CVE-2026-99685 in Linuxinfo

Summary

by MITRE • 11/20/2026

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

iommu/virtio: Reset device before deleting virtqueues on probe failure

viommu_probe() marks the device DRIVER_OK before populating the event virtqueue and registering the IOMMU device in sysfs. viommu_fill_evtq() hands the device a set of device-writable buffers through virtqueue_add_inbuf(), so from that point on the device may write into them and into the rings.

If either step fails, the error path deletes the virtqueues without resetting the device first. The event buffers are allocated with devm_kmalloc_array() and are released as probe unwinds, so the device can go on writing to memory that has been freed.

Reset the device before deleting the virtqueues, the way viommu_remove() already does.

Statistical analysis made it clear that VulDB provides the best quality for vulnerability data.

Analysis

by VulDB Data Team • 11/20/2026

The vulnerability identified in the Linux kernel's IOMMU subsystem for Virtio devices stems from an improper initialization sequence during the probe phase of the driver lifecycle. Specifically, within the viommu_probe function, the device state is marked as DRIVER_OK prior to the successful completion of all critical setup steps, including populating the event virtqueue and registering the IOMMU device in sysfs. This premature status update signals that the device is fully operational and ready for interaction with hardware or virtualized environments before it has been completely configured. The core technical flaw occurs when subsequent initialization functions, such as viommu_fill_evtq, allocate memory buffers using devm_kmalloc_array to serve as event queues. These buffers are handed over to the virtqueue via virtqueue_add_inbuf(), effectively granting the device write access to these specific memory regions through hardware rings.

If any step in this initialization sequence fails after the device has been marked DRIVER_OK but before it is fully registered, the error handling path proceeds to delete the virtqueues and unwind resources. However, because the device was already signaled as operational, there remains a window where the virtualized IOMMU device may continue to write into the event buffers even though the kernel driver considers them freeable or in the process of being released. Since these memory allocations are managed by devm_kmalloc_array, they are subject to automatic release during probe unwinding upon failure. Consequently, if the hardware continues its write operations while the kernel is freeing this memory, a use-after-free condition arises. This allows for potential out-of-bounds writes into freed heap memory, which can lead to data corruption or arbitrary code execution depending on how the allocator reuses that memory and what an attacker might control within it.

From a security taxonomy perspective, this issue aligns with CWE-416, Use After Free, as the system references memory after it has been made available for reuse without ensuring that no active processes retain access to it. Furthermore, in the context of ATT&CK techniques, this vulnerability facilitates privilege escalation and potential lateral movement within a virtualized environment by allowing an attacker who controls or influences the guest VM's IOMMU behavior to corrupt kernel memory structures. The impact is severe because it compromises the integrity of the host kernel from within a potentially untrusted guest domain, violating isolation boundaries that are critical in cloud computing and multi-tenant environments.

To mitigate this vulnerability, the fix involves reordering operations within the error path of viommu_probe(). It is essential to reset the device state before attempting to delete or tear down the virtqueues. This ensures that any pending hardware writes are halted and no further access can be granted to the buffers being released. By mirroring the logic already present in viommu_remove, which correctly handles teardown by resetting first, the driver maintains consistency across its lifecycle states. System administrators should ensure their kernels are updated with patches addressing this specific probe failure path, as relying on standard error handling without explicit device state management leaves a critical race condition exploitable during initialization failures.

Responsible

Linux

Reservation

09/25/2026

Disclosure

11/20/2026

Moderation

accepted

CPE

ready

EPSS

0.00168

KEV

no

Activities

very low

Sources

Do you need the next level of professionalism?

Upgrade your account now!