llama.cpp versions b1886 through 0.9.0 contain an integer overflow vulnerability in the LLaMA-Android JNI wrapper. Specifically, the new_1batch() function multiplies sizeof(llama_seq_id) by the n_seq_max parameter, which is attacker-controlled, without checking for overflow. This results in a heap buffer allocation wraparound, causing insufficient memory allocation. An attacker can exploit this by supplying a crafted n_seq_max value via a malicious model file or JNI call, triggering heap corruption that may lead to denial of service or arbitrary code execution on affected Android applications.

Exploitation of this vulnerability can cause heap corruption due to insufficient memory allocation, potentially resulting in denial of service or arbitrary code execution on Android applications using the vulnerable LLaMA-Android JNI wrapper. No known exploits in the wild have been reported.

Patch status is not yet confirmed — check the vendor advisory for current remediation guidance. Until a patch is available, avoid processing untrusted model files or JNI calls that could supply malicious n_seq_max values to the LLaMA-Android binding.