CVE-2026-54234
Açıklama (NVD)
vLLM is a high-throughput and memory-efficient inference and serving engine for LLMs. Prior to 0.24.0, a frontend-legal multi-request speculative decoding workload can cause the rejection sampler to produce a recovered token equal to the model vocabulary size boundary value, which is then converted to negative one when the engine selects the next live token for a request and is written back into the drafter's input ids; that out-of-vocabulary value is later consumed by the model's embedding and attention path and crashes the engine worker with a GPU device-side assertion. The same triggering request sequence is reachable through the public gRPC Generate and Abort endpoints, so a remote client that can send generation requests can crash the shared engine worker, aborting concurrent requests and causing a service-wide denial of service for other clients of the deployment until the worker is restarted. This issue is fixed in version 0.24.0.
Yayın: 07.07.2026 · Son değişiklik: 25.10.2026 · Zayıflık: CWE-20, CWE-1284
CVSS 3.1 metrikleri
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
| Saldırı vektörü | Ağ (uzaktan) |
| Karmaşıklık | Düşük |
| Gereken yetki | Yok |
| Kullanıcı etkileşimi | Gerekmez |
| Kapsam | Değişmez |
| Gizlilik | Yok |
| Bütünlük | Yok |
| Erişilebilirlik | Yüksek |
Hesaplanan temel puan: 7.5. CVSS hesaplayıcıda incele