CVE-2025-49847 SUSE CVE security@suse.de SUSE Security Team SUSE CVE-2025-49847 Interim 1 2 2025-07-12T06:34:16Z current 2025-06-18T23:13:02Z 2025-07-12T06:34:16Z cve-database/bin/generate-cvrf-cve.pl 2020-12-27T01:00:00Z CVE-2025-49847 llama.cpp is an inference of several LLM models in C/C++. Prior to version b5662, an attacker‐supplied GGUF model vocabulary can trigger a buffer overflow in llama.cpp's vocabulary‐loading code. Specifically, the helper _try_copy in llama.cpp/src/vocab.cpp: llama_vocab::impl::token_to_piece() casts a very large size_t token length into an int32_t, causing the length check (if (length < (int32_t)size)) to be bypassed. As a result, memcpy is still called with that oversized size, letting a malicious model overwrite memory beyond the intended buffer. This can lead to arbitrary memory corruption and potential code execution. This issue has been patched in version b5662. The CVRF data is provided by SUSE under the Creative Commons License 4.0 with Attribution (CC-BY-4.0). https://www.suse.com/support/security/rating/ SUSE Security Ratings openSUSE Tumbleweed ggml-devel-5699-1.1 libggml-5699-1.1 libggml-base-5699-1.1 libggml-cpu-5699-1.1 libggml-opencl-5699-1.1 libggml-vulkan-5699-1.1 libllama-5699-1.1 libmtmd-5699-1.1 llamacpp-5699-1.1 llamacpp-devel-5699-1.1 ggml-devel-5699-1.1 as a component of openSUSE Tumbleweed libggml-5699-1.1 as a component of openSUSE Tumbleweed libggml-base-5699-1.1 as a component of openSUSE Tumbleweed libggml-cpu-5699-1.1 as a component of openSUSE Tumbleweed libggml-opencl-5699-1.1 as a component of openSUSE Tumbleweed libggml-vulkan-5699-1.1 as a component of openSUSE Tumbleweed libllama-5699-1.1 as a component of openSUSE Tumbleweed libmtmd-5699-1.1 as a component of openSUSE Tumbleweed llamacpp-5699-1.1 as a component of openSUSE Tumbleweed llamacpp-devel-5699-1.1 as a component of openSUSE Tumbleweed llama.cpp is an inference of several LLM models in C/C++. Prior to version b5662, an attacker‐supplied GGUF model vocabulary can trigger a buffer overflow in llama.cpp's vocabulary‐loading code. Specifically, the helper _try_copy in llama.cpp/src/vocab.cpp: llama_vocab::impl::token_to_piece() casts a very large size_t token length into an int32_t, causing the length check (if (length < (int32_t)size)) to be bypassed. As a result, memcpy is still called with that oversized size, letting a malicious model overwrite memory beyond the intended buffer. This can lead to arbitrary memory corruption and potential code execution. This issue has been patched in version b5662. CVE-2025-49847 openSUSE Tumbleweed:ggml-devel-5699-1.1 openSUSE Tumbleweed:libggml-5699-1.1 openSUSE Tumbleweed:libggml-base-5699-1.1 openSUSE Tumbleweed:libggml-cpu-5699-1.1 openSUSE Tumbleweed:libggml-opencl-5699-1.1 openSUSE Tumbleweed:libggml-vulkan-5699-1.1 openSUSE Tumbleweed:libllama-5699-1.1 openSUSE Tumbleweed:libmtmd-5699-1.1 openSUSE Tumbleweed:llamacpp-5699-1.1 openSUSE Tumbleweed:llamacpp-devel-5699-1.1 important