Scholarlink

Root-Cause Analysis (RCA) is crucial for finding software security flaws from fuzzing results. However, automating this process is considered challenging. In particular, it is well known that statistical RCA techniques are quite slow, frequently requiring hours to thoroughly examine a crash. In this paper, we propose a LOC-based instruction selection method for statistical root cause analysis. Our method improves both the effectiveness and the efficiency of AURORA, a novel binary-only method for automated RCA. LOC-based outperforms AURORA on most of the targets. Especially, our LOC-based method is much faster and more effective than AURORA on all complex, time-consuming targets.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

LOC-Based Instruction Selection for Statistical Root Cause Analysis

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Duc Anh Bui,
  • Thanh Le Dinh

摘要

Root-Cause Analysis (RCA) is crucial for finding software security flaws from fuzzing results. However, automating this process is considered challenging. In particular, it is well known that statistical RCA techniques are quite slow, frequently requiring hours to thoroughly examine a crash. In this paper, we propose a LOC-based instruction selection method for statistical root cause analysis. Our method improves both the effectiveness and the efficiency of AURORA, a novel binary-only method for automated RCA. LOC-based outperforms AURORA on most of the targets. Especially, our LOC-based method is much faster and more effective than AURORA on all complex, time-consuming targets.