Description: llama.cpp is an inference of several LLM models in C/C++. In commits 55d4206c8 and prior, the n_discard parameter is parsed directly from JSON input in the llama.cpp server's completion endpoints without validation to ensure it's non-negative. When a negative value is supplied and the context fills up, llama_memory_seq_rm/add receives a reversed range and negative offset, causing out-of-bounds memory writes in the token evaluation loop. This deterministic memory corruption can crash the process or enable remote code execution (RCE). There is no fix at the time of publication.
The **CVSS Base Score** is a score from **0 to 10** that represents the intrinsic severity of a vulnerability. A higher score indicates greater severity.
Database CWE: v4.18
CWE-787: Out-of-bounds Write ↗
The product writes data past the end, or before the beginning, of the intended buffer.
Fonte: MITRE CWE
The **EPSS (Exploit Prediction Scoring System)** is a score from **0 to 1** that indicates the **probability** that a vulnerability will be exploited in the real world in the next 30 days. A higher value indicates a greater likelihood of exploitation.
The **Percentile** indicates how much higher this vulnerability's EPSS score is compared to all other vulnerabilities in the EPSS database. For example, a percentile of 0.90 (90%) means that 90% of vulnerabilities have an EPSS score equal to or lower than the current one.
*Data updated as of: 2026-01-24
The **CISA KEV Catalog** lists vulnerabilities that have been **actively exploited in the real world**. If a CVE is present in this catalog, it indicates that the threat is immediate and mitigation should be a top priority.
CVE **CVE-2026-21869** is not present in the CISA KEV Catalog. This indicates that it is not currently classified by CISA as an actively exploited vulnerability.
No articles present on Red Hot Cyber.
No results found on GitHub for this CVE.