file without an entry in the rollback journal that can
** restore the database to its original form. Two conditions must be
** met before writing to the database files. (1) the database must be
** locked. (2) we know that the original page content is fully synced
** in the main journal either because the page is not in cache or else
** the page is marked as needSync==0.
**
** 2008-04-14: When attempting to vacuum a corrupt database file, it
** is possible to fail a statement on a database that does not yet exist.
** Do not attempt to write if database file has never been opened.
*/
if( pagerUseWal(pPager) ){
pPg = 0;
}else{
assert( jrnlOpen==0 || ( sqlite3OsDeviceCharacteristics(pPager->jfd) &
SQLITE_IOCAP_UNDELETABLE_WHEN_OPEN
));
*pExists = 0;
if( !jrnlOpen ){
rc = sqlite3OsAccess(pVfs, pPager->zJournal, SQLITE_ACCESS_EXISTS, &exists);
}
if( rc==SQLITE_OK && exists ){
int locked = 0; /* True if some process holds a RESERVED lock */
/* Race condition here: Another process might have been holding the
** the RESERVED lock and have a journal open at the sqlite3OsAccess()
** call above, but then delete the journal and drop the lock before
** we get to the following sqlite3OsCheckReservedLock() call. If that
** is the case, this routine might think there is a hot journal when
** in fact there is none. This results in a false-positive which will
** be dealt with by the playback routine. Ticket #3883.
*/
rc = sqlite3OsCheckReservedLock(pPager->fd, &locked);
if( rc==SQLITE_OK && !locked ){
Pgno nPage; /* Number of pages in database file */
rc = pagerPagecount(pPager, &nPage);
if( rc==SQLITE_OK ){
/* If the database is zero pages in size, that means that either (1) the
** journal is a remnant from a prior database with the same name where
** the database file but not the journal was deleted, or (2) the initial
** transaction that populates a new database is being rolled back.
** In either case, the journal file can be deleted. However, take care
** not to delete the journal file if it is already open due to
lockIdx-WAL_READ_LOCK(0));
return zName;
}
}
#endif /*defined(SQLITE_TEST) || defined(SQLITE_DEBUG) */
/*
** Set or release locks on the WAL. Locks are either shared or exclusive.
** A lock cannot be moved directly between shared and exclusive - it must go
** through the unlocked state first.
**
** In locking_mode=EXCLUSIVE, all of these routines become no-ops.
*/
static int walLockShared(Wal *pWal, int lockIdx){
int rc;
if( pWal->exclusiveMode ) return SQLITE_OK;
rc = sqlite3OsShmLock(pWal->pDbFd, lockIdx, 1,
SQLITE_SHM_LOCK | SQLITE_SHM_SHARED);
WALTRACE(("WAL%p: acquire SHARED-%s %s\n", pWal,
walLockName(lockIdx), rc ? "failed" : "ok"));
** the necessary locks, this routine returns SQLITE_BUSY.
*/
static int walIndexRecover(Wal *pWal){
int rc; /* Return Code */
i64 nSize; /* Size of log file */
u32 aFrameCksum[2] = {0, 0};
int iLock; /* Lock offset to lock for checkpoint */
int nLock; /* Number of locks to hold */
/* Obtain an exclusive lock on all byte in the locking range not already
** locked by the caller. The caller is guaranteed to have locked the
** WAL_WRITE_LOCK byte, and may have also locked the WAL_CKPT_LOCK byte.
** If successful, the same bytes that are locked here are unlocked before
** this function returns.
*/
assert( pWal->ckptLock==1 || pWal->ckptLock==0 );
assert( WAL_ALL_BUT_WRITE==WAL_WRITE_LOCK+1 );
assert( WAL_CKPT_LOCK==WAL_ALL_BUT_WRITE );
assert( pWal->writeLock );
iLock = WAL_ALL_BUT_WRITE + pWal->ckptLock;
nLock = SQLITE_SHM_NLOCK - iLock;
rc = walLockExclusive(pWal, iLock, nLock);
if( rc ){
** so it takes care to hold an exclusive lock on the corresponding
** WAL_READ_LOCK() while changing values.
*/
static int walTryBeginRead(Wal *pWal, int *pChanged, int useWal, int cnt){
volatile WalCkptInfo *pInfo; /* Checkpoint information in wal-index */
u32 mxReadMark; /* Largest aReadMark[] value */
int mxI; /* Index of largest aReadMark[] value */
int i; /* Loop counter */
int rc = SQLITE_OK; /* Return code */
assert( pWal->readLock<0 ); /* Not currently locked */
/* Take steps to avoid spinning forever if there is a protocol error.
**
** Circumstances that cause a RETRY should only last for the briefest
** instances of time. No I/O or other system calls are done while the
** locks are held, so the locks should not be held for very long. But
** if we are unlucky, another process that is holding a lock might get
** paged out or take a page-fault that is time-consuming to resolve,
** during the few nanoseconds that it is holding the lock. In that case,
** it might take longer than normal for the lock to free.
** This code really belongs in btree.c. But btree.c is getting too
** big and we want to break it down some. This packaged seemed like
** a good breakout.
*/
#ifndef SQLITE_OMIT_SHARED_CACHE
#if SQLITE_THREADSAFE
/*
** Obtain the BtShared mutex associated with B-Tree handle p. Also,
** set BtShared.db to the database handle associated with p and the
** p->locked boolean to true.
*/
static void lockBtreeMutex(Btree *p){
assert( p->locked==0 );
assert( sqlite3_mutex_notheld(p->pBt->mutex) );
assert( sqlite3_mutex_held(p->db->mutex) );
sqlite3_mutex_enter(p->pBt->mutex);
p->pBt->db = p->db;
p->locked = 1;
}
/*
** Release the BtShared mutex associated with B-Tree handle p and
** clear the p->locked boolean.
*/
static void SQLITE_NOINLINE unlockBtreeMutex(Btree *p){
BtShared *pBt = p->pBt;
assert( p->locked==1 );
assert( sqlite3_mutex_held(pBt->mutex) );
assert( sqlite3_mutex_held(p->db->mutex) );
assert( p->db==pBt->db );
sqlite3_mutex_leave(pBt->mutex);
p->locked = 0;
}
/* Forward reference */
static void SQLITE_NOINLINE btreeLockCarefully(Btree *p);
/*
** Enter a mutex on the given BTree object.
**
** If the object is not sharable, then no mutex is ever required
** and this routine is a no-op. The underlying mutex is non-recursive.
** But we keep a reference count in Btree.wantToLock so the behavior
** of this interface is recursive.
**
** To avoid deadlocks, multiple Btrees are locked in the same order
** by all database connections. The p->pNext is a list of other
** Btrees belonging to the same database connection as the p Btree
** which need to be locked after p. If we cannot get a lock on
** p, then first unlock all of the others on p->pNext, then wait
** for the lock to become available on p, then relock all of the
** subsequent Btrees that desire a lock.
*/
SQLITE_PRIVATE void sqlite3BtreeEnter(Btree *p){
/* Some basic sanity checking on the Btree. The list of Btrees
** connected by pNext and pPrev should be in sorted order by
** Btree.pBt value. All elements of the list should belong to
** the same connection. Only shared Btrees are on the list. */
assert( p->pNext==0 || p->pNext->pBt>p->pBt );
assert( p->pPrev==0 || p->pPrev->pBt<p->pBt );
assert( p->pNext==0 || p->pNext->db==p->db );
assert( p->pPrev==0 || p->pPrev->db==p->db );
assert( p->sharable || (p->pNext==0 && p->pPrev==0) );
/* Check for locking consistency */
assert( !p->locked || p->wantToLock>0 );
assert( p->sharable || p->wantToLock==0 );
/* We should already hold a lock on the database connection */
assert( sqlite3_mutex_held(p->db->mutex) );
/* Unless the database is sharable and unlocked, then BtShared.db
** should already be set correctly. */
assert( (p->locked==0 && p->sharable) || p->pBt->db==p->db );
if( !p->sharable ) return;
p->wantToLock++;
if( p->locked ) return;
btreeLockCarefully(p);
}
/* This is a helper function for sqlite3BtreeLock(). By moving
** complex, but seldom used logic, out of sqlite3BtreeLock() and
** into this routine, we avoid unnecessary stack pointer changes
** and thus help the sqlite3BtreeLock() routine to run much faster
** in the common case.
*/
static void SQLITE_NOINLINE btreeLockCarefully(Btree *p){
Btree *pLater;
/* In most cases, we should be able to acquire the lock we
** want without having to go through the ascending lock
** procedure that follows. Just be sure not to block.
*/
if( sqlite3_mutex_try(p->pBt->mutex)==SQLITE_OK ){
p->pBt->db = p->db;
p->locked = 1;
return;
}
/* To avoid deadlock, first release all locks with a larger
** BtShared address. Then acquire our lock. Then reacquire
** the other BtShared locks that we used to hold in ascending
** order.
*/
for(pLater=p->pNext; pLater; pLater=pLater->pNext){
assert( pLater->sharable );
assert( pLater->pNext==0 || pLater->pNext->pBt>pLater->pBt );
assert( !pLater->locked || pLater->wantToLock>0 );
if( pLater->locked ){
unlockBtreeMutex(pLater);
}
}
lockBtreeMutex(p);
for(pLater=p->pNext; pLater; pLater=pLater->pNext){
if( pLater->wantToLock ){
lockBtreeMutex(pLater);
}
}
}
}
#ifndef NDEBUG
/*
** Return true if the BtShared mutex is held on the btree, or if the
** B-Tree is not marked as sharable.
**
** This routine is used only from within assert() statements.
*/
SQLITE_PRIVATE int sqlite3BtreeHoldsMutex(Btree *p){
assert( p->sharable==0 || p->locked==0 || p->wantToLock>0 );
assert( p->sharable==0 || p->locked==0 || p->db==p->pBt->db );
assert( p->sharable==0 || p->locked==0 || sqlite3_mutex_held(p->pBt->mutex) );
assert( p->sharable==0 || p->locked==0 || sqlite3_mutex_held(p->db->mutex) );
return (p->sharable==0 || p->locked);
}
#endif
#ifndef SQLITE_OMIT_INCRBLOB
/*
** Enter and leave a mutex on a Btree given a cursor owned by that
** Btree. These entry points are used by incremental I/O and can be
** omitted if that module is not used.
*/
*/
SQLITE_PRIVATE int sqlite3SchemaMutexHeld(sqlite3 *db, int iDb, Schema *pSchema){
Btree *p;
assert( db!=0 );
if( pSchema ) iDb = sqlite3SchemaToIndex(db, pSchema);
assert( iDb>=0 && iDb<db->nDb );
if( !sqlite3_mutex_held(db->mutex) ) return 0;
if( iDb==1 ) return 1;
p = db->aDb[iDb].pBt;
assert( p!=0 );
return p->sharable==0 || p->locked==1;
}
#endif /* NDEBUG */
#else /* SQLITE_THREADSAFE>0 above. SQLITE_THREADSAFE==0 below */
/*
** The following are special cases for mutex enter routines for use
** in single threaded applications that use shared cache. Except for
** these two routines, all mutex operations are no-ops in that case and
** are null #defines in btree.h.
**
/* This routine is a no-op if the shared-cache is not enabled */
if( !p->sharable ){
return SQLITE_OK;
}
/* If some other connection is holding an exclusive lock, the
** requested lock may not be obtained.
*/
if( pBt->pWriter!=p && (pBt->btsFlags & BTS_EXCLUSIVE)!=0 ){
sqlite3ConnectionBlocked(p->db, pBt->pWriter->db);
return SQLITE_LOCKED_SHAREDCACHE;
}
for(pIter=pBt->pLock; pIter; pIter=pIter->pNext){
/* The condition (pIter->eLock!=eLock) in the following if(...)
** statement is a simplification of:
**
** (eLock==WRITE_LOCK || pIter->eLock==WRITE_LOCK)
**
** since we know that if eLock==WRITE_LOCK, then no other connection
** may hold a WRITE_LOCK on any table in this file (since there can
** only be a single writer).
*/
assert( pIter->eLock==READ_LOCK || pIter->eLock==WRITE_LOCK );
assert( eLock==READ_LOCK || pIter->pBtree==p || pIter->eLock==READ_LOCK);
if( pIter->pBtree!=p && pIter->iTable==iTab && pIter->eLock!=eLock ){
sqlite3ConnectionBlocked(p->db, pIter->pBtree->db);
if( eLock==WRITE_LOCK ){
assert( p==pBt->pWriter );
pBt->btsFlags |= BTS_PENDING;
}
return SQLITE_LOCKED_SHAREDCACHE;
}
}
return SQLITE_OK;
}
#endif /* !SQLITE_OMIT_SHARED_CACHE */
** The call to sqlite3BtreeRollback() drops any table-locks held by
** this handle.
*/
sqlite3BtreeRollback(p, SQLITE_OK, 0);
sqlite3BtreeLeave(p);
/* If there are still other outstanding references to the shared-btree
** structure, return now. The remainder of this procedure cleans
** up the shared-btree.
*/
assert( p->wantToLock==0 && p->locked==0 );
if( !p->sharable || removeFromSharingList(pBt) ){
/* The pBt is no longer on the sharing list, so we can access
** it without having to hold the mutex.
**
** Clean out and delete the BtShared object.
*/
assert( !pBt->pCursor );
sqlite3PagerClose(pBt->pPager);
if( pBt->xFreeSchema && pBt->pSchema ){
pBt->xFreeSchema(pBt->pSchema);
}
sqlite3DbFree(0, pBt->pSchema);
freeTempSpace(pBt);
sqlite3_free(pBt);
}
#ifndef SQLITE_OMIT_SHARED_CACHE
assert( p->wantToLock==0 );
assert( p->locked==0 );
if( p->pPrev ) p->pPrev->pNext = p->pNext;
if( p->pNext ) p->pNext->pPrev = p->pPrev;
#endif
sqlite3_free(p);
return SQLITE_OK;
}
/*
** Change the limit on the number of pages allowed in the cache.
#endif
}
/*
** Get a reference to pPage1 of the database file. This will
** also acquire a readlock on that file.
**
** SQLITE_OK is returned on success. If the file is not a
** well-formed database file, then SQLITE_CORRUPT is returned.
** SQLITE_BUSY is returned if the database is locked. SQLITE_NOMEM
** is returned if we run out of memory.
*/
static int lockBtree(BtShared *pBt){
int rc; /* Result code from subfunctions */
MemPage *pPage1; /* Page 1 of the database file */
int nPage; /* Number of pages in the database */
int nPageFile = 0; /* Number of pages in the database file */
int nPageHeader; /* Number of pages in the database according to hdr */
assert( sqlite3_mutex_held(pBt->mutex) );
**
** sqlite3BtreeCreateTable()
** sqlite3BtreeCreateIndex()
** sqlite3BtreeClearTable()
** sqlite3BtreeDropTable()
** sqlite3BtreeInsert()
** sqlite3BtreeDelete()
** sqlite3BtreeUpdateMeta()
**
** If an initial attempt to acquire the lock fails because of lock contention
** and the database was previously unlocked, then invoke the busy handler
** if there is one. But if there was previously a read-lock, do not
** invoke the busy handler - just return SQLITE_BUSY. SQLITE_BUSY is
** returned when there is already a read-lock in order to avoid a deadlock.
**
** Suppose there are two processes A and B. A has a read lock and B has
** a reserved lock. B tries to promote to exclusive but is blocked because
** of A's read lock. A tries to promote to reserved but is blocked by B.
** One or the other of the two processes must give way or there can be
** no progress. By returning SQLITE_BUSY and not invoking the busy callback
** when A already has a read lock, we encourage A to give up and let B
** proceed.
*/
SQLITE_PRIVATE int sqlite3BtreeBeginTrans(Btree *p, int wrflag){
sqlite3 *pBlock = 0;
BtShared *pBt = p->pBt;
int rc = SQLITE_OK;
}else if( wrflag>1 ){
BtLock *pIter;
for(pIter=pBt->pLock; pIter; pIter=pIter->pNext){
if( pIter->pBtree!=p ){
pBlock = pIter->pBtree->db;
break;
}
}
}
if( pBlock ){
sqlite3ConnectionBlocked(p->db, pBlock);
rc = SQLITE_LOCKED_SHAREDCACHE;
goto trans_begun;
}
#endif
/* Any read-only or read-write transaction implies a read-lock on
** page 1. So if some other shared-cache client already has a write-lock
** on page 1, the transaction cannot be opened. */
rc = querySharedCacheTableLock(p, MASTER_ROOT, READ_LOCK);
if( SQLITE_OK!=rc ) goto trans_begun;
/* It is illegal to drop a table if any cursors are open on the
** database. This is because in auto-vacuum mode the backend may
** need to move another root-page to fill a gap left by the deleted
** root page. If an open cursor was using this page a problem would
** occur.
**
** This error is caught long before control reaches this point.
*/
if( NEVER(pBt->pCursor) ){
sqlite3ConnectionBlocked(p->db, pBt->pCursor->pBtree->db);
return SQLITE_LOCKED_SHAREDCACHE;
}
rc = btreeGetPage(pBt, (Pgno)iTable, &pPage, 0);
if( rc ) return rc;
rc = sqlite3BtreeClearTable(p, iTable, 0);
if( rc ){
releasePage(pPage);
return rc;
}
sqlite3_backup *p; /* Value to return */
#ifdef SQLITE_ENABLE_API_ARMOR
if( !sqlite3SafetyCheckOk(pSrcDb)||!sqlite3SafetyCheckOk(pDestDb) ){
(void)SQLITE_MISUSE_BKPT;
return 0;
}
#endif
/* Lock the source database handle. The destination database
** handle is not locked in this routine, but it is locked in
** sqlite3_backup_step(). The user is required to ensure that no
** other thread accesses the destination handle for the duration
** of the backup operation. Any attempt to use the destination
** database connection while a backup is in progress may cause
** a malfunction or a deadlock.
*/
sqlite3_mutex_enter(pSrcDb->mutex);
sqlite3_mutex_enter(pDestDb->mutex);
if( pSrcDb==pDestDb ){
/* If the source pager is currently in a write-transaction, return
** SQLITE_BUSY immediately.
*/
if( p->pDestDb && p->pSrc->pBt->inTransaction==TRANS_WRITE ){
rc = SQLITE_BUSY;
}else{
rc = SQLITE_OK;
}
/* Lock the destination database, if it is not locked already. */
if( SQLITE_OK==rc && p->bDestLocked==0
&& SQLITE_OK==(rc = sqlite3BtreeBeginTrans(p->pDest, 2))
){
p->bDestLocked = 1;
sqlite3BtreeGetMeta(p->pDest, BTREE_SCHEMA_VERSION, &p->iDestSchema);
}
/* If there is no open read-transaction on the source database, open
** one now. If a transaction is opened here, then it will be closed
** before this function exits.
for(i=0; i<nDb; i++){
if( i!=1 && DbMaskTest(p->lockMask,i) && ALWAYS(aDb[i].pBt!=0) ){
sqlite3BtreeEnter(aDb[i].pBt);
}
}
}
#endif
#if !defined(SQLITE_OMIT_SHARED_CACHE) && SQLITE_THREADSAFE>0
/*
** Unlock all of the btrees previously locked by a call to sqlite3VdbeEnter().
*/
SQLITE_PRIVATE void sqlite3VdbeLeave(Vdbe *p){
int i;
sqlite3 *db;
Db *aDb;
int nDb;
if( DbMaskAllZero(p->lockMask) ) return; /* The common case */
db = p->db;
aDb = db->aDb;
nDb = db->nDb;
assert( db->nVdbeRead>=db->nVdbeWrite );
assert( db->nVdbeWrite>=0 );
}
p->magic = VDBE_MAGIC_HALT;
checkActiveVdbeCnt(db);
if( p->db->mallocFailed ){
p->rc = SQLITE_NOMEM;
}
/* If the auto-commit flag is set to true, then any locks that were held
** by connection db have now been released. Call sqlite3ConnectionUnlocked()
** to invoke any required unlock-notify callbacks.
*/
if( db->autoCommit ){
sqlite3ConnectionUnlocked(db);
}
assert( db->nVdbeActive>0 || db->autoCommit==0 || db->nStatement==0 );
return (p->rc==SQLITE_BUSY ? SQLITE_BUSY : SQLITE_OK);
}
/*
** Each VDBE holds the result of the most recent sqlite3_step() call
** in p->rc. This routine sets that result back to SQLITE_OK.
** P3==0 means the main database, P3==1 means the database used for
** temporary tables, and P3>1 means used the corresponding attached
** database. Give the new cursor an identifier of P1. The P1
** values need not be contiguous but all P1 values should be small integers.
** It is an error for P1 to be negative.
**
** If P5!=0 then use the content of register P2 as the root page, not
** the value of P2 itself.
**
** There will be a read lock on the database whenever there is an
** open cursor. If the database was unlocked prior to this instruction
** then a read lock is acquired as part of this instruction. A read
** lock allows other processes to read the database but prohibits
** any other process from modifying the database. The read lock is
** released when all cursors are closed. If this instruction attempts
** to get a read lock but fails, the script terminates with an
** SQLITE_BUSY error code.
**
** The P4 value may be either an integer (P4_INT32) or a pointer to
** a KeyInfo structure (P4_KEYINFO). If it is a pointer to a KeyInfo
** structure, then said structure defines the content and collating
**
** Obtain a lock on a particular table. This instruction is only used when
** the shared-cache feature is enabled.
**
** P1 is the index of the database in sqlite3.aDb[] of the database
** on which the lock is acquired. A readlock is obtained if P3==0 or
** a write lock if P3==1.
**
** P2 contains the root-page of the table to lock.
**
** P4 contains a pointer to the name of the table being locked. This is only
** used to generate an error message if the lock cannot be obtained.
*/
case OP_TableLock: {
u8 isWriteLock = (u8)pOp->p3;
if( isWriteLock || 0==(db->flags&SQLITE_ReadUncommitted) ){
int p1 = pOp->p1;
assert( p1>=0 && p1<db->nDb );
assert( DbMaskTest(p->btreeMask, p1) );
assert( isWriteLock==0 || isWriteLock==1 );
rc = sqlite3BtreeLockTable(db->aDb[p1].pBt, pOp->p2, isWriteLock);
if( (rc&0xFF)==SQLITE_LOCKED ){
const char *z = pOp->p4.z;
sqlite3SetString(&p->zErrMsg, db, "database table is locked: %s", z);
}
}
break;
}
#endif /* SQLITE_OMIT_SHARED_CACHE */
#ifndef SQLITE_OMIT_VIRTUALTABLE
/* Opcode: VBegin * * * P4 *
**
** P4 may be a pointer to an sqlite3_vtab structure. If so, call the
SortSubtask *pTask,
const char *zEvent
){
i64 t;
int iTask = (pTask - pTask->pSorter->aTask);
sqlite3OsCurrentTimeInt64(pTask->pSorter->db->pVfs, &t);
fprintf(stderr, "%lld:bg%d %s\n", t, iTask, zEvent);
}
static void vdbeSorterBlockDebug(
SortSubtask *pTask,
int bBlocked,
const char *zEvent
){
if( bBlocked ){
i64 t;
sqlite3OsCurrentTimeInt64(pTask->pSorter->db->pVfs, &t);
fprintf(stderr, "%lld:main %s\n", t, zEvent);
}
}
#else
# define vdbeSorterWorkDebug(x,y)
# define vdbeSorterRewindDebug(y)
# define vdbeSorterPopulateDebug(x,y)
# define vdbeSorterBlockDebug(x,y,z)
if( i<2 ){
sqlite3_snprintf(sizeof(zErr),zErr, "cannot detach database %s", zName);
goto detach_error;
}
if( !db->autoCommit ){
sqlite3_snprintf(sizeof(zErr), zErr,
"cannot DETACH database within transaction");
goto detach_error;
}
if( sqlite3BtreeIsInReadTrans(pDb->pBt) || sqlite3BtreeIsInBackup(pDb->pBt) ){
sqlite3_snprintf(sizeof(zErr),zErr, "database %s is locked", zName);
goto detach_error;
}
sqlite3BtreeClose(pDb->pBt);
pDb->pBt = 0;
pDb->pSchema = 0;
sqlite3ResetAllSchemasOfConnection(db);
return;
detach_error:
pParse->explain = (u8)explainFlag;
pParse->nVar = 0;
}
#ifndef SQLITE_OMIT_SHARED_CACHE
/*
** The TableLock structure is only used by the sqlite3TableLock() and
** codeTableLocks() functions.
*/
struct TableLock {
int iDb; /* The database containing the table to be locked */
int iTab; /* The root page of the table to be locked */
u8 isWriteLock; /* True for write lock. False for a read lock */
const char *zName; /* Name of the table */
};
/*
** Record the fact that we want to lock a table at run-time.
**
** The table to be locked has root page iTab and is found in database iDb.
** A read or a write lock can be taken depending on isWritelock.
**
** This routine just records the fact that the lock is desired. The
** code to make the lock occur is generated by a later call to
** codeTableLocks() which occurs during sqlite3FinishCoding().
*/
SQLITE_PRIVATE void sqlite3TableLock(
Parse *pParse, /* Parsing context */
int iDb, /* Index of the database containing the table to lock */
int iTab, /* Root page number of the table to be locked */
u8 isWriteLock, /* True for a write lock */
const char *zName /* Name of the table to be locked */
){
Parse *pToplevel = sqlite3ParseToplevel(pParse);
int i;
int nBytes;
TableLock *p;
assert( iDb>=0 );
for(i=0; i<pToplevel->nTableLock; i++){
p = &pToplevel->aTableLock[i];
if( p->iDb==iDb && p->iTab==iTab ){
p->iTab = iTab;
p->isWriteLock = isWriteLock;
p->zName = zName;
}else{
pToplevel->nTableLock = 0;
pToplevel->db->mallocFailed = 1;
}
}
/*
** Code an OP_TableLock instruction for each table locked by the
** statement (configured by calls to sqlite3TableLock()).
*/
static void codeTableLocks(Parse *pParse){
int i;
Vdbe *pVdbe;
pVdbe = sqlite3GetVdbe(pParse);
assert( pVdbe!=0 ); /* sqlite3GetVdbe cannot fail: VDBE already allocated */
for(i=0; i<pParse->nTableLock; i++){
sqlite3_limit(db, SQLITE_LIMIT_WORKER_THREADS, -1));
break;
}
#if defined(SQLITE_DEBUG) || defined(SQLITE_TEST)
/*
** Report the current state of file logs for all databases
*/
case PragTyp_LOCK_STATUS: {
static const char *const azLockName[] = {
"unlocked", "shared", "reserved", "pending", "exclusive"
};
int i;
sqlite3VdbeSetNumCols(v, 2);
pParse->nMem = 2;
sqlite3VdbeSetColName(v, 0, COLNAME_NAME, "database", SQLITE_STATIC);
sqlite3VdbeSetColName(v, 1, COLNAME_NAME, "status", SQLITE_STATIC);
for(i=0; i<db->nDb; i++){
Btree *pBt;
const char *zState = "unknown";
int j;
** but it does *not* override schema lock detection, so this all still
** works even if READ_UNCOMMITTED is set.
*/
for(i=0; i<db->nDb; i++) {
Btree *pBt = db->aDb[i].pBt;
if( pBt ){
assert( sqlite3BtreeHoldsMutex(pBt) );
rc = sqlite3BtreeSchemaLocked(pBt);
if( rc ){
const char *zDb = db->aDb[i].zName;
sqlite3ErrorWithMsg(db, rc, "database schema is locked: %s", zDb);
testcase( db->flags & SQLITE_ReadUncommitted );
goto end_prepare;
}
}
}
sqlite3VtabUnlockList(db);
pParse->db = db;
pParse->nQueryLoop = 0; /* Logarithmic, so 0 really means 1 */
sqlite3_mutex_leave(db->mutex);
assert( rc==SQLITE_OK || *ppStmt==0 );
return rc;
}
/*
** Rerun the compilation of a statement after a schema change.
**
** If the statement is successfully recompiled, return SQLITE_OK. Otherwise,
** if the statement cannot be recompiled because another connection has
** locked the sqlite3_master table, return SQLITE_LOCKED. If any other error
** occurs, return SQLITE_SCHEMA.
*/
SQLITE_PRIVATE int sqlite3Reprepare(Vdbe *p){
int rc;
sqlite3_stmt *pNew;
const char *zSql;
sqlite3 *db;
assert( sqlite3_mutex_held(sqlite3VdbeDb(p)->mutex) );
zSql = sqlite3_sql((sqlite3_stmt *)p);
}
rc = execSql(db, pzErrMsg, zSql);
if( db->nDb>nDb ){
pDb = &db->aDb[db->nDb-1];
assert( strcmp(pDb->zName,"vacuum_db")==0 );
}
if( rc!=SQLITE_OK ) goto end_of_vacuum;
pTemp = db->aDb[db->nDb-1].pBt;
/* The call to execSql() to attach the temp database has left the file
** locked (as there was more than one active statement when the transaction
** to read the schema was concluded. Unlock it here so that this doesn't
** cause problems for the call to BtreeSetPageSize() below. */
sqlite3BtreeCommit(pTemp);
nRes = sqlite3BtreeGetReserve(pMain);
/* A VACUUM cannot change the pagesize of an encrypted database. */
#ifdef SQLITE_HAS_CODEC
if( db->nextPagesize ){
extern void sqlite3CodecGetKey(sqlite3*, int, void**, int*);
if( !sqlite3SafetyCheckOk(db) || zName==0 ) return SQLITE_MISUSE_BKPT;
#endif
return createModule(db, zName, pModule, pAux, xDestroy);
}
/*
** Lock the virtual table so that it cannot be disconnected.
** Locks nest. Every lock should have a corresponding unlock.
** If an unlock is omitted, resources leaks will occur.
**
** If a disconnect is attempted while a virtual table is locked,
** the disconnect is deferred until all locks have been removed.
*/
SQLITE_PRIVATE void sqlite3VtabLock(VTable *pVTab){
pVTab->nRef++;
}
/*
** pTab is a pointer to a Table structure representing a virtual-table.
** Return a pointer to the VTable object used by connection db to access
** Return a static string that describes the kind of error specified in the
** argument.
*/
SQLITE_PRIVATE const char *sqlite3ErrStr(int rc){
static const char* const aMsg[] = {
/* SQLITE_OK */ "not an error",
/* SQLITE_ERROR */ "SQL logic error or missing database",
/* SQLITE_INTERNAL */ 0,
/* SQLITE_PERM */ "access permission denied",
/* SQLITE_ABORT */ "callback requested query abort",
/* SQLITE_BUSY */ "database is locked",
/* SQLITE_LOCKED */ "database table is locked",
/* SQLITE_NOMEM */ "out of memory",
/* SQLITE_READONLY */ "attempt to write a readonly database",
/* SQLITE_INTERRUPT */ "interrupted",
/* SQLITE_IOERR */ "disk I/O error",
/* SQLITE_CORRUPT */ "database disk image is malformed",
/* SQLITE_NOTFOUND */ "unknown operation",
/* SQLITE_FULL */ "database or disk is full",
/* SQLITE_CANTOPEN */ "unable to open database file",
/* SQLITE_PROTOCOL */ "locking protocol",
/* SQLITE_EMPTY */ "table contains no data",
** This file contains the implementation of the sqlite3_unlock_notify()
** API method and its associated functionality.
*/
/* Omit this entire file if SQLITE_ENABLE_UNLOCK_NOTIFY is not defined. */
#ifdef SQLITE_ENABLE_UNLOCK_NOTIFY
/*
** Public interfaces:
**
** sqlite3ConnectionBlocked()
** sqlite3ConnectionUnlocked()
** sqlite3ConnectionClosed()
** sqlite3_unlock_notify()
*/
#define assertMutexHeld() \
assert( sqlite3_mutex_held(sqlite3MutexAlloc(SQLITE_MUTEX_STATIC_MASTER)) )
/*
** Head of a linked list of all sqlite3 objects created by this process
** for which either sqlite3.pBlockingConnection or sqlite3.pUnlockConnection
** is not NULL. This variable may only accessed while the STATIC_MASTER
** mutex is held.
*/
static sqlite3 *SQLITE_WSD sqlite3BlockedList = 0;
#ifndef NDEBUG
/*
** This function is a complex assert() that verifies the following
** properties of the blocked connections list:
**
** 1) Each entry in the list has a non-NULL value for either
** pUnlockConnection or pBlockingConnection, or both.
**
** 2) All entries in the list that share a common value for
** xUnlockNotify are grouped together.
**
** 3) If the argument db is not NULL, then none of the entries in the
** blocked connections list have pUnlockConnection or pBlockingConnection
** set to db. This is used when closing connection db.
*/
static void checkListProperties(sqlite3 *db){
sqlite3 *p;
for(p=sqlite3BlockedList; p; p=p->pNextBlocked){
int seen = 0;
sqlite3 *p2;
/* Verify property (1) */
assert( p->pUnlockConnection || p->pBlockingConnection );
/* Verify property (2) */
for(p2=sqlite3BlockedList; p2!=p; p2=p2->pNextBlocked){
if( p2->xUnlockNotify==p->xUnlockNotify ) seen = 1;
assert( p2->xUnlockNotify==p->xUnlockNotify || !seen );
assert( db==0 || p->pUnlockConnection!=db );
assert( db==0 || p->pBlockingConnection!=db );
}
}
}
#else
# define checkListProperties(x)
#endif
/*
** Remove connection db from the blocked connections list. If connection
** db is not currently a part of the list, this function is a no-op.
*/
static void removeFromBlockedList(sqlite3 *db){
sqlite3 **pp;
assertMutexHeld();
for(pp=&sqlite3BlockedList; *pp; pp = &(*pp)->pNextBlocked){
if( *pp==db ){
*pp = (*pp)->pNextBlocked;
break;
}
}
}
/*
** Add connection db to the blocked connections list. It is assumed
** that it is not already a part of the list.
*/
static void addToBlockedList(sqlite3 *db){
sqlite3 **pp;
assertMutexHeld();
for(
pp=&sqlite3BlockedList;
*pp && (*pp)->xUnlockNotify!=db->xUnlockNotify;
pp=&(*pp)->pNextBlocked
);
db->pNextBlocked = *pp;
*pp = db;
}
/*
** Obtain the STATIC_MASTER mutex.
*/
static void enterMutex(void){
sqlite3_mutex_enter(sqlite3MutexAlloc(SQLITE_MUTEX_STATIC_MASTER));
checkListProperties(0);
}
** Register an unlock-notify callback.
**
** This is called after connection "db" has attempted some operation
** but has received an SQLITE_LOCKED error because another connection
** (call it pOther) in the same process was busy using the same shared
** cache. pOther is found by looking at db->pBlockingConnection.
**
** If there is no blocking connection, the callback is invoked immediately,
** before this routine returns.
**
** If pOther is already blocked on db, then report SQLITE_LOCKED, to indicate
** a deadlock.
**
** Otherwise, make arrangements to invoke xNotify when pOther drops
** its locks.
**
** Each call to this routine overrides any prior callbacks registered
** on the same "db". If xNotify==0 then any prior callbacks are immediately
** cancelled.
*/
SQLITE_API int sqlite3_unlock_notify(
sqlite3 *db,
void (*xNotify)(void **, int),
void *pArg
){
int rc = SQLITE_OK;
sqlite3_mutex_enter(db->mutex);
enterMutex();
if( xNotify==0 ){
removeFromBlockedList(db);
db->pBlockingConnection = 0;
db->pUnlockConnection = 0;
db->xUnlockNotify = 0;
db->pUnlockArg = 0;
}else if( 0==db->pBlockingConnection ){
/* The blocking transaction has been concluded. Or there never was a
** blocking transaction. In either case, invoke the notify callback
** immediately.
*/
xNotify(&pArg, 1);
}else{
sqlite3 *p;
for(p=db->pBlockingConnection; p && p!=db; p=p->pUnlockConnection){}
if( p ){
rc = SQLITE_LOCKED; /* Deadlock detected. */
}else{
db->pUnlockConnection = db->pBlockingConnection;
db->xUnlockNotify = xNotify;
db->pUnlockArg = pArg;
removeFromBlockedList(db);
addToBlockedList(db);
}
}
leaveMutex();
assert( !db->mallocFailed );
sqlite3ErrorWithMsg(db, rc, (rc?"database is deadlocked":0));
sqlite3_mutex_leave(db->mutex);
return rc;
}
/*
** This function is called while stepping or preparing a statement
** associated with connection db. The operation will return SQLITE_LOCKED
** to the user because it requires a lock that will not be available
** until connection pBlocker concludes its current transaction.
*/
SQLITE_PRIVATE void sqlite3ConnectionBlocked(sqlite3 *db, sqlite3 *pBlocker){
enterMutex();
if( db->pBlockingConnection==0 && db->pUnlockConnection==0 ){
addToBlockedList(db);
}
db->pBlockingConnection = pBlocker;
leaveMutex();
}
/*
** This function is called when
** the transaction opened by database db has just finished. Locks held
** by database connection db have been released.
**
** This function loops through each entry in the blocked connections
** list and does the following:
**
** 1) If the sqlite3.pBlockingConnection member of a list entry is
** set to db, then set pBlockingConnection=0.
**
** 2) If the sqlite3.pUnlockConnection member of a list entry is
** set to db, then invoke the configured unlock-notify callback and
** set pUnlockConnection=0.
**
** 3) If the two steps above mean that pBlockingConnection==0 and
** pUnlockConnection==0, remove the entry from the blocked connections
** list.
*/
SQLITE_PRIVATE void sqlite3ConnectionUnlocked(sqlite3 *db){
void (*xUnlockNotify)(void **, int) = 0; /* Unlock-notify cb to invoke */
int nArg = 0; /* Number of entries in aArg[] */
sqlite3 **pp; /* Iterator variable */
void **aArg; /* Arguments to the unlock callback */
void **aDyn = 0; /* Dynamically allocated space for aArg[] */
void *aStatic[16]; /* Starter space for aArg[]. No malloc required */
aArg = aStatic;
enterMutex(); /* Enter STATIC_MASTER mutex */
/* This loop runs once for each entry in the blocked-connections list. */
for(pp=&sqlite3BlockedList; *pp; /* no-op */ ){
sqlite3 *p = *pp;
/* Step 1. */
if( p->pBlockingConnection==db ){
p->pBlockingConnection = 0;
}
/* Step 2. */
if( p->pUnlockConnection==db ){
assert( p->xUnlockNotify );
aDyn = aArg = pNew;
}else{
/* This occurs when the array of context pointers that need to
** be passed to the unlock-notify callback is larger than the
** aStatic[] array allocated on the stack and the attempt to
** allocate a larger array from the heap has failed.
**
** This is a difficult situation to handle. Returning an error
** code to the caller is insufficient, as even if an error code
** is returned the transaction on connection db will still be
** closed and the unlock-notify callbacks on blocked connections
** will go unissued. This might cause the application to wait
** indefinitely for an unlock-notify callback that will never
** arrive.
**
** Instead, invoke the unlock-notify callback with the context
** array already accumulated. We can then clear the array and
** begin accumulating any further context pointers without
** requiring any dynamic allocation. This is sub-optimal because
** it means that instead of one callback with a large array of
** context pointers the application will receive two or more
aArg[nArg++] = p->pUnlockArg;
xUnlockNotify = p->xUnlockNotify;
p->pUnlockConnection = 0;
p->xUnlockNotify = 0;
p->pUnlockArg = 0;
}
/* Step 3. */
if( p->pBlockingConnection==0 && p->pUnlockConnection==0 ){
/* Remove connection p from the blocked connections list. */
*pp = p->pNextBlocked;
p->pNextBlocked = 0;
}else{
pp = &p->pNextBlocked;
}
}
if( nArg!=0 ){
xUnlockNotify(aArg, nArg);
}
sqlite3_free(aDyn);
leaveMutex(); /* Leave STATIC_MASTER mutex */
}
/*
** This is called when the database connection passed as an argument is
** being closed. The connection is removed from the blocked list.
*/
SQLITE_PRIVATE void sqlite3ConnectionClosed(sqlite3 *db){
sqlite3ConnectionUnlocked(db);
enterMutex();
removeFromBlockedList(db);
checkListProperties(db);
leaveMutex();
}
#endif
/************** End of notify.c **********************************************/
/************** Begin file fts3.c ********************************************/
/*
** 2006 Oct 10
**
** New error codes may be added in future versions of SQLite.
**
** See also: [extended result code definitions]
*/
#define SQLITE_OK 0 /* Successful result */
/* beginning-of-error-codes */
#define SQLITE_ERROR 1 /* SQL error or missing database */
#define SQLITE_INTERNAL 2 /* Internal logic error in SQLite */
#define SQLITE_PERM 3 /* Access permission denied */
#define SQLITE_ABORT 4 /* Callback routine requested an abort */
#define SQLITE_BUSY 5 /* The database file is locked */
#define SQLITE_LOCKED 6 /* A table in the database is locked */
#define SQLITE_NOMEM 7 /* A malloc() failed */
#define SQLITE_READONLY 8 /* Attempt to write a readonly database */
#define SQLITE_INTERRUPT 9 /* Operation terminated by sqlite3_interrupt()*/
#define SQLITE_IOERR 10 /* Some kind of disk I/O error occurred */
#define SQLITE_CORRUPT 11 /* The database disk image is malformed */
#define SQLITE_NOTFOUND 12 /* Unknown opcode in sqlite3_file_control() */
#define SQLITE_FULL 13 /* Insertion failed because database is full */
#define SQLITE_CANTOPEN 14 /* Unable to open the database file */
#define SQLITE_PROTOCOL 15 /* Database lock protocol error */
#define SQLITE_EMPTY 16 /* Database is empty */
** this file-control is NULL. However, if the database file is being synced
** as part of a multi-database commit, the argument points to a nul-terminated
** string containing the transactions master-journal file name. VFSes that
** do not need this signal should silently ignore this opcode. Applications
** should not call [sqlite3_file_control()] with this opcode as doing so may
** disrupt the operation of the specialized VFSes that do require it.
**
** <li>[[SQLITE_FCNTL_COMMIT_PHASETWO]]
** The [SQLITE_FCNTL_COMMIT_PHASETWO] opcode is generated internally by SQLite
** and sent to the VFS after a transaction has been committed immediately
** but before the database is unlocked. VFSes that do not need this signal
** should silently ignore this opcode. Applications should not call
** [sqlite3_file_control()] with this opcode as doing so may disrupt the
** operation of the specialized VFSes that do require it.
**
** <li>[[SQLITE_FCNTL_WIN32_AV_RETRY]]
** ^The [SQLITE_FCNTL_WIN32_AV_RETRY] opcode is used to configure automatic
** retry counts and intervals for certain disk I/O operations for the
** windows [VFS] in order to provide robustness in the presence of
** anti-virus programs. By default, the windows VFS will retry file read,
** file write, and file delete operations up to 10 times, with a delay
** <ul>
** <li> SQLITE_SHM_LOCK | SQLITE_SHM_SHARED
** <li> SQLITE_SHM_LOCK | SQLITE_SHM_EXCLUSIVE
** <li> SQLITE_SHM_UNLOCK | SQLITE_SHM_SHARED
** <li> SQLITE_SHM_UNLOCK | SQLITE_SHM_EXCLUSIVE
** </ul>
**
** When unlocking, the same SHARED or EXCLUSIVE flag must be supplied as
** was given on the corresponding lock.
**
** The xShmLock method can transition between unlocked and SHARED or
** between unlocked and EXCLUSIVE. It cannot transition between SHARED
** and EXCLUSIVE.
*/
#define SQLITE_SHM_UNLOCK 1
#define SQLITE_SHM_LOCK 2
#define SQLITE_SHM_SHARED 4
#define SQLITE_SHM_EXCLUSIVE 8
/*
** CAPI3REF: Maximum xShmLock index
**
SQLITE_API int sqlite3_complete16(const void *sql);
/*
** CAPI3REF: Register A Callback To Handle SQLITE_BUSY Errors
** KEYWORDS: {busy-handler callback} {busy handler}
**
** ^The sqlite3_busy_handler(D,X,P) routine sets a callback function X
** that might be invoked with argument P whenever
** an attempt is made to access a database table associated with
** [database connection] D when another thread
** or process has the table locked.
** The sqlite3_busy_handler() interface is used to implement
** [sqlite3_busy_timeout()] and [PRAGMA busy_timeout].
**
** ^If the busy callback is NULL, then [SQLITE_BUSY]
** is returned immediately upon encountering the lock. ^If the busy callback
** is not NULL, then the callback might be invoked with two arguments.
**
** ^The first argument to the busy handler is a copy of the void* pointer which
** is the third argument to sqlite3_busy_handler(). ^The second argument to
** the busy handler callback is the number of times that the busy handler has
**
** The presence of a busy handler does not guarantee that it will be invoked
** when there is lock contention. ^If SQLite determines that invoking the busy
** handler could result in a deadlock, it will go ahead and return [SQLITE_BUSY]
** to the application instead of invoking the
** busy handler.
** Consider a scenario where one process is holding a read lock that
** it is trying to promote to a reserved lock and
** a second process is holding a reserved lock that it is trying
** to promote to an exclusive lock. The first process cannot proceed
** because it is blocked by the second and the second process cannot
** proceed because it is blocked by the first. If both processes
** invoke the busy handlers, neither will make any progress. Therefore,
** SQLite returns [SQLITE_BUSY] for the first process, hoping that this
** will induce the first process to release its read lock and allow
** the second process to proceed.
**
** ^The default busy callback is NULL.
**
** ^(There can only be a single busy handler defined for each
** [database connection]. Setting a new busy handler clears any
** previously set handler.)^ ^Note that calling [sqlite3_busy_timeout()]
**
** A busy handler must not close the database connection
** or [prepared statement] that invoked the busy handler.
*/
SQLITE_API int sqlite3_busy_handler(sqlite3*, int(*)(void*,int), void*);
/*
** CAPI3REF: Set A Busy Timeout
**
** ^This routine sets a [sqlite3_busy_handler | busy handler] that sleeps
** for a specified amount of time when a table is locked. ^The handler
** will sleep multiple times until at least "ms" milliseconds of sleeping
** have accumulated. ^After at least "ms" milliseconds of sleeping,
** the handler returns 0 which causes [sqlite3_step()] to return
** [SQLITE_BUSY].
**
** ^Calling this routine with an argument less than or equal to zero
** turns off all busy handlers.
**
** ^(There can only be a single busy handler for a particular
** [database connection] at any given moment. If another busy handler
** CAPI3REF: Online Backup API.
**
** The backup API copies the content of one database into another.
** It is useful either for creating backups of databases or
** for copying in-memory databases to or from persistent files.
**
** See Also: [Using the SQLite Online Backup API]
**
** ^SQLite holds a write transaction open on the destination database file
** for the duration of the backup operation.
** ^The source database is read-locked only while it is being read;
** it is not locked continuously for the entire backup operation.
** ^Thus, the backup may be performed on a live source database without
** preventing other database connections from
** reading or writing to the source database while the backup is underway.
**
** ^(To perform a backup operation:
** <ol>
** <li><b>sqlite3_backup_init()</b> is called once to initialize the
** backup,
** <li><b>sqlite3_backup_step()</b> is called one or more times to transfer
** the data between the two databases, and finally
** errors are considered fatal.)^ The application must accept
** that the backup operation has failed and pass the backup operation handle
** to the sqlite3_backup_finish() to release associated resources.
**
** ^The first call to sqlite3_backup_step() obtains an exclusive lock
** on the destination file. ^The exclusive lock is not released until either
** sqlite3_backup_finish() is called or the backup operation is complete
** and sqlite3_backup_step() returns [SQLITE_DONE]. ^Every call to
** sqlite3_backup_step() obtains a [shared lock] on the source database that
** lasts for the duration of the sqlite3_backup_step() call.
** ^Because the source database is not locked between calls to
** sqlite3_backup_step(), the source database may be modified mid-way
** through the backup process. ^If the source database is modified by an
** external process or via a database connection other than the one being
** used by the backup operation, then the backup will be automatically
** restarted by the next call to sqlite3_backup_step(). ^If the source
** database is modified by the using the same database connection as is used
** by the backup operation, then the backup database is automatically
** updated at the same time.
**
** [[sqlite3_backup_finish()]] <b>sqlite3_backup_finish()</b>
** ^This API may be used to register a callback that SQLite will invoke
** when the connection currently holding the required lock relinquishes it.
** ^This API is only available if the library was compiled with the
** [SQLITE_ENABLE_UNLOCK_NOTIFY] C-preprocessor symbol defined.
**
** See Also: [Using the SQLite Unlock Notification Feature].
**
** ^Shared-cache locks are released when a database connection concludes
** its current transaction, either by committing it or rolling it back.
**
** ^When a connection (known as the blocked connection) fails to obtain a
** shared-cache lock and SQLITE_LOCKED is returned to the caller, the
** identity of the database connection (the blocking connection) that
** has locked the required resource is stored internally. ^After an
** application receives an SQLITE_LOCKED error, it may call the
** sqlite3_unlock_notify() method with the blocked connection handle as
** the first argument to register for a callback that will be invoked
** when the blocking connections current transaction is concluded. ^The
** callback is invoked from within the [sqlite3_step] or [sqlite3_close]
** call that concludes the blocking connections transaction.
**
** ^(If sqlite3_unlock_notify() is called in a multi-threaded application,
** there is a chance that the blocking connection will have already
** concluded its transaction by the time sqlite3_unlock_notify() is invoked.
** If this happens, then the specified callback is invoked immediately,
** from within the call to sqlite3_unlock_notify().)^
**
** ^If the blocked connection is attempting to obtain a write-lock on a
** shared-cache table, and more than one other connection currently holds
** a read-lock on the same table, then SQLite arbitrarily selects one of
** the other connections to use as the blocking connection.
**
** ^(There may be at most one unlock-notify callback registered by a
** blocked connection. If sqlite3_unlock_notify() is called when the
** blocked connection already has a registered unlock-notify callback,
** then the new callback replaces the old.)^ ^If sqlite3_unlock_notify() is
** called with a NULL pointer as its second argument, then any existing
** unlock-notify callback is canceled. ^The blocked connections
** unlock-notify callback may also be canceled by closing the blocked
** connection using [sqlite3_close()].
**
** The unlock-notify callback is not reentrant. If an application invokes
** any sqlite3_xxx API functions from within an unlock-notify callback, a
** crash or deadlock may be the result.
**
** ^Unless deadlock is detected (see below), sqlite3_unlock_notify() always
** returns SQLITE_OK.
**
** <b>Callback Invocation Details</b>
**
** When an unlock-notify callback is registered, the application provides a
** single void* pointer that is passed to the callback when it is invoked.
** However, the signature of the callback function allows SQLite to pass
** it an array of void* context pointers. The first argument passed to
** an unlock-notify callback is a pointer to an array of void* pointers,
** and the second is the number of entries in the array.
**
** When a blocking connections transaction is concluded, there may be
** more than one blocked connection that has registered for an unlock-notify
** callback. ^If two or more such blocked connections have specified the
** same callback function, then instead of invoking the callback function
** multiple times, it is invoked once with the set of void* context pointers
** specified by the blocked connections bundled together into an array.
** This gives the application an opportunity to prioritize any actions
** related to the set of unblocked database connections.
**
** <b>Deadlock Detection</b>
**
** Assuming that after registering for an unlock-notify callback a
** database waits for the callback to be issued before taking any further
** action (a reasonable assumption), then using this API may cause the
** application to deadlock. For example, if connection X is waiting for
** connection Y's transaction to be concluded, and similarly connection
** Y is waiting on connection X's transaction, then neither connection
** will proceed and the system may remain deadlocked indefinitely.
**
** To avoid this scenario, the sqlite3_unlock_notify() performs deadlock
** detection. ^If a given call to sqlite3_unlock_notify() would put the
** system in a deadlocked state, then SQLITE_LOCKED is returned and no
** unlock-notify callback is registered. The system is said to be in
** a deadlocked state if connection A has registered for an unlock-notify
** callback on the conclusion of connection B's transaction, and connection
** B has itself registered for an unlock-notify callback when connection
** A's transaction is concluded. ^Indirect deadlock is also detected, so
** the system is also considered to be deadlocked if connection B has
** registered for an unlock-notify callback on the conclusion of connection
** C's transaction, where connection C is waiting on connection A. ^Any
** number of levels of indirection are allowed.
**
** <b>The "DROP TABLE" Exception</b>
**
** When a call to [sqlite3_step()] returns SQLITE_LOCKED, it is almost
** always appropriate to call sqlite3_unlock_notify(). There is however,
** one exception. When executing a "DROP TABLE" or "DROP INDEX" statement,
** SQLite checks if there are any currently executing SELECT statements
** invoked immediately. If the application then re-attempts the "DROP TABLE"
** or "DROP INDEX" query, an infinite loop might be the result.
**
** One way around this problem is to check the extended error code returned
** by an sqlite3_step() call. ^(If there is a blocking connection, then the
** extended error code is set to SQLITE_LOCKED_SHAREDCACHE. Otherwise, in
** the special "DROP TABLE/INDEX" case, the extended error code is just
** SQLITE_LOCKED.)^
*/
SQLITE_API int sqlite3_unlock_notify(
sqlite3 *pBlocked, /* Waiting connection */
void (*xNotify)(void **apArg, int nArg), /* Callback function to invoke */
void *pNotifyArg /* Argument to pass to xNotify */
);
/*
** CAPI3REF: String Comparison
**
** ^The [sqlite3_stricmp()] and [sqlite3_strnicmp()] APIs allow applications
** and extensions to compare the contents of two buffers containing UTF-8
t/08_busy.t view on Meta::CPAN
dbfile => 'foo',
RaiseError => 1,
PrintError => 0,
AutoCommit => 0,
);
my $dbfile = dbfile('foo');
# NOTE: Let's make it clear what we're doing here.
# $dbh starts locking with the first INSERT statement.
# $dbh2 tries to INSERT, but as the database is locked,
# it starts waiting. However, $dbh won't release the lock.
# Eventually $dbh2 gets timed out, and spits an error, saying
# the database is locked. So, we don't need to let $dbh2 wait
# too much here. It should be timed out anyway.
ok($dbh2->$call_func(300, 'busy_timeout'));
ok($dbh->do("CREATE TABLE Blah ( id INTEGER, val VARCHAR )"));
ok($dbh->commit);
ok($dbh->do("INSERT INTO Blah VALUES ( 1, 'Test1' )"));
eval {
$dbh2->do("INSERT INTO Blah VALUES ( 2, 'Test2' )");
};
ok($@);
t/08_busy.t view on Meta::CPAN
$dbh->commit;
ok($dbh2->do("INSERT INTO Blah VALUES ( 2, 'Test2' )"));
$dbh2->commit;
$dbh2->disconnect;
undef($dbh2);
# NOTE: The second test is to see what happens if a lock is
# is released while waiting. When both parent and child are
# ready, the database is locked by the child. The parent
# starts waiting for a long enough time (apparently we need
# to wait much longer than we expected, as testers may use
# very slow (virtual) machines to test, but don't worry,
# it's only for the slowest environment). After a short sleep,
# the child commits and releases the lock. Eventually the parent
# notices that, and does the pended INSERT (hopefully before
# it is timed out). As both the parent and the child wait till
# both are ready, we don't need to sleep for a long time.
pipe(READER, WRITER);
my $pid = fork;
t/47_execute.t view on Meta::CPAN
dbfile => 'foo',
RaiseError => 1,
PrintError => 0,
AutoCommit => 0,
);
my $dbfile = dbfile('foo');
# NOTE: Let's make it clear what we're doing here.
# $dbh starts locking with the first INSERT statement.
# $dbh2 tries to INSERT, but as the database is locked,
# it starts waiting. However, $dbh won't release the lock.
# Eventually $dbh2 gets timed out, and spits an error, saying
# the database is locked. So, we don't need to let $dbh2 wait
# too much here. It should be timed out anyway.
ok($dbh->$call_func(300, 'busy_timeout'));
ok($dbh2->$call_func(300, 'busy_timeout'));
$dbh->do("CREATE TABLE Blah ( id INTEGER )");
$dbh->do("INSERT INTO Blah VALUES ( 1 )");
$dbh->commit;
my $sth;
ok($sth = $dbh->prepare("SELECT id FROM Blah"));
$sth->execute;
( run in 1.083 second using v1.01-cache-2.11-cpan-49f99fa48dc )