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.
CVE-2026-43629
CVSS base score
8.1 (v3.1)
Severity
HIGH
Attack vector
Network
Weakness
CWE-190, CWE-787
Last modified
October 29, 2026
CVSS vector
CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:H/I:H/A:H
Affected products
ggml llama.cpp
References
- github.com Patch, Third Party Advisory
- www.vulncheck.com Mitigation, Patch, Third Party Advisory