ggml-org llama.cpp src/llama-vocab.cpp llama_vocab::tokenize memory corruption

Summaryinfo

A vulnerability was found in ggml-org llama.cpp. It has been classified as critical. This affects the function llama_vocab::tokenize of the file src/llama-vocab.cpp. This manipulation causes memory corruption. This vulnerability is tracked as CVE-2025-52566. The attack is restricted to local execution. No exploit exists. Applying a patch is the recommended action to fix this issue.

Detailsinfo

A vulnerability has been found in ggml-org llama.cpp (unknown version) and classified as critical. Affected by this vulnerability is the function llama_vocab::tokenize of the file src/llama-vocab.cpp. The manipulation with an unknown input leads to a memory corruption vulnerability. The CWE definition for the vulnerability is CWE-119. The product performs operations on a memory buffer, but it can read from or write to a memory location that is outside of the intended boundary of the buffer. As an impact it is known to affect confidentiality, integrity, and availability. The summary by CVE is:

llama.cpp is an inference of several LLM models in C/C++. Prior to version b5721, there is a signed vs. unsigned integer overflow in llama.cpp's tokenizer implementation (llama_vocab::tokenize) (src/llama-vocab.cpp:3036) resulting in unintended behavior in tokens copying size comparison. Allowing heap-overflowing llama.cpp inferencing engine with carefully manipulated text input during tokenization process. This issue has been patched in version b5721.

The advisory is shared at github.com. This vulnerability is known as CVE-2025-52566 since 06/18/2025. The exploitation appears to be easy. An attack has to be approached locally. Technical details are known, but no exploit is available.

The vulnerability scanner Nessus provides a plugin with the ID 271154 (Linux Distros Unpatched Vulnerability : CVE-2025-52566), which helps to determine the existence of the flaw in a target environment.

Applying the patch dd6e6d0b6a4bbe3ebfc931d1eb14db2f2b1d70af is able to eliminate this problem. The bugfix is ready for download at github.com.

The vulnerability is also documented in the databases at Tenable (271154) and EUVD (EUVD-2025-19074). If you want to get the best quality for vulnerability data then you always have to consider VulDB.

Productinfo

Vendor

• ggml-org

Name

• llama.cpp

License

• open-source

Website

• Product: https://github.com/ggml-org/llama.cpp/

CPE 2.3info

• 🔍
• 🔍

CPE 2.2info

• 🔍
• 🔍

CVSSv4info

CVSSv3info

CVSSv2info

Exploitinginfo

Threat Intelligenceinfo

Countermeasuresinfo

Timelineinfo

Sourcesinfo

Entryinfo

If you want to get the best quality for vulnerability data then you always have to consider VulDB.

Discussion