CVE-2026-34159

Published Apr 1, 2026

Last updated a month ago

Overview

AI description

Automated description summarized from trusted sources.

CVE-2026-34159 describes a vulnerability found in the `llama.cpp` project, specifically within its RPC backend. The flaw exists in the `deserialize_tensor()` function, which fails to perform bounds validation when a tensor's `buffer` field is set to zero. This oversight allows an unauthenticated attacker to read and write arbitrary process memory by sending specially crafted `GRAPH_COMPUTE` messages. When exploited in conjunction with pointer leaks from `ALLOC_BUFFER`/`BUFFER_GET_BASE` messages, this vulnerability can lead to a complete bypass of Address Space Layout Randomization (ASLR) and enable remote code execution. The attack requires only TCP access to the RPC server port and no authentication. The issue has been addressed in `llama.cpp` version b8492.

Description
llama.cpp is an inference of several LLM models in C/C++. Prior to version b8492, the RPC backend's deserialize_tensor() skips all bounds validation when a tensor's buffer field is 0. An unauthenticated attacker can read and write arbitrary process memory via crafted GRAPH_COMPUTE messages. Combined with pointer leaks from ALLOC_BUFFER/BUFFER_GET_BASE, this gives full ASLR bypass and remote code execution. No authentication required, just TCP access to the RPC server port. This issue has been patched in version b8492.
Source
security-advisories@github.com
NVD status
Analyzed
Products
llama.cpp

Risk scores

CVSS 3.1

Type
Secondary
Base score
9.8
Impact score
5.9
Exploitability score
3.9
Vector string
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Severity
CRITICAL

Weaknesses

security-advisories@github.com
CWE-119

Social media

Hype score
Not currently trending

Configurations