Highly Configurable Systems - October 2016
Public Audience
Purpose: To highlight progress. Information is generally at a higher level which is accessible to the interested public.
PI(s): Jurgen Pfeffer
Co-PI(s): Christian Kastner
1) HARD PROBLEM(S) ADDRESSED (with short descriptions)
Scalability and compositionality
We address scalability of assurances for highly configurable systems with exponentially growing configuration spaces. A compositional analysis of
options will allow to scale the analysis; for this it's important to investigate how options are implemented and how they interact. In addition, modular and timely recertification of changes and variations is essential to make security judgements scale in practice.
2) PUBLICATIONS
G. Ferreira, M. Malik, C. Kastner, J. Pfeffer, and S. Apel. Do #ifdefs Influence the Occurrence of Vulnerabilities? An Empirical Study of the Linux Kernel. In Proceedings of the 20th International Software Product Line Conference (SPLC), New York, NY: ACM Press, September 2016.
Jens Meinicke, Chu-Pan Wong, Christian Kastner, Thomas Thum, Gunter Saake. On Essential Configuration Complexity: Measuring Interactions In Highly-Configurable Systems. In Proceedings Int'l Conf. Automated Software Engineering (ASE). 2016
C. Bogart, C. Kastner, J. Herbsleb, and F. Thung. How to Break an API: Cost Negotiation and Community Values in Three Software Ecosystems. In Proceedings of the ACM SIGSOFT Symposium on the Foundations of Software Engineering (FSE), New York, NY: ACM Press, November 2016.
3) KEY HIGHLIGHTS
* Interviewed 6 additional developers, reviewers, and policy makers regarding software security and safety certification practices using Common Criteria and DO178. Coded and classified issues, preparing discussion regarding security-related recertification and compositional certification.
* Studied attack vectors and mitigation through malicious package updates in software ecosystems of open source software, particularly Node.js/npm.
Measured attack surface and evaluated feasibility of soundly verifying updates of small packages as free of certain attacks.
*With our collaborators at Passau, we started a project to study the social dynamics of developer networks to better understand possible coordination problems or blindspots that cause failures to successfully address vulnerabilities in highly configurable systems. Using data from github and email networks used in the development process, we have so far investigated whether we can explain the growth of developer networks through preferential attachment, replicating a proposed network growth model but using a more rigorous maximum likelihood-based approach for statistical inference. We also have begun to explore specifications in Stochastic Actor-Oriented Models, a class of statistical network models that are on the cutting edge of being able to model the co-evolution of behavior and network structure. These will allow us to consider preferential attachment as a proposed explanation for network growth along with controls for reciprocity, transitivity, and node attributes.