CVSS 3.1
Вероятност да бъде експлоатирана в следващите 30 дни (EPSS)
Атакува се
Време до експлоатация
Атакува се по мрежата, без особени условия, с обикновен потребителски вход и без действие от потребителя.
- Експлоатация: няма данни за експлоатация
- Атаката не може да се автоматизира
Описание
Оригинално описание
vLLM is an inference and serving engine for large language models. Prior to 0.32.0, the Phi-4 multimodal audio encoder in phi4mm_audio.py and phi4mm_utils.py uses eager attention by default, allowing a remote client to submit one legal long audio clip that causes MultiHeadedAttention.forward_attention to materialize and retain full attention matrices in self.attn. The scaled dot-product attention branch does not materialize these matrices. The resulting quadratic GPU-memory use can cause an out-of-memory condition and terminate EngineCore for all users. This issue is fixed in version 0.32.0.
Засегнати продукти
- vllm-project vllm: < 0.32.0
Вид слабост
История
- 26 октомври 2026 Публикувано
Други уязвимости в vllm-project vllm
Първоизточници
- https://www.cve.org/CVERecord?id=CVE-2026-115913
- https://nvd.nist.gov/vuln/detail/CVE-2026-115913
- https://github.com/vllm-project/vllm/security/advisories/GHSA-8397-8qf6-f2gm
- https://github.com/vllm-project/vllm/pull/60614
- https://github.com/vllm-project/vllm/commit/3d52578e13454de0bb8352cac988347cdcc89928
- https://github.com/vllm-project/vllm/releases/tag/v0.32.0