In the Linux kernel, a bug in the iommufd/viommu component caused the group lock to remain held when allocating a virtual device if the requested size exceeded the core structure size. The code path skipped releasing the igroup mutex, leading to a deadlock that stalls all subsequent vDEVICE operations on that group. This defect is logged with WARN_ON_ONCE and is fixed by redirecting to out_unlock_igroup. The vulnerability is limited to the kernel's handling of virtual device allocations and can be triggered by an attacker controlling a driver that performs such an allocation.

The flaw affects any Linux kernel configuration that includes the iommufd and viommu modules. All kernel versions that ship with the original, unpatched code are potentially impacted, regardless of distribution. The missing CPE indicates that the kernel species itself is the affected product, and no specific vendor version numbers are given.

The absence of an EPSS score and no KEV listing mean the publicly known exploitation potential is uncertain, but the nature of the bug—an unreleased kernel lock causing a local deadlock—suggests a moderate to high severity if an attacker can trigger the allocation path. The CVSS score is not provided, but a deadlock in kernel code can lead to denial of service for affected users or services, and may allow privilege escalation if the timeout or retry logic is abused. Since the bug requires access to a driver that can invoke vdevice allocation, the attack vector is likely local or via a hostile device driver installation. The vulnerability remains critical while unpatched.