Information Exposure Affecting tritonserver-backend-vllm-cuda-13.0 package, versions <25.11-r12

Note: Versions mentioned in the description apply only to the upstream tritonserver-backend-vllm-cuda-13.0 package and not the tritonserver-backend-vllm-cuda-13.0 package as distributed by Chainguard. See How to fix? for Chainguard relevant fixed versions and status.

vLLM is an inference and serving engine for large language models (LLMs). From 0.5.5 until 0.23.1rc0, integer truncation of tensor dimensions in vLLM's GGUF dequantize kernels (csrc/quantization/gguf/gguf_kernel.cu) causes partial tensor processing. The output tensor is allocated at full size via torch::empty (uninitialized memory), but the dequantize CUDA kernel processes only a truncated number of elements. The unfilled portion of the output tensor retains whatever was previously in GPU memory. In multi-tenant inference deployments, this residual GPU memory may contain tensor data from other users' inference requests, constituting information disclosure. This vulnerability is fixed in 0.23.1rc0.

• https://github.com/vllm-project/vllm/commit/f219788f91952827132fa4fdf916427cd20d225e
• https://github.com/vllm-project/vllm/pull/44971
• https://github.com/vllm-project/vllm/security/advisories/GHSA-5jv2-g5wq-cmr4