llamaCVEVulnerability Intelligence
/
Advisory · GitHub · PyPI · vllm

GHSA-Q97C-P368-75C4

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

Parche disponible
Descargar PNG JSON
CVSS
—
medium
Fuente: —
ExplotaciónSin evidenciaNo figura en CISA KEV
Exploits públicos0Ninguno conocido
Parche1 versiónPrimer parche a -10 d de la divulgación
Exposición22 dDesde la divulgación, 28 oct 2026, 4:15 p. m.
Detección19 dDesde la divulgación hasta su ingreso al índice
Timeline

Ciclo de vida de la vulnerabilidad

Hoy
Divulgación
Publicada
Modificada
Explotación
Remediación
Parche 0.32.0
Monitoreo
Detectada
18 oct 26
28 oct 26
29 oct 26
17 nov 26
20 nov 26
18 oct 26
26 oct 26
3 nov 26
11 nov 26
20 nov 26
FechaFaseEventoDetalle
Remediación Parche 0.32.0 vllm · -10 d desde divulgación
Divulgación Publicada GHSA-Q97C-P368-75C4 divulgada
Divulgación Modificada Última actualización del registro
Monitoreo Detectada Ingresó al índice
Resumen
Descripción del advisory
Texto original · GitHub · en inglés

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 and takes the engine down. Any client that can send pre-tokenized input (`/v1/completions` with `prompt` as token ids, `/tokenize`-fed flows, or the `kv_transfer_params["prompt_token_ids"]` reuse path) can trigger it.

## Root cause
`vllm/v1/engine/input_processor.py` accepts a token id when `id <= max(tokenizer.max_token_id, model_config.get_vocab_size() - 1)`.
`vllm/tokenizers/hf.py` computes `max_token_id = max(max(vocab.values()), tokenizer.vocab_size)`. For Llama-2 tokenizers `max(vocab.values()) = 31999` but `tokenizer.vocab_size = 32000`, so `max_token_id = 32000` — one greater than the largest valid id.
The engine bound therefore becomes `max(32000, 31999) = 32000`, and id 32000 is accepted.
The model's embedding has `pad_vocab_size(32000, 64) = 32000` rows (no padding), valid indices 0..31999. At `VocabParallelEmbedding.forward` with `tp_size == 1` there is no masking, so index 32000 hits `F.embedding` out of range.

## Affected
- tp=1 (single GPU per model — common per-copy layout). tp>1 masks via `get_masked_input_and_mask`, so the id is clamped, not an OOB — likely wrong output rather than a crash.
- Models whose vocab is a multiple of 64 with `tokenizer.vocab_size` one past the max real id (Llama-2 family, vocab 32000, is the clear case). Qwen2.5 is NOT affected (padded headroom).

## Suggested fix
max_token_id in vllm/tokenizers/hf.py now uses tokenizer.vocab_size - 1, so an id equal to the vocabulary size fails input validation instead of reaching the embedding. Fixed in MR59491 https://github.com/vllm-project/vllm/pull/59491, released in v0.32.0.

## Reporter
Shijie Lyu (GitHub shijie-lyu lshjhf@gmail.com). Reported privately per SECURITY.

This advisory was fixed in MR59491 https://github.com/vllm-project/vllm/pull/59491.

Versiones

Rangos afectados

Afectada Corregida Última
vllm< 0.32.0
0.32.0
0.34.0
Remediación

Estado del parche por paquete

Paquete / productoEstadoCorregida enDesde divulgaciónÚltima versión
pypivllm
Corregida 0.32.0 -10 d 0.34.0 no afectada
Versiones corregidas0.32.0
Explotación

Arsenal público

Sin exploits, módulos ni PoC públicos conocidos a la fecha de verificación.

Referencias
Advisoryhttps://nvd.nist.gov/vuln/detail/CVE-2026-115916
Otrahttps://github.com/vllm-project/vllm/security/advisories/GHSA-q97c-p368-75c4
Otrahttps://github.com/vllm-project/vllm/pull/59491
Otrahttps://github.com/vllm-project/vllm/commit/37cecc8141e2dcba9dd534897a164d4baaf27229
Otrahttps://github.com/vllm-project/vllm
Otrahttps://github.com/vllm-project/vllm/releases/tag/v0.32.0

Las referencias externas se muestran para copiar: cve.llama.ec no lo redirige a otros sitios.