Reading into a buffer and writing from it at the same time

Hello,

I'm programming an embedded device which
receives data over USB device and saves it to
a file. Currently it isn't done in an effective way:

     byte[] readBuffer = new byte[_readBufferSize];
     .....
     while (true) {
                    int bytesRead = in.read(readBuffer, offset,
readBuffer.length);
                    if (bytesRead == -1) {
                        break;
                    }
                    out.write(readBuffer, 0, bytesRead);
     }

- because at any point the device
either reads from USB or writes to a file.

I would like to introduce a second thread
which would take over writing to the file.

So I've created this class for the buffer:

    public CyclicBuffer(int len)
    {
        _buf = new byte[len];
        _head = 0;
        // the buffer is empty
        _tail = _head;
    }

    public read(InputStream in) throws IOException
    {
        int nbytes;

/*
 * _buf: [_][_][_][x][x][x][x][_][_][_][_]
 * ^ ^ ^
 * | | |
 * _tail _head _buf.length
 */
        if (_tail <= _head)
            nbytes = in.read(_buf, _head, _buf.length - _head);
/*
 * _buf: [x][x][_][_][_][_][_][_][x][x][x]
 * ^ ^ ^
 * | | |
 * _head _tail _buf.length
 */
        else
            nbytes = in.read(_buf, _head, _tail - _head);

        if (nbytes <= 0)
            return -1;

        // advance to the next free position in buffer
        _head = (_head + nbytes) % _buf.length;
    }

    public write(OutputStream out) throws IOException
    {
        int nbytes;

        if (_tail == _head)
            throw new IllegalStateException();
/*
 * _buf: [_][_][_][x][x][x][x][_][_][_][_]
 * ^ ^ ^
 * | | |
 * _tail _head _buf.length
 */
        else if (_tail < _head)
            nbytes = out.write(_buf, _head, _head - _tail);
/*
 * _buf: [x][x][_][_][_][_][_][_][x][x][x]
 * ^ ^ ^
 * | | |
 * _head _tail _buf.length
 */
        else
            nbytes = out.write(_buf, _head, _buf.length - _tail);

        if (nbytes <= 0)
            return -1;

        // advance to the next occupied position in buffer
        _tail = (_tail + nbytes) % _buf.length;
    }
}

My problem is that I can't wrap my head around -
where should I insert wait() and notify() calls
and on what objects, so that 2 threads can work
with this buffer simultaneously.

Also I wonder what length should I select for
this cyclic buffer, provided that reading and
writing take approximately same amount of
time (I've measured it in profiler) and that
USB data arrives in _readBufferSize chunks.

Thank you for any hints
Alex