CVE-2024-21836: Integer Overflow
First published: Mon Feb 26 2024(Updated: )
A heap-based buffer overflow vulnerability exists in the GGUF library header.n_tensors functionality of llama.cpp Commit 18c2e17. A specially crafted .gguf file can lead to code execution. An attacker can provide a malicious file to trigger this vulnerability.
Credit: talos-cna@cisco.com
| Affected Software | Affected Version | How to fix |
|---|---|---|
| llama.cpp GGUF library | ||
| Llama.cpp | <2024-01-09 |
Never miss a vulnerability like this again
Sign up to SecAlerts for real-time vulnerability data matched to your software, aggregated from hundreds of sources.
Frequently Asked Questions
What is the severity of CVE-2024-21836?
CVE-2024-21836 is classified as a high-severity vulnerability due to the potential for heap-based buffer overflow and possible code execution.
How do I fix CVE-2024-21836?
To mitigate CVE-2024-21836, users should apply the latest security patches provided by the developers of the GGUF library as soon as they are available.
What causes CVE-2024-21836?
CVE-2024-21836 is caused by a flaw in the header.n_tensors functionality of the GGUF library in llama.cpp that can be exploited through specially crafted .gguf files.
Who is affected by CVE-2024-21836?
Any user or application utilizing the GGUF library in llama.cpp is at risk of exploitation from CVE-2024-21836.
Can CVE-2024-21836 be exploited remotely?
Yes, CVE-2024-21836 can be exploited remotely if an attacker provides a malicious .gguf file to the affected application.
- agent/references
- agent/first-publish-date
- agent/type
- agent/description
- agent/author
- collector/mitre-cve
- source/MITRE
- agent/source
- agent/weakness
- agent/last-modified-date
- agent/severity
- agent/guess-ai
- agent/software-canonical-lookup
- agent/software-canonical-lookup-request
- agent/event
- agent/softwarecombine
- agent/tags
- collector/nvd-api
- source/NVD
- vendor/llama.cpp
- canonical/llama.cpp gguf library
- vendor/ggerganov
- canonical/llama.cpp