Traditionally, microcontroller-based systems have not placed much emphasis on security. They have usually been thought of as isolated, disconnected from the world, and not very vulnerable, just because of the difficulty in accessing them. The Internet of Things has changed this. Now, code running on small microcontrollers often has access to the internet, or at least to other devices (that may themselves have vulnerabilities). Given the volume they are often deployed at, uncontrolled access can be devastating 1.
This document describes the requirements and process for ensuring security is addressed within the Zephyr project. All code submitted should comply with these principles.
Much of this document comes from [CIIBPB].
Introduction and Scope¶
This document covers guidelines for the Zephyr Project, from a security perspective. Many of the ideas contained herein are captured from other open source efforts.
We begin with an overview of secure design as it relates to Zephyr. This is followed by a section on Secure development knowledge, which gives basic requirements that a developer working on the project will need to have. This section gives references to other security documents, and full details of how to write secure software are beyond the scope of this document. This section also describes vulnerability knowledge that at least one of the primary developers should have. This knowledge will be necessary for the review process described below this.
Following this is a description of the review process used to incorporate changes into the Zephyr codebase. This is followed by documentation about how security-sensitive issues are handled by the project.
Finally, the document covers how changes are to be made to this document.
Designing an open software system such as Zephyr to be secure requires adhering to a defined set of design standards. In [SALT75], the following, widely accepted principles for protection mechanisms are defined to help prevent security violations and limit their impact:
Open design as a design guideline incorporates the maxim that protection mechanisms cannot be kept secret on any system in widespread use. Instead of relying on secret, custom-tailored security measures, publicly accepted cryptographic algorithms and well established cryptographic libraries shall be used.
Economy of mechanism specifies that the underlying design of a system shall be kept as simple and small as possible. In the context of the Zephyr project, this can be realized, e.g., by modular code [PAUL09] and abstracted APIs.
Complete mediation requires that each access to every object and process needs to be authenticated first. Mechanisms to store access conditions shall be avoided if possible.
Fail-safe defaults defines that access is restricted by default and permitted only in specific conditions defined by the system protection scheme, e.g., after successful authentication. Furthermore, default settings for services shall be chosen in a way to provide maximum security. This corresponds to the “Secure by Default” paradigm [MS12].
Separation of privilege is the principle that two conditions or more need to be satisfied before access is granted. In the context of the Zephyr project, this could encompass split keys [PAUL09].
Least privilege describes an access model in which each user, program, and thread, shall have the smallest possible subset of permissions in the system required to perform their task. This positive security model aims to minimize the attack surface of the system.
Least common mechanism specifies that mechanisms common to more than one user or process shall not be shared if not strictly required. The example given in [SALT75] is a function that should be implemented as a shared library executed by each user and not as a supervisor procedure shared by all users.
Psychological acceptability requires that security features are easy to use by the developers in order to ensure their usage and the correctness of its application.
In addition to these general principles, the following points are specific to the development of a secure RTOS:
Complementary Security/Defense in Depth: do not rely on a single threat mitigation approach. In case of the complementary security approach, parts of the threat mitigation are performed by the underlying platform. In case such mechanisms are not provided by the platform, or are not trusted, a defense in depth [MS12] paradigm shall be used.
Less commonly used services off by default: to reduce the exposure of the system to potential attacks, features or services shall not be enabled by default if they are only rarely used (a threshold of 80% is given in [MS12]). For the Zephyr project, this can be realized using the configuration management. Each functionality and module shall be represented as a configuration option and needs to be explicitly enabled. Then, all features, protocols, and drivers not required for a particular use case can be disabled. The user shall be notified if low-level options and APIs are enabled but not used by the application.
Change management: to guarantee a traceability of changes to the system, each change shall follow a specified process including a change request, impact analysis, ratification, implementation, and validation phase. In each stage, appropriate documentation shall be provided. All commits shall be related to a bug report or change request in the issue tracker. Commits without a valid reference shall be denied.
Secure development knowledge¶
The Zephyr project must have at least one primary developer who knows how to design secure software.
This requires understanding the following design principles, including the 8 principles from [SALT75]:
economy of mechanism (keep the design as simple and small as practical, e.g., by adopting sweeping simplifications)
fail-safe defaults (access decisions shall deny by default, and projects’ installation shall be secure by default)
complete mediation (every access that might be limited must be checked for authority and be non-bypassable)
open design (security mechanisms should not depend on attacker ignorance of its design, but instead on more easily protected and changed information like keys and passwords)
separation of privilege (ideally, access to important objects should depend on more than one condition, so that defeating one protection system won’t enable complete access. For example, multi-factor authentication, such as requiring both a password and a hardware token, is stronger than single-factor authentication)
least privilege (processes should operate with the least privilege necessary)
least common mechanism (the design should minimize the mechanisms common to more than one user and depended on by all users, e.g., directories for temporary files)
psychological acceptability (the human interface must be designed for ease of use - designing for “least astonishment” can help)
limited attack surface (the set of the different points where an attacker can try to enter or extract data)
input validation with whitelists (inputs should typically be checked to determine if they are valid before they are accepted; this validation should use whitelists (which only accept known-good values), not blacklists (which attempt to list known-bad values)).
A “primary developer” in a project is anyone who is familiar with the project’s code base, is comfortable making changes to it, and is acknowledged as such by most other participants in the project. A primary developer would typically make a number of contributions over the past year (via code, documentation, or answering questions). Developers would typically be considered primary developers if they initiated the project (and have not left the project more than three years ago), have the option of receiving information on a private vulnerability reporting channel (if there is one), can accept commits on behalf of the project, or perform final releases of the project software. If there is only one developer, that individual is the primary developer.
At least one of the primary developers must know of common kinds of errors that lead to vulnerabilities in this kind of software, as well as at least one method to counter or mitigate each of them.
Examples (depending on the type of software) include SQL injection, OS injection, classic buffer overflow, cross-site scripting, missing authentication, and missing authorization. See the CWE/SANS top 25 or OWASP Top 10 for commonly used lists.
Zephyr Security Subcommittee¶
There shall be a “Zephyr Security Subcommittee”, responsible for enforcing this guideline, monitoring reviews, and improving these guidelines.
This team will be established according to the Zephyr Project charter.
The Zephyr project shall use a code review system that all changes are required to go through. Each change shall be reviewed by at least one primary developer that is not the author of the change. This developer shall determine if this change affects the security of the system (based on their general understanding of security), and if so, shall request the developer with vulnerability knowledge, or the secure designer to also review the code. Any of these individuals shall have the ability to block the change from being merged into the mainline code until the security issues have been addressed.
Issues and Bug Tracking¶
The Zephyr project shall have an issue tracking system (such as GitHub) that can be used to record and track defects that are found in the system.
Because security issues are often sensitive, this issue tracking system shall have a field to indicate a security issue. Setting this field shall result in the issue only being visible to the Zephyr Security Subcommittee. In addition, there shall be a field to allow the Zephyr Security Subcommittee to add additional users that will have visibility to a given issue.
This embargo, or limited visibility, shall only be for a fixed duration, with a default being a project-decided value. However, because security considerations are often external to the Zephyr project itself, it may be necessary to increase this embargo time. The time necessary shall be clearly annotated in the issue itself.
The list of issues shall be reviewed at least once a month by the Zephyr Security Subcommittee. This review should focus on tracking the fixes, determining if any external parties need to be notified or involved, and determining when to lift the embargo on the issue. The embargo should not be lifted via an automated means, but the review team should avoid unnecessary delay in lifting issues that have been resolved.