Eslam Elnikety

Anjo Vahldiek-Oberwagner, Eslam Elnikety, Nuno O. Duarte et al.

Isolating sensitive state and data can increase the security and robustness of many applications. Examples include protecting cryptographic keys against exploits like OpenSSL's Heartbleed bug or protecting a language runtime from native libraries written in unsafe languages. When runtime references across isolation boundaries occur relatively infrequently, then conventional page-based hardware isolation can be used, because the cost of kernel- or hypervisor-mediated domain switching is tolerable. However, some applications, such as the isolation of cryptographic session keys in network-facing services, require very frequent domain switching. In such applications, the overhead of kernel- or hypervisor-mediated domain switching is prohibitive. In this paper, we present ERIM, a novel technique that provides hardware-enforced isolation with low overhead on x86 CPUs, even at high switching rates (ERIM's measured overhead is less than 1% for 100,000 switches per second). The key idea is to combine protection keys (MPKs), a feature recently added to x86 that allows protection domain switches in userspace, with binary inspection to prevent circumvention. We show that ERIM can be applied with little effort to new and existing applications, doesn't require compiler changes, can run on a stock Linux kernel, and has low runtime overhead even at high domain switching rates.

Eslam Elnikety, Deepak Garg, Peter Druschel

Data retrieval systems such as online search engines and online social networks must comply with the privacy policies of personal and selectively shared data items, regulatory policies regarding data retention and censorship, and the provider's own policies regarding data use. Enforcing these policies is difficult and error-prone. Systematic techniques to enforce policies are either limited to type-based policies that apply uniformly to all data of the same type, or incur significant runtime overhead. This paper presents Shai, the first system that systematically enforces data-specific policies with near-zero overhead in the common case. Shai's key idea is to push as many policy checks as possible to an offline, ahead-of-time analysis phase, often relying on predicted values of runtime parameters such as the state of access control lists or connected users' attributes. Runtime interception is used sparingly, only to verify these predictions and to make any remaining policy checks. Our prototype implementation relies on efficient, modern OS primitives for sandboxing and isolation. We present the design of Shai and quantify its overheads on an experimental data indexing and search pipeline based on the popular search engine Apache Lucene.