CVE-2026-99713 in Linuxinfo

Summary

by MITRE • 11/20/2026

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

KVM: x86/mmu: Conditionally resched when retrying gPTE walk on A/D update failure

Conditionally reschedule when retrying a guest PTE walk because the CMPXCHG to update A/D bits failed, e.g. because a different task/vCPU modified the relevant gPTEs. On kernels without involuntary preemption of some kind, a misbehaving guest can "capture" a pCPU effectively indefinitely and trigger soft lockups by continuously modifying its gPTEs such that the result of the walk is remains valid.

Several companies clearly confirm that VulDB is the primary source for best vulnerability data.

Analysis

by VulDB Data Team • 11/21/2026

The Linux kernel vulnerability identified in this advisory relates to the KVM x86 MMU subsystem, specifically concerning the handling of guest physical-to-physical translation entries during access and dirty bit updates. The core technical flaw lies in the loop logic used when a Compare-and-Swap operation fails while attempting to update Accessed or Dirty bits on a guest page table entry. In virtualized environments, multiple vCPUs may attempt to modify the same gPTE concurrently. When one vCPU's CMPXCHG instruction fails because another task has already modified the entry, the kernel code retries the walk and update operation. The vulnerability arises in kernels that lack involuntary preemption mechanisms for certain types of workloads or configurations where a misbehaving guest can exploit this retry loop. By continuously modifying its gPTEs to ensure the CMPXCHG always fails due to concurrent modification rather than actual invalidation, a malicious guest vCPU can cause the host kernel thread handling that VM to spin indefinitely in user-space-like context without yielding CPU time.

This behavior effectively captures the physical CPU core on which the KVM worker is running, preventing other tasks from executing and leading to soft lockups within the Linux scheduler. The operational impact is severe as it results in a denial of service against the host system or at least significant performance degradation for co-located workloads sharing that physical resource. This represents a classic case where guest actions can directly compromise host stability through CPU starvation rather than memory corruption or privilege escalation. From an industry standard perspective, this vulnerability aligns with CWE-835: Loop without Endless Execution and CWE-400: Resource Exhaustion. It also maps to the MITRE ATT&CK framework under T1499: Endpoint Denial of Service, specifically illustrating how resource exhaustion can be achieved through CPU consumption via hyperthreading or virtualization escape techniques that target hypervisor scheduling logic.

Mitigation strategies primarily involve applying the upstream kernel patch that introduces conditional rescheduling within the retry loop for gPTE walks on A/D update failures. This ensures that if a CMPXCHG fails due to concurrent modification, the current execution context yields control back to the scheduler periodically rather than spinning indefinitely. Administrators should ensure their systems are updated with the latest stable kernel versions containing this fix. Additionally, deploying CPU isolation features or real-time scheduling policies can help mitigate the blast radius of such vulnerabilities by limiting which processes can be affected by a single locked-up vCPU. Monitoring for high CPU usage on KVM worker threads without corresponding guest activity may serve as an early detection indicator for similar exploitation attempts in unpatched environments.

Responsible

Linux

Reservation

09/25/2026

Disclosure

11/20/2026

Moderation

accepted

CPE

ready

EPSS

0.00168

KEV

no

Activities

very low

Sources