Automated enforcement for relaxed information release with reference points

Abstract

Language-based information flow security is a promising approach for enforcement of strong security and protection of the data confidentiality for the end-to-end communications. Here, noninterference is the standard and most restricted security property that completely forbids confidential data from being released to public context. Although this baseline property has been extensively enforced in various cases, there are still many programs, which are considered secure enough, violating this property in some way. In order to control the information release in these programs, the predetermined ways should be specified by means of which confidential data can be released. These intentional releases, also called declassifications, are regulated by several more relaxed security properties than noninterference. The security properties for controlled declassification have been developed on different dimensions with declassification goals. However, the mechanisms used to enforce these properties are still unaccommodating, unspecific, and insufficiently studied. In this work, a new security property, the Relaxed Release with Reference Points (R3P), is presented to limit the information that can be declassified in a program. Moreover, a new mechanism using reachability analysis has been proposed for the pushdown system to enforce R3P on programs. In order to show R3P is competent for use, it has been proved that it complies with the well-known prudent principles of declassification, and in addition finds some restrictions on our security policy. The widespread usage, precision, efficiency, and the influencing factors of our enforcement have been evaluated.