Description
vLLM before 0.29.0 validates allowed_token_ids against tokenizer length instead of model output logits width in SamplingParams._validate_allowed_token_ids(). Attackers can supply token IDs above the output vocabulary that pass validation, causing LogitBiasState to corrupt GPU logits state and allow concurrent requests to sample tokens outside their allowlists.
Affected products
Version statuses are supplied by the assigning organization. Consult the original record for detailed ranges and platform conditions.
vllm-project vllm
Default version status: unaffected
0 to < 0.29.0affected0.29.0unaffected
pkg:pypi/vllmReferences
- https://github.com/vllm-project/vllm/pull/49080 ↗
- https://github.com/vllm-project/vllm ↗
- https://github.com/vllm-project/vllm/commit/5b0e5b69ac1a3884a6479c9537789c95263cc804 ↗
- https://github.com/vllm-project/vllm/blob/v0.28.0/vllm/sampling_params.py#L881-L903 ↗
- https://github.com/vllm-project/vllm/blob/v0.28.0/vllm/v1/worker/gpu/sample/logit_bias.py#L179-L191 ↗
- https://www.vulncheck.com/advisories/vllm-before-0.29.0-cross-request-logits-corruption-via-allowed-token-ids ↗