arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1808.01246cs.PLcs.CR
收藏

证书增强型数据流分析

Certificate Enhanced Data-Flow Analysis

Mohamed Nassim Seghir

摘要:

Proof-carrying-code被提出作为一种解决方案,以确保双方之间的信任关系:(重量级)分析器和(轻量级)检查器。分析器验证给定应用程序与指定属性的一致性,并生成证明分析结果有效性的证书。那么,检查器只需测试证明的一致性,而不是构造证明。我们开始研究这种技术在数据流分析中的适用性。特别是,我们想知道分析器和检查器之间是否存在显著的性能差异。因此,我们开发了一个工具,称为DCert,实现了一个跨过程的上下文和流敏感的数据流分析器和检查器的Android。将我们的工具应用于现实世界的大型应用程序,我们发现检查的速度比验证快8倍。这一重要的时间增益意味着为应用商店上的应用程序配备证书的潜力,这些证书可以在计算和存储资源有限的移动设备上进行检查。我们描述了我们的实施和报告的实验结果。

英文摘要:

Proof-carrying-code was proposed as a solution to ensure a trust relationship between two parties: a (heavyweight) analyzer and a (lightweight) checker. The analyzer verifies the conformance of a given application to a specified property and generates a certificate attesting the validity of the analysis result. It suffices then for the checker just to test the consistency of the proof instead of constructing it. We set out to study the applicability of this technique in the context of data- flow analysis. In particular, we want to know if there is a significant performance difference between the analyzer and the checker. Therefore, we developed a tool, called DCert, implementing an inter-procedural context and flow-sensitive data-flow analyzer and checker for Android. Applying our tool to real-world large applications, we found out that checking can be up to 8 times faster than verification. This important gain in time suggests a potential for equipping applications on app stores with certificates that can be checked on mobile devices which are limited in computation and storage resources. We describe our implementation and report on experimental results.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑