漏洞信息
尽管我们使用了先进的大模型技术,但其输出仍可能包含不准确或过时的信息。神龙努力确保数据的准确性,但请您根据实际情况进行核实和判断。
Vulnerability Title
vLLM before 0.28.0 Denial of Service via audio extraction
Vulnerability Description
vLLM versions >=0.10.2 and <0.28.0 do not apply any audio decode-size or duration limit when extracting audio from video input for NanoNemotronVL models. In nano_nemotron_vl.py, _extract_audio_from_videos calls load_audio_pyav(BytesIO(video_bytes)) without the max_duration_s or max_decode_bytes parameters, so neither VLLM_MAX_AUDIO_DECODE_DURATION_S nor VLLM_MAX_AUDIO_DECODE_BYTES is enforced (unlike the direct audio upload path in AudioMediaIO). When a NanoNemotronVL model is served with use_audio_in_video=True, an attacker who supplies a small, highly compressed video as multimodal input can force the server to allocate gigabytes of memory during audio decoding, resulting in a denial of service. Fixed in vLLM 0.28.0.
CVSS Information
CVSS:3.1/AV:L/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
Vulnerability Type
未加控制的资源消耗(资源穷尽)
Vulnerability Title
vLLM 资源管理错误漏洞
Vulnerability Description
vLLM是vLLM团队开源的一个适用于 LLM 的高吞吐量和内存高效推理和服务引擎。 vLLM 0.10.2版本至0.28.0之前版本存在资源管理错误漏洞,该漏洞源于从视频输入提取音频时未限制音频解码大小或时长,_extract_audio_from_videos调用load_audio_pyav时未传入max_duration_s和max_decode_bytes参数,攻击者提供小型高压缩视频可导致服务器在音频解码期间分配大量内存,造成拒绝服务。
CVSS Information
N/A
Vulnerability Type
N/A