EAR: Debugging Libraries through API Call Traces
Debugging software library bugs triggered by client code is challenging, as isolating the execution of the library from that of the client’s is often non-trivial. For instance, in the context of KLEE, crash and performance bugs in constraint solvers can be difficult to isolate and report to solver developers when they get triggered only by a highly specific sequence of API calls along a path explored by KLEE.
To address this problem, we propose an approach and tool called eXact Reproducer Suite (XRS). XRS is capable of recording the precise sequence of API calls made to a target library, together with their arguments and return values. This enables developers to automatically diagnose library bugs and performance issues found during complex client executions. We applied XRS to KLEE and STP, demonstrating its utility in practice.
Fri 16 OctDisplayed time zone: Amsterdam, Berlin, Bern, Rome, Stockholm, Vienna change
10:30 - 10:50 | |||
10:30 5mPoster | Automatic Unit-Test Generation Based on Symbolic Execution KLEE Tigran Avagyan RAU, Levon Shahbazyan Center of Advanced Software Technologies, Russian-Armenian University, Hovhannes Movsisyan RAU, Hayk Aslanyan RAU | ||
10:35 5mPoster | EAR: Debugging Libraries through API Call Traces KLEE | ||
10:40 5mPoster | Interprocedural Symbolic Static Analysis for Go KLEE | ||