Track vulnerabilities like this one. Sign up free to get alerted when software you use is affected.
9.9

CVE-2026-118375: llama.cpp can be crashed or hijacked by remote users

CVE-2026-118375 · published 8 days ago
Summary

Versions of llama.cpp between the indicated revisions have a flaw in the network code that lets an unauthenticated remote system send special data and overwrite memory on the server. This can cause the server to stop working or let an attacker run their own code. Upgrade to the fixed version or apply the vendor's patch as soon as possible.

What to do
  • Update ggml-org llama.cpp to version b11865 or later.
Affected software
Ecosystem VendorProductAffected versions
– ggml-org llama.cpp < b11865
Debian:14 debian llama.cpp All versions
Ubuntu:26.10 canonical llama.cpp All versions
Original advisory text
llama.cpp b8841 before b11865 Stack-based Buffer Overflow via ggml-rpc RDMA rdma_recv()
llama.cpp b8841 before b11865 contains an out-of-bounds write vulnerability in the ggml-rpc RDMA transport rdma_recv() function that allows unauthenticated peers to overflow the rpc-server stack. Attackers advertising RDMA capabilities in the HELLO message can send up to 256 KiB into a 1-byte stack variable in rpc_serve_client(), crashing the server or potentially executing code.
Fix within
Internet-facing 14 days
Internal At next upgrade
  • Not known to be exploited
  • Needs hands-on effort to exploit
  • Gives an attacker full control
Severity
9.9 Critical
Exploitation
<1% chance of attack within 30 days
Type
CWE-787Out-of-bounds Write
Timeline
Published28 Oct 2026
Updated4 Nov 2026
First seen28 Oct 2026
Track software like this
Free during beta