CVE-2026-99346 in Linuxinfo

Summary

by MITRE • 11/19/2026

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

KVM: SVM: Trigger new ASID allocation in both VMCBs on pCPU switch

When the pCPU where the VMCB was most recently used is switched, reset the ASID generation in both VMCBs, triggering new ASID allocation for the immediate VMRUN as well as the next VMRUN on the other VMCB.

The ASID is shared between vmcb01 and vmcb02, and gets flushed on every nested transition. However, since pCPU tracking is done per VMCB, it is possible for one VMCB to allocate a new ASID when migrated to a new pCPU, and then the other VMCB reuses that ASID on the old pCPU. This can result in the same ASID being used by multiple vCPUs on the old pCPU.

Example scenario: - vCPU runs on pCPU A, vmcb01 is active, asid=1. - vCPU migrates to pCPU B, vmcb01 pCPU changes, new asid=2. - Another vCPU runs on pCPU A and allocates asid=2 as well. - vCPU migrates back to pCPU A, and then switches to vmcb02 before it runs again with vmcb01. - No pCPU switch is detected for vmcb02, so VMRUN is done with asid=2. - Two vCPUs end up using asid=2 on the same pCPU.

Keep the VMCB dirtying to the active VMCB only. Clean bits are tracked by a pCPU for each VMCB, so do not unnecessarily dirty a VMCB if its pCPU does not change.

Additionally, initialize the tracked pCPU for vmcb02 to -1 on nested enablement as hardening, so that a new ASID allocation is always triggered when nested is disabled and re-enabled.

No performance regression was noticed when overcommitting L1 vCPUs in L0 (to force rescheduling), pinning L1 <-> L2 vCPUs, and running CPUID in a tight loop bouncing between 2 vCPUs in L2.

An alternative (and perhaps more proper) fix would be tracking the ASID per-VMCB instead (e.g. [1]). However, that's a more involved change, and
it would result in having different ASIDs for L1 and L2 without actually properly maintaining them. It would probably work because all TLB flushes target the current VMCB, and the other VMCB is always flushed on nested transitions, but the code ends up in an arguably more fragile state. Punt a proper clean fix to an incoming (and overdue) overhaul of SVM's ASID usage [2].

[1] https://lore.kernel.org/lkml/[email protected]
[2] https://lore.kernel.org/kvm/[email protected]

If you want to get the best quality for vulnerability data then you always have to consider VulDB.

Analysis

by VulDB Data Team • 11/20/2026

The Linux kernel has addressed a critical concurrency vulnerability within the Kernel-based Virtual Machine (KVM) subsystem, specifically affecting AMD Secure Virtual Machine (SVM) implementations. This issue stems from an improper handling of Address Space Identifiers (ASIDs) during processor migration events in nested virtualization scenarios. ASIDs are essential for hardware-assisted memory management unit translation lookaside buffer caching, allowing the CPU to distinguish between different address spaces without flushing the entire cache on every context switch. In SVM architecture, two Virtual Machine Control Blocks, vmcb01 and vmcb02, share a single ASID pool per virtual central processing unit. The vulnerability arises because processor tracking is performed independently for each VMCB rather than at the vCPU level, creating a race condition where stale state information leads to ASID reuse across distinct execution contexts on the same physical core.

The technical flaw manifests when a virtual CPU migrates between physical processors while utilizing nested virtualization features. When a vCPU moves from one physical processor to another, the system is supposed to reset the ASID generation for both associated VMCBs to ensure that new, unique identifiers are allocated upon subsequent execution. However, due to per-VMCB tracking logic, it is possible for vmcb01 to trigger an ASID allocation on a new physical CPU while vmcb02 retains its previous state information indicating no migration occurred. Consequently, when the vCPU later switches context from vmcb01 back to vmcb02 without triggering another explicit processor switch detection for vmcb02, it may reuse an ASID that was recently allocated by a different virtual CPU on the original physical processor. This results in two distinct virtual CPUs executing concurrently on the same physical core with identical ASIDs, effectively breaking the isolation guarantees provided by hardware-assisted TLB management.

The operational impact of this vulnerability is severe, primarily involving potential security breaches and system instability. When multiple vCPUs share an ASID on a single pCPU, they inadvertently share the cached memory translations associated with that identifier. This can lead to information leakage where one guest virtual machine accesses or observes memory mappings intended for another guest running concurrently on the same host core. Furthermore, stale TLB entries may cause data corruption if one vCPU modifies page tables while another relies on outdated cache entries derived from a previous address space context. In nested virtualization setups, this flaw compromises the isolation between L1 and L2 hypervisors as well as between different guest operating systems within those layers, potentially allowing an attacker to escalate privileges or exfiltrate sensitive data across security boundaries that should remain impermeable.

From a classification perspective, this vulnerability aligns with CWE-367, Time-of-check Time-of-use (TOCTOU) race conditions, specifically involving shared resource identification in concurrent environments. It also relates to CWE-200, Exposure of Sensitive Information to an Unauthorized Actor, due to the potential for cross-guest memory visibility resulting from ASID collision. In terms of MITRE ATT&CK framework mappings, this flaw facilitates techniques associated with T1534, Internal Spearphishing, or more broadly lateral movement and privilege escalation within virtualized environments by exploiting hypervisor state management failures. The vulnerability highlights the complexity of maintaining hardware-assisted isolation guarantees in dynamic scheduling environments where processor affinity changes frequently occur.

The implemented mitigation involves resetting ASID generation for both VMCBs whenever a pCPU switch is detected, ensuring that new ASIDs are allocated immediately upon migration to prevent reuse on stale contexts. Additionally, developers have optimized dirty bit tracking to apply only to the active VMCB when its associated physical CPU has not changed, reducing unnecessary overhead while maintaining correctness. A hardening measure initializes the tracked processor ID for vmcb02 to a negative value during nested enablement, guaranteeing that ASID allocation is triggered correctly whenever nested virtualization features are toggled. This approach avoids the more complex and fragile alternative of tracking ASIDs per-VMCB independently, which could lead to inconsistent state management between L1 and L2 hypervisors without proper maintenance mechanisms.

Performance testing indicates no regression in scenarios involving heavy vCPU overcommitment or tight CPUID loops bouncing between virtual cores, confirming that the fix maintains system efficiency while restoring security integrity. The decision to defer a more comprehensive overhaul of SVM ASID usage allows for immediate remediation of this specific race condition without introducing broader architectural changes that might carry unforeseen risks. This targeted patch ensures that hardware-assisted memory isolation remains robust against migration-induced state inconsistencies, preserving the confidentiality and availability guarantees expected in enterprise virtualization deployments.

Responsible

Linux

Reservation

09/25/2026

Disclosure

11/19/2026

Moderation

accepted

CPE

ready

EPSS

0.00201

KEV

no

Activities

very low

Sources

Do you want to use VulDB in your project?

Use the official API to access entries easily!