Логотип exploitDog
Консоль
Логотип exploitDog

exploitDog

github логотип

GHSA-rh4j-5rhw-hr54

Опубликовано: 27 янв. 2025
Источник: github
Github: Прошло ревью
CVSS3: 7.5

Описание

vllm: Malicious model to RCE by torch.load in hf_model_weights_iterator

Description

The vllm/model_executor/weight_utils.py implements hf_model_weights_iterator to load the model checkpoint, which is downloaded from huggingface. It use torch.load function and weights_only parameter is default value False. There is a security warning on https://pytorch.org/docs/stable/generated/torch.load.html, when torch.load load a malicious pickle data it will execute arbitrary code during unpickling.

Impact

This vulnerability can be exploited to execute arbitrary codes and OS commands in the victim machine who fetch the pretrained repo remotely.

Note that most models now use the safetensors format, which is not vulnerable to this issue.

References

Пакеты

Наименование

vllm

pip
Затронутые версииВерсия исправления

< 0.7.0

0.7.0

EPSS

Процентиль: 51%
0.00274
Низкий

7.5 High

CVSS3

Дефекты

CWE-502

Связанные уязвимости

CVSS3: 7.5
redhat
11 месяцев назад

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.

CVSS3: 7.5
nvd
11 месяцев назад

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.

CVSS3: 7.5
debian
11 месяцев назад

vLLM is a library for LLM inference and serving. vllm/model_executor/w ...

EPSS

Процентиль: 51%
0.00274
Низкий

7.5 High

CVSS3

Дефекты

CWE-502