The CVE Binary Tool helps you determine if your system includes known vulnerabilities. You can scan binaries for over 200 common, vulnerable components (openssl, libpng, libxml2, expat and others), or if you know the components used, you can get a list of known vulnerabilities associated with an SBOM or a list of components and versions.
cve-bin-tool has an existing fuzz testing setup which is based on Google Atheris. One of the areas it doesn't yet cover is the files used by the language list parsers. These are typically lists of 3rd party components/requirements written in a format to a specific packaging tool for a specific programming language. These may be lists of requirements generated by a human, or they could be generated by a tool.
This particular request is to fuzz the Debian parser which uses control files, but requests will be filed for the other parsers as well. You can see which ones are listed under the security tag.
Why?
Regular fuzz testing can help us find bugs and potential security issues in parsing . While we hope users aren't going to be regularly scanning malicious control files we'd still like to be able to handle things correctly if a file is really malformed.
Be aware that Atheris and its requirements can be a bit finicky to set up and last time we ran a big fuzzing campaign only some versions of Python in some environments actually worked easily. If you find any issues with following the setup docs, or manage to find good workarounds for an environment we haven't mentioned, please file issues or make a PR to add them to our docs.
Create a new proto file (or files) to generate fuzzed control files and add them to our proto files directory: https://github.com/intel/cve-bin-tool/tree/main/fuzz/proto_files. It's ok to have tests against files that are completely garbage, but probably the most interesting bugs will come from files that mostly look correct, and the proto setup will help you do that. If you're not sure how any of this works, you may find it useful to read this primer on structure-aware fuzzing
Make a python file to call your fuzzer. Here's what the cyclonedx fuzzer looks like, as an example. Yours may be considerably different -- feel free to search for other examples and read the Atheris/libfuzzer/protobuf-mutator docs to help you figure out what you need.
Description
cve-bin-tool has an existing fuzz testing setup which is based on Google Atheris. One of the areas it doesn't yet cover is the files used by the language list parsers. These are typically lists of 3rd party components/requirements written in a format to a specific packaging tool for a specific programming language. These may be lists of requirements generated by a human, or they could be generated by a tool.
This particular request is to fuzz the Debian parser which uses
control
files, but requests will be filed for the other parsers as well. You can see which ones are listed under the security tag.Why?
Regular fuzz testing can help us find bugs and potential security issues in parsing . While we hope users aren't going to be regularly scanning malicious
control
files we'd still like to be able to handle things correctly if a file is really malformed.How should I do this?
control
files and add them to our proto files directory: https://github.com/intel/cve-bin-tool/tree/main/fuzz/proto_files. It's ok to have tests against files that are completely garbage, but probably the most interesting bugs will come from files that mostly look correct, and the proto setup will help you do that. If you're not sure how any of this works, you may find it useful to read this primer on structure-aware fuzzing