Worm containment must be automatic because worms can spread too fast for humans to respond. Recent work proposed network-level techniques to automate worm containment; these techniques have limitations because there is no information about the vulnerabilities exploited by worms at the network level. We propose Vigilante, a new end-to-end architecture to contain worms automatically that addresses these limitations. In Vigilante, hosts detect worms by instrumenting vulnerable programs to analyze infection attempts. We introduce dynamic data-flow analysis: a broad-coverage host-based algorithm that can detect unknown worms by tracking the flow of data from network messages and disallowing unsafe uses of this data. We also show how to integrate other host-based detection mechanisms into the Vigilante architecture. Upon detection, hosts generate self-certifying alerts (SCAs), a new type of security alert that can be inexpensively verified by any vulnerable host. Using SCAs, hosts can cooperate to contain an outbreak, without having to trust each other. Vigilante broadcasts SCAs over an overlay network that propagates alerts rapidly and resiliently. Hosts receiving an SCA protect themselves by generating filters with vulnerability condition slicing: an algorithm that performs dynamic analysis of the vulnerable program to identify control-flow conditions that lead

Attackers and defenders of computer systems both strive to gain complete control over the system. To maximize their control, both attackers and defenders have migrated to low-level, operating system code. In this paper, we assume the perspective of the attacker, who is trying to run malicious software and avoid detection. By assuming this perspective, we hope to help defenders understand and defend against the threat posed by a new class of rootkits. We evaluate a new type of malicious software that gains qualitatively more control over a system. This new type of malware, which we call a virtual-machine based rootkit (VMBR), installs a virtual-machine monitor underneath an existing operating system and hoists the original operating system into a virtual machine. Virtual-machine based rootkits are hard to detect and remove because their state cannot be accessed by software running in the target system. Further, VMBRs support general-purpose malicious services by allowing such services to run in a separate operating system that is protected from the target system. We evaluate this new threat by implementing two proof-of-concept VMBRs. We use our proof-of-concept VMBRs to subvert Windows XP and Linux target systems, and we implement four example malicious services using the VMBR platform. Last, we use what we learn from our proof-of-concept VMBRs to explore ways to defend against this new threat. We discuss possible ways to detect and prevent VMBRs, and we implement a defense strategy suitable for protecting systems against this threat. 1.

Vulnerability-driven filtering of network data can offer a fast and easy-to-deploy alternative or intermediary to software patching, as exemplified in Shield [43]. In this paper, we take Shield’s vision to a new domain, inspecting and cleansing not just static content, but also dynamic content. The dynamic content we target is the dynamic HTML in web pages, which have become a popular vector for attacks. The key challenge in filtering dynamic HTML is that it is undecidable to statically determine whether an embedded script will exploit the browser at run-time. We avoid this undecidability problem by rewriting web pages and any embedded scripts into safe equivalents, inserting checks so that the filtering is done at run-time. The rewritten pages contain logic for recursively applying run-time checks to dynamically generated or modified web content, based on known vulnerabilities. We have built and evaluated BrowserShield, a system that performs this dynamic instrumentation of embedded scripts, and that admits policies for customized run-time actions like vulnerabilitydriven filtering. 1

Spyware is a class of malicious code that is surreptitiously installed on victims ’ machines. Once active, it silently monitors the behavior of users, records their web surfing habits, and steals their passwords. Current anti-spyware tools operate in a way similar to traditional virus scanners. That is, they check unknown programs against signatures associated with known spyware instances. Unfortunately, these techniques cannot identify novel spyware, require frequent updates to signature databases, and are easy to evade by code obfuscation. In this paper, we present a novel dynamic analysis approach that precisely tracks the flow of sensitive information as it is processed by the web browser and any loaded browser helper objects. Using the results of our analysis, we can identify unknown components as spyware and provide comprehensive reports on their behavior. The techniques presented in this paper address limitations of our previous work on spyware detection and significantly improve the quality and richness of our analysis. In particular, our approach allows a human analyst to observe the actual flows of sensitive data in the system. Based on this information, it is possible to precisely determine which sensitive data is accessed and where this data is sent to. To demonstrate the effectiveness of the detection and the comprehensiveness of the generated reports, we evaluated our system on a substantial body of spyware and benign samples. 1

In a virtualized environment, the VMM is the system’s primary resource manager. Some services usually implemented at the OS layer, like I/O scheduling or certain kinds of security monitoring, are therefore more naturally implemented inside the VMM. Implementing such services at the VMM layer can be complicated by the lack of OS and application-level knowledge within a VMM. This paper describes techniques that can be used by a VMM to independently overcome part of the “semantic gap ” separating it from the guest operating systems it supports. These techniques enable the VMM to track the existence and activities of operating system processes. Antfarm is an implementation of these techniques that works without detailed knowledge of a guest’s internal architecture or implementation. An evaluation of Antfarm for two virtualization environments and two operating systems shows that it can accurately infer process events while incurring only a small 2.5 % runtime overhead in the worst case. To demonstrate the practical benefits of process information in a VMM we implement an anticipatory disk scheduler at the VMM level. This case study shows that significant disk throughput improvements are possible in a virtualized environment by exploiting process information within a VMM. 1

Exploits for new vulnerabilities, especially when incorporated within a fast spreading worm, can compromise nearly all vulnerable hosts within a short amount of time. This problem demonstrates the need for fast defenses which can react to a new vulnerability quickly. In addition, a realistic defense system should (a) not require source code since in practice most vulnerable systems do not have source code access nor is there adequate time to involve the software vendor, (b) be accurate, i.e., have a negligible false positive rate and low false negative rate, and (c) be efficient, i.e., add little overhead to normal program execution.

An alarming trend in malware attacks is that they are armed with stealthy techniques to detect, evade, and subvert malware detection facilities of the victim. On the defensive side, a fundamental limitation of traditional host-based anti-malware systems is that they run inside the very hosts they are protecting (“in the box”), making them vulnerable to counter-detection and subversion by malware. To address this limitation, recent solutions based on virtual machine (VM) technologies advocate placing the malware detection facilities outside of the protected VM (“out of the box”). However, they gain tamper resistance at the cost of losing the native, semantic view of the host which is enjoyed by the “in the box” approach, thus leading to a technical challenge known as the semantic gap. In this paper, we present the design, implementation, and evaluation of VMwatcher – an “out-of-the-box” approach that overcomes the semantic gap challenge. A new technique called guest view casting is developed to systematically reconstruct internal semantic views (e.g., files, processes, and kernel modules) of a VM from the outside in a non-intrusive manner. Specifically, the new technique casts semantic definitions of guest OS data structures and functions on virtual machine monitor (VMM)-level VM states, so that the semantic view can be reconstructed. With the semantic gap bridged, we identify two unique malware detection capabilities: (1) view comparison-based malware detection and its demonstration in rootkit detection and (2) “out-of-the-box” deployment of hostbased anti-malware software with improved detection accuracy and tamper-resistance. We have implemented a proof-of-concept prototype on both Linux and Windows platforms and our experimental results with real-world malware, including elusive kernel-level rootkits, demonstrate its practicality and effectiveness.

Abstract. Kernel rootkits pose a significant threat to computer systems as they run at the highest privilege level and have unrestricted access to the resources of their victims. Many current efforts in kernel rootkit defense focus on the detection of kernel rootkits – after a rootkit attack has taken place, while the smaller number of efforts in kernel rootkit prevention exhibit limitations in their capability or deployability. In this paper we present a kernel rootkit prevention system called NICKLE which addresses a common, fundamental characteristic of most kernel rootkits: the need for executing their own kernel code. NICKLE is a lightweight, virtual machine monitor (VMM) based system that transparently prevents unauthorized kernel code execution for unmodified commodity (guest) OSes. NICKLE is based on a new scheme called memory shadowing, wherein the trusted VMM maintains a shadow physical memory for a running VM and performs real-time kernel code authentication so that only authenticated kernel code will be stored in the shadow memory. Further, NICKLE transparently routes guest kernel instruction fetches to the shadow memory at runtime. By doing so, NICKLE guarantees that only the authenticated kernel code will be executed, foiling the kernel rootkit’s attempt to strike in the first place. We have implemented NICKLE in three VMM platforms: QEMU+KQEMU, VirtualBox, and VMware Workstation. Our experiments with 23 real-world kernel rootkits targeting the Linux or Windows OSes demonstrate NICKLE’s effectiveness. Furthermore, our performance evaluation shows that NICKLE introduces small overhead to the VMM platform. 1

Analyzing the behavior of running programs has a wide variety of compelling applications, from intrusion detection and prevention to bug discovery. Unfortunately, the high runtime overheads imposed by complex analysis techniques makes their deployment impractical in most settings. We present a virtual machine based architecture called Aftersight ameliorates this, providing a flexible and practical way to run heavyweight analyses on production workloads. Aftersight decouples analysis from normal execution by logging nondeterministic VM inputs and replaying them on a separate analysis platform. VM output can be gated on the results of an analysis for intrusion prevention or analysis can run at its own pace for intrusion detection and best effort prevention. Logs can also be stored for later analysis offline for bug finding or forensics, allowing analyses that would otherwise be unusable to be applied ubiquitously. In all cases, multiple analyses can be run in parallel, added on demand, and are guaranteed not to interfere with the running workload. We present our experience implementing Aftersight as part of the VMware virtual machine platform and using it to develop a realtime intrusion detection and prevention system, as well as an an offline system for bug detection, which we used to detect numerous novel and serious bugs in VMware ESX Server, Linux, and Windows applications.

Debugging and profiling large-scale distributed applications is a daunting task. We present Friday, a system for debugging distributed applications that combines deterministic replay of components with the power of symbolic, low-level debugging and a simple language for expressing higher-level distributed conditions and actions. Friday allows the programmer to understand the collective state and dynamics of a distributed collection of coordinated application components. To evaluate Friday, we consider several distributed problems, including routing consistency in overlay networks, and temporal state abnormalities caused by route flaps. We show via micro-benchmarks and larger-scale application measurement that Friday can be used interactively to debug large distributed applications under replay on common hardware. 1