A Framework for Securing Software Update Systems
The Update Framework (TUF) helps developers to secure new or existing software update systems, which are often found to be vulnerable to many known attacks. TUF addresses this widespread problem by providing a comprehensive, flexible security framework that developers can integrate with any software update system. The framework can be easily integrated (or implemented in the native programming languages of these update systems) due to its concise, self-contained architecture and specification.
What Is a Software Update System?
Generally, a software update system is an application (or part of an application) running on a client system that obtains and installs software. These systems typically update the applications installed on client systems to introduce new features, enhancements, and security fixes.
Three major classes of software update systems are:
- Application updaters which are used by applications to update themselves. For example, Firefox updates itself through its own application updater.
- Library package managers such as those offered by many programming languages for installing additional libraries. These are systems such as Python's pip/easy_install + PyPI, Perl's CPAN, Ruby's RubyGems, and PHP's Composer.
- System package managers used by operating systems to update and install all of the software on a client system. Debian's APT, Red Hat's YUM, and openSUSE's YaST are examples of these.
There are literally thousands of different software update systems in common use today. (In fact the average Windows user has about two dozen different software updaters on their machine!)
We are building a library that can be universally (and in most cases transparently) used to secure software update systems.
On the surface, the update procedure followed by a software update system can be regarded as straightforward. Obtaining and installing an update just means:
- Knowing when an update exists.
- Downloading the update.
- Applying the changes introduced by the update.
The problem with this view is that it is only straightforward when there are no malicious parties involved throughout the update procedure. If an attacker is trying to interfere with these seemingly simple steps, there is plenty that they can do.
TUF is designed to perform the first two steps of the above update procedure, while guarding against the majority of attacks that malicious actors have at their disposal; especially those attacks that are overlooked by security-conscious developers.
Let's assume you take the approach that most systems do (at least, the ones that even try to be secure). You download both the file you want and a cryptographic signature of the file. You already know which key you trust to make the signature. You check that the signature is correct and was made by this trusted key. All seems well, right? Wrong. You are still at risk in many ways, including:
- An attacker keeps giving you the same file, so you never realize there is an update.
- An attacker gives you an older, insecure version of a file that you already have, so you download that one and blindly use it thinking it's newer.
- An attacker gives you a newer version of a file you have but it's not the newest one. It's newer to you, but it may be insecure and exploitable by the attacker.
- An attacker compromises the key used to sign these files and now you download a malicious file that is properly signed.
These are just some of the attacks software update systems are vulnerable to when only using signed files. See Security for a full list of attacks and updater weaknesses TUF is designed to prevent.
The following papers provide detailed information on securing software updater systems, TUF's design and implementation details, attacks on package managers, and package management security:
- Diplomat: Using Delegations to Protect Community Repositories
- Survivable Key Compromise in Software Update Systems
- A Look In the Mirror: Attacks on Package Managers
- Package Management Security
What TUF Does
In order to securely download and verify target files, TUF requires a few extra files to exist on a repository. These are called metadata files. TUF metadata files contain additional information, including information about which keys are trusted, the cryptographic hashes of files, signatures on the metadata, metadata version numbers, and the date after which the metadata should be considered expired.
When a software update system using TUF wants to check for updates, it asks TUF to do the work. That is, your software update system never has to deal with this additional metadata or understand what's going on underneath. If TUF reports back that there are updates available, your software update system can then ask TUF to download these files. TUF downloads them and checks them against the TUF metadata that it also downloads from the repository. If the downloaded target files are trustworthy, TUF hands them over to your software update system. See Metadata for more information and examples.
TUF specification document is also available:
TUF Home Page
The home page for the TUF project can be found at: https://updateframework.com
Please visit https://groups.google.com/forum/?fromgroups#!forum/theupdateframework if you would like to contact the TUF team. Questions, feedback, and suggestions are welcomed in this low-volume mailing list.
A group feed is available at: https://groups.google.com/forum/feed/theupdateframework/msgs/atom.xml?num=50
pip - installing and managing Python packages (recommended) Installing from Python Package Index (https://pypi.python.org/pypi). Note: Please use "pip install --no-use-wheel tuf" if your version of pip <= 1.5.6 $ pip install tuf Installing from local source archive. $ pip install <path to archive> Or from the root directory of the unpacked archive. $ pip install .
Instructions for Contributors
Note: Development of TUF occurs on the "develop" branch of this repository.
To facilitate development and installation of edited version of the code base,
developers are encouraged to install Virtualenv,
which is a tool to create isolated Python environments. It includes
setuptools, Python packages that can be used to
install TUF and its dependencies. All installation methods of
virtualenv are outlined in the installation
and instructions for installing locally from source are provided here:
$ curl -O https://pypi.python.org/packages/source/v/virtualenv/virtualenv-15.0.3.tar.gz $ tar xvfz virtualenv-15.0.3.tar.gz $ cd virtualenv-15.0.3 $ python virtualenv.py myVE
Before installing TUF, a couple of its Python dependencies have non-Python dependencies of their own that should installed first. PyCrypto and PyNaCl (third-party dependencies needed by the repository tools) require Python and FFI (Foreign Function Interface) development header files. Debian-based distributions can install these header libraries with apt (Advanced Package Tool.)
$ apt-get install build-essential libssl-dev libffi-dev python-dev
Fedora-based distributions can install these libraries with dnf.
$ dnf install libffi-devel redhat-rpm-config openssl-devel
OS X users can install these header libraries with the Homebrew package manager.
$ brew install python $ brew install libffi
Installation of minimal, optional, development, and testing requirements can then be accomplished with one command:
$ pip install -r dev-requirements.txt
The Update Framework's unit tests can be executed by invoking tox. All supported Python versions are tested, but must already be installed locally.
TUF has four major classes of users: clients, for whom TUF is largely transparent; mirrors, who will (in most cases) have nothing at all to do with TUF; upstream servers, who will largely be responsible for care and feeding of repositories; and integrators, who do the work of putting TUF into existing projects.
An integration requires importing a single module into the new or existing software updater and calling particular methods to perform updates. Generating metadata files stored on upstream servers can be handled by repository tools that we provide for this purpose.
This material is based upon work supported by the National Science Foundation under Grant No. CNS-1345049 and CNS-0959138. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.