Transcript Malware

CIT 380: Securing Computer
Systems
Malware
CIT 380: Securing Computer Systems
Slide #1
Topic
1.
Types of Malware
1.
2.
3.
4.
5.
2.
3.
4.
5.
6.
7.
Trojan Horses
Viruses
Worms
Backdoors
Rootkits
Self-Protection Mechanisms.
Payloads.
Malware Interactions.
Detecting Malware.
Defending against Malware.
The changing Malware environment.
CIT 380: Securing Computer Systems
Slide #2
Types of Malware
Trojan Horse
Tricks user into executing malicious code.
Virus
When run by user, copies self into other files.
Worm
Copies self from computer to computer.
Backdoors
Leaves opening for attacker to gain access.
Rootkits
Hides attacker activities from system administrators.
CIT 380: Securing Computer Systems
Slide #3
What about Spyware?
Malware by any other name…
– Corporate malware.
– Presents legal issues for anti-malware software.
CIT 380: Securing Computer Systems
Slide #4
Trojan Horse
Program with both an
overt and covert effect
– Displays expected
behavior when user
executes.
– Covert effect (executed
with user’s privileges)
violates security policy.
Attacker:
cat >ls
cp /bin/sh /tmp/.xxsh
chmod u+s,o+x /tmp/.xxsh
rm ./ls
ls $*
^D
Victim:
ls
CIT 380: Securing Computer Systems
Slide #5
Virus
Self-replicating code
– Propagating (replicating) Trojan horse.
– Inserts (possibly evolved) copy into other files.
Virus Pseudocode:
If spread condition then
Foreach target-file
if not infected then copy virus to target-file
Perform (malicious) action
Execute normal code
CIT 380: Securing Computer Systems
Slide #6
Types of Viruses
1. Boot Sector
–
–
–
When system boots, code in boot sector executed.
Propagate by altering boot disk creation.
Uncommon today because of low use of boot
floppies, but some Vista laptops shipped with one.
2. Executable
–
–
–
Infects executable programs (e.g., COM, EXE).
Executes when infected program is run.
Virus usually runs first, then runs original code.
3. Dynamic Library
–
–
Infected dynamicly linked libraries (DLLs.)
Executed when any program uses infected DLL.
CIT 380: Securing Computer Systems
Slide #7
Types of Viruses
4. Device Driver
–
–
Infects loadable device driver.
Executes in kernel mode.
5. Virtual Machine (.NET)
–
–
Infects .NET MSIL binaries.
Portable: compiled to native code by CLR.
6. Archive Infectors
–
–
Inserts Trojan horse into ZIP files.
Uses social engineering techniques to get user to run.
CIT 380: Securing Computer Systems
Slide #8
Types of Viruses
7. Macro Virus
–
–
–
Infects embedded interpreted code.
Needs interpreter like sh, MS Word macro.
Can infect executables or data files
•
–
Executables must invoke appropriate interpreter.
Most modern data formats support some type
of scripting, including
•
•
•
Microsoft Office
Windows Help files
HTML: VBScript, JScript
CIT 380: Securing Computer Systems
Slide #9
Infection Methods
1. Overwriting
–
–
Overwrites program code with virus.
Breaks infected program.
2. Appending
–
–
Append virus code to executable.
Insert JMP at beginning of executable.
3. Prepending
–
–
Insert virus code at beginning of executable.
Shift original code to follow virus.
CIT 380: Securing Computer Systems
Slide #10
Infection Methods
4. Parasitic
–
–
Inserts virus code at beginning of executable.
Shifts beginning of program to end of file.
5. Cavity
–
–
Insert virus code into unused blocks of file.
Insert JMP at beginning of executable.
6. Fractionated Cavity
–
–
Fragment virus; inject into multiple cavities.
Loader reads fragments into continuous
memory.
CIT 380: Securing Computer Systems
Slide #11
Infection Methods
7. Compressing
–
–
Compresses executable to make space.
Inserts virus and decompression code.
8. Fragmenting
–
–
–
Dynamically fragment virus.
Insert fragments by overwriting or shifting code.
Fragments JMP/CALL each other.
9. Companion
–
–
Infects COM file of same name as EXE file.
Infects alternate data stream of Win32 file.
CIT 380: Securing Computer Systems
Slide #12
In-Memory Strategies
Direct Action
– Virus runs only when infected code is run.
Memory Resident
–
–
–
–
–
Remains active in memory after application terminates.
Interrupt hook (TSR) in DOS.
Kernel-mode rootkit techniques under modern OSes.
Can infect any program that runs after virus.
Example: Jerusalem Virus (Danube variant)
• Multipartite TSR virus.
• Infects all executables except command.com.
• Also infects boot sector.
• Deletes files on Friday the 13th.
CIT 380: Securing Computer Systems
Slide #13
Worms
Copies self from one computer to another
Self-replicating: No user action required unlike
virus or Trojan horse programs.
Spreads via network protocols
ex: SMTP (email), fingerd, MS SQL
CIT 380: Securing Computer Systems
Slide #14
History of Worms
Morris Worm Nov 1988 Disabled most of Internet
using multiple vectors.
Melissa
Mar 1999 MS Word macro virus
spread via Outlook email.
Code Red
Aug 2001 IIS Buffer overflow.
Code Green
Slammer
Sobig
Sep 2001 Removed Code Red II and
patched vulnerability.
Jan 2003 SQL Server worm infected
entire Internet <1 hr.
Jun 2003 Spam zombie botnet; RCI.
CIT 380: Securing Computer Systems
Slide #15
Worm Components
1.
2.
3.
4.
5.
Vector
Propagation Engine
Target Selection
Scanning Engine
Payload
CIT 380: Securing Computer Systems
Slide #16
Vector
Software to gain access to target host.
Common vectors:
–
–
–
–
–
Buffer overflow exploits.
Network file sharing, both NFS/SMB and P2P.
Social-engineering via email or IM.
Weak passwords.
Parasitism: target backdoors and worm flaws.
CIT 380: Securing Computer Systems
Slide #17
Propagation Engine
Transfers worm to host exploited by vector.
– Small worms like Slammer included in vector.
Worm Propagation Methods:
–
–
–
–
FTP
HTTP
SMB
TFTP
CIT 380: Securing Computer Systems
Slide #18
Remote Control Interface
RCI allows creator to control infected hosts.
Many worms do not have a RCI.
May be a well-known backdoor program.
Common remote control features:
Start/stop infecting new targets.
Download new vectors.
Download new target selectors.
Download new payloads.
CIT 380: Securing Computer Systems
Slide #19
Target Selection
Selecting targets for potential infection.
E-mail address harvesting
– Address books.
– Parse disk files.
– Search news groups.
Network share enumeration
– Check for filesystems shared with other systems.
Network scanning
– Target hosts on current network and connected nets.
– Randomized scanning of Internet space.
Web searching
– Search Google for addresses or vulnerable software.
CIT 380: Securing Computer Systems
Slide #20
Scanning Engine
Check targets for vulnerabilities.
– If vector small, scanning can be skipped.
Scan for vulnerable services.
– Like targeted nmap port scan.
OS Check
– Check for correct OS for vector to work.
Version checking.
– Check version of target software.
– May customize vector based on information.
CIT 380: Securing Computer Systems
Slide #21
Morris Worm
• First Internet Worm: November 1988
• Multi-architecture: Sun, VAX
• Multi-vector
– sendmail (debug backdoor)
– fingerd (buffer overflow)
– rsh (open .rhosts; password cracking)
CIT 380: Securing Computer Systems
Slide #22
Morris Worm
Spreading algorithm
Local network topology: gateways, neighbors.
Used users’ .rhosts, .forward files.
Limited reinfection rate.
Detection Avoidance
Forged process listing as (sh).
Removed created files quickly after use.
CIT 380: Securing Computer Systems
Slide #23
Morris Worm
Resource Requirements
Disk Space.
C compiler and linker.
Network connection to parent computer.
Problems
Didn’t limit re-infections.
Saturated CPU, network resources.
CIT 380: Securing Computer Systems
Slide #24
Malware Self-Protection
Anti-debugging
Detect/disable debuggers when used to analyze code.
Attack anti-malware tools
Disable anti-malware tools upon infection.
Kill processes or destroy/modify signatures.
API checksums
Avoid having UNIX/Win32 API calls in code.
Store checksums of API names and search for match.
Code obfuscation
Use unusual tricks and unused code to avoid dissassembly
and prevent quick analysis of purpose.
Self-modifying code.
CIT 380: Securing Computer Systems
Slide #25
Self-Protection
Compression
Code looks almost random; size is smaller.
Use unusual executable packers to avoid analysis.
Data encryption
Encrypt strings, hostnames, IP addresses to avoid
detection.
Embedding
Use multiple levels of executable packers like UPX.
Scanners have to understand and have time to parse
and decompress each file format.
CIT 380: Securing Computer Systems
Slide #26
Self-Protection
Entry-Point Obscuring
Changing initial code or entry point easy to notice.
Alter program code to gain control randomly.
Host morphing
Alter host file during infection to prevent removal.
CIT 380: Securing Computer Systems
Slide #27
Self-Protection: Encryption
Encrypt all code except small decryptor.
– Note that copy protected files will have similar
decryptors to prevent analysis too.
– Often uses multiple decryptors.
– Change encryption key dynamically.
Random Decryption Algorithm (RDA)
– Choose random key for encryption.
– Brute force search for key to decrypt.
– Slows VMs/debuggers used for analysis.
CIT 380: Securing Computer Systems
Slide #28
Self-Protection: Polymorphism
Alter malware code with each infection.
–
–
–
–
–
Cannot be detected by signature scanning.
May alter decryptor only or entire code.
Insert junk instructions that do nothing.
Fragment and rearrange order of code.
Alternate sets of instructions for the same task.
• Ex: SUB -1 instead of ADD 1
– Randomize names in macro viruses.
CIT 380: Securing Computer Systems
Slide #29
Case Study: Zmist
EPO, encrypted, polymorphic virus.
Code integration
Decompiles PE files to smallest elements.
Inserts virus randomly into existing code.
Rebuilds executable.
Polymorphic decryptor
Inserted as random fragments linked by JMPs.
Randomizes self with ETG engine.
CIT 380: Securing Computer Systems
Slide #30
Payloads
Accidentally destructive.
Replication damages data due or exhausts system resources
due to malware bugs.
Ex: Morris Worm reinfected hosts, using all CPU.
Nondestructive.
Displays message, graphics, sound, or open CD door.
Ex: Christma worm on IBM network in 1987.
Destructive.
Triggers randomly or on some event or machine type.
Deletes files or overwrites data.
Hardware destroyers: overwrite BIOS.
CIT 380: Securing Computer Systems
Slide #31
Payloads
Denial of Service
Sometimes accidental due to high network use.
Launch DDOS attack with all infected systems.
Data Theft
Phishing scams and spyware.
Encryptors (ransomware)
Encrypts user data.
Ex: One_Half encrypts disk; enables access while running.
Ex: AIDS Info: encrypts disk and holds for ransom.
Spam
Use network of infected systems to launder spam email.
Ex: Sobig worm.
CIT 380: Securing Computer Systems
Slide #32
Malware Interactions
What happens when a virus infects a worm?
Typically both propagate.
May use each other’s self-protection techniques.
What if anti-virus software removes a virus?
Likely leaves unknown virus/worm alone.
Partial removal can mutate the malware into a new form.
Competition and Parasitism
Malware may remove competing malware.
May exploit backdoors/RCI left by previous malware.
May infect competing malware, hijacking its propagation.
CIT 380: Securing Computer Systems
Slide #33
Theory of Malicious Code
Theorem 1: It is undecidable whether an arbitrary
program contains a computer virus.
Proof:
Define virus v as TM program that copies v to other parts
of the tape, while not overwriting any part of v.
Reduce to Halting Problem: T’ running code V’
reproduces V iff running T on V halts.
Theorem 2: It is undecidable whether an arbitrary
program contains malicious logic.
CIT 380: Securing Computer Systems
Slide #34
Detecting Malware
Signature-based
– Look for known patterns in malicious code.
– Defeated by polymorphic viruses.
Smart scanning
– Skips junk instructions inserted by poly engines.
– Skips whitespace/case changes in macro viruses.
Decryption
– Brute-forces simple XOR-based encryption.
– Checks decrypted text against small virus sig to
decide whether has plaintext or not.
CIT 380: Securing Computer Systems
Slide #35
Detecting Malware
Code Emulation
– Execute potential malware on VM.
– Scan VM memory after certain # iterations.
– Watch instructions for decryptor profile.
Code Optimization.
– Optimize away junk instructions and odd
techniques used by polymorphic viruses.
CIT 380: Securing Computer Systems
Slide #36
Detecting Malware
Heuristics
–
–
–
–
–
Code execution starts in last section.
Suspicious code redirection.
Suspicious section ACLs or size.
Suspicious library routine imports.
Hard-coded pointers into OS kernel.
Neural Network Heuristics
– IBM researchers trained neural net to recognize
difficult polymorphic viruses.
– Released in Symantec antivirus.
CIT 380: Securing Computer Systems
Slide #37
Detecting Malware
Behavior-based
– Watch for known actions from malicious code.
– Network access signature of worm.
– Unexpected use of dangerous system calls.
Integrity Checking
–
–
–
–
Host-based Intrusion Detection System.
Record MAC, size, dates, ACL of files.
Periodically check for changes.
ex: Tripwire, AIDE, Osiris
CIT 380: Securing Computer Systems
Slide #38
Defences: Data vs. Code
Separate data and instructions
– Virus treats program as data
• Writes self to file.
– Virus treats program as instructions
• Virus executes when program is run.
– Solution: Treat all programs as data until
trusted authority marks as executable.
• Development difficult when compilers can’t produce
executable code.
CIT 380: Securing Computer Systems
Slide #39
Defences: Information Flow
Limit Information Flow
– Virus executes with user’s identity.
– Soln: Limit information flow between users.
• Set flow distance to be one for users A, B, C.
• A creates virus (fd=0), B executes it (fd=1).
• C cannot execute B’s infected program (fd=2).
– Indirect virus spread limited.
– How can we track information flow?
CIT 380: Securing Computer Systems
Slide #40
Defences: Least Privilege
Limit programs to least privilege needed
example: SELinux
Mail virus example
1.
2.
3.
4.
Virus arrives via email.
Virus exploits bug in email client to execute.
Virus saves self to file in Startup folder.
Virus infects Office documents.
How least privilege would stop
– Mail application cannot create virus binaries.
– Mail application cannot write to Startup folder.
– Mail application cannot write to Office documents.
CIT 380: Securing Computer Systems
Slide #41
Defences: Sandboxes
Execute code in protected sandbox or VM.
Virtual Browser Appliance
Linux guest running Firefox under VMWare.
Infections can only attack VM, not real host.
Reset VM to initial state if infected.
CIT 380: Securing Computer Systems
Slide #42
Defences: Anomaly Detection
Validate program actions with policy
Limit access to system calls.
Example: systrace.
Check statistical characteristics.
Programmer style.
Compare source code with object.
Statistics of write frequencies, program
executions.
CIT 380: Securing Computer Systems
Slide #43
Defences: Counter-worms
Worm that removes other worms from net.
Nachi/Welchia
•
•
•
•
Multi-vector W32 worm
Nachi.A removes W32/Blaster worm
Nachi.B removes W32/MyDoom worm
Installed MSRPC DCOM patch to prevent future
infections from Blaster.
• Removes self after 2004.
Side-effects
• Infected Diebold ATMs
• Worm traffic DOSed Internet, esp Microsoft.
CIT 380: Securing Computer Systems
Slide #44
Fast Worms
Slammer Worm Characteristics
–
–
–
–
–
–
–
–
Attacked MS SQL servers.
Worm is single 404-bye UDP packet.
Random-scan (PRNG bugs limited.)
Limited by network bandwidth, not latency.
Observed scan rate of 26,000 hosts/second.
Infected 90% of vulnerable hosts in 10 min.
Too fast for humans to react.
Shutdown 13,000 Bank of America ATMs due
to compromising db servers, heavy traffic.
CIT 380: Securing Computer Systems
Slide #45
Profitable Malware
Sobig
– W32 worm using email/network share vectors.
– Contains upgrade mechanism
• Worm checked sites every few minutes.
• When site valid, downloaded code.
• Later variants could update upgrade server list.
– Downloaded payload from upgrade mechanism
• Key logger.
• Wingate proxy server (for spam proxying.)
CIT 380: Securing Computer Systems
Slide #46
Profitable Malware
Trojans
Backdoor.Lala transfers authentication cookies
for eBay, PayPal, etc. to maker.
PWSteal.Bancos automates phishing by
displaying fake web pages when browser goes
to certain bank sites.
Spyware and Adware
More than ever using Trojan techniques.
Win32/Bube virus exploits IE flaw and acts as a
virus infecting IE, then downloads adware.
CIT 380: Securing Computer Systems
Slide #47
Mobile Malware
2004: Cabir virus infecting Symbian OS mobile
phones using Bluetooth appeared in June.
2005: Commwarrior-A worm spreads to Symbian
series 60 phones via phone’s MMS.
Around a 1000 pieces of mobile malware exist.
For Blackberries and Palm Pilots too.
Expect more as smart phones become common.
CIT 380: Securing Computer Systems
Slide #48
Offline Impact
Davis-Besse nuclear power plant
Slammer infected Plant Process Computer and Safety
Parameter Display System (Jan 2003.)
Analog backups unaffected.
Infected contractor’s network, then moved through T1
line that bypassed plant firewall.
Seattle 911 system
Slammer disabled computer systems.
Dispatchers reverted to manual systems.
2003 Blackout
Blaster infected First Energy systems.
CIT 380: Securing Computer Systems
Slide #49
Modern Malware is
Stealthy: rootkit techniques common.
Targeted: targets smaller banks and countries,
leverages current events:
– January: Storm Worm appears via email with subject
“230 dead as storm batters Europe.”
– February: Miami Dolphins Stadium site hacked before
superbowl so that it would infect browsers with trojan
that grabbed WoW data.
Blended: combine trojan, virus, worm features.
Web-based: use web for delivery and update.
Profit-driven: the goal is to make money.
CIT 380: Securing Computer Systems
Slide #50
References
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
Ross Anderson, Security Engineering, Wiley, 2001.
Matt Bishop, Computer Security: Art and Science, Addison-Wesley, 2003.
William Cheswick, Steven Bellovin, and Avriel Rubin, Firewalls and Internet Security, 2nd
edition, 2003.
Fred Cohen, http://www.all.net/books/virus/part1.html, 1984.
Simson Garfinkel, Gene Spafford, and Alan Schartz, Practical UNIX and Internet Security, 3rd
edition, O’Reilly & Associates, 2003.
Alexander Gostev, “Malware Evolution: January - March 2005,”
http://www.viruslist.com/en/analysis?pubid=162454316, April 18 2005.
Elias Levy, “Crossover: Online Pests Plaguing the Offline World,” IEEE Security & Privacy,
2003.
Stuart McClure, Joel Scambray, George Kurtz, Hacking Exposed, 5th edition, McGraw-Hill,
2003.
Hilarie Orman, “The Morris Worm: A Fifteen-Year Perspective,” IEEE Security & Privacy, 2003
Cyrus Peikari and Anton Chuvakin, Security Warrior, O’Reilly & Associates, 2003.
Ed Skoudis, Counter Hack Reloaded, Prentice Hall, 2006.
Ed Skoudis and Lenny Zeltser, Malware: Fighting Malicious Code, Prentice Hall, 2003.
Staniford, Stuart, Paxson, Vern, and Weaver, Nicholas, ‘How to 0wn the Internet in Your Spare
Time,” Proceedings of the 11th USENIX Security Symposium, 2002
Peter Szor, The Art of Computer Virus Research and Defense, Addison-Wesley, 2005.
Trend Micro, “1H2007 Threat Roundup,”
http://us.trendmicro.com/imperia/md/content/us/pdf/threats/securitylibrary/1h_2007_threat_round
up_final_jul2007.pdf, 2007.
CIT 380: Securing Computer Systems
Slide #51