CVE-2026-116180

Published: Ott 25, 2026 Last Modified: Ott 25, 2026
ExploitDB:
Other exploit source:
Google Dorks:
HIGH 8,3
Source: [email protected]
Attack Vector: network
Attack Complexity: low
Privileges Required: low
User Interaction: none
Confidentiality: N/A
Integrity: N/A
Availability: N/A
HIGH 7,7
Source: [email protected]
Attack Vector: network
Attack Complexity: low
Privileges Required: low
User Interaction: none
Scope: changed
Confidentiality: none
Integrity: none
Availability: high

Description

vLLM from 0.8.5 before 0.31.0 has a denial of service flaw. LogprobsTensors.empty_cpu() in vllm/v1/outputs.py allocates an uninitialized int32 tensor, and the V1 model runner leaves prompt positions reported as computed by a KV connector unwritten. A client sharing the engine can send prompt_logprobs (or echo with logprobs) against a connector-cached prefix, making tokenizer decode raise OverflowError, which crashes EngineCore and drops all co-tenant requests.

908

Use of Uninitialized Resource

Incomplete
Abstraction: Base Structure: Simple
Common Consequences
Security Scopes Affected:
Confidentiality Availability
Potential Impacts:
Read Memory Read Application Data Dos: Crash, Exit, Or Restart
Applicable Platforms
All platforms may be affected
Likelihood of Exploit: Medium
View CWE Details
https://github.com/vllm-project/vllm/security/advisories/GH…
https://github.com/vllm-project/vllm/security/advisories/GHSA-rhcx-5729-88vg
https://www.vulncheck.com/advisories/vllm-from-0.8.5-before…
https://www.vulncheck.com/advisories/vllm-from-0.8.5-before-0.31.0-cross-tenant…