Live Exposure Defense: From CVE to Confirmed Exposure in 12 Hours – See more

Back
New CVE Detected

CVE-2026-93436 – Memory Exhaustion (DoS) – vLLM ≤ 0.29.0 (Disaggregated Deployments)

Be the first to know when new zero-days emerge:

Subscribe to Email Alerts
Subscribe to RSS Feed

Summary

CVE-2026-93436 is a memory exhaustion (denial-of-service) vulnerability in vLLM, the open-source large language model inference and serving engine. Affecting versions up to and including 0.29.0, the flaw resides in the NIXL-based disaggregated (prefill/decode-split) deployment mode, where the decode-side worker fails to release tracking metadata for inference requests that are rejected or aborted. An unauthenticated remote attacker can exploit this by repeatedly submitting requests with max_tokens=0, causing worker memory to grow unbounded until the process crashes or must be restarted. The issue is rated HIGH severity (CVSS 8.7 under CVSS 4.0).

Technical details

  • Root cause: In vLLM’s NIXL KV-connector push worker (vllm/distributed/kv_transfer/kv_connector/v1/nixl/push_worker.py), when a decode-side request is rejected before scheduling, the worker enqueues an "empty recv" operation to release the paired prefiller. This empty-recv metadata is stored in the _recving_metadata tracking dictionary, but because no actual data transfer ever completes for it, the entry is never cleaned up.
  • Trigger conditions: Occurs specifically in disaggregated (prefill/decode split) deployments using the NIXL KV connector. Sending completion/chat requests with max_tokens=0 causes the request to be aborted with do_remote_prefill still active, generating an orphaned empty-recv entry each time.
  • Attack vector: Network-based, no authentication or user interaction required (AV:N/AC:L/AT:N/PR:N/UI:N). An attacker only needs API access to the inference endpoint.
  • Impact: Repeated exploitation causes the _recving_metadata dictionary to grow without bound, exhausting worker memory over time and leading to worker crash/restart — a sustained availability (denial-of-service) impact with no confidentiality or integrity loss (VA:H under CVSS 4.0; A:H under CVSS 3.1).

Affected software

  • vllm-project/vllm: all versions from the start of the project through 0.29.0 (inclusive), when deployed in disaggregated prefill/decode mode using the NIXL KV connector.

Severity

  • CVSS v3.1 Base Score: 7.5 (HIGH) — CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
  • CVSS v4.0 Base Score: 8.7 (HIGH) — CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:H/SC:N/SI:N/SA:N
  • Weakness: CWE-401 (Missing Release of Memory after Effective Lifetime)

Mitigation and recommended actions

  • Immediate: Upgrade vLLM to a version containing the fix in PR #55677, which ensures empty-receive metadata with no local blocks is no longer persisted in the push worker’s tracking state, and that the pull worker drops metadata for empty-block notifications unless a real transfer was already registered.
  • If immediate patching is not possible:
    • Monitor decode-worker memory consumption in disaggregated NIXL deployments and set up alerting/auto-restart on abnormal growth.
    • Rate-limit or filter inbound requests that specify max_tokens=0, or reject such requests at a reverse proxy/gateway in front of the inference API.
    • Where feasible, temporarily avoid disaggregated (prefill/decode-split) NIXL deployment mode until patched, using standard (non-disaggregated) serving instead.
    • Restrict inference API access to trusted networks/clients to reduce exposure to unauthenticated abuse.

References

Are you exposed?

Get a free report of your organization’s exposure to this CVE and threat

  • Mapping of all Assets with this Technology
  • Identification of potentially exposed assets to this CVE
  • Confirmation of verified exploitable assets
Get Exposure Report

How IONIX’s External Exposure Management Platform Detects and Validates
Zero-Days to Shrink MTTR

1

Map your entire attack surface (continously)

IONIX uses multi-factor discovery methods, including DNS analysis, certificate mapping, metadata inspection, and more, to automatically map every internet-facing asset across your environment. This includes cloud instances, third-party platforms, shadow IT, and even forgotten infrastructure that traditional tools miss.

2

Monitor for new CVEs

Dozens of threat intel feeds using agentic technology are continuously analyzed to detect the appearance of proof-of-concept code, exploit kits, and indicators of active targeting. IONIX goes further by applying AI to proactively evaluate whether emerging vulnerabilities are likely to be exploited, even before PoCs go public.

3

Identify Potential External Exposures

Not all CVEs matter. IONIX filters vulnerabilities by asking attacker-centric questions: Can it be reached from the internet? Does it require authentication? Is it being exploited in the wild? This dramatically reduces noise and focuses teams on threats that can actually be weaponized.

4

Create Safe, Scalable Exploit Validations

IONIX transforms real-world PoCs into safe, non-intrusive test payloads that can be run in production environments without disruption. These simulations are precisely targeted to the systems that are vulnerable, ensuring rapid validation without unnecessary load.

5

Execute Exploit Validations

By combining context about software stack, versioning, exposure status, and reachability, IONIX ensures that only the right payloads are executed against the right assets, maximizing efficiency and minimizing risk.

6

Drive Fast and Actionable Remediation

Results are routed through integrations with ticketing, SOAR, and SIEM tools. Issues are written in plain language, bundled into remediation clusters, and prioritized based on asset criticality, exploitability, and blast radius. This shortens mean time to remediation (MTTR) and empowers teams to act with confidence.

Are you exposed?

Get a free report of your organization’s exposure to this CVE and threat

  • Mapping of all Assets with this Technology
  • Identification of potentially exposed assets to this CVE
  • Confirmation of verified exploitable assets
Get Exposure Report
Close

Subscribe to Threat Center RSS

Copy/paste the link below into your preferred RSS reader or follow these instructions to subscribe to Slack alerts.

Open RSS feedSlack setup instructions

Close

Get Real-Time CVE Alerts to Your Email

Be the first to know when new zero-days emerge