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

Back
New CVE Detected

CVE-2026-93688 – Unauthenticated Memory Exhaustion (DoS) – SGLang ≤ 0.5.19

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

Subscribe to Email Alerts
Subscribe to RSS Feed

Summary

CVE-2026-93688 is a resource-exhaustion vulnerability (denial of service) affecting SGLang, an open-source LLM serving framework, when deployed in prefill/decode disaggregation mode with the Mooncake KV transfer backend. SGLang versions through 0.5.19 fail to validate bootstrap_room values submitted to the decode engine, allowing an unauthenticated attacker to force unbounded memory allocation on the prefill process until it is killed by the operating system. The flaw carries a CVSS v4.0 base score of 8.7 (High).

Technical details

  • Root cause: In SGLang’s Mooncake KV-transfer connector, the prefill-side bootstrap control thread unconditionally creates a transfer_infos[room] entry for every bootstrap_room value it receives, with no validation that the room corresponds to a legitimate, in-flight request.
  • Trigger conditions: SGLang must be running in prefill/decode (PD) disaggregation mode using the Mooncake KV transfer backend; the decode engine’s HTTP API must be network-reachable.
  • Attack vector: An attacker sends repeated POST /generate requests to the decode engine, each carrying a unique, attacker-chosen bootstrap_room value. These orphaned rooms are never cleaned up because the existing cleanup path only fires when KV chunks are actually received, which orphan rooms never get.
  • Impact: Sustained injection of unique room identifiers causes memory on the prefill process to grow without bound until the kernel OOM killer terminates it, taking down the entire PD-disaggregation deployment. Recovery requires a full restart of the prefill process, and the attack can be repeated indefinitely. No confidentiality or integrity impact has been reported.

Affected software

  • SGLang (sgl-project/sglang), all versions through 0.5.19, when configured for prefill/decode disaggregation with the Mooncake KV transfer backend.

Severity

  • CVSS v4.0: 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
  • CVSS v3.1: 7.5 (High) — CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H
  • CWE-770: Allocation of Resources Without Limits or Throttling

Mitigation and recommended actions

  • Immediate: As of publication, no official SGLang release resolves this issue; a fix is proposed in an open, unmerged pull request against the project’s Mooncake connector (adding admission control, bounded LRU tracking of terminal rooms, and a TTL-based sweeper for orphaned entries). Track the upstream repository and upgrade to a patched release as soon as one is published.
  • If no patch is available:
    • Do not expose the decode engine’s HTTP endpoint (/generate) directly to untrusted networks; restrict access to the PD-disaggregation cluster to trusted internal callers only, via network segmentation, firewall rules, or a service mesh/VPN.
    • Place an authenticating reverse proxy or API gateway in front of decode engine endpoints so only authorized clients can submit generation requests.
    • Monitor prefill process memory usage and configure alerting/auto-restart on abnormal growth, and consider running the prefill process under a memory cgroup limit to contain impact and speed recovery.
    • Where feasible, disable or avoid running PD-disaggregation with the Mooncake backend on internet-facing deployments until an upstream fix is available.

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