Type Abstraction for Relaxed Noninterference * - Archive ouverte HAL Access content directly
Conference Papers Year : 2017

Type Abstraction for Relaxed Noninterference *


Information-flow security typing statically prevents confidential information to leak to public channels. The fundamental information flow property, known as noninterference, states that a public observer cannot learn anything from private data. As attractive as it is from a theoretical viewpoint, noninterference is impractical: real systems need to intentionally declassify some information, selectively. Among the different information flow approaches to declassification, a particularly expressive approach was proposed by Li and Zdancewic, enforcing a notion of relaxed noninterference by allowing programmers to specify declassification policies that capture the intended manner in which public information can be computed from private data. This paper shows how we can exploit the familiar notion of type abstraction to support expressive declassification policies in a simpler, yet more expressive manner. In particular, the type-based approach to declassification—which we develop in an object-oriented setting—addresses several issues and challenges with respect to prior work, including a simple notion of label ordering based on subtyping, support for recursive declassification policies, and a local, modular reasoning principle for relaxed noninterference. This work paves the way for integrating declassification policies in practical security-typed languages.
Fichier principal
Vignette du fichier
LIPIcs-ECOOP-2017-7(1).pdf (668.11 Ko) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-01637023 , version 1 (17-11-2017)
hal-01637023 , version 2 (12-02-2019)



Raimil Cruz, Tamara Rezk, Bernard P. Serpette, Éric Tanter. Type Abstraction for Relaxed Noninterference *. ECOOP 2017 - 31st European Conference on Object-Oriented Programming, Jun 2017, Barcelona, Spain. pp.1-27, ⟨10.4230/LIPIcs.ECOOP.2017.53⟩. ⟨hal-01637023v2⟩
359 View
803 Download



Gmail Mastodon Facebook X LinkedIn More