---
id: "CVE-2026-98440"
kind: "vulnerability"
aliases: ["CVE-2026-98440"]
title: "Linux KVM arm64: shadow stage-2 structures freed too early, leaving stale references"
severity: "unknown"
cvss: null
epss: null
exploited: false
status: "open"
category: "runtime"
products: []
ecosystems: []
tags: ["arm64", "kvm", "linux", "nested-virtualization", "runtime"]
not_affected_if: ["default-config"]
published_at: "2026-10-18T16:16:54+00:00"
first_seen_at: "2026-10-23T21:57:15+00:00"
updated_at: "2026-10-25T05:27:32+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-98440", "first_seen_at": "2026-10-23T21:57:15+00:00"}]
---

# Linux KVM arm64: shadow stage-2 structures freed too early, leaving stale references

In the Linux kernel's arm64 KVM nested-virtualization code, shadow stage-2 structures were freed during a flush callback while other parts of the system could still reference them. The fix delays freeing them until the VM is destroyed, and the description does not state a confirmed attacker outcome.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- unknown

## Fix

- unknown

## Timeline

- 2026-10-23T21:57:15+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-98440

## From the sources

In the Linux kernel, the following vulnerability has been resolved:

KVM: arm64: nv: Delay freeing of shadow S2 structures until VM destruction

We free the shadow S2 structures from kvm_arch_flush_shadow_all(), which
is a Bad Idea(tm). Freeing the page tables is fair game (this is what
this callback is for), but freeing the container that could still be
referenced by another part of the system is not great.

Instead, grow separate destructors that gets called when we tear the VM
down for good. From there, we can nuke both the individual MMUs as well
as the global array that points to them, safe in the knowledge that the
vcpus themselves have been destroyed already.

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_