llama.cpp memory exposure via token IDs in server completion requests
ggml-org llama.cpp
Affected versions
| Product | Affected | Fixed in |
|---|---|---|
| ggml-org llama.cpp |
| No fix yet |
Original description
llama.cpp before b3701 contains an out-of-bounds read vulnerability in llama_decode_internal and llama_encode_internal that allows remote unauthenticated attackers to read memory by supplying unvalidated token IDs. Attackers can send integer-array prompts to the llama-server /completion endpoint with IDs outside 0 to n_vocab to read embedding memory out of bounds or crash the process.
From the CVE record · VulnCheck