Vulnerability in ggml-org llama.cpp

[ Description ]Official description

llama.cpp before b11393 contains a use-after-free and double free vulnerability in common_chat_peg_mapper::map that allows unauthenticated remote attackers to corrupt heap memory via a dangling current_tool pointer. Attackers can submit a chat_parser in a POST /completion request emitting a tool-id after a tool-close tag to crash llama-server and shape a heap write primitive.

[ Fixes ]Fixes and patches

No fix or interim patch is linked from our sources yet. Check the vendor references below and limit exposure of affected systems until a fix is available.

[ References ]References

• github.com/ggml-org/llama.cpp
• github.com/ggml-org/llama.cpp/blob/bed0a856606ee4a24a164066f73d2379447033f5/common/chat…
• github.com/ggml-org/llama.cpp/blob/bed0a856606ee4a24a164066f73d2379447033f5/tools/serve…
• github.com/ggml-org/llama.cpp/commit/dbe4c3ed42343f0a7ba0fd7e808ffeaca404d29f
• github.com/ggml-org/llama.cpp/pull/29942
• www.vulncheck.com/advisories/llama-cpp-before-b11393-use-after-free-via-common-chat-peg…