vLLM through 0.29.0 fails to properly clean up decode-side metadata for rejected inference requests in prefill/decode disaggregated deployments. Remote attackers can submit requests with max_tokens=0 to exhaust decode-worker memory without bound until the worker restarts.
01
describes (1)
Related CVEs.
| ID | severity | exploitable | ||
|---|---|---|---|---|
| CVE-2026-93436US | high7.5 | — | ||
vLLM through 0.29.0 fails to properly clean up decode-side metadata for rejected inference requests in prefill/decode disaggregated deployments. Remote attackers can submit request… | ||||
02
references
External citations — patches, advisories, analysis.
- https://github.com/vllm-project/vllmPACKAGE
- https://github.com/vllm-project/vllm/blob/v0.29.0/vllm/distributed/kv_transfer/kv_connector/v1/nixl/push_worker.py#L162-L181WEB
- https://github.com/vllm-project/vllm/pull/55677FIX
- https://www.vulncheck.com/advisories/vllm-through-0.29.0-memory-exhaustion-via-rejected-requestsADVISORY
scry
Advisories like this name the exposed products attackers go after on your perimeter. See what is exposed on your perimeter with a free passive scan.
Free scan