Heap-Based Buffer Overflow Vulnerability in GGUF Library Header
CVE-2024-21836
8.8HIGH
What is CVE-2024-21836?
A vulnerability has been identified within the GGUF library, specifically in the header.n_tensors functionality of llama.cpp, where a heap-based buffer overflow can occur. This issue arises when processing specially crafted .gguf files, potentially allowing for remote code execution. An attacker could exploit this vulnerability by supplying a maliciously designed file, thus compromising the integrity and security of applications utilizing this library. The risk posed by this vulnerability necessitates immediate attention to ensure appropriate mitigation measures are implemented.
Affected Version(s)
llama.cpp Commit 18c2e17
References
CVSS V3.1
Score:
8.8
Severity:
HIGH
Confidentiality:
High
Integrity:
High
Availability:
High
Attack Vector:
Network
Attack Complexity:
Low
Privileges Required:
None
User Interaction:
Required
Scope:
Unchanged
Timeline
Vulnerability published
Vulnerability Reserved
Credit
Discovered by Francesco Benvenuto of Cisco Talos.
