Source Code Review

Dominion Democracy Suite 4.14-A.1 Voting System with Adjudication

Version 2.4

Report Date: 2014-11-14

Version: 1.04
Status: RELEASED
Classification: Public

atsec information security corporation

9130 Jollyville Road, Suite 260
Austin, TX 78759
Tel: +1 512 615 7300
Fax: +1 512 615 7301
www.atsec.com

Last update: 2014-11-14 Classification: Public Status: RELEASED

Version: 1.04 ©2014 atsec information security corporation Page 1 of 12
Trademarks
atsec and the atsec logo is a registered trademark of atsec information security corporation.
The Freeman, Craft, McGregor Group (FCMG) logo is a trademark of FCMG.
Dominion Democracy, Democracy Suite Election Management System, Election Event Definition, Results
Tally and Reporting, and ImageCast are trademarks of Dominion Voting Systems, Inc.
Microsoft, Microsoft SQL Server, Microsoft .NET, Windows 2007 and Visual C# are registered trademarks
of Microsoft Corporation.
MITRE is a registered trademark of The MITRE Corporation.
IText is a registered trademark by iText Group NV.

Last update: 2014-11-14 Classification: Public Status: RELEASED

Version: 1.04 ©2014 atsec information security corporation Page 2 of 12
Table of Contents
1 Executive Summary ......................................................................................................................................4
2 Introduction ...................................................................................................................................................5
2.1 Scope and Basis ................................................................................................................................5
2.2 Inputs ..................................................................................................................................................5
2.3 Threat Model ......................................................................................................................................6
2.4 Methodology .......................................................................................................................................6
2.4.1 Published vulnerabilities ..........................................................................................................6
2.4.2 Code quality .............................................................................................................................6
2.4.3 Design ......................................................................................................................................7
2.4.4 Cryptography ...........................................................................................................................7
2.4.5 Backdoors ................................................................................................................................7
2.4.6 Summary of the results ............................................................................................................8
3 Results ..........................................................................................................................................................9
Glossary .........................................................................................................................................................11
References ....................................................................................................................................................12

Last update: 2014-11-14 Classification: Public Status: RELEASED

Version: 1.04 ©2014 atsec information security corporation Page 3 of 12
1 Executive Summary
This report was prepared by atsec information security corporation to review aspects of the security and
integrity of the Dominion Democracy Suite 4.14-A.1 Voting System with Adjudication Version 2.4.
This report identifies the security vulnerabilities that might be exploited to alter vote results, critical election
data such as audit logs, or to conduct a denial of service attack on the Adjudication system that were found
through static code review and by searches of public vulnerability sources.
The atsec team identified sixteen potential vulnerabilities in the system; all were categorized as having low
levels of severity. All but two of the sixteen potential vulnerabilities involve non-conformances to one of the
following coding or cryptographic standards:
• 2005 Voluntary Voting System Guidelines (specifically sections 5 and 7 of Volume
I and section 5 of Volume II)
• StyleCop
• FIPS 140-2

The introduction to this report describes the basis of the source code review performed. In Chapter 3, we
present the detailed results of the analysis.

Last update: 2014-11-14 Classification: Public Status: RELEASED

Version: 1.04 ©2014 atsec information security corporation Page 4 of 12
2 Introduction
The goal of this project was to provide test support services to assist the California Secretary of State
(SOS) with the re-evaluation of Dominion Democracy Suite 4.14-A.1 Voting System, specifically the new
Adjudication Version 2.4 component for its suitability for use in the State of California in accordance with
Elections Code sections 19001 et seq.
This report has been prepared in support of a contract awarded to Freeman, Craft, McGregor Group, Inc.
by atsec information security corporation as a result of the review of the security and integrity of the revised
component, Adjudication System Version 2.4, a part of the Dominion Democracy Suite 4.14-A.1 Voting
1
System.
The source code review was performed by the following atsec consultants:
• Hedy Leung
• King Ables
• Lou Losee
• Swapneela Unkule
These individuals have prior experience of testing voting systems, and have not been involved in the
federal level testing for the EAC of the subject voting system.

2.1 Scope and Basis

The Dominion Voting Systems Democracy Suite Version 4.14-A.1 Voting System with Adjudication Version
2.4 (hereafter referred to as the “voting system” or simply as the “system”) is a paper ballot-based, optical
scan voting system. The system hardware consists of four major components:
• The Election Management System (EMS)
• ImageCast Evolution (ICE) precinct scanner with optional ballot marking capabilities
• ImageCast Central (ICC) central count scanner
• Adjudication system
atsec performed the code review on the basis of the Statement of Work between Freeman, Craft,
McGregor Group Inc. #14S52049 with the State of California, which states that review includes evaluating
the security of the Adjudication System as it is allowed to be configured for use by the State of California
(hereafter referred to as “the California configuration”).
The threat model given in section 2.3 below describes the basis of atsec’s examination.

2.2 Inputs
The reviewers were provided with a set of documents associated with the system that were used to
support the results described in this report. These documents are listed in the References chapter below.
These documents were examined during the source code review in order to understand the voting
system’s architecture and design and to support the identification of any discrepancies between the
documentation and the source code.
The reviewers were also provided with the source code for the following Adjudication components.
◦ Adjudication Services (server) version 2.4.1.3201
◦ Adjudication (client) version 2.4.1.3201

1
The system was previously examined by the EAC with certification ID: EAC Certification Number:
DemSuite-4-14-A.1:
http://www.eac.gov/testing_and_certification/certified_voting_systems.aspx#Dominion_DemocracySuite414A.1Mod

Last update: 2014-11-14 Classification: Public Status: RELEASED

Version: 1.04 ©2014 atsec information security corporation Page 5 of 12
 ◦ EMS Adjudication Services (Integration point with EMS) version 4.14.37

2.3 Threat Model

This assessment is centered on the threat model prescribed in the Statement of Work. The system is
expected to counter the following attacks:
• Alter vote results
• Alter critical election data, such as audit logs
• Conduct a denial of service attack on the voting system
To the extent possible, vulnerabilities found have been reported with an indication of whether the
exploitation of the vulnerability would require access by the:
• Elections official insider: Wide range of knowledge of the voting machine design and
configuration. May have unrestricted access to the machine for long periods of time. Their
designated activities include:
◦ Set up and pre-election procedures
◦ Election operation
◦ Post-election processing of results
◦ Archiving and storage operations
• Vendor insider: Has great knowledge of the voting machine design and configuration. They have
unlimited access to the machine before it is delivered to the purchaser and, thereafter, may have
unrestricted access when performing warranty and maintenance service, and when providing
election administration services.
Identified potential vulnerabilities are described along with the anticipated factors necessary to mount an
attack. The atsec team did not attempt to demonstrate the exploitability of any identified potential
vulnerabilities.

2.4 Methodology
The atsec team used the following methodology for the source code review.

2.4.1 Published vulnerabilities

The reviewers searched the MITRE CVE database for potential vulnerabilities in the system. Although
these lists may not have entries for the voting system itself, constituent software that the voting system
uses may contain vulnerabilities. For the current scope of project, the review team identified that the
Adjudication System is based on a C#/.NET environment and conducted searches for vulnerabilities
related to these components. Searches for vulnerabilities for C#, .NET, xca, BouncyCastle, iTextSharp,
and System.Security.Cryptography identified only one potential vulnerability that might pertain to the
software under review.

2.4.2 Code quality

While performing the examination of the code for other activities, the reviewers identified and recorded
areas within the code base that demonstrate poor code quality. Although poor code quality does not
necessarily represent vulnerabilities, it is a weakness and may lead to the introduction of vulnerabilities.
The following coding standards were used during this analysis:
• 2005 Voluntary Voting System Guidelines [VVSG1], [VVSG2] and supplemental interpretation
statements found at:
http://www.eac.gov/testing_and_certification/request_for_interpretations1.aspx

Last update: 2014-11-14 Classification: Public Status: RELEASED

Version: 1.04 ©2014 atsec information security corporation Page 6 of 12
 • StyleCop Coding Standard found at: https://stylecop.codeplex.com
• The CERT Oracle Secure Coding Standard for Java [CERTJ]
The Adjudication System does not contain Java source code, but C# has similarities to Java, so the Java
coding standard still provides useful input into the analysis of code quality.

2.4.3 Design
The source code review team utilized the provided usage and installation guidance, source code, and any
other provided material as well as publicly available information in order to construct an understanding of
the architecture and design of the voting system. This included discovering the external interfaces and their
associated security mechanisms and controls, particularly as much information as possible was gathered
to support conclusions regarding the ability for a threat agent to tamper with or circumvent security
controls.
The provided design description also provided a mapping of the high-level features and interfaces of the
product to the features and interfaces implementation.
Interfaces represent the primary attack surface of the voting system. Interfaces can include web-based
interfaces, native graphic user interfaces, command line interfaces, or technical interfaces that are not
designed for direct user interaction (e.g., database connections). Each of these interfaces was examined to
identify the security controls that counter the threats.
Secure interfaces also depend on filtering out poorly structured or corrupt data. The review team
specifically checked for input validation mechanisms and determined if related attacks, such as command
injection are possible.

2.4.4 Cryptography
While cryptography is often the hardest security mechanism to break directly by brute-force, misuse of
cryptographic primitives or implementation errors can render that protection weak or non-existent. The
review team identified use sites of cryptography throughout the source code and determined if its use is
appropriate for the given purpose. For example, using a cryptographic hash function to protect passwords
is appropriate while using an encryption algorithm with a hard-coded key is not. The cryptographic
primitives used in the source code are AES, HMAC-SHA-256, SHA-1, MD5, and RSA key generation.
Note that in a code review, there is no effective way to test the correctness of implementation of any
cryptographic algorithm. We recommend that all cryptographic algorithms be functionally tested, such as
with the Cryptographic Algorithm Validation System (CAVS) test which is part of the NIST Cryptographic
Algorithm Validation Program.

2.4.5 Backdoors
Those with malicious intent who also have access to the Adjudication System during development may be
able to place backdoors into the source code so that they could gain unauthorized access to the
Adjudication System during operation.
Backdoors are extremely hard to find because a seasoned programmer can obfuscate code to look benign.
The atsec team would like to stress that, when facing a competent and sufficiently motivated malicious
developer, it is extremely difficult to prove that all backdoors in a system have been identified. The famous
Turing award lecture by Ken Thompson in 1984 entitled Reflections on Trusting Trust [TRUST]
demonstrated how fundamentally easy it is to undermine all security mechanisms when the developers
cannot be trusted. This voting system is no exception. The current scope of the project is the Adjudication
System, which consists of total 617 C# source files.
A full backdoor analysis is also impractical in a short period of time because a deep understanding of the
data structures and code design is required to be able to recognize functionality that is out of place.
Penetration test on a running system and observation of data modification and movement is also helpful.
For this report, the reviewers are only able to examine the source code and look for signs of obfuscation or
strange functionality.

Last update: 2014-11-14 Classification: Public Status: RELEASED

Version: 1.04 ©2014 atsec information security corporation Page 7 of 12
The review team marked the areas of poor code quality and use of cryptographic operations for further
scrutiny. For example, a particular area of code that has poor code quality and accesses sensitive
information such as authentication credentials is identified as a likely candidate for a hidden backdoor. The
reviewers treated such areas by considering the threat of introduced backdoors in addition to unintentional
implementation flaws.

2.4.6 Summary of the results

A summary of the results is listed in Chapter 3. Each result contains:
• A description of the identified potential vulnerability or weakness.
• An assessment of what threats are involved in the possible exploitation of the vulnerability or
weakness.
• A categorization of the result, which can be:
◦ A weakness in the source code.
Weaknesses are issues identified in the source code that are not directly exploitable but may
indicate the existence of exploitable vulnerabilities within the source code.

◦ A nonconformity in the code quality standards.

Nonconformities do not necessarily imply weaknesses, though the rationale for the
requirement is often based on preventing weaknesses.

◦ A potential vulnerability in the source code.

Potential vulnerabilities cannot be fully verified, but one or more conditions for the vulnerability
have been observed.

◦ A vulnerability in the source code.

The reviewers have either shown or referenced other parties who have asserted the identified
vulnerability to be exploitable.

• A severity level assigned to the result, which can be one of:

◦ Low severity.
Low severity implies the impact to the product is low, is already mitigated by the system, or the
difficulty in exploitation would likely require indefinite access to the systems, expert knowledge
of the system, or would require cost prohibitive resources.

◦ Medium severity.
Medium severity implies either the impact of exploitation to the product would be significant, or
the difficulty in exploitation would likely require extended access to the systems, informed
knowledge of the system, or would require significant resources.

◦ High severity.
High severity implies either the impact of exploitation to the product would result in complete
compromise of security, or the difficulty in exploitation would likely require little to no access or
knowledge of the systems or little to no resources.

Last update: 2014-11-14 Classification: Public Status: RELEASED

Version: 1.04 ©2014 atsec information security corporation Page 8 of 12
3 Results
The following table summarizes the results that arose from the source code review team's assessment of
the Adjudication component system. Potential exploitation of a weakness or vulnerability and type of
attacker is noted where applicable.
Table 1: Summary of Potential Vulnerabilities in the Adjucation System

ID Description Assessment Categorization

1 Catchall catch-blocks in try- Catchall catch-blocks may not Type: Weakness

catch statements. handle some exceptions Severity: Low
appropriately. Maliciously
crafted input may cause denial
of service or otherwise
undefined behavior.

2 Try-catch statements do not Uncaught exceptions are Type: Weakness

handle all potential thrown to the calling function. Severity: Low
exceptions. Maliciously crafted input may
cause a denial of service.

3 Switch statements do not When no default cases exist, Type: Weakness; VVSG
have default cases. control may pass through the nonconformity
switch statement without Severity: Low
proper processing.

4 Code extends beyond 80- The source code often Type: VVSG nonconformity
character width limit exceeds the limit by only a few Severity: Low
specified by VVSG. characters. In some more rare
cases, it extends further. The
reviewers do not consider this
a security related issue and did
not find that it detracts from
readability.

5 Initialize every variable upon Instances were found where Type: VVSG nonconformity
declaration, and comment its variables are not initialized. Severity: Low
use.

6 Member variables of a class Instances were found where Type: VVSG nonconformity
must be initialized in the the member variables are not Severity: Low
class constructor(s), either initialized.
directly or indirectly

7 Mixed-mode operations exist VVSG states mixed-mode Type: VVSG nonconformity

counter to VVSG operations should be avoided Severity: Low
requirement. or at least clearly explained if
necessary, instances of mixed-
mode operations were found.

8 No detection of overflows in The source code does not Type: Weakness, VVSG
arithmetic performed on vote appear to check for arithmetic nonconformity
counters. overflows anywhere within the Severity: Low
source tree.

Last update: 2014-11-14 Classification: Public Status: RELEASED

Version: 1.04 ©2014 atsec information security corporation Page 9 of 12
ID Description Assessment Categorization

9 Restriction on code size by For the current scope of code, Type: VVSG nonconformity
VVSG requirement stating: the reviewer found that 435 Severity: Low
no more than 50% of all files (70%) are greater than 60
modules exceeding 60 lines lines (VVSG limit is 50%), 267
in length, no more than 5% files (43%) are greater than
of all modules exceeding 120 120 lines (VVSG limit is 5%),
lines in length, and no and 139 files (23%) are greater
modules exceeding 240 lines than 240 lines (VVSG limit is
in length. 0%).

10 SA1501: A statement that is An instance was found where Type: StyleCop nonconformity
wrapped in opening and the statement and the curly Severity: Low
closing curly brackets must braces are on the same line.
be written on a single line.

11 SA1124: Do not place a Code containing ‘#region’ was Type: StyleCop nonconformity
region anywhere within the found. Severity: Low
code.

12 SA1120: When the code An instance was found where Type: StyleCop nonconformity
contains a C# comment it only comment ‘//’ was present Severity: Low
must contain text. without any text.

13 SA1122: The code cannot Two instances were found Type: StyleCop nonconformity
contain empty strings. where a variable was set to “”. Severity: Low

14 The MD5 is not a FIPS One instance of use of MD5 Type: Potential vulnerability, FIPS
Approved algorithm and was found. nonconformity
should not be used. Severity: Low

15 The SHA-1 and MD5 are not One instance of use of SHA-1 Type: Potential vulnerability, FIPS
FIPS Approved Key and MD5 for key generation nonconformity
Generation methods and was found. Severity: Low
should not be used.

16 The Rijndael implementation Several uses of the Rijndael Type: Weakness, FIPS
in Microsoft Cryptographic algorithm were found. nonconformity
library is in non-conformance Severity: Low
to FIPS-197, AES
implementation should be
used instead.

Last update: 2014-11-14 Classification: Public Status: RELEASED

Version: 1.04 ©2014 atsec information security corporation Page 10 of 12
Glossary
AES Advanced Encryption Standard ICC ImageCast Central
API Application Programming Interface ICE ImageCast Evolution
ATI Audio Tactile Interface ICP ImageCast Precinct
AVS Accessible Voting Station IP Internet Protocol
CAVP Cryptographic Algorithm Validation IV Initialization Vector
Program
LAN Local Area Network
CBC Cipher Block Chaining
LDF Log Data File
CMVP Cryptographic Module Validation
MCF Machine Context File
Program
NAS Network Attached Storage
COTS Commercial Off the Shelf
NIST National Institute of Science and
CSP Critical Security Parameter
Technology
CVE Common Vulnerability and
OS Operating System
Exposures
PC Personal Computer
CWE Common Weakness Enumeration
RNG Random Number Generator
DCF Device Configuration File
RRH Result Receiver Host
DCM Data Center Manager
RSA Rivest, Shamir, and Adelman
ECDSA Elliptic Curve Digital Signature
Algorithm RTM Result Transfer Manager
EED Election Event Designer RTR Results Tally and Reporting
EMS Electronic Management System SHA Secure Hash Algorithm
FIPS Federal Information Processing TCP Transmission Control Protocol
Standard
USB Universal Serial Bus
HMAC Hash Message Authentication
VIF Voter Information File
Code
VVSG Voluntary Voter System Guidelines
HTTP Hyper Text Transfer Protocol

HTTPS Hyper Text Transfer Protocol

Secure

Last update: 2014-11-14 Classification: Public Status: RELEASED

Version: 1.04 ©2014 atsec information security corporation Page 11 of 12
References
Documentation provided for the source code review included Dominion Democracy Suite product
documentation, as well as other publically available standards documents. The atsec source team also
consulted other publically available documents.
Dominion Democracy Suite General Documentation
[DEMC] Dem Suite EMS Coding Standard C#.docx Revision 0.0.98.
Dominion Democracy Suite EMS Documentation
[EMSFUN] 2.03 - EMS Functionality Description Version: 4.14.A-1::253 September 19, 2014
Dominion Democracy Suite Adjudication Documentation
[ADJSDS] 2.05 - Adjudication Software Design and Specification, Version: 4.14.D::25, Dominion
Voting Systems, September 18, 2014.
[ASMM] 2.09 – Adjudication System Maintenance Manual, version 4.14.D::9, September 18,
2014.
[DSIAP] Democracy Suite ImageCast Adjudication Installation and Configuration Procedures,
Version: 4.14.C::48, July 17, 2014
Public Documents
[CERTJ] Long, et al., The CERT Oracle Secure Coding Standard for Java, Addison-Wesley,
Upper Saddle River, NJ, 2012.
[FCAM] Xie, T., Liy, F., Feng, D.,: Fast Collision Attack on MD5. Cryptology ePrint Archive,
Report 2013/170 (2013),
https://eprint.iacr.org/2013/170.pdf.
[FIPS140-2C] “Annex C: Approved Random Number Generators for FIPS PUB 140-2, Security
Requirements for Cryptographic Modules,” Information Technology Laboratory,
National Institute of Technology, February, 2012.
[NFR] SANS Institute InfoSec Reading Room, .NET Framework Rootkits: Backdoors inside
your Framework, November, 2008,
http://www.sans.org/reading-room/whitepapers/windowsnet/dotnet-framework-rootkits-
backdoors-framework-32954.
[SP800131A] NIST Special Publication 800-131A, “Transitions: Recommendation for Transitioning
the Use of Cryptographic Algorithms and Key Lengths,” U. S. Department of
Commerce and National Institute of Standards and Technology, January, 2011.
[TRUST] Thompson, Ken, “Reflections on Trusting Trust,” Communications of the ACM, Volume
27, Number 8, August, 1984, http://dl.acm.org/citation.cfm?id=358210.
[VVSG1] United States Election Assistance Commission, 2005 Voluntary Voting System
Guidelines, Volume 1, Version 1.0, 2005.
[VVSG2] United States Election Assistance Commission, 2005 Voluntary Voting System
Guidelines, Volume 2, Version 1.0, 2005.

Last update: 2014-11-14 Classification: Public Status: RELEASED

Version: 1.04 ©2014 atsec information security corporation Page 12 of 12