## vLLM: Out-of-vocabulary prompt token id passes input validation and triggers a CUDA device-side assert (engine crash / DoS)

- 種別: advisory
- 重大度: medium
- 関連技術: python
- CVE: CVE-2026-115916
- 公開日時: 2026-10-28T21:15:08.000Z
- 情報源: GitHub Advisory Database (https://github.com/advisories/GHSA-q97c-p368-75c4)
- パーマリンク: https://vulnpulse.app/items/ghsa-ghsa-q97c-p368-75c4

影響: pip:vllm < 0.32.0 → 0.32.0
For a model whose vocabulary size is an exact multiple of the embedding padding size (64) — e.g. any Llama-2-family model, vocab 32000 — a single request carrying a pre-tokenized prompt with token id equal to the vocab size (32000) passes vLLM's input-id validation but is one past the last embedding row. At tensor-parallel size 1 the embedding lookup does not mask indices, so the id reaches F.embedding out of range and raises a CUDA device-side assert, which poisons the engine's CUDA context an…