A Datalog Semantics for Paralocks

Paper i proceeding, 2013

Broberg and Sands (POPL’10) introduced a logic-based policy language, Paralocks, suitable for static information-flow control in programs. Although Paralocks comes with a precise information-flow semantics for programs, the logic-based semantics of policies, describing how policies are combined and compared, is less well developed. This makes the algorithms for policy comparison and computation ad-hoc, and their security guarantees less intuitive. In this paper we provide a new semantics for Paralocks policies based on Datalog. By doing so we are able to show that the ad-hoc semantics from earlier work coincides with the natural Datalog interpretation. Furthermore we show that by having a Datalog-inspired semantics, we can borrow language extensions and algorithms from Datalog for the benefit of Paralocks. We explore how these extensions and algorithms interact with the design and implementation of Paragon, a language combining Paralocks with Java.