Skip to content
Tech News
← Back to articles

Researchers disclose KVM flaw allowing VM escape to host memory on AMD servers

read original more articles
GoKawiil Brief

Security researchers demonstrated a vulnerability in Linux's KVM hypervisor, tracked as CVE-2026-99312, that let a virtual machine with root access read and write physical host memory on AMD CPUs with nested virtualization enabled. They showed a read exploit on a DigitalOcean droplet, retrieving the host's kernel version and roughly 127GB of host RAM, and a write exploit on their own bare-metal AMD server, writing 32 bytes into live host kernel memory. The flaw reportedly existed in the kernel for nearly four years and was introduced by an earlier security fix.

Why It Matters

GoKawiil's interpretation of the reporting above, not reported fact.

An attacker who can write arbitrary data into host kernel memory has what researchers describe as the basic building block for full code execution on the host, which would let a single malicious tenant potentially access or tamper with every other customer's VM on a shared machine. This undercuts the core promise of cloud isolation that providers like DigitalOcean sell, and suggests other multi-tenant platforms running vulnerable KVM/AMD configurations could be similarly exposed until patched.

Key Takeaways

Source: terra.security — Written By, 2026-11-18

Published there as: “CVE-2026-99312: KVM Guest-to-Host Escape”

Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.
Keep reading
digitalocean 5 briefs → kvm 4 briefs →