The attack is described in a paper entitled "Cross-VM Side Channels and Their Use to Extract Private Keys," authored by Yinqian Zhang, a PhD. student at the University of North Carolina at Chapel Hill; UNC professor Michael K. Reiter; Thomas Ristenpart, an assistant professor at University of Wisconsin-Madison; and Ari Juels, chief scientist at EMC's RSA security division.
According to the paper (PDF), the group was able to demonstrate an attack in a lab environment that allowed a malicious virtual machine (VM) to extract a private ElGamal decryption key from a co-resident virtual machine running Gnu Privacy Guard, which implements the OpenPGP email encryption standard.
"This attack is the first such attack demonstrated on a symmetric multiprocessing system virtualized using a modern VMM (Xen)," the paper states. "Such systems are very common today, ranging from desktops that use virtualization to sandbox application or OS compromises, to clouds that co-locate the workloads of mutually distrustful customers. Constructing such a side-channel requires overcoming challenges including core migration, numerous sources of channel noise, and the difficulty of preempting the victim with sufficient frequency to extract fine-grained information from it."
"The upshot is that isolation in public clouds is imperfect and can potentially be breached," Juels told Dark Reading. "So highly sensitive workloads should not be placed in a public cloud. Our attack is the first solid confirmation of a long hypothesized attack vector."
The underlying vulnerable code is in the most recent version of the libgcrypt library, according to the report.
"Specifically, we show that the attacker VM's monitoring of a victim's repeated exponentiations over the course of a few hours provides it enough information to reconstruct the victim's 457-bit private exponent accompanying a 4096-bit modulus with very high accuracy—so high that the attacker was then left to search fewer than 10, 000 possible exponents to find the right one," the researchers write.
The attack is fairly elaborate, Juels says. The attack environment is extremely noisy, making the attacker's job similar to reconstructing a conversation while listening through the wall with a stereo blaring on the other side, he explains. But then, the goal was ambitious: to extract a complete cryptographic key.
"To glean other secrets may be easier," he says. "I think it's unlikely that there will be a rash of private-key thefts, but attacks are always subject to improvement and automation. So the bottom line remains: Co-residency is not safe for highly sensitive workloads."
"A key point here is that the attacker doesn't need to compromise the VM she's using," he adds. "She can simply pay for and launch her own VMs in the cloud to mount the attack. In other words, she doesn't need to compromise any software. The harder part is getting the attack VM to sit on the same host as the victim."
Have a comment on this story? Please click "Add Your Comment" below. If you'd like to contact Dark Reading's editors directly, send us a message.