GPU VulnDB

Database/Kernel, userspace & hypervisor

Linux kernel KVM/arm64: VNCR invalidation races nested page fault and installs a stale TLB entry

CVSS 9.3CVE-2026-89916Kernel, userspace & hypervisorcurated

Impact

A VNCR TLB invalidation on one vCPU can race a second vCPU that is walking the stage-1 page tables for the same page. The faulting vCPU can insert a TLB entry for a translation that the other vCPU has just invalidated, so a nested guest keeps a mapping the guest hypervisor believes it has torn down - a stale-translation window between nested guests on the same host. The fix makes invalidate_vncr_va() bump mmu_invalidate_seq so the fault is replayed. Applies only to arm64 hosts running nested virtualization; a host with no nested guests never enters this path.

Who can reach it

Local, from a nested guest (or a guest hypervisor) on an arm64 KVM host with nested virtualization enabled. Requires the ability to run a VM on the host; no host authentication beyond that.

What to do

Take a stable kernel containing the linked commits and reboot each arm64 host, draining or migrating guests first. No mitigation short of disabling nested virtualization is described. The record gives commits, not fixed version numbers.

References

Related entries

All Kernel, userspace & hypervisor entries

This entry is curated: imported from vendor advisories with machine assistance, not yet individually verified. Confirm against your vendor's advisory before acting, and report anything wrong.