EARs in the wild

Abstract

Execution After Redirect vulnerabilities-logic flaws in web applications where unintended code is executed after a redirect-have received little attention from the research community. In fact, we found a research paper that incorrectly modeled the redirect semantics, causing their static analysis to miss EAR vulnerabilities.To understand the breadth and scope of EARs in the real world, we performed a large-scale analysis to determine the prevalence of EARs on the Internet. We crawled 8,097,283 URLs from 255,957 domains. We employ a black-box approach that finds EARs which manifest themselves by information leakage in the HTTP redirect response. For this type of EAR, we developed a classification system that discovered 2,173 security-critical EARs among 416 domains. This result shows that EARs are a serious and prevalent problem on the Internet today and deserve future research attention.