A flaw in the AMD GPU driver of the Linux kernel allows pending RCU callbacks to be executed after the driver module has been unloaded, resulting in a use‑after‑free condition. The unchecked callback execution accesses memory that has already been freed, causing a kernel panic and disruption of system operation. The official description explicitly notes that inserting an rcu_barrier() in the module exit path now prevents this late callback execution, thereby mitigating the crash.

All Linux kernel builds that include the amdgpu driver are affected until patched, regardless of distribution or kernel version. If a system loads the amdgpu module and subsequently unloads it before the infrastructure is patched, the vulnerability can be exercised.

The vulnerability is a use‑after‑free that can be triggered by a local, privileged attacker who has the ability to load and unload kernel modules. The CVSS score of 5.5 reflects a moderate impact; the EPSS score of < 1% indicates a very low likelihood of exploitation. Based on the description, it is inferred that the attacker would need privileged access to perform the module unload. The crash caused by the UAF could compromise system stability, but the CVE description does not confirm any escalation or data‑exfiltration beyond the crash.