Curing Blue Pill virtualisation attacks
North Carolina State University researchers develop security software HyperSafe
By Joab Jackson | Published: 09:52, 10 May 2010
Two researchers from North Carolina State University have developed software that they say can protect virtualisation hypervisors from malicious "Blue Pill" rootkit threats.
"HyperSafe enables the hypervisor self-protection from code injection attempts," said Xuxian Jiang, an assistant professor of computer science at NCSU.
Jiang, along with his PhD. student Zhi Wang, developed the software, called HyperSafe, with funding from the US Army Research Office and the National Science Foundation.
He plans to present a paper describing the work, "HyperSafe: A Lightweight Approach to Provide Lifetime Hypervisor Control-Flow Integrity," at the IEEE Symposium On Security And Privacy, which will take place in Oakland, California, on May 18.
Security buffs might remember Jiang's earlier work. Last year, HookSafe, kernel root kit protection software he helped devise, attracted a bit of attention in the security community.
The new software borrows applies some of the kernel protection ideas developed for HookSafe to the hypervisor.
Typically, a hypervisor attack will exploit a vulnerability, such as a buffer overflow, to inject malicious code into hypervisor. In Blue Pill attacks, for instance, a rootkit is installed that can intercept all the calls and redirect them, unbeknownst to the operating system.
"With the lockdown technique in HyperSafe, we can effectively block such attempts," Jiang said. Thus far, none of the commercial virtualisation vendors have offered a cure for Blue Pill-styled attacks, Jiang said.
The software secures hypervisors with two techniques. One is something called "non-bypassable memory lockdown," which secures page table memory in such a way that it can only be altered by the hypervisor administrator. This means that no new executable code can be exerted, nor could existing code be altered.
Secondly, the software also protects the function pointers (also widely known as hooks) in the hypervisor from being compromised. Again, only the hypervisor administrator can alter the settings.
Currently, the HyperSafe prototype works with BitVisor and Xen, though should also be easily ported to other Type-I hypervisors such as VMware ESX and Microsoft Hyper-V. In a production setting, hypervisors would have to be modified to include the protective code.
Although Jiang has no immediate plans for commercialising the technology, he said he would be welcome to working with any of the key players in the market.
