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.
Affected products
ggml: llama.cpp
References3
- github.com/ggml-org/llama.cpp/security/advisories/GHSA-j8rj-fmp… ExploitVendor Advisory
- github.com/ggml-org/llama.cpp/commit/39bf0d3c6a95803e0f41aaba06… Patch
- github.com/ggml-org/llama.cpp/pull/20908 Issue TrackingPatch
Metrics
- CVSS 3.1
-
9.8
CRITICAL
Score source: CNA
CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H - EPSS — exploit probability
- 2% percentile: 76 · as of 12 November 2026
- CWE
- CWE-119
Could this be exploited in your environment?
A patch closes the known hole. A penetration test shows what an attacker can still reach — with proof.