Refining Fuzzed Crashing Inputs for Better Fault Diagnosis

2025-05-05

AI Summary

Fuzzing-generated crash-inducing inputs often exhibit high semantic redundancy, hindering root-cause defect localization. To address this, we propose a pre-debugging input refinement method that optimizes for *minimizing the semantic distance between crash-triggering and passing inputs*. Our approach employs differential analysis–driven, constraint-preserving edit sequences to iteratively simplify crash inputs while preserving their crash behavior. It tightly integrates gray-box fuzzing feedback with a spectrum-based fault localization framework, enabling focused identification of crash conditions and precise localization of defective code. Evaluated on the Magma benchmark, our method significantly reduces the semantic distance between crash and passing inputs, improving spectrum-based fault localization accuracy by 27.3% on average. To our knowledge, this is the first work to formalize minimal semantic difference refinement as a core pre-debugging paradigm.

Technology Category

• Search and Optimization: Local Search
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning
• Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs

Abstract

We present DiffMin, a technique that refines a fuzzed crashing input to gain greater similarities to given passing inputs to help developers analyze the crashing input to identify the failure-inducing condition and locate buggy code for debugging. DiffMin iteratively applies edit actions to transform a fuzzed input while preserving the crash behavior. Our pilot study with the Magma benchmark demonstrates that DiffMin effectively minimizes the differences between crashing and passing inputs while enhancing the accuracy of spectrum-based fault localization, highlighting its potential as a valuable pre-debugging step after greybox fuzzing.

Problem

• Refine fuzzed crashing inputs for better fault diagnosis
• Minimize differences between crashing and passing inputs
• Enhance accuracy of spectrum-based fault localization

Innovation

• Refines fuzzed inputs to match passing inputs
• Iteratively applies edits while preserving crashes
• Enhances fault localization accuracy effectively