SHF: EAGER: Collaborative Research: Mapping Software Analysis Problems to Efficient and Accurate Constraints

SHF：EAGER：协作研究：将软件分析问题映射到高效、准确的约束

• 批准号: 1449626
• 负责人: Matthew Dwyer
• 金额: $ 7.5万
• 依托单位: University of Nebraska-Lincoln
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2014
• 资助国家: 美国
• 起止时间: 2014-08-01 至 2016-07-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1449626&HistoricalAwards=false
• 关键词: SHF; EAGER; Collaborative; Research; Mapping

Techniques for finding faults in software systems, such as crashes, security vulnerabilities, and deadlocks, have become increasingly powerful over the past two decades. This is due in no small part to the development of efficient automated satisfiability solvers. The interest in applying these solvers to an ever wider class of software analysis applications has pushed solvers to their limits. As a result, analysis developers are currently forced to approximate analysis?s queries to make use of existing solvers. Because of this software analyses can mistakenly diagnose an error, miss reporting a true error, and suffer unnecessarily poor performance. This research seeks to establish accuracy as an important missing dimension of solver support and its success will lead to broader and more cost-effective use of solvers to produce high-quality software.This project is the first to systematically explore and link the accuracy requirements of a software analysis to the accuracy provided by a solver. This project does this by exploring approaches to specify the accuracy requirements of solver clients and detect, recover and report solution accuracy for integer and string constraints. These capabilities are being implemented in an existing solver interface framework, called Green, which is applied to perform symbolic execution of Java programs, using Symbolic Pathfinder. The project will evaluate the extent to this approach simplifies client analysis development, enables clients to use a variety of solvers - even those that do not perfectly match accuracy requirements, and improves analysis performance.

在过去的二十年中，在软件系统（例如崩溃，安全漏洞和僵局）中找到故障的技术已经变得越来越强大。 这在很大程度上归功于有效的自动化可满足器的发展。将这些求解器应用于更广泛的软件分析应用程序的兴趣将求解器推向了限制。结果，目前，分析开发人员被迫近似分析的查询来利用现有求解器。 由于这种软件分析，可能会错误地诊断出错误，因此错过报告真实错误，并且性能不必要。这项研究旨在将准确性确立为求解器支持的重要缺失维度，其成功将导致更广泛，更具成本效益的求解器来生产高质量的软件。该项目是第一个系统地探索并将软件分析的准确性要求与求解器提供的准确性联系起来的项目。该项目通过探索方法来指定求解器客户端的准确性要求，并检测，恢复和报告解决方案的准确性，以实现整数和字符串约束的准确性。 这些功能是在称为green的现有求解器接口框架中实现的，该框架被应用于使用符号pathfinder执行Java程序的符号执行。 该项目将评估这种方法的程度，简化了客户分析的开发，使客户能够使用各种求解器 - 即使是那些无法完全匹配准确性要求的求解器，并提高了分析性能。