CVE-2025-38104

Name of the Vulnerable Software and Affected Versions Linux kernel versions prior to 6.12.0

Description The issue is related to the use of a mutex for RLCG register access in the Linux kernel's amdgpu driver, which can lead to priority inversion in SRIOV environments. This occurs when a low-priority task holds a mutex that a high-priority task needs, causing conflicts and potentially leading to incorrect operations. The amdgpu virt rlcg reg rw function is particularly critical in this context, as it attempts to acquire the mutex. The call stack indicates that this function is invoked from amdgpu sriov wreg, which in turn is called from gmc v11 0 flush gpu tlb. The bug is identified by the message "[ BUG: Invalid wait context ]", which signifies that a thread is trying to acquire a mutex while in a context that does not allow it to sleep.

Recommendations To resolve the issue, update the Linux kernel to a version that includes the fix for the RLCG register access priority inversion. As a temporary workaround, consider disabling the amdgpu virt rlcg reg rw function until a patch is available. Restrict access to the vulnerable module amdgpu to minimize the risk of exploitation. Avoid using the amdgpu sriov wreg function in the affected API endpoint until the issue is resolved.