vLLM 至 0.29.0 版本存在一个内存损坏漏洞,位于 Triton 的 中。该漏洞表现为:提示词 token ID 在对惩罚(penalty)相关的提示词存在位图(prompt-presence bitset)进行索引时,未对词表大小进行边界检查。攻击者可以提交多模态音频请求,其中包含等于词表大小的 token,从而引发越界写操作,导致并发请求的采样器(sampler)状态被破坏,并改变重复惩罚(repetition penalty)的行为。
| Vendor | Product | Version Range | Status |
|---|---|---|---|
| vllm-project | vllm | ≤ 0.29.0 |
affected |
Although we use advanced large model technology, its output may still contain inaccurate or outdated information.Shenlong tries to ensure data accuracy, but please verify and judge based on the actual situation.
| Vendor | Product | Affected Versions | CPE | Subscribe |
|---|---|---|---|---|
| vllm-project | vllm | 0 ~ 0.29.0 | - |
|
| # | POC Description | Source Link | Shenlong Link |
|---|
No public POC found.
Login to generate AI POC| CVE-2026-93592 | 7.5 HIGH | vLLM before 0.28.0 Denial of Service via negative token ID |
| CVE-2026-93840 | 3.7 LOW | vLLM before 0.29.0 Cross-Request Logits Corruption via allowed_token_ids |
No comments yet