Using hypervisor to provide data secrecy for user applications on a per-page basis

摘要

It is our pleasure to welcome you to the Fourth ACM SIGPLAN/SIGOPS Conference on Virtual Execution Environments (VEE 08), in Seattle, USA. VEE is the premier venue for presenting and learning about new research on all aspects of virtualization, and we hope that strength will be further enhanced with this year's program. >VEE is a unique conference in that it brings together diverse elements of virtual machine research, most notably operating system (OS) virtualization and virtual machines for implementing programming languages (PL). Reflecting the broad scope of the community, VEE this year is being co-located with ASPLOS, a cross-disciplinary conference that spans PL, OS and architecture research. Although VEE has its roots in the PL community, since 2007 it has had two program chairs - one each from the OS and PL communities. This year, for the first time, half of the program committee comes from each of these two communities. Also for the first time, over 50% of the submitted papers (and about two-thirds of the accepted papers) fell in the broad area of OS virtualization. >A total of 57 papers were submitted to VEE this year. This number was lower than for VEE 2007, perhaps because there was a short gap of less than 7 months between the 2007 and 2008 submission deadlines, in order to co-locate VEE 2008 with ASPLOS (VEE 2007 was co-located with PLDI in June). An informal classification of the submissions by the program chairs showed about 31 on OS-virtualization, 25 on PL virtualization (two of which had a significant OS component), and one unrelated to virtualization. This was a significant change from last year, when about two-thirds of the submissions were PL-related. One possible reason is that OS virtualization has recently been growing fast as a field of research, driven in part by strong interest from industry. >The program committee met in person near Chicago's O'Hare airport for a full day to discuss the papers, except for two members who phoned in. Every paper was discussed at least briefly at the meeting. The decision for every paper but one had a clear consensus, an important goal we set beforehand, and the one case that required a vote was not a close call. Overall, 18 papers were accepted, an acceptance rate of 32%. The committee explicitly decided to select papers purely on their individual merits rather than to strive for a balanced program between the OS and PL areas. The result is that twelve of the selected papers are related to OS virtualization and six to PL virtualization.