Wednesday, August 25, 2010

Analyzing Side Channel Attacks on Embedded Systems

General embedded systems based on micro-controller and complex processors:
-USB sticks
-Car locks
-Remote access tokens
-Mobile devices
-Game consoles
-Multi-media chipsets for pay-TV

Think of Security:
-What is the threat from side channel analysis to embedded systems?
-How does it compare with attacks on smart cards?
-What are the future developments?

Attacking Side Channels
-Power consumption
-Electro-Magnetic radiation

Power/EM traces
-Signal leakage from busses, registers, ALUs, etc.

Statistical data detection
-Where is data processed in presence of noise?
-Collect many traces with different data (n > 1000)
-Assume data values are:
    known (e.g. algorithm input or output)
    uniformly random (typical for crypto)
-We focus on one bit of one variable in the process

Differential trace
-Input: n traces with known variable (e.g. input or output)
-Output: 1 trace with indication where bit causes trace differences

Purpose of Side Channel Attacks on Embedded Systems
-Retrieve secrets (Key, PIN, Unlock code)
-Reverse engineer (Program flow, Crypto protocol, Algorithm)

Why Side Channel Attacks are interesting? If side channel threats depends on:
-Physical access?
-Access time window?
-Interfacing and control?
-Exploitation equipment $?

A device becomes interesting when:
-It contains a secret
-It contains a feature that can be unlocked
-Logical or physical access to internals is hard

Typical Side Channel Attack Example

Typical Prerequisites
-Access to side channel
-Access to input or output data
-Minimize noise in side channel
-Time measurement of operation (trigger)
-Link data to operation

Processor comparison with Smart Card
Acquisition comparison with Smart Card

Test vs. Attack
-An attacker needs to turn a vulnerability into an exploit
-A tester needs to gain insight in attacker cost efficiently
-How to create the optimal environment to discover a vulnerability?

General aspects of testing
-Controlling the crypto
-Linking data with measurements
-Efficiency of acquisition
-Increased speed versus increased complexity

Timing analysis
-Peripheral outputs assist (example XBOX 360)
-Exploiting runtime access (cache)
-Increasing accuracy with EM and power
-Timing is a risk in many software implementations: both crypto and comparisons

XBOX 360 with Backdoor

 -XBOX 360 has a secure boot chain
-First boot loader security implemented with a HMAC-SHA1
-Hash secret key + boot loader with SHA1
-Compare 16 bytes result with stored 16 bytes
-Comparison is per byte -> timing attack
-Implementation in this infectus board:
    It can modify stored HMAC-SHA1 value in NAND flash
    Observes timing of diagnostic POST byte on PCB
    Reset CPU with nTRST
-Brute forcing 16*128 = 2048 values on average takes about 2 hrs

Power analysis
-Tapping power or supplying it
-Reaching rails
-Identifying the correct supply rail
-Disabling power domains
-Disabling peripherals
-All require more detailed knowledge on target

EM (Electro Magnetic) Analysis
-EM signal adds dimension
-How to locate?
-When can EM be better?
-EMA is an active research topic
-EM seems to add most when target operation is small relative to overall chip

Threat and Impact
-Few countermeasures
-Significant leakage
-Fast acquisition
-Required level of control
-Attacks needed to achieve control
-High noise level, increased acquisition times

-Random Interrupts
-Data / Key masking

-Randomizing flow
-Blinding / Masking
-Protocol design

Monday, August 9, 2010

Scanning SS7 Networks and Telecom Backbones

Historic View
-Phreaking is a term for the action of making a telephone system do something that it normally should not allow.
-Telecommunications security problems started in the 1960’s when the hackers of the time started to discover ways to abuse the telephone company.
-Discovery and exploration of features of telecommunications systems.
-Controlling Network Elements (NE) in a way that was not planned by its designers.
-Abusing weaknesses of protocols, systems and applications in telephone networks.

Fraud Implanted by
-Blue Box
-Internal Fraud

-US: 911, Europe: 112
-How much lost revenue is one minute of downtime?

Today's View
-SIP account hacking, remind the "Calling Cards" fraud?
-VoIP GW hacking, remind the "PBX hacking"?
-Signaling hacking directly on SS7 – SIGTRAN level

SS7 Attacks Scenarios
-Theft of service, interception of calling cards numbers, privacy concerns
-Introduce harmful packets into the national and global SS7 networks
-Get control of call processing, get control of accounting reports
-Obtain credit card numbers, non-listed numbers, etc.
-Messages can be read, altered, injected or deleted
-Denial of service, security triplet replay to compromise authentication
-Annoyance calls, free calls, disruption of emergency services
-Capture of gateways, rerouting of call traffic
-Disruption of service to large parts of the network
-Call processing exposed through Signaling Control Protocol
-Announcement service exposed to IP through RTP
-Disclosure of bearer channel traffic

Telecom Backbone

Discovering The Backbone
-Europe / US: CLEC vs ILEC

New services and new business partners
-Premium numbers, SMS providers, etc.

Push toward an “All IP” infrastructure
-Management network
-SIGTRAN (SS7 over IP)

-Formerly, the walled garden

-Hard to make it reliable (QoS, SBCs)

SS7 and IP
-There is also exponential growth in the use of interconnection between the telecommunication networks and the Internet, for example with VoIP protocols (e.g. SIP, SCTP, M3UA, etc.)
-The IT community now has many protocol converters for conversion of SS7 data to IP, primarily for the transportation of voice and data over the IP networks. In addition new services such  as those based on IN will lead to a growing use of the SS7 network for general data transfers.
-There have been a number of incidents from accidental action on SS7, which have damaged a  network. To date, there have been very few deliberate actions. Far from VoIP here. 

Attacking SIGTRAN with SCTPscan (
Where implementation diverge from RFCs
-RFC says "hosts should never answer to INIT packets on non-existings ports".
-Syn scanning is slow when no RST

Below the IDS
-How many firewall logs dropped SCTP packets?
-How many IDS(s) watch for SCTP socket evil content?
-Example: - Real life distributed IDS, Hundreds of thousands of IP scanned, nor detected neither reported as scanner.

INIT vs SHUTDOWN_ACK Packet Scanning
From RFC 2960
-8.4 Handle "Out of the blue" Packets
-An SCTP packet is called an "out of the blue" (OOTB) packet if it is correctly formed, i.e., passed the  receiver's Adler-32 / CRC-32 check (see Section 6.8), but the receiver is not able to identify the association to which this packet belongs.
-The receiver of an OOTB packet MUST do the following:
"If the packet contains a SHUTDOWN ACK chunk, the receiver should respond to the sender of the OOTB packet with a SHUTDOWN COMPLETE."

-New way to elicit answers even if not answering ABORTs to INITs targeted at not-opened port.

SCTP ports (-sS) Stealth Scanning
root@bt:~/sctp# ./sctpscan-v11 --scan --autoportscan -r
Netscanning with Crc32 checksumed packet SCTP present on port 2905 SCTP present on port 7102 SCTP present on port 7103 SCTP present on port 7105 SCTP present on port 7551 SCTP present on port 7701 SCTP present on port 7800 SCTP present on port 8001 SCTP present on port 2905

SCTP Stack Fingerprinting
-SCTP stack reliability
-Robustness testing (stress testing)
-QA of a few stacks
-Fuzzing built-in SCTPscan
-Discrepancies in SCTP answer packets
-Different stack behaviours
-Much more states than TCP=opportunities
-Cookie randomness

Credits: Philippe Langlois, P1 Security (

Monday, August 2, 2010

Using DAVIX For Security Visualization (revised)

Information visualization
-Visualize large collections of abstract data

Scientific visualization
-Representation of data with geometric structure

Visualization Concept
-Analyzing floods of data in tabular or textual form is tedious
-Humans must sequentially scan such data
-Visualization exploits the human's visual perceptive capabilities and parallel processing Size, Shape, Distance, and Color
-Easy to spot patterns and irregularities

Data types supported
Has a sequence e.g. day of week
Has no sequence e.g. types of fishes
Can be measured e.g. length, time, weight, temperature, speed

Visualization Effectiveness
-Each data type has its most effective way of visualization

Information Visualization Process

DAVIX Linux Distribution (
-Provide the audience with a workable and integrated tools set
-Enable them to immediately start with security visualization
-Motivate them to contribute to the security visualization community

Tools Available
-Network Tools (Argus, Snort, Wireshark)
-Logging (syslog-ng)
-Fetching Data (wget, ftp, scp)

-Shell Tools (awk, grep, sed)
-Visualization Preprocessing (AfterGlow, LGL)
-Extraction (Chaosreader)
-Data Enrichment (geoiplookup, whois, gwhois)

-Network Traffic (EtherApe, InetVis, tnv)
-Generic (AfterGlow, Cytoscape, Graphviz, LGL Viewer, Mondrian, R Project, Treemap)

Interface Transport
-Each visualization tool has its own file format interfaces
-Data must be converted to match the import interfaces
-These adapters are mostly self-written snippets of code

Important Note:
All the images presented in this post are intellectual property of the copyright owner (