os_unix.c

sqlite-3.4.1,嵌入式数据库.是一个功能强大的开源数据库,给学习和研发以及小型公司的发展带来了全所未有的好处.

  int r;
  fd = open(zDirname, O_RDONLY|O_BINARY, 0);
  OSTRACE3("DIRSYNC %-3d (%s)\n", fd, zDirname);
  if( fd<0 ){
    return SQLITE_CANTOPEN; 
  }
  r = fsync(fd);
  close(fd);
  SimulateIOError( r=1 );
  if( r ){
    return SQLITE_IOERR_DIR_FSYNC;
  }else{
    return SQLITE_OK;
  }
#endif
}

/*
** Truncate an open file to a specified size
*/
static int unixTruncate(OsFile *id, i64 nByte){
  int rc;
  assert( id );
  rc = ftruncate(((unixFile*)id)->h, (off_t)nByte);
  SimulateIOError( rc=1 );
  if( rc ){
    return SQLITE_IOERR_TRUNCATE;
  }else{
    return SQLITE_OK;
  }
}

/*
** Determine the current size of a file in bytes
*/
static int unixFileSize(OsFile *id, i64 *pSize){
  int rc;
  struct stat buf;
  assert( id );
  rc = fstat(((unixFile*)id)->h, &buf);
  SimulateIOError( rc=1 );
  if( rc!=0 ){
    return SQLITE_IOERR_FSTAT;
  }
  *pSize = buf.st_size;
  return SQLITE_OK;
}

/*
** 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, return
** non-zero.  If the file is unlocked or holds only SHARED locks, then
** return zero.
*/
static int unixCheckReservedLock(OsFile *id){
  int r = 0;
  unixFile *pFile = (unixFile*)id;

  assert( pFile );
  sqlite3OsEnterMutex(); /* Because pFile->pLock is shared across threads */

  /* Check if a thread in this process holds such a lock */
  if( pFile->pLock->locktype>SHARED_LOCK ){
    r = 1;
  }

  /* Otherwise see if some other process holds it.
  */
  if( !r ){
    struct flock lock;
    lock.l_whence = SEEK_SET;
    lock.l_start = RESERVED_BYTE;
    lock.l_len = 1;
    lock.l_type = F_WRLCK;
    fcntl(pFile->h, F_GETLK, &lock);
    if( lock.l_type!=F_UNLCK ){
      r = 1;
    }
  }
  
  sqlite3OsLeaveMutex();
  OSTRACE3("TEST WR-LOCK %d %d\n", pFile->h, r);

  return r;
}

/*
** Lock the file with the lock specified by parameter locktype - one
** of the following:
**
**     (1) SHARED_LOCK
**     (2) RESERVED_LOCK
**     (3) PENDING_LOCK
**     (4) EXCLUSIVE_LOCK
**
** Sometimes when requesting one lock state, additional lock states
** are inserted in between.  The locking might fail on one of the later
** transitions leaving the lock state different from what it started but
** still short of its goal.  The following chart shows the allowed
** transitions and the inserted intermediate states:
**
**    UNLOCKED -> SHARED
**    SHARED -> RESERVED
**    SHARED -> (PENDING) -> EXCLUSIVE
**    RESERVED -> (PENDING) -> EXCLUSIVE
**    PENDING -> EXCLUSIVE
**
** This routine will only increase a lock.  Use the sqlite3OsUnlock()
** routine to lower a locking level.
*/
static int unixLock(OsFile *id, int locktype){
  /* The following describes the implementation of the various locks and
  ** lock transitions in terms of the POSIX advisory shared and exclusive
  ** lock primitives (called read-locks and write-locks below, to avoid
  ** 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.
  ** This property is used by the algorithm for rolling back a journal file
  ** after a crash.
  **
  ** An EXCLUSIVE lock, obtained after a PENDING lock is held, is
  ** implemented by obtaining a write-lock on the entire 'shared byte
  ** range'. Since all other locks require a read-lock on one of the bytes
  ** within this range, this ensures that no other locks are held on the
  ** database. 
  **
  ** The reason a single byte cannot be used instead of the 'shared byte
  ** range' is that some versions of windows do not support read-locks. By
  ** locking a random byte from a range, concurrent SHARED locks may exist
  ** even if the locking primitive used is always a write-lock.
  */
  int rc = SQLITE_OK;
  unixFile *pFile = (unixFile*)id;
  struct lockInfo *pLock = pFile->pLock;
  struct flock lock;
  int s;

  assert( pFile );
  OSTRACE7("LOCK    %d %s was %s(%s,%d) pid=%d\n", pFile->h,
      locktypeName(locktype), locktypeName(pFile->locktype),
      locktypeName(pLock->locktype), pLock->cnt , getpid());

  /* If there is already a lock of this type or more restrictive on the
  ** OsFile, do nothing. Don't use the end_lock: exit path, as
  ** sqlite3OsEnterMutex() hasn't been called yet.
  */
  if( pFile->locktype>=locktype ){
    OSTRACE3("LOCK    %d %s ok (already held)\n", pFile->h,
            locktypeName(locktype));
    return SQLITE_OK;
  }

  /* Make sure the locking sequence is correct
  */
  assert( pFile->locktype!=NO_LOCK || locktype==SHARED_LOCK );
  assert( locktype!=PENDING_LOCK );
  assert( locktype!=RESERVED_LOCK || pFile->locktype==SHARED_LOCK );

  /* This mutex is needed because pFile->pLock is shared across threads
  */
  sqlite3OsEnterMutex();

  /* Make sure the current thread owns the pFile.
  */
  rc = transferOwnership(pFile);
  if( rc!=SQLITE_OK ){
    sqlite3OsLeaveMutex();
    return rc;
  }
  pLock = pFile->pLock;

  /* If some thread using this PID has a lock via a different OsFile*
  ** handle that precludes the requested lock, return BUSY.
  */
  if( (pFile->locktype!=pLock->locktype && 
          (pLock->locktype>=PENDING_LOCK || locktype>SHARED_LOCK))
  ){
    rc = SQLITE_BUSY;
    goto end_lock;
  }

  /* If a SHARED lock is requested, and some thread using this PID already
  ** has a SHARED or RESERVED lock, then increment reference counts and
  ** return SQLITE_OK.
  */
  if( locktype==SHARED_LOCK && 
      (pLock->locktype==SHARED_LOCK || pLock->locktype==RESERVED_LOCK) ){
    assert( locktype==SHARED_LOCK );
    assert( pFile->locktype==0 );
    assert( pLock->cnt>0 );
    pFile->locktype = SHARED_LOCK;
    pLock->cnt++;
    pFile->pOpen->nLock++;
    goto end_lock;
  }

  lock.l_len = 1L;

  lock.l_whence = SEEK_SET;

  /* A PENDING lock is needed before acquiring a SHARED lock and before
  ** acquiring an EXCLUSIVE lock.  For the SHARED lock, the PENDING will
  ** be released.
  */
  if( locktype==SHARED_LOCK 
      || (locktype==EXCLUSIVE_LOCK && pFile->locktype<PENDING_LOCK)
  ){
    lock.l_type = (locktype==SHARED_LOCK?F_RDLCK:F_WRLCK);
    lock.l_start = PENDING_BYTE;
    s = fcntl(pFile->h, F_SETLK, &lock);
    if( s==(-1) ){
      rc = (errno==EINVAL) ? SQLITE_NOLFS : SQLITE_BUSY;
      goto end_lock;
    }
  }


  /* If control gets to this point, then actually go ahead and make
  ** operating system calls for the specified lock.
  */
  if( locktype==SHARED_LOCK ){
    assert( pLock->cnt==0 );
    assert( pLock->locktype==0 );

    /* Now get the read-lock */
    lock.l_start = SHARED_FIRST;
    lock.l_len = SHARED_SIZE;
    s = fcntl(pFile->h, F_SETLK, &lock);

    /* Drop the temporary PENDING lock */
    lock.l_start = PENDING_BYTE;
    lock.l_len = 1L;
    lock.l_type = F_UNLCK;
    if( fcntl(pFile->h, F_SETLK, &lock)!=0 ){
      rc = SQLITE_IOERR_UNLOCK;  /* This should never happen */
      goto end_lock;
    }
    if( s==(-1) ){
      rc = (errno==EINVAL) ? SQLITE_NOLFS : SQLITE_BUSY;
    }else{
      pFile->locktype = SHARED_LOCK;
      pFile->pOpen->nLock++;
      pLock->cnt = 1;
    }
  }else if( locktype==EXCLUSIVE_LOCK && pLock->cnt>1 ){
    /* We are trying for an exclusive lock but another thread in this
    ** same process is still holding a shared lock. */
    rc = SQLITE_BUSY;
  }else{
    /* The request was for a RESERVED or EXCLUSIVE lock.  It is
    ** assumed that there is a SHARED or greater lock on the file
    ** already.
    */
    assert( 0!=pFile->locktype );
    lock.l_type = F_WRLCK;
    switch( locktype ){
      case RESERVED_LOCK:
        lock.l_start = RESERVED_BYTE;
        break;
      case EXCLUSIVE_LOCK:
        lock.l_start = SHARED_FIRST;
        lock.l_len = SHARED_SIZE;
        break;
      default:
        assert(0);
    }
    s = fcntl(pFile->h, F_SETLK, &lock);
    if( s==(-1) ){
      rc = (errno==EINVAL) ? SQLITE_NOLFS : SQLITE_BUSY;
    }
  }
  
  if( rc==SQLITE_OK ){
    pFile->locktype = locktype;
    pLock->locktype = locktype;
  }else if( locktype==EXCLUSIVE_LOCK ){
    pFile->locktype = PENDING_LOCK;
    pLock->locktype = PENDING_LOCK;
  }

end_lock:
  sqlite3OsLeaveMutex();
  OSTRACE4("LOCK    %d %s %s\n", pFile->h, locktypeName(locktype), 
      rc==SQLITE_OK ? "ok" : "failed");
  return rc;
}

/*
** Lower the locking level on file descriptor pFile to locktype.  locktype
** 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.
*/
static int unixUnlock(OsFile *id, int locktype){
  struct lockInfo *pLock;
  struct flock lock;
  int rc = SQLITE_OK;
  unixFile *pFile = (unixFile*)id;

  assert( pFile );
  OSTRACE7("UNLOCK  %d %d was %d(%d,%d) pid=%d\n", pFile->h, locktype,
      pFile->locktype, pFile->pLock->locktype, pFile->pLock->cnt, getpid());

  assert( locktype<=SHARED_LOCK );
  if( pFile->locktype<=locktype ){
    return SQLITE_OK;
  }
  if( CHECK_THREADID(pFile) ){
    return SQLITE_MISUSE;
  }
  sqlite3OsEnterMutex();
  pLock = pFile->pLock;
  assert( pLock->cnt!=0 );
  if( pFile->locktype>SHARED_LOCK ){
    assert( pLock->locktype==pFile->locktype );
    if( locktype==SHARED_LOCK ){
      lock.l_type = F_RDLCK;
      lock.l_whence = SEEK_SET;
      lock.l_start = SHARED_FIRST;
      lock.l_len = SHARED_SIZE;
      if( fcntl(pFile->h, F_SETLK, &lock)==(-1) ){
        /* This should never happen */
        rc = SQLITE_IOERR_RDLOCK;
      }
    }
    lock.l_type = F_UNLCK;
    lock.l_whence = SEEK_SET;
    lock.l_start = PENDING_BYTE;
    lock.l_len = 2L;  assert( PENDING_BYTE+1==RESERVED_BYTE );
    if( fcntl(pFile->h, F_SETLK, &lock)!=(-1) ){
      pLock->locktype = SHARED_LOCK;
    }else{
      rc = SQLITE_IOERR_UNLOCK;  /* This should never happen */
    }
  }
  if( locktype==NO_LOCK ){
    struct openCnt *pOpen;

    /* Decrement the shared lock counter.  Release the lock using an
    ** OS call only when all threads in this same process have released
    ** the lock.
    */
    pLock->cnt--;
    if( pLock->cnt==0 ){
      lock.l_type = F_UNLCK;
      lock.l_whence = SEEK_SET;
      lock.l_start = lock.l_len = 0L;
      if( fcntl(pFile->h, F_SETLK, &lock)!=(-1) ){
        pLock->locktype = NO_LOCK;
      }else{
        rc = SQLITE_IOERR_UNLOCK;  /* This should never happen */
      }
    }

    /* Decrement the count of locks against this same file.  When the
    ** count reaches zero, close any other file descriptors whose close
    ** was deferred because of outstanding locks.
    */
    pOpen = pFile->pOpen;
    pOpen->nLock--;
    assert( pOpen->nLock>=0 );
    if( pOpen->nLock==0 && pOpen->nPending>0 ){
      int i;
      for(i=0; i<pOpen->nPending; i++){
        close(pOpen->aPending[i]);
      }
      free(pOpen->aPending);
      pOpen->nPending = 0;
      pOpen->aPending = 0;
    }
  }
  sqlite3OsLeaveMutex();
  pFile->locktype = locktype;
  return rc;
}

/*
** Close a file.
*/
static int unixClose(OsFile **pId){
  unixFile *id = (unixFile*)*pId;

  if( !id ) return SQLITE_OK;
  unixUnlock(*pId, NO_LOCK);
  if( id->dirfd>=0 ) close(id->dirfd);
  id->dirfd = -1;
  sqlite3OsEnterMutex();

  if( id->pOpen->nLock ){
    /* If there are outstanding locks, do not actually close the file just
    ** yet because that would clear those locks.  Instead, add the file
    ** descriptor to pOpen->aPending.  It will be automatically closed when
    ** the last lock is cleared.
    */
    int *aNew;
    struct openCnt *pOpen = id->pOpen;
    aNew = realloc( pOpen->aPending, (pOpen->nPending+1)*sizeof(int) );
    if( aNew==0 ){
      /* If a malloc fails, just leak the file descriptor */