LiSTT: An Investigation into Unsound-Incomplete Yet Practical Result Yielding Static Taintflow Analysis

Vulnerability analysis is an important component of software assurance practices. One of its most challenging issues is to find software flaws that could be exploited by malicious users. A necessary condition is the existence of some tainted information flow between tainted input sources and vulnerable functions. Finding the existence of such a taint flow dynamically is an expensive and nondeterministic process. On the other hand, though static analysis may explore (theoretically) all the tainted paths, scalability is an issue, especially in the view of complete- and soundness. In this paper, we explore the possibilities of making static analysis scalable, by compromising its complete- and soundness properties and yet making it effective in detecting taint flows that lead to vulnerability exploitation. This technique is based on a combination of call graph slicing and data-flow analysis. A prototype tool has been developed, and we give experimental results showing that this approach is effective on large applications.