Minkyu Jung (KAIST), Soomin Kim (KAIST), HyungSeok Han (KAIST), Jaeseung Choi (KAIST), Sang Kil Cha (KAIST)

Current binary analysis research focuses mainly on the back-end, but not on the front-end. However, we note that there are several key design points in the front-end that can greatly improve the efficiency of binary analyses. To demonstrate our idea, we design and implement B2R2, a new binary analysis platform that is fast with regard to lifting binary code and evaluating the corresponding IR. Our platform is written purely in F#, a functional programming language, without any external dependencies. Thus, it naturally supports pure parallelism. B2R2’s IR embeds metadata in its language for speeding up dataflow analyses, and it is designed to be efficient for evaluation. Therefore, any binary analysis technique can benefit from our IR design. We discuss our design decisions to build an efficient binary analysis front-end, and summarize lessons learned. We also make our source code public on GitHub.

View More Papers

Investigating Graph Embedding Neural Networks with Unsupervised Features Extraction...

Luca Massarelli (Sapienza University of Rome), Giuseppe A. Di Luna (CINI - National Laboratory of Cybersecurity), Fabio Petroni (Independent Researcher), Leonardo Querzoni (Sapienza University of Rome), Roberto Baldoni (Italian Presidency of Ministry Council)

Read More

BAR2019 Keynote Talk

Dustin Fraze, Program Manager, DARPA I2O

Read More

Evaluating Disassembly Ground Truth Through Dynamic Tracing (abstract)

Lambang Akbar (National University of Singapore), Yuancheng Jiang (National University of Singapore), Roland H.C. Yap (National University of Singapore), Zhenkai Liang (National University of Singapore), Zhuohao Liu (National University of Singapore)

Read More