program registration submissions committee lodging portland

Efficient Software Model Checking of Data Structure Properties

Efficient Software Model Checking of Data Structure Properties

Research Paper

Thursday, Oct 26, from 13:30 to 15:00

This paper presents novel language and analysis techniques that significantly speed up software model checking of data structure properties. Consider checking a red-black tree implementation. Traditional software model checkers systematically generate all red-black tree states (within some given bounds) and check every red-black tree operation (such as insert, delete, or lookup) on every red-black tree state. Our key idea is as follows. As our checker checks a red-black tree operation $o$ on a red-black tree state s, it uses program analysis techniques to identify other red-black tree states s1, s2, ..., sk on which the operation o behaves similarly. Our analyses guarantee that if o executes correctly on s, then o will execute correctly on every s'i. Our checker therefore does not need to check o on any s'i once it checks o on s. It thus safely prunes those state transitions from its search space, while still achieving complete test coverage within the bounded domain. Our preliminary results show orders of magnitude improvement over previous approaches. We believe our techniques can make software model checking significantly faster, and thus enable checking of much larger programs and complex program properties than currently possible.

Paul Darga, University of Michigan
Chandrasekhar Boyapati, University of Michigan

Research Papers in the same session
Related Onward! Papers
Related Panels
Related Practitioner Reports
Related Research Papers
Related Tutorials
Related Workshops

While Space Available
program registration submissions committee lodging portland
For comments and questions about the web site
please contact us at
© 2005 OOPSLA