CVE-2026-69147

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 mediaiokwargs.video.videobackend to pynvvideocodec, and MediaConnector.fetchvideo forwards that choice to VideoMediaIO even when startup configuration...

EUVD-2026-80935

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 mediaiokwargs.video.videobackend to pynvvideocodec, and MediaConnector.fetchvideo forwards that choice to VideoMediaIO even when startup configuration...

CVE-2019-10520

An unprivileged application can allocate GPU memory by calling memory allocation ioctl function and can exhaust all the memory which results in out of memory in Snapdragon Mobile, Snapdragon Voice & Music in QCS405, SD 210/SD 212/SD 205, SD 665, SD 675, SD 712 / SD 710 / SD 670, SD 730, SD 845 / ...

CVE-2019-10520

The CVE-2019-10520 issue is a local, memory-allocator-based vulnerability described as an unprivileged app being able to allocate GPU memory via a memory allocation ioctl, potentially exhausting all memory and causing out-of-memory on Snapdragon devices (Mobile/Voice & Music) across multiple SDM/...