Control Dependencies in Interpretive Systems

Interpreters and just-in-time (JIT) compilers are ubiquitous in modern computer systems, making it important to have good program analyses for reasoning about such systems. Control dependence, which plays a fundamental role in a number of program analyses, is an important contender in this regard. Existing algorithms for (dynamic) control dependence analysis do not take into account some important runtime characteristics of interpretive computations, and as a result produce results that may be imprecise and/or unsound. This paper describes a new notion of control dependence and analysis algorithm for interpretive systems. This significantly improves dynamic control dependence information, with corresponding improvements in client analyses such as dynamic program slicing and reverse engineering. To the best of our knowledge, this is the first proposal to reason about low-level dynamic control dependencies in interpretive systems in the presence of dynamic code generation and optimization.