vLLM is an inference and serving engine for large language models (LLMs). Prior to 0.22.0, vLLMs revision pinning controls do not consistently apply to all artifacts loaded for a model. A deployment that supplies –revision or –code-revision can still load dynamic code, GGUF files, image processors, retrieval side weights, or same-repository subfolder weights/config from an unpinned/default revision. This is a supply-chain integrity issue for pinned vLLM deployments. Operators can believe they are serving a reviewed model revision while vLLM resolves behavior-affecting nested or sibling artifacts outside that reviewed revision. This vulnerability is fixed in 0.22.0.
Weakness
The product does not sufficiently verify the origin or authenticity of data, in a way that causes it to accept invalid data.
Affected Software
| Name | Vendor | Start Version | End Version |
|---|---|---|---|
| Vllm | Vllm | * | 0.22.0 (excluding) |
Related Attack Patterns
- https://cwe.mitre.org/data/definitions/111.html
- https://cwe.mitre.org/data/definitions/141.html
- https://cwe.mitre.org/data/definitions/142.html
- https://cwe.mitre.org/data/definitions/148.html
- https://cwe.mitre.org/data/definitions/218.html
- https://cwe.mitre.org/data/definitions/384.html
- https://cwe.mitre.org/data/definitions/385.html
- https://cwe.mitre.org/data/definitions/386.html
- https://cwe.mitre.org/data/definitions/387.html
- https://cwe.mitre.org/data/definitions/388.html
- https://cwe.mitre.org/data/definitions/665.html
- https://cwe.mitre.org/data/definitions/701.html