First York Doctoral Symposium on Computing

The term ”shape analysis” refers to fully automatic techniques for discovering the shape of data structures used by programs and thereby verifying the memory safety of those programs. Although many shape analysis techniques have been proposed for the past 10 years, the field of shape analysis is still considered to be in an early academic stage; the techniques are not mature enough for being used in practice. One of the major problems is that all the existing shape analysis techniques do not scale. None of those techniques have been applied to programs more than 1000 LOC. In this talk, I will describe our effort for addressing this scalability issue. My talk will focus on two crucial factors for scalability, exploiting the locality of programs and using the join operator in abstract interpretation. This is join work with Oukseh Lee, Cristiano Calcagno, Dino Distefano and Peter O’Hearn.