InternLM 的 LMDeploy 在 0.17.0 及之前版本中,当以 DistServe 的预填充/解码分离(prefill/decode disaggregation)模式运行时,由于代理层(proxy)使用的是面向用户的会话 ID,而非内部调度器的密钥,导致无法正确释放调度器会话。未经身份验证的攻击者可以向代理端点发送补全请求(completion requests),从而累积未释放的调度器元数据和内存,直至预填充工作进程因内存溢出而被系统终止(out-of-memory killed)。
Although we use advanced large model technology, its output may still contain inaccurate or outdated information.Shenlong tries to ensure data accuracy, but please verify and judge based on the actual situation.
| # | POC Description | Source Link | Shenlong Link |
|---|
No public POC found.
Login to generate AI POCNo comments yet