Graduate Research Talk: A Technique for Recovering Efficiently Computed Dynamic Calling Contexts

Abstract

Runtime systems for bug detection and debugging tools have long used stack traces to provide detailed information about a running program. The cost of computing stack traces is high, but not a concern for a bug detection actions that halt the program. We present an alternative technique for inexpensively computing and maintaining context-sensitive information during long-running dynamic program analyses. We have successfully implemented our technique and will show positive preliminary results.