Critical Linux KVM/arm64 Vulnerability Lets Attackers Escape VMs, Gain Host Access

Key Takeaways

• A critical vulnerability, CVE-2026-89775, has been discovered in the Linux kernel’s KVM/arm64 virtualization component.
• The flaw allows attackers to escape an ARM64 virtual machine and gain unauthorized access to the host system.
• Environments with nested virtualization enabled are primarily at risk, including multi-tenant cloud platforms and systems allowing untrusted users to create VMs.
• The vulnerability has been patched in the mainline Linux kernel; administrators must update their systems.

A significant security flaw, identified as CVE-2026-89775, has been uncovered in the Linux kernel’s KVM/arm64 virtualization module. This vulnerability poses a severe threat by potentially enabling an attacker to break out of an ARM64 virtual machine and compromise the underlying host system.

The issue is particularly concerning for KVM/arm64 deployments where nested virtualization is active. Such configurations are common in cloud computing environments that serve multiple tenants and in systems where users can provision their own virtual machines, thereby creating a substantial risk of lateral movement from a guest to the host infrastructure.

Security researcher Hyunwoo Kim reported this vulnerability, attributing it to a type truncation error within the KVM/arm64 stage-1 page-table walk process. This flaw impacts how the kernel calculates the necessary size of a memory region for invalidation within the virtual CPU’s pseudo-Translation Lookaside Buffer (pseudo-TLB).

How the Vulnerability Works

Typically, KVM is responsible for invalidating outdated memory translations whenever memory mappings change. This crucial step prevents a guest virtual machine from accessing memory that the host kernel has already released, relocated, or reallocated. However, in the vulnerable code path, the calculation for the invalidation size can erroneously return zero. This zero value, intended to signify an unknown memory size, is then misinterpreted by the VNCR pseudo-TLB invalidation logic as a valid, albeit empty, range size.

This misinterpretation leads to the complete skipping of the essential invalidation operation. Consequently, stale memory access information persists, making it available to a malicious guest VM. This condition can allow a previously freed host memory page to remain mapped as writable at a fixed host-kernel address. A malicious guest could then achieve 64-bit read and write access to this page without triggering any traps or virtual machine exits, effectively granting direct manipulation of host memory from within the guest environment.

Impact on Cloud and Local Systems

The implications of this vulnerability are particularly profound for public cloud providers utilizing ARM64 infrastructure. An attacker capable of creating a virtual instance with nested virtualization enabled could potentially exploit this flaw to traverse from their guest system to the host, thereby breaching the isolation boundary that separates tenant VMs from the cloud provider’s core infrastructure. Such a breach could lead to unauthorized access to other tenant data or critical cloud services.

Furthermore, the vulnerability also introduces a risk of local privilege escalation. The disclosure highlights that some Linux distributions, including certain Red Hat Enterprise Linux configurations, might expose /dev/kvm with world-writable permissions (0666). In such scenarios, an unprivileged local user could leverage the flaw, in conjunction with enabled nested virtualization, to gain root privileges on the host system.

The affected Linux kernel code was introduced by commit 7270cc9157f47 on May 14, 2025, and the fix was integrated upstream via commit 8053393680d4 on August 6, 2026.

What You Should Do

• Update Kernels Immediately: System administrators should prioritize updating to a Linux kernel release that includes the official patch for CVE-2026-89775 as soon as it is made available by their distribution vendor.
• Review Nested Virtualization: Organizations operating ARM64 KVM hosts should assess whether nested virtualization is strictly necessary. Disabling this feature in environments where it is not required can significantly reduce exposure until patches are fully deployed.
• Cloud Provider Actions: Cloud providers must prioritize patching their shared infrastructure, meticulously evaluate tenant access to nested virtualization capabilities, and verify permissions on /dev/kvm to mitigate potential local attack vectors.

Disclaimer: HackersRadar reports on cybersecurity threats and incidents for informational and awareness purposes only. We do not engage in hacking activities, data exfiltration, or the hosting or distribution of stolen or leaked information. All content is based on publicly available sources.