Isn't the problem with fsync() that you can't wait for specific IO operations? You're waiting for every single operation on the file descriptor. If anything happens between your write and the call to fsync(), you're going to get delayed by the additional operation that managed to get through.
Modern storage hardware has multiple IO queues with operations in flight at any given time, so I'm not sure to what extent this applies. It may be that only operations that are directly relevant to any given range of data need to be waited for.
