Rootkits: Subverting the Windows Kernel

Preface

A rootkit is a set of programs and code that allows a permanent and undetectable presence on a computer.

We became interested in rootkits because of our professional work in computer security, but the pursuit of the subject quickly expanded into a personal mission (also known as late nights and weekends). This led Hoglund to found rootkit.com, a forum devoted to reverse engineering and rootkit development. Both of us are deeply involved with rootkit.com. Butler first contacted Hoglund online through this Web site because Butler had a new and powerful rootkit called FU that needed testing,¹ Butler sent Hoglund some source code and a pre-compiled binary. However, by accident, he did not send Hoglund the source code to the kernel driver. To Butler's amazement, Hoglund just loaded the pre-compiled rootkit onto his workstation without question, and reported back that FU seemed to be working fine! Our trust in one another has only grown since then.²

Both of us have long been driven by an almost perverse need to reverse-engineer the Windows kernel. It's like when someone says we can't do something—then we accomplish it. It is very satisfying learning how so-called computer security products work and finding ways around them. This inevitably leads to better protection mechanisms.

The fact that a product claims to provide some level of protection does not necessarily mean it actually does. By playing the part of an attacker, we are always at an advantage. As the attacker we must think of only one thing that a defender didn't consider. Defenders, on the other hand, must think of every possible thing an attackermight do. The numbers work in the attacker's favor.

We teamed up a few years ago to offer the training class "Offensive Aspects of Rootkit Technology." This training started as a single day of material that since has grown to include hundreds of pages of notes and example code. The material for the class eventually became the foundation for this book. We now offer the rootkit training class several times a year at the Black Hat security conference, and also privately.

After training for awhile, we decided to deepen our relationship, and we now work together at HBGary, Inc. At HBGary, we tackle very complex rootkit problems on a daily basis. In this book, we use our experience to cover the threats that face Windows users today, and likely will only increase in the future.

This book is intended for those who are interested in computer security and want a truer perspective concerning security threats. A lot has been written on how intruders gain access to computer systems, but little has been said regarding what can happen once an intruder gains that initial access. Like the title implies, this book will cover what an intruder can do to cover her presence on a compromised machine.

We believe that most software vendors, including Microsoft, do not take rootkits seriously. That is why we are publishing this book. The material in this book is not groundbreaking for someone who has worked with rootkits or operating systems for years—but for most people this book should prove that rootkits are a serious threat. It should prove that your virus scanner or desktop firewall is never good enough. It should prove that a rootkit can get into your computer and stay there for years without you ever knowing about it.

To best convey rootkit information, we wrote most of this book from an attacker's perspective; however, we end the book on a defensive posture. As you begin to learn your attackers' goals and techniques, you will begin to learn your own system's weaknesses and how to mitigate its shortcomings. Reading this book will help you improve the security of your system or help you make informed decisions when it comes to purchasing security software.

As all of the code samples are written in C, you will gain more insight if you already understand basic C concepts—the most important one being pointers. If you have no programming knowledge, you should still be able to follow along and understand the threats without needing to understand the particular implementation details. Some areas of the book draw on principles from the Windows device driver architecture, but experience writing device drivers is not required. We will walk you through writing your first Windows device driver and build from there.

This book covers Windows rootkits, although most of the concepts apply to other operating systems as well, such as LINUX. We focus on kernel rootkits because these are the most difficult to detect. Many public rootkits for Windows are userland rootkits³ because these are the easiest to implement, since they do not involve the added complexity of understanding how the undocumented kernel works.

This book is not about specific real-world rootkits. Rather, it teaches the generic approaches used by all rootkits. In each chapter, we introduce a basic technique, explain its purposes, and show how it's implemented using code examples. Armed with this information, you should be able to expand the examples in a million different ways to perform a variety of tasks. When working in the kernel, you are really limited only by your imagination.

You can download most of the code in this book from rootkit.com. Throughout the book, we will reference the particular URL for each individual example. Other rootkit authors also publish research at rootkit.com that you may find useful for keeping up with the latest discoveries.

1. Butler was not interested in rootkits for malicious purposes. He was instead fascinated with the power of kernel modifications. This led Butler to develop one of the first rootkit-detection programs, VICE.

2. Hoglund still wonders, from time to time, whether that original version of FU is still running on his workstation.

3. Userland rootkits are rootkits that do not employ kernel-level modifications, but instead rely only upon user-program modifications.