In order to better understand memory resource management, I have been reading Kit Colbert and Rajesh Venkatasubramanian's VMworld presentation (TA21 - Understanding "Host" and "Guest" Memory Usage And Related Memory Management Concepts) and would like to know:

Where common memory pages exist across multiple VMs, in order for transparent page sharing to take place, do the pages need to be loaded into the same base memory address on each guest?

Or, to put it another way, can transparent page sharing be used to optimize files which were rebased by the guest OS upon loading?

Also, once the Balloon driver has pinned memory within the guest and passed it back to the host, how does the guest (Windows) OS or application take back this memory when required?

Windows guests, if that makes any difference.

Thanks.