SGLang through 0.5.22 with the Mooncake disaggregation backend accepts unauthenticated decode registration messages on the Prefill rank ZMQ PULL socket, allowing denial of service. Network-adjacent attackers can send many registrations with unique mooncake_session_id values and large nested-buffer frames to exhaust Prefill memory until the scheduler is OOM-killed.
Fix
Nothing published yet: no patch, vendor advisory or mitigation reference on NVD, no vendor advisory in the News for this CVE, not in CISA's KEV, no fixed package version. BaselineWatch does not guess at one.
Change history 1
- First seenCVSS 7.1NVD modified 2026-10-19 14:17
Affected (CPE criteria) 0
NVD has not attached CPE configurations yet.
References 6
- https://github.com/sgl-project/sglang/blob/v0.5.22/python/sglang/srt/disaggregation/common/utils.py#L47-L69
- https://github.com/sgl-project/sglang/blob/v0.5.22/python/sglang/srt/disaggregation/mooncake/conn.py#L2885-L2935
- https://github.com/sgl-project/sglang/blob/v0.5.22/python/sglang/srt/disaggregation/mooncake/conn.py#L889-L909
- https://github.com/sgl-project/sglang/issues/45012
- https://www.vulncheck.com/advisories/sglang-through-0.5.22-memory-exhaustion-via-unauthenticated-mooncake-decode-registrations
- https://github.com/sgl-project/sglang/issues/45012