Vulnerability details

llama.cpp builds b4882 through b9058 contain a heap buffer overflow vulnerability in the KV cache state restore path where the state_read_data() function computes write size without overflow checking, allowing attackers with write access to the slot_save_path directory to corrupt heap memory. Attackers can craft malicious state files where cell_count multiplication overflows or exceeds tensor buffer allocation to write attacker-controlled bytes past buffer boundaries, potentially resulting in heap metadata corruption, model weight corruption, or arbitrary code execution via function pointer overwrite.

• https://github.com/Vladimir-tokarev-cyera/llama-cpp-security-patches mitigation
• https://www.vulncheck.com/advisories/llama-cpp-b4882-b9058-buffer-overflow-in-kv-cache-state-restore third-party-advisory