vLLM: Request-selected PyNvVideoCodec GPU decode bypasses static VRAM reservation
No exploitation in the wild has been reported. EPSS puts the chance of exploitation in the next 30 days at 1%. No fixed version is recorded.
vLLM is an inference and serving engine for large language models. Prior to 0.28.0, request bodies for Chat Completions and Responses can set media_io_kwargs.video.video_backend to pynvvideocodec, and MediaConnector.fetch_video forwards that choice to VideoMediaIO even when startup configuration selected a software decoder. The engine's _reserve_mm_ipc_gpu_memory logic budgets decoder memory only from static configuration, so the request-selected VIDEO_LOADER_REGISTRY backend can create a CUDA context, decoder surfaces, and decoded-frame allocations that were not removed from the engine's KV-cache budget. An attacker able to submit video requests to a video-capable GPU deployment with PyNvVideoCodec installed can exhaust shared GPU memory, causing request failures, worker crashes, or denial of service. The first release containing the fix is version 0.28.0.
Affected
Vendors, products and packages with version ranges
| Vendor | Product | Versions | Fixed |
|---|---|---|---|
| vllm-project | <0.28.0 |
Timeline
Every event on this vulnerability
Embed
A live severity badge for an advisory, README or ticket