CyberRota Analysis
AI-GeneratedvLLM versions prior to 0.29.0 are vulnerable to a resource-limit bypass due to an issue in the PyNvVideoCodec decoder, allowing unauthenticated attackers to manipulate sampler subclasses in video requests. This exploitation can lead to exceeding configured decoder limits, resulting in potential GPU memory exhaustion. Organizations utilizing vLLM for video processing should prioritize updating to mitigate this low-severity vulnerability.
Public Exploit Signal
A public exploit, PoC, GitHub repository or Metasploit reference was detected for this CVE.
Note: these links are listed for security research and verification purposes only.
Original NVD Description
vLLM before 0.29.0 contains a resource-limit bypass vulnerability in PyNvVideoCodec decoder allocation where sampler subclass shadowing allows independent counter increments. Unauthenticated attackers can select different sampler subclasses in video requests to exceed configured decoder limits and exhaust unaccounted GPU memory.
Related CVEs
Other vulnerabilities affecting the same vendor(s)