os_unix.c

最新的sqlite3.6.2源代码

**    PENDING -> EXCLUSIVE
**
** This routine will only increase a lock.  Use the sqlite3OsUnlock()
** routine to lower a locking level.
*/
static int unixLock(sqlite3_file *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
  ** unixFile, do nothing. Don't use the end_lock: exit path, as
  ** enterMutex() 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
  */
  enterMutex();

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

  /* If some thread using this PID has a lock via a different unixFile*
  ** 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) ){
      int tErrno = errno;
      rc = sqliteErrorFromPosixError(tErrno, SQLITE_IOERR_LOCK);
      if( IS_LOCK_ERROR(rc) ){
        pFile->lastErrno = tErrno;
      }
      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 ){
    int tErrno = 0;
    assert( pLock->cnt==0 );
    assert( pLock->locktype==0 );

    /* Now get the read-lock */
    lock.l_start = SHARED_FIRST;
    lock.l_len = SHARED_SIZE;
    if( (s = fcntl(pFile->h, F_SETLK, &lock))==(-1) ){
      tErrno = errno;
    }
    /* 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 ){
      if( s != -1 ){
        /* This could happen with a network mount */
        tErrno = errno; 
        rc = sqliteErrorFromPosixError(tErrno, SQLITE_IOERR_UNLOCK); 
        if( IS_LOCK_ERROR(rc) ){
          pFile->lastErrno = tErrno;
        }
        goto end_lock;
      }
    }
    if( s==(-1) ){
      rc = sqliteErrorFromPosixError(tErrno, SQLITE_IOERR_LOCK);
      if( IS_LOCK_ERROR(rc) ){
        pFile->lastErrno = tErrno;
      }
    }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) ){
      int tErrno = errno;
      rc = sqliteErrorFromPosixError(tErrno, SQLITE_IOERR_LOCK);
      if( IS_LOCK_ERROR(rc) ){
        pFile->lastErrno = tErrno;
      }
    }
  }
  
  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:
  leaveMutex();
  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(sqlite3_file *id, int locktype){
  struct lockInfo *pLock;
  struct flock lock;
  int rc = SQLITE_OK;
  unixFile *pFile = (unixFile*)id;
  int h;

  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;
  }
  enterMutex();
  h = pFile->h;
  pLock = pFile->pLock;
  assert( pLock->cnt!=0 );
  if( pFile->locktype>SHARED_LOCK ){
    assert( pLock->locktype==pFile->locktype );
    SimulateIOErrorBenign(1);
    SimulateIOError( h=(-1) )
    SimulateIOErrorBenign(0);
    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(h, F_SETLK, &lock)==(-1) ){
        int tErrno = errno;
        rc = sqliteErrorFromPosixError(tErrno, SQLITE_IOERR_RDLOCK);
        if( IS_LOCK_ERROR(rc) ){
          pFile->lastErrno = tErrno;
        }
				goto end_unlock;
      }
    }
    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(h, F_SETLK, &lock)!=(-1) ){
      pLock->locktype = SHARED_LOCK;
    }else{
      int tErrno = errno;
      rc = sqliteErrorFromPosixError(tErrno, SQLITE_IOERR_UNLOCK);
      if( IS_LOCK_ERROR(rc) ){
        pFile->lastErrno = tErrno;
      }
			goto end_unlock;
    }
  }
  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;
      SimulateIOErrorBenign(1);
      SimulateIOError( h=(-1) )
      SimulateIOErrorBenign(0);
      if( fcntl(h, F_SETLK, &lock)!=(-1) ){
        pLock->locktype = NO_LOCK;
      }else{
        int tErrno = errno;
				rc = sqliteErrorFromPosixError(tErrno, SQLITE_IOERR_UNLOCK);
        if( IS_LOCK_ERROR(rc) ){
          pFile->lastErrno = tErrno;
        }
        pLock->cnt = 1;
				goto end_unlock;
      }
    }

    /* 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.
    */
    if( rc==SQLITE_OK ){
      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]);
        }
        sqlite3_free(pOpen->aPending);
        pOpen->nPending = 0;
        pOpen->aPending = 0;
      }
    }
  }
	
end_unlock:
  leaveMutex();
  if( rc==SQLITE_OK ) pFile->locktype = locktype;
  return rc;
}

/*
** This function performs the parts of the "close file" operation 
** common to all locking schemes. It closes the directory and file
** handles, if they are valid, and sets all fields of the unixFile
** structure to 0.
*/
static int closeUnixFile(sqlite3_file *id){
  unixFile *pFile = (unixFile*)id;
  if( pFile ){
    if( pFile->dirfd>=0 ){
      close(pFile->dirfd);
    }
    if( pFile->h>=0 ){
      close(pFile->h);
    }
    OSTRACE2("CLOSE   %-3d\n", pFile->h);
    OpenCounter(-1);
    memset(pFile, 0, sizeof(unixFile));
  }
  return SQLITE_OK;
}

/*
** Close a file.
*/
static int unixClose(sqlite3_file *id){
  if( id ){
    unixFile *pFile = (unixFile *)id;
    unixUnlock(id, NO_LOCK);
    enterMutex();
    if( pFile->pOpen && pFile->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 = pFile->pOpen;
      aNew = sqlite3_realloc(pOpen->aPending, (pOpen->nPending+1)*sizeof(int) );
      if( aNew==0 ){
        /* If a malloc fails, just leak the file descriptor */
      }else{
        pOpen->aPending = aNew;
        pOpen->aPending[pOpen->nPending] = pFile->h;
        pOpen->nPending++;
        pFile->h = -1;
      }
    }
    releaseLockInfo(pFile->pLock);
    releaseOpenCnt(pFile->pOpen);
    closeUnixFile(id);
    leaveMutex();
  }
  return SQLITE_OK;
}


#ifdef SQLITE_ENABLE_LOCKING_STYLE
#pragma mark AFP Support

/*
 ** The afpLockingContext structure contains all afp lock specific state
 */
typedef struct afpLockingContext afpLockingContext;
struct afpLockingContext {
  unsigned long long sharedLockByte;
  const char *filePath;
};

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)

/* 
 ** Return SQLITE_OK on success, SQLITE_BUSY on failure.
 */