CVE-2026-80728

A flaw was found in the Linux kernel's AMD GPU (amdgpu) driver. During device teardown, a timing issue can occur where memory used by the driver is released prematurely. This can happen during a device probe failure or rollback. This premature release leads to the driver attempting to write to an invalid memory location, causing a kernel page fault and system instability or a crash (Denial of Service).

An AMDGPU probe failure or rollback can hit a kernel page fault because the aperture mapping may be released before the DRM device release path finishes. IP software finalization callbacks such as vcn_v4_0_sw_fini can then access fw_shared through a stale pointer derived from aper_base_kaddr after the mapping was already unmapped. For the CVSS the PR:L is used in the paranoid score because some deployments may delegate GPU reset, device binding, or driver control to a service account, container root, or reduced capability administrator rather than full host root. The issue is not network reachable and requires local control over the affected device or driver lifecycle. Impact is at least denial of service via kernel crash. In the worst still defensible case it may have limited confidentiality or integrity impact because the bug involves a stale kernel mapping pointer and a write in teardown context, so it should be reviewed manually.

The CVSS score(s) listed for this vulnerability do not reflect the associated product's status, and are included for informational purposes to better understand the severity of this vulnerability.

Mitigation for this issue is either not available or the currently available options don't meet the Red Hat Product Security criteria comprising ease of use and deployment, applicability to widespread installation base or stability.