Bill Marczak, Nicholas Weaver with Jakub Dalek, Roya Ensafi, David Fiflield, Sarah McKune, Arn Rey, John Scot-Railton, Ronald Deibert, Vern Paxson; China’s Great Cannon; Citizen Lab, Munk School fo Global Affairs; University of Toronto; 2015-04; 19 pages.
A cookie can contain a “secure” flag, indicating that it should be only sent over an HTTPS connection. Yet there is no corresponding flag to indicate how a cookie was set: attackers who act as a man-in-the-middle even temporarily on an HTTP session can inject cookies which will be attached to subsequent HTTPS connections. Similar attacks can also be launched by a web attacker from a related domain. Although an acknowledged threat, it has not yet been studied thoroughly. This paper aims to fill this gap with an in-depth empirical assessment of cookie injection attacks. We find that cookie-related vulnerabilities are present in important sites (such as Google and Bank of America), and can be made worse by the implementation weaknesses we discovered in major web browsers (such as Chrome, Firefox, and Safari). Our successful attacks have included privacy violation, online victimization, and even financial loss and account hijacking. We also discuss mitigation strategies such as HSTS, possible browser changes, and present a proof-of-concept browser extension to provide better cookie isolation between HTTP and HTTPS, and between related domains.
804060 – Cookies set via HTTP requests may be used to bypass HTTPS and reveal private information; CERT
Vern Paxson, University of California, Berkeley, and International Computer Science Institute
Mihai Christodorescu, Qualcomm Research
Mobin Javed, University of California, Berkeley
Josyula Rao, Reiner Sailer, Douglas Lee Schales, and Marc Ph. Stoecklin, IBM Research
Kurt Thomas, University of California, Berkeley
Wietse Venema, IBM Research
Nicholas Weaver, International Computer Science Institute and University of California, San Diego
DNS queries represent one of the most common forms of network traffic, and likely the least blocked by sites. As such, DNS provides a highly attractive channel for attackers who wish to communicate surreptitiously across a network perimeter, and indeed a variety of tunneling toolkits exist. We develop a novel measurement procedure that fundamentally limits the amount of information that a domain can receive surreptitiously through DNS queries to an upper bound specified by a site’s security policy, with the exact setting representing a tradeoff between the scope of potential leakage versus the quantity of possible detections that a site’s analysts must investigate.
Rooted in lossless compression, our measurement procedure is free from false negatives. For example, we address conventional tunnels that embed the payload in the query names, tunnels that repeatedly query a fixed alphabet of domain names or varying query types, tunnels that embed information in query timing, and communication that employs combinations of these. In an analysis of 230 billion lookups from real production networks, our procedure detected 59 confirmed tunnels. For the enterprise datasets with lookups by individual clients, detecting surreptitious communication that exceeds 4 kB/day imposes an average analyst burden of 1–2 investigations/week.