MEDIUM Vulnerabilities #Linux #Vulnerability

CVE-2026-64561: Zapscape KVM Flaw Allows Guest VM Escape

2026-08-07 02:08 UTC 3 min read Runtime Rebel Intel

This article was written by a language model from the source above and was not reviewed by a human before publication. Verify anything operational against the original. Editorial policy

Key points

Immediate impact: Privileged L1 guest kernel code can escape KVM isolation and execute arbitrary commands on the underlying Linux host.
Affected systems: Linux kernel versions 5.9 through recent stable releases running KVM/x86 with nested virtualization exposed to untrusted guests.
Remediation: Update to a patched stable kernel version or apply vendor-backported fixes immediately.

Overview of Zapscape (CVE-2026-64561)

A recently disclosed vulnerability in the Linux kernel’s Kernel-based Virtual Machine (KVM) subsystem, tracked as CVE-2026-64561 and colloquially known as Zapscape, allows an attacker with kernel privileges inside an L1 guest virtual machine to escape isolation and achieve code execution on the host. According to a report by The Hacker News, the security issue impacts environments where nested virtualization is exposed to untrusted guests.

Security researcher Hyunwoo Kim discovered and reported the flaw, noting that successful exploitation grants root-equivalent capabilities on the host system. While public proof-of-concept code has been released targeting AMD nested SVM/NPT on Linux 7.1.3, researchers emphasize that weaponising the bug in production cloud environments requires adapting the exploit to specific host configurations and memory backends.

Technical Analysis of the KVM Shadow MMU Flaw

The vulnerability stems from a stale-root check ordering error within KVM’s shadow memory management unit (MMU) bookkeeping, classified by Red Hat as CWE-825 (expired pointer dereference). During guest-triggered page fault handling, KVM can reclaim MMU pages, invalidating the shadow MMU root page currently utilized by the fault-handling path.

Because the recursive zap path fails to re-verify the root state after making additional MMU pages available, KVM continues operating under the invalidated root. Child shadow pages inherit this invalid state and remain active on KVM’s active MMU page list. Subsequent cleanup operations can attach the same list link to multiple lists simultaneously, freeing the page while dangling list references remain and causing a use-after-free condition.

Prerequisites for Exploitation

Successful execution of the exploit chain relies on specific environmental conditions:

Guest Privileges: Attackers require L1 kernel privileges (typically guest root access) to trigger the fault handling sequence.
Hardware Configuration: On Intel systems, the configuration requires both EPT page-walk length 4 and 5 to be exposed to the L1 guest. AMD hardware does not have an equivalent prerequisite.
Software Versions: Affected upstream versions include Linux 5.9 up to unpatched stable releases, with fixes subsequently backported to stable trees such as 6.6.148, 6.12.101, 6.18.42, 7.1.6, and 7.2-rc5.

Mitigations and Patching Guidance

Upstream maintainers addressed the issue via commit 2abd5287f083, which adjusts the stale-root check ordering. The patch positions the check after make_mmu_pages_available(), forcing KVM to restart the fault using RET_PF_RETRY if reclaim operations invalidate the current root, thereby preventing invalid mapping creation.

Administrators managing multi-tenant hypervisors or environments exposing nested virtualization should prioritize the following remediation steps:

Audit hypervisor configurations to determine if nested virtualization is exposed to untrusted tenants.
Apply the latest kernel security updates provided by upstream maintainers or downstream enterprise distributions.
Verify that vendor packages contain backported fixes even if the core version string remains unchanged.

#Linux #Vulnerability

X/Twitter LinkedIn Reddit HN

← Back to Articles