Romain Malmain (EURECOM), Andrea Fioraldi (EURECOM), Aurelien Francillon (EURECOM)

Despite QEMU’s popularity for binary-only fuzzing, the fuzzing community faces challenges like the proliferation of hard-to-maintain QEMU forks and the lack of an up-to-date, flexible framework well-integrated with advanced fuzzing engines. This leads to a gap in emulation-based fuzzing tools that are both maintainable and fuzzing-oriented.

To cope with that, we present LIBAFL QEMU, a library written in Rust that provides an interface for fuzzing-based emulation by wrapping around QEMU, in both system and user mode. We focus on addressing the limitations of existing QEMU forks used in fuzzing by offering a well-integrated, maintainable and up-to-date solution. In this paper, we detail the design, implementation, and practical challenges of LIBAFL QEMU, including its APIs and fuzzing capabilities and we showcase the library’s use in two case studies: fuzzing an Android library and a Windows kernel driver.

We compare the fuzzers written for these 2 targets with the state-of-the-art, AFL++ qemu mode for the Android library, and KAFL for the Windows driver. For the former, we show that LIBAFL QEMU outperforms AFL++ qemu mode both in terms of speed and coverage. For the latter, despite KAFL being built above hardware-based virtualization instead of emulation, we show we can run complex targets such as Windows and still reach comparable performance, with an overhead expected by a software emulator.

View More Papers

DEMASQ: Unmasking the ChatGPT Wordsmith

Kavita Kumari (Technical University of Darmstadt, Germany), Alessandro Pegoraro (Technical University of Darmstadt), Hossein Fereidooni (Technische Universität Darmstadt), Ahmad-Reza Sadeghi (Technical University of Darmstadt)

Read More

Rapid Vulnerability Mitigation with Security Workarounds

Zhen Huang (Pennsylvania State University), Gang Tan (Pennsylvania State University)

Read More

Scrappy: SeCure Rate Assuring Protocol with PrivacY

Kosei Akama (Keio University), Yoshimichi Nakatsuka (ETH Zurich), Masaaki Sato (Tokai University), Keisuke Uehara (Keio University)

Read More

ORL-AUDITOR: Dataset Auditing in Offline Deep Reinforcement Learning

Linkang Du (Zhejiang University), Min Chen (CISPA Helmholtz Center for Information Security), Mingyang Sun (Zhejiang University), Shouling Ji (Zhejiang University), Peng Cheng (Zhejiang University), Jiming Chen (Zhejiang University), Zhikun Zhang (CISPA Helmholtz Center for Information Security and Stanford University)

Read More