ALEXKIDD-FUZZER: Kernel Fuzzing Guided by Symbolic Information
Primary Investigator:
Research Independant
Kyungtae Kim, Byoungyoung Lee
Abstract
Black-box and white-box fuzzing (i.e. symbolic execution) are both getting popular for software testing. However, both of them have severe limitations that prevent maximizing code coverage.
We design ALEXKIDD-FUZZER, which overcomes limitation of such fuzzing and symbolic execution.
We first employ general fuzzing mechanism such that feasible execution paths are explored at a rapid pace.
Furthermore, during fuzzing execution, we allow concolic engine to guide the fuzzer to make unreachable-code reachable.