CVE-2025-24357
vllm: Malicious model to RCE by torch.load in hf_model_weights_iterator
7.5
HIGH
CVSS 3.1
EPSS 0.69%
描述
vLLM is a library for LLM inference and serving. vllm/model_executor/weight_utils.py implements hf_model_weights_iterator to load the model checkpoint, which is downloaded from huggingface. It uses the torch.load function and the weights_only parameter defaults to False. When torch.load loads malicious pickle data, it will execute arbitrary code during unpickling. This vulnerability is fixed in v0.7.0.
如何修補 CVE-2025-24357
要修補 CVE-2025-24357,請將受影響套件升級到下列已修補版本。
- —升級至 0.7.0 或更新版本
- —升級至 d3d6bb13fb62da3234addf6574922a4ec0513d04 或更新版本
CVE-2025-24357 正在被利用嗎?
低 — EPSS 為 0.7%,目前沒有觀察到大規模利用活動。
受影響套件(2)
- from 0, < 0.7.0
- from 0, < d3d6bb13fb62da3234addf6574922a4ec0513d04 | from 0, < 0.7.0
CVSS 分數
| 來源 | 版本 | 嚴重程度 | 向量 |
|---|---|---|---|
| osv | CVSS 3.1 | HIGH7.5 | CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:H/A:H |