Security Flaw in vLLM Inference Engine by vLLM Project
CVE-2025-47277

9.8CRITICAL

Key Information:

Vendor

Vllm-project

Status
Vllm
Vendor
CVE Published:
20 May 2025

What is CVE-2025-47277?

The vLLM inference engine can expose a security risk when using the PyNcclPipe KV cache transfer integration, particularly in versions 0.6.5 through 0.8.4. This vulnerability allows the TCPStore interface to listen on all network interfaces instead of being restricted to a private network as intended, potentially enabling unauthorized access. The issue arises when the --kv-ip parameter, meant for private communication, is misconfigured. A patch in version 0.8.5 resolves this by ensuring the TCPStore binds to a specified private interface, thereby enhancing security against unwanted access.

Affected Version(s)

vllm >= 0.6.5, < 0.8.5

References

https://github.com/vllm-project/vllm/security/advisories/...
x_refsource_CONFIRM
https://github.com/vllm-project/vllm/pull/15988
x_refsource_MISC
https://github.com/vllm-project/vllm/commit/0d6e187e88874...
x_refsource_MISC

CVSS V3.1

Score:
9.8
Severity:
CRITICAL
Confidentiality:
High
Integrity:
High
Availability:
High
Attack Vector:
Network
Attack Complexity:
Low
Privileges Required:
None
User Interaction:
None
Scope:
Unchanged

Timeline

  • Vulnerability published

  • Vulnerability Reserved

.
CVE-2025-47277 : Security Flaw in vLLM Inference Engine by vLLM Project