Function Renaming in Reverse Engineering of Embedded Device Firmware with ChatGPT
This program is tentative and subject to change.
Firmware reverse engineering is essential for uncovering internal mechanisms and identifying security vulnerabilities. While restoring the structural elements of code is achievable, the absence of function names significantly complicates the process of analyzing and understanding firmware logic. Recognizing the capabilities of large language models (LLMs) in code generation, this paper explores their potential in automating the renaming of functions. We introduce FirmNamer, a prototype system designed to streamline the labor-intensive task of analyzing decompiled code and assigning logical function names. FirmNamer achieves this by constructing dynamic prompts for LLMs based on extracted function code and contextual information. Through extensive evaluation, FirmNamer demonstrates superior performance in function renaming, achieving functional precision and semantic precision of 86.6% and 49%, respectively, outperforming existing state-of-the-art approaches such as DeGPT, DEBIN, NFRE, NERO, and SYMLM.
This program is tentative and subject to change.
Wed 15 OctDisplayed time zone: Perth change
13:40 - 15:20 | LLMs for Program Analysis and Verification ILMPL at Orchid East Chair(s): Guannan Wei Tufts University | ||
13:40 15mTalk | Function Renaming in Reverse Engineering of Embedded Device Firmware with ChatGPT LMPL Puzhuo Liu Ant Group & Tsinghua University, Peng Di Ant Group & UNSW Sydney, Yu Jiang Tsinghua University | ||
13:55 15mTalk | Enhancing Semantic Understanding in Pointer Analysis Using Large Language Models LMPL Baijun Cheng Peking University, Kailong Wang Huazhong University of Science and Technology, Ling Shi Nanyang Technological University, Haoyu Wang Huazhong University of Science and Technology, Yao Guo Peking University, Ding Li Peking University, Xiangqun Chen Peking University | ||
14:10 15mTalk | Improving SAST Detection Capability with LLMs and Enhanced DFA LMPL Yuan Luo Tencent Security Yunding Lab, Zhaojun Chen Tencent Security Yunding Lab, Yuxin Dong Peking University, Haiquan Zhang Tencent Security Yunding Lab, Yi Sun Tencent Security Yunding Lab, Fei Xie Tencent Security Yunding Lab, Zhiqiang Dong Tencent Security Yunding Lab | ||
14:25 15mTalk | ClearAgent: Agentic Binary Analysis for Effective Vulnerability Detection LMPL Xiang Chen The Hong Kong University of Science and Technology, Anshunkang Zhou The Hong Kong University of Science and Technology, Chengfeng Ye The Hong Kong University of Science and Technology, Charles Zhang The Hong Kong University of Science and Technology | ||
14:40 15mTalk | CG-Bench: Can Language Models Assist Call Graph Construction in the Real World? LMPL Ting Yuan , Wenrui Zhang Huawei Technologies Co., Ltd, Dong Chen Huawei, Jie Wang Huawei Technologies Co., Ltd Pre-print | ||
14:55 20mTalk | Beyond Static Pattern Matching? Rethinking Automatic Cryptographic API Misuse Detection in the Era of LLMs LMPL |