OpenSSF

Authors: Anne Bertucio, Christopher Robinson, David Wheeler, OpenSSF Vulnerability Disclosure WG members

https://github.com/ossf/oss-vulnerability-guide/blob/main/maintainer-guide.md

Vulnerability disclosure is the process of reporting, remediating, and communicating the details of a discovered vulnerability.  This is a critical component of software security both for the software communities that create the code as well as the downstream consumers that ingest and use it. It is so critical in fact, that it was one of the requirements of a recent United States Executive Order on improving software supply chain security. Vulnerability disclosure takes an organized effort on both the software maintainers and security researchers (referred to as “finders”). Within open source projects, this effort typically falls to the project maintainers.

A common saying in the vulnerability disclosure and incident response field is to, “have a plan before you need a plan.” Many open source maintainers have little-to-no familiarity with what a vulnerability disclosure plan should be. Maintainers are experts at creatively solving problems through code, not necessarily at being experts in the area of software security.  While many may have familiarity with secure coding concepts, they have little to no time for creating and drafting a plan for their project. The end result is open source projects without vulnerability disclosure policies, finders without directions on how to report, and users without a clear way to get information on vulnerabilities that may affect them.

Today the OpenSSF is releasing a guide and resources on coordinated vulnerability disclosure (CVD) for open source projects.  This guide was created by the OpenSSF Vulnerability Disclosure Working Group and has been informed by broadly-accepted industry good practices around CVD. The guide takes maintainers through CVD from pre-report preparations to publicly disclosing vulnerabilities, and puts the steps of CVD in the context of open source software development. The guide also includes commonly-needed policy and communication templates, such as a security policy (frequently referred to as a SECURITY.md), embargo notifications, and disclosure announcements. 

The Open Source ecosystem is broad and diverse. While projects may need to modify the resources for their project, the OpenSSF hopes that this encourages project maintainers who are unfamiliar with vulnerability disclosure to learn and adopt CVD for their projects, and simplifies implementation for the disclosure-familiar. These tools and practices can help improve the overall security and awareness of every community that integrates them on whatever level the project can.

This guide borrows the approaches of other open source project disclosure efforts: the Google Guide to CVD for OSS projects, the OpenStack Vulnerability Management Process, and the Kubernetes Security and Disclosure Process. 

This CVD guide is just one of many projects that the Open Source Security Foundation is actively working on to improve security within the OSS ecosystem. The OpenSSF is focused on the incredibly broad spectrum of open source software and seeks to improve the lives of developers, projects, and end-consumers of these fantastic communities.

The guide and resources are available on the OpenSSF GitHub.

Authors: Mike Maraya, Jeff Mendoza

We’re excited to announce Allstar, a GitHub app that provides automated continuous enforcement of security best practices for GitHub projects. With Allstar, owners can check for security policy adherence, set desired enforcement actions, and continuously enact those enforcements when triggered by a setting or file change in the organization or project repository. Allstar will help the open source community proactively reduce security risk while adding as little friction as possible.

Allstar is a companion to Security Scorecards, an automated tool that assesses risk to a repository and its dependencies. Security Scorecards checks a number of important heuristics (currently 18), such as whether the project uses branch protection, cryptographically signs release artifacts, or requires code review. From these scores, users can understand specific areas to improve in order to strengthen the security posture of their project. From here, Allstar takes the next step and allows maintainers to opt into automated enforcement of specific checks. If your repository fails a particular check that you enable, Allstar intervenes to make the necessary changes to remediate the issue, avoiding the extra effort of regular manual fixes. In short, Security Scorecards helps you measure your current security posture against where you want to be; Allstar helps you get there.

Continuous Automated Enforcement

Allstar works by continuously checking expected GitHub API states and repository file contents (repository settings, branch settings, workflow settings) against defined security policies and applying enforcement actions (filing issues, changing the settings) when expected states do not match the policies. The continuous nature of the enforcement protects against stealthy attacks that human enforcement might not notice: Allstar will detect and respond to a policy violation if someone, for example, temporarily disables branch protections in order to commit a malicious change before reenabling the protections. 

OpenSSF runs an Allstar instance that anyone can install and use. However, you can create and run your own Allstar instance for security or customization reasons.

User-Defined Enforcement Actions

Allstar lets you pick the enforcement actions that make sense for the organization, the repository, and the specific policies you’ve enabled. The following enforcement actions are available today, with more planned for the future:

• Log the security policy adherence failure with no additional action
• Open a GitHub issue
• Revert the modified GitHub policy setting to match the original Allstar configuration

Security Policy Enforcements Available Today

A limited number of security policy checks are currently enforced by Allstar, with additional policies planned in the coming months. Here’s what’s up and running so far:

Branch Protection

Branch protection sets requirements before a collaborator can push changes to a branch in your repository. Allstar can enforce the following requirements:

• Require approval on pull requests, which helps meet the code review requirement for Supply-chain Levels for Software Artifacts (SLSA)
• Set a number of required pull request approvals
• Dismiss stale pull request approvals
• Block force pushes

Security Policy

A defined policy for responsible vulnerability disclosure helps protect the users of your project, ensuring that you have a chance to remediate an issue before public disclosure. Allstar can enforce the presence of a security policy file (SECURITY.md).

Outside Collaborator Administrators

Allstar can enforce a requirement that users with administrator privileges on a repository be members of the owning organization. It can also disallow push access for outside collaborators. 

Binary Artifacts

Binary artifacts in a repository are threat vectors that cannot be accurately reviewed by a human. Allstar will detect these and alert the user if found.

What’s Next

Here are some of the enforcements we’re looking to build in future releases:

Automatic Dependency Update

Security vulnerabilities are regularly discovered and fixed in open source packages. Automatically updating your dependencies helps keep known vulnerabilities out of your project. Allstar will be able to ensure that automatic dependency updates via Dependabot or Renovate are enabled on your repository.

Frozen Dependencies

Automatic incorporation of new dependency versions without review is an attack vector. A lock file or similar language-specific pinning file can protect against a compromised dependency release making its way into your project. Allstar will be able to detect and enforce the presence of language-specific dependency pinning.

Get Involved

Allstar is still in the early stages of development, so we welcome adoption and community feedback. You can get started using Allstar and help improve it by submitting issues and/or pull requests for new additions. We look forward to rolling out more enforcements; in the meanwhile, taking simple steps like enforcing code review and setting branch protections can make a significant difference in protecting against supply-chain attacks. Taking these fundamental actions together can help raise the bar for security standards in open source software.

Our communities take security seriously and have been instrumental in creating the tools and standards that every organization needs to comply with the recent US Executive Order

Overview

The US White House recently released its Executive Order (EO) on Improving the Nation’s Cybersecurity (along with a press call) to counter “persistent and increasingly sophisticated malicious cyber campaigns that threaten the public sector, the private sector, and ultimately the American people’s security and privacy.”

In this post, we’ll show what the Linux Foundation’s communities have already built that support this EO and note some other ways to assist in the future. But first, let’s put things in context.

The Linux Foundation’s Open Source Security Initiatives In Context

We deeply care about security, including supply chain (SC) security. The Linux Foundation is home to some of the most important and widely-used OSS, including the Linux kernel and Kubernetes. The LF’s previous Core Infrastructure Initiative (CII) and its current Open Source Security Foundation (OpenSSF) have been working to secure OSS, both in general and in widely-used components. The OpenSSF, in particular, is a broad industry coalition “collaborating to secure the open source ecosystem.”

The Software Package Data Exchange (SPDX) project has been working for the last ten years to enable software transparency and the exchange of software bill of materials (SBOM) data necessary for security analysis. SPDX, recognized and implemented as ISO/IEC standard 5962:2021, is supported by global companies with massive supply chains, and has a large open and closed source tooling support ecosystem. SPDX already meets the requirements of the executive order for SBOMs.

Finally, several LF foundations have focused on the security of various verticals. For example,  LF Public Health and LF Energy have worked on security in their respective sectors. Our cloud computing industry collaborating within CNCF has also produced a guide for supporting software supply chain best practices for cloud systems and applications.

Given that context, let’s look at some of the EO statements (in the order they are written) and how our communities have invested years in open collaboration to address these challenges.

Best Practices

The EO 4(b) and 4(c) says that

The “Secretary of Commerce [acting through NIST] shall solicit input from the Federal Government, private sector, academia, and other appropriate actors to identify existing or develop new standards, tools, and best practices for complying with the standards, procedures, or criteria [including] criteria that can be used to evaluate software security, include criteria to evaluate the security practices of the developers and suppliers themselves, and identify innovative tools or methods to demonstrate conformance with secure practices [and guidelines] for enhancing software supply chain security.” Later in EO 4(e)(ix) it discusses “attesting to conformity with secure software development practices.”

The OpenSSF’s CII Best Practices badge project specifically identifies best practices for OSS, focusing on security and including criteria to evaluate the security practices of developers and suppliers (it has over 3,800 participating projects). LF is also working with SLSA (currently in development) as potential additional guidance focused on addressing supply chain issues further.

Best practices are only useful if developers understand them, yet most software developers have never received education or training in developing secure software. The LF has developed and released its Secure Software Development Fundamentals set of courses available on edX to anyone at no cost. The OpenSSF Best Practices Working Group (WG) actively works to identify and promulgate best practices. We also provide a number of specific standards, tools, and best practices, as discussed below.

Encryption and Data Confidentiality

The EO 3(d) requires agencies to adopt “encryption for data at rest and in transit.” Encryption in transit is implemented on the web using the TLS (“https://”) protocol, and Let’s Encrypt is the world’s largest certificate authority for TLS certificates.

In addition, the LF Confidential Computing Consortium is dedicated to defining and accelerating the adoption of confidential computing. Confidential computing protects data in use (not just at rest and in transit) by performing computation in a hardware-based Trusted Execution Environment. These secure and isolated environments prevent unauthorized access or modification of applications and data while in use.

Supply Chain Integrity

The EO 4(e)(iii) states a requirement for

 “employing automated tools, or comparable processes, to maintain trusted source code supply chains, thereby ensuring the integrity of the code.” 

The LF has many projects that support SC integrity, in particular:

• in-toto is a framework specifically designed to secure the integrity of software supply chains.
• The Update Framework (TUF) helps developers maintain the security of software update systems, and is used in production by various tech companies and open source organizations.  
• Uptane is a variant of TUF; it’s an open and secure software update system design which protects software delivered over-the-air to the computerized units of automobiles.
• sigstore is a project to provide a public good / non-profit service to improve the open source software supply chain by easing the adoption of cryptographic software signing (of artifacts such as release files and container images) backed by transparency log technologies (which provide a tamper-resistant public log). 
• We are also funding focused work on tools to ease signature and verify origins, e.g., we’re working to extend git to enable pluggable support for signatures, and the patatt tool provides an easy way to provide end-to-end cryptographic attestation to patches sent via email.
• OpenChain (ISO 5230) is the International Standard for open source license compliance. Application of OpenChain requires identification of OSS components. While OpenChain by itself focuses more on licenses, that identification is easily reused to analyze other aspects of those components once they’re identified (for example, to look for known vulnerabilities).

Software Bill of Materials (SBOMs) support supply chain integrity; our SBOM work is so extensive that we’ll discuss that separately.

Software Bill of Materials (SBOMs)

Many cyber risks come from using components with known vulnerabilities. Known vulnerabilities are especially concerning in key infrastructure industries, such as the national fuel pipelines,  telecommunications networks, utilities, and energy grids. The exploitation of those vulnerabilities could lead to interruption of supply lines and service, and in some cases, loss of life due to a cyberattack.

One-time reviews don’t help since these vulnerabilities are typically found after the component has been developed and incorporated. Instead, what is needed is visibility into the components of the software environments that run these key infrastructure systems, similar to how food ingredients are made visible.

A Software Bill of Materials (SBOM) is a nested inventory or a list of ingredients that make up the software components used in creating a device or system. This is especially critical as it relates to a national digital infrastructure used within government agencies and in key industries that present national security risks if penetrated. The use of SBOMs would improve understanding of the operational and cyber risks of those software components from their originating supply chain.

The EO has extensive text about requiring a software bill of materials (SBOM) and tasks that depend on SBOMs:

• EO 4(e) requires providing a purchaser an SBOM “for each product directly or by publishing it on a public website” and “ensuring and attesting… the integrity and provenance of open source software used within any portion of a product.” 
• It also requires tasks that typically require SBOMs, e.g., “employing automated tools, or comparable processes, that check for known and potential vulnerabilities and remediate them, which shall operate regularly….” and “maintaining accurate and up-to-date data, provenance (i.e., origin) of software code or components, and controls on internal and third-party software components, tools, and services present in software development processes, and performing audits and enforcement of these controls on a recurring basis.” 
• EO 4(f) requires publishing “minimum elements for an SBOM,” and EO 10(j) formally defines an SBOM as a “formal record containing the details and supply chain relationships of various components used in building software…  The SBOM enumerates [assembled] components in a product… analogous to a list of ingredients on food packaging.”

The LF has been developing and refining SPDX for over ten years; SPDX is used worldwide and is approved as ISO/IEC International Standard 5962:2021.  SPDX is a file format that identifies the software components within a larger piece of computer software and metadata such as the licenses of those components. SPDX 2.2 already supports the current guidance from the National Telecommunications and Information Administration (NTIA) for minimum SBOM elements. Some ecosystems have ecosystem-specific conventions for SBOM information, but SPDX can provide information across all arbitrary ecosystems.

SPDX is real and in use today, with increased adoption expected in the future. For example:

• An NTIA “plugfest” demonstrated ten different producers generating SPDX. SPDX supports acquiring data from different sources (e.g., source code analysis, executables from producers, and analysis from third parties). 
• A corpus of some LF projects with SPDX source SBOMs is available. 
• Various LF projects are working to generate binary SBOMs as part of their builds, including yocto and Zephyr. 
• To assist with further SPDX adoption, the LF is paying to write SPDX plugins for major package managers.

Vulnerability Disclosure

No matter what, some vulnerabilities will be found later and need to be fixed. EO 4(e)(viii) requires “participating in a vulnerability disclosure program that includes a reporting and disclosure process.” That way, vulnerabilities that are found can be reported to the organizations that can fix them. 

The CII Best Practices badge passing criteria requires that OSS projects specifically identify how to report vulnerabilities to them. More broadly, the OpenSSF Vulnerability Disclosures Working Group is working to help “mature and advocate well-managed vulnerability reporting and communication” for OSS. Most widely-used Linux distributions have a robust security response team, but the Alpine Linux distribution (widely used in container-based systems) did not. The Linux Foundation and Google funded various improvements to Alpine Linux, including a security response team.

We hope that the US will update its Vulnerabilities Equities Process (VEP) to work more cooperatively with commercial organizations, including OSS projects, to share more vulnerability information. Every vulnerability that the US fails to disclose is a vulnerability that can be found and exploited by attackers. We would welcome such discussions.

Critical Software

It’s especially important to focus on critical software — but what is critical software? EO 4(g) requires the executive branch to define “critical software,” and 4(h) requires the executive branch to “identify and make available to agencies a list of categories of software and software products… meeting the definition of critical software.”

Linux Foundation and the Laboratory for Innovation Science at Harvard (LISH) developed the report Vulnerabilities in the Core,’ a Preliminary Report and Census II of Open Source Software, which analyzed the use of OSS to help identify critical software. The LF and LISH are in the process of updating that report. The CII identified many important projects and assisted them, including OpenSSL (after Heartbleed), OpenSSH,  GnuPG, Frama-C, and the OWASP Zed Attack Proxy (ZAP). The OpenSSF Securing Critical Projects Working Group has been working to better identify critical OSS projects and to focus resources on critical OSS projects that need help. There is already a first-cut list of such projects, along with efforts to fund such aid.

Internet of Things (IoT)

Unfortunately, internet-of-things (IoT) devices often have notoriously bad security. It’s often been said that “the S in IoT stands for security.” 

EO 4(s) initiates a pilot program to “educate the public on the security capabilities of Internet-of-Things (IoT) devices and software development practices [based on existing consumer product labeling programs], and shall consider ways to incentivize manufacturers and developers to participate in these programs.” EO 4(t) states that such “IoT cybersecurity criteria” shall “reflect increasingly comprehensive levels of testing and assessment.”

The Linux Foundation develops and is home to many of the key components of IoT systems. These include:

• The Linux kernel, used by many IoT devices. 
• The yocto project, which creates custom Linux-based systems for IoT and embedded systems. Yocto supports full reproducible builds. 
• EdgeX Foundry, which is a flexible OSS framework that facilitates interoperability between devices and applications at the IoT edge, and has been downloaded millions of times. 
• The Zephyr project, which provides a real-time operating system (RTOS) used by many for resource-constrained IoT devices and is able to generate SBOM’s automatically during build. Zephyr is one of the few open source projects that is a CVE Numbering Authority.
• The seL4 microkernel, which is the most assured operating system kernel in the world; it’s notable for its comprehensive formal verification.

Security Labeling

EO 4(u) focuses on identifying:

“secure software development practices or criteria for a consumer software labeling program [that reflects] a baseline level of secure practices, and if practicable, shall reflect increasingly comprehensive levels of testing and assessment that a product may have undergone [and] identify, modify, or develop a recommended label or, if practicable, a tiered software security rating system.”

The OpenSSF’s CII Best Practices badge project (noted earlier) specifically identifies best practices for OSS development, and is already tiered (passing, silver, and gold). Over 3,800 projects currently participate.

There are also a number of projects that relate to measuring security and/or broader quality:

• Community Health Analytics Open Source Software (CHAOSS) focuses on creating analytics and metrics to help define community health and identify risk 
• The OpenSSF Security Metrics Project, which is in the process of development, was created to collect, aggregate, analyze, and communicate relevant security data about open source projects.
• The OpenSSF Security Reviews initiative provides a collection of security reviews of open source software.
• The OpenSSF Security Scorecards provide a set of automated pass/fail checks to provide a quick review of arbitrary OSS.

Conclusion

The Linux Foundation (LF) has long been working to help improve the security of open source software (OSS), which powers systems worldwide. We couldn’t do this without the many contributions of time, money, and other resources from numerous companies and individuals; we gratefully thank them all.  We are always delighted to work with anyone to improve the development and deployment of open source software, which is important to us all.

David A. Wheeler, Director of Open Source Supply Chain Security at the Linux Foundation