In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu: fix ACP MFD device leak on init failure acp_hw_init() registers ACP… (CVE-2026-98779)

The Linux kernel's drm/amdgpu driver had a flaw in the acp_hw_init() function where ACP child devices were registered with mfd_add_devices() before being attached to the ACP power domain and hardware initialization. If attaching a child device to the power domain failed or if ACP reset or clock enable operations timed out, the failure path only freed some allocations but left the registered child platform devices still registered and unreleased, causing a device leak. The fix involves removing children from the power domain and unregistering the MFD devices on failures occurring after mfd_add_devices() succeeds, while retaining existing cleanup for mfd_add_devices() failures. This issue was identified by static analysis and confirmed by manual review.

The vulnerability causes a resource leak in the Linux kernel's drm/amdgpu driver by leaving ACP MFD child devices registered after initialization failures. This could lead to resource exhaustion or instability in the affected kernel subsystem. There is no indication of direct code execution or privilege escalation from this issue. No known exploits in the wild have been reported.

A fix has been implemented in the Linux kernel to properly unregister ACP MFD child devices on initialization failure paths after mfd_add_devices() succeeds. Users should apply the official kernel update that includes this fix. Since this is a kernel-level resource management issue, updating to a patched kernel version is the recommended remediation.