Servers and applications
CVE-2026-100647
vLLM DoS via unbounded `cache_salt` length: multi-hundred-MB salt stalls the single EngineCore scheduler thread (CPU exhaustion)
- Severity
- Medium 6.9
- Published
- 12 Sept 2026
- Last changed
- 6 Oct 2026
- Added to UpHeron
- 3 Nov 2026
- Fixed in
- 0.29.0
- Exploit chance
- 0.3% in the next 30 daysHigher than 29% of CVEs. EPSS by FIRST, 19 Nov 2026
- Weakness
- Improper input validation (CWE-20)Uncontrolled resource consumption (CWE-400)Allocation without limits (CWE-770)
- Attack
- Over the network · No login needed · No user action needed
- Attacker can
- Slow it down
- CVSS vector
CVSS: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- Also known as
- GHSA-wpww-v874-ph2p
- Sources
- NVD, Project advisories (GitHub)
Description
An unauthenticated remote attacker can exhaust the CPU of the single EngineCore scheduler thread and stall every request on the server by sending HTTP requests with an oversized cache_salt parameter. The cache_salt field — accepted on the OpenAI-compatible chat completions, completions, responses and pooling endpoints and on the Anthropic /v1/messages endpoint — is validated only with min_length=1 and has no maximum length. The full salt string is embedded verbatim into block-0's prefix-cache extra keys and hashed with sha256(pickle.dumps(...)) on vLLM's single EngineCore scheduler thread during request construction, so each request costs O(salt bytes) of serialization plus SHA-256. Measured end-to-end on a live server, a 10 MB salt adds ~58 ms of request latency, 100 MB ~599 ms, and 200 MB ~1195 ms (baseline ~9 ms); under 6 concurrent 100 MB-salt attackers, unrelated ordinary requests are stalled to p90 ≈ 1.8 s / max ≈ 2.5 s (idle median 8.5 ms).
Affected software
| Software | Affected versions |
|---|---|
vLLMServers and applications · cpe:a:vllm:vllm |
|
vLLMServers and applications · github:vllm-project/vllm |
|
About vLLM
open-source software for large language model inference
- Docker Hub pulls
- 44,195,225Docker Hub, 15 Nov 2026
- GitHub stars
- 96,943GitHub, 15 Nov 2026
- Forks
- 24,836GitHub, 15 Nov 2026
- Latest release
- v0.33.0rc3GitHub, 15 Nov 2026
- Released
- 2 Nov 2026GitHub
- License
- Apache Software License 2.0Wikidata
- First published on GitHub
- 9 Feb 2023GitHub
Website Source code Docker image
Vulnerabilities in vLLM
- In our list
- 169
- Published in the last 12 months
- 150
- Critical, last 12 months
- 6
- High, last 12 months
- 63
- Known exploited
- 0
- CVE-2026-130686vLLM 0.6.0 before 0.33.0 Resource Exhaustion via Multi-Image RequestsHigh 8.7 18 Nov 2026
- CVE-2026-130684vLLM before 0.33.0 Path Traversal via speculators_config verifier.name_or_pathMedium 5.3 18 Nov 2026
- CVE-2026-130685vLLM 0.28.0 before 0.33.0 Unauthenticated DoS via Out-of-Range Prompt Token IDHigh 8.7 18 Nov 2026
- GHSA-xgh8-qwjh-q853/v1/systemone bypasses the --trust-request-chat-template gateMedium 6.5 18 Nov 2026
- CVE-2026-129356vllm-project vllm Common Prefix single_type_kv_cache_manager.py race conditionLow 2.3 15 Nov 2026
- CVE-2026-126394vLLM 0.30.0 before 0.31.0 Resource Exhaustion via OpenPangu Video Backend num_framesMedium 6.9 13 Nov 2026
References
- NVD https://nvd.nist.gov/vuln/detail/CVE-2026-100647
- CVE.org https://www.cve.org/CVERecord?id=CVE-2026-100647
- GitHub Advisory Database https://github.com/advisories/GHSA-wpww-v874-ph2p
- github.com https://github.com/vllm-project/vllm/security/advisories/GHSA-wpww-v874-ph2p