Linux KVM/arm64 Vulnerability Lets Attackers Escape Virtual Machines and Gain Host Access

A newly disclosed Linux kernel vulnerability, tracked as CVE-2026-89775, could let attackers escape an ARM64 virtual machine and access the underlying host system.

The issue affects KVM/arm64 environments where nested virtualization is enabled, creating a serious risk for multi-tenant cloud infrastructure and systems that allow untrusted users to create virtual machines.

Security researcher Hyunwoo Kim reported that the flaw stems from a type truncation issue in the KVM/arm64 stage-1 page-table walk process.

The bug affects how the kernel calculates the size of a memory region it must invalidate from the virtual CPU’s pseudo Translation Lookaside Buffer, or pseudo-TLB.

Under normal conditions, KVM must invalidate stale memory translations after memory mappings change. This prevents a guest virtual machine from continuing to access memory that has been released, moved, or reassigned by the host kernel.

In the vulnerable code path, however, the affected size calculation can return 0, a value intended to mean that the memory size is unknown.
The VNCR pseudo-TLB invalidation logic incorrectly treats the 0 value as a valid range size.

This produces an empty invalidation range, causing the required invalidation operation to be skipped entirely. As a result, stale memory access information can remain available to a malicious guest.

Linux KVM/arm64 Vulnerability

According to the disclosure, a freed host memory page can remain mapped as writable at a fixed host-kernel address. A malicious guest can then obtain 64-bit read and write access to that page without triggering a trap or virtual machine exit. This behavior gives an attacker a path to manipulate host memory directly from inside a guest environment.

The vulnerability is particularly significant in public-cloud deployments using ARM64 infrastructure. An attacker who can create an instance with nested virtualization capabilities may be able to move from the guest system to the host.

If successful, the attack could compromise the isolation boundary that normally separates one customer’s virtual machine from the cloud provider’s underlying infrastructure.

The issue may also create a local privilege-escalation risk. The report noted that some distributions, including Red Hat Enterprise Linux configurations, can expose /dev/kvm with world-writable permissions set to 0666.

With nested virtualization enabled, an unprivileged local user may be able to exploit the flaw to gain root privileges on the host. The vulnerability affects Linux kernel code introduced by commit 7270cc9157f47, dated May 14, 2025, and was fixed upstream by commit 8053393680d4, dated August 6, 2026.

Administrators should update to a kernel release containing the mainline fix as soon as it becomes available from their distribution vendor.

Organizations operating ARM64 KVM hosts should also review whether nested virtualization is necessary. Turning off the feature where it is not required can reduce exposure while patches are deployed.

Cloud providers should prioritize patching shared infrastructure, assess tenant access to nested virtualization, and verify permissions on /dev/kvm to limit local attack opportunities. The disclosure confirms that the Linux mainline kernel includes a patch.

Cut every SOC alert investigation by 21 min. Power your SOC with instant IOC context for immediate response: Integrate TI Lookup in your SOC