Linux Kernel: KVM arm64 privilege escalation in MOPS handling (CVE-2026-64555)
A vulnerability in the Linux kernel KVM implementation for arm64 allows an attacker within a nested-capable KVM guest to trigger improper SPSR_EL2 state handling, potentially leading to host-level privilege escalation or a hypervisor panic.
CVE search metadata
CVE search record: CVE-2026-64555. Severity: high. CVSS: 8.8. EPSS: 0.13%. KEV: no. Product: Linux Kernel. Brief: Linux Kernel: KVM arm64 privilege escalation in MOPS handling (CVE-2026-64555). Brief link: https://feed.craftedsignal.io/briefs/2026-11-linux-kvm-mops-cve-2026-64555/
Triage facts
Revision 1- Applicability
- Linux Kernel
- Evidence
- Unknown
- Severity
- High
- Sources
- 2 public sources
- Freshness
- Revision 1Nov 17, 19:50 UTC
- Vulnerabilities
- 1 CVE
- Source quality
- A2Roughly even chance
Overview
A vulnerability exists in the Linux kernel KVM for arm64 systems within the kvm_hyp_handle_mops() function. The issue arises from incorrect handling of synthetic vcpu_cpsr() state during MOPS (Memory Operations) exception rewinding. Under nested virtualization, the kernel fails to translate the synthetic state before writing it to hardware, leading to corruption of hardware SPSR_EL2 registers.
Attack Chain
The attacker must be executing code inside a KVM/arm64 guest. A guest hypervisor at vEL2 can trigger the MOPS exception by executing a main/epilogue MOPS instruction with ill-formed alignment/size parameters. This occurs during normal vCPU execution without host-side interaction.
Impact
Successful exploitation allows a low-privileged guest user to gain execution at EL2, granting access to the entire host kernel address space and the memory of other guests. Triggering the flaw can also cause a host-wide hypervisor panic, resulting in a denial-of-service for all hosted virtual machines.
Recommendation
Users should apply patches for the following affected version ranges: Linux kernel >= 6.7, < 6.12.97; >= 6.13, < 6.18.40; and >= 6.19, < 7.1.5.