Lune

USENIX Security2023顶会

Multiview: Finding Blind Spots in Access-Deny Issues Diagnosis

Bingyu Shen, Tianyi Shan, Yuanyuan Zhou

出版方
2023年份
2顶会引用

摘要

Access-deny issues are hard to fix because it implies both availability and security requirements. On one hand, system administrators (sysadmins) need to make a change quickly to enable legitimate access. On the other hand, sysadmins need to make sure the change does not allow excessive access. Fulfilling the second requirement on security is especially challenging because it highly requires the sysadmins' knowledge of the system environments and security context. Blind spots in knowledge and system settings may hinder sysadmins from finding solutions that align with the security context. Insecure fixes can over-grant permissions, which may only get noticed after the security vulnerability gets exploited. This paper aims to help sysadmins reduce blind spots in diagnosis by providing multiple directions to resolve accessdeny issues. We propose a system, called Multiview, that automatically mutates the configurations to explore possible directions to fix the access-deny issue and lets the configuration changes in each direction grant as few permissions as possible. Multiview provides a detailed diagnosis report, including access-control configurations that are related to the denial, possible configuration changes in different directions to allow the request, as well as the impact on the accesscontrol state of the entire system. We conducted a user study to evaluate Multiview with 20 participants on five real-world access-deny issues. Multiview can reduce the percentage of insecure fixes from 44.0% to 2.0% and reduce the diagnosis time by 62.0% on average. We also evaluated Multiview on 112 real-world failure cases from eight different systems and server applications, and it can successfully diagnose 89 of them. Multiview accurately identifies the failure-causing configurations and provides possible directions to each access-deny issue within one minute.

问问这篇 Paper

智能体会读完全文。

Lune 把这篇 Paper 索引到了每一个公式,引用它的顶会 Paper 也一样。你提问,回答直接引用原文。

可以从这些问题问起

智能体调用

Luneget_paper_fulltext

在 Lune 里问

免费开始,无需绑卡

引用它的顶会 Paper2

问问它们各自怎么用它

它引用的顶会 Paper5

相关 Paper

黄昏的海面,两侧是细线勾勒的悬崖