Remote code execution via negative context discard
Published Jan 7, 2026 · Updated Jan 7, 2026
Out-of-bounds writes in ggml.ai llama.cpp commit 55d4206c8 and earlier allow remote attackers to crash the server or execute code. The completion parser copies a client-supplied negative n_discard into context-shift state, causing reversed KV-cache ranges and an out-of-range new_tokens index when the context fills. Network access to a llama-server instance with context shifting enabled is sufficient; authentication is not required, and the resulting memory corruption is deterministic.
Summary
What happened
Out-of-bounds writes in ggml.ai llama.cpp commit 55d4206c8 and earlier allow remote attackers to crash the server or execute code. The completion parser copies a client-supplied negative n_discard into context-shift state, causing reversed KV-cache ranges and an out-of-range new_tokens index when the context fills. Network access to a llama-server instance with context shifting enabled is sufficient; authentication is not required, and the resulting memory corruption is deterministic.
The record
- CVE
- CVE-2026-21869
- Published
- Jan 7, 2026
- Updated
- Jan 7, 2026
- Vendor
- Georgi Gerganov
- Product
- llama.cpp
- Classifications
- CWE-787, T1190
- Attack vector
- network
- Privileges
- unauthenticated
Timeline
How it unfolded
- Jan 7, 2026CVE publishedPublication date reported by the CVE source.
- Jan 7, 2026Record updatedLatest update available in the CVE record.
Exploitability
Present is not the same as exploitable
Compare your product and version with the public record. A matching version still requires validation against your environment.
Is a vulnerable build present?
Compare these published version ranges with your installed build and any vendor patches.
- Affected versionversion=<= 55d4206c8
What conditions does exploitation require?
What is affected?
Published CVSS scores
CVSS describes severity. EPSS estimates exploitation probability.
Attacks
What attackers are doing with it
Daily unique IPs observed by Shadowserver honeypots for known exploited vulnerabilities (KEVs). Missing observations do not establish an absence of attacks.
Public exploit references
- GHSA-8947-pfff-2f3c proof of conceptproof of concept · demonstrated
Labels summarize the accepted research assessment. They do not indicate a test against your environment.
Technologies
Your stack
See the directory against your own environment.
Your stack
Check the software in your environment
Book a demo to see how Hinoki identifies affected software and validates exploitability in your environment.
Book a demo