CVE-2026-100647

[CVE catalog]

Severity

Medium · CVSS 6.9CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:L/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X

Published

Sep 13, 2026, 10:00:00 PM UTC

Updated

Sep 26, 2026, 2:16:47 PM UTC

Source

[NVD] · [CVE record]

EPSS

0.35% · Oct 27, 2026

A flaw was found in vLLM. An unauthenticated remote attacker can cause a Denial of Service (DoS) by sending HTTP requests containing an excessively large cache_salt parameter to exposed API endpoints. Because this input lacks maximum length validation, processing the oversized data consumes excessive CPU resources on the core scheduler thread. This stalling prevents legitimate requests from being processed, resulting in high latency or complete service unavailability for concurrent users. vllm: vLLM: Denial of Service via unbounded cache_salt parameter This vulnerability is rated as Moderate severity because, while it can be triggered remotely without authentication to degrade service availability, it causes a transient scheduling delay without leading to host compromise, data modification, or information disclosure. In Red Hat OpenShift AI and Red Hat AI inference deployments, vLLM processes client prompts through OpenAI-compatible interfaces where prefix caching can evaluate request metadata. If an attacker submits excessively large salt attributes across concurrent API calls, the serialization and hashing workload monopolizes the core scheduling thread, creating significant latency spikes and temporary starvation for benign inference jobs. Because the failure mode is confined to application responsiveness and does not crash the underlying infrastructure or affect confidential model weights, the overall risk remains moderate. The CVSS score(s) listed for this vulnerability do not reflect the associated product's status, and are included for informational purposes to better understand the severity of this vulnerability.