CVE-2025-68230
Description
In the Linux kernel, the following vulnerability has been resolved:
drm/amdgpu: fix gpu page fault after hibernation on PF passthrough
On PF passthrough environment, after hibernate and then resume, coralgemm will cause gpu page fault.
Mode1 reset happens during hibernate, but partition mode is not restored on resume, register mmCP_HYP_XCP_CTL and mmCP_PSP_XCP_CTL is not right after resume. When CP access the MQD BO, wrong stride size is used, this will cause out of bound access on the MQD BO, resulting page fault.
The fix is to ensure gfx_v9_4_3_switch_compute_partition() is called when resume from a hibernation. KFD resume is called separately during a reset recovery or resume from suspend sequence. Hence it's not required to be called as part of partition switch.
(cherry picked from commit 5d1b32cfe4a676fe552416cb5ae847b215463a1a)
AI Insight
LLM-synthesized narrative grounded in this CVE's description and references.
A GPU page fault can occur in AMD GPU driver after hibernation on PF passthrough environments due to incorrect partition mode restoration.
The Linux kernel's AMD GPU driver (amdgpu) contains a vulnerability that can trigger a GPU page fault after hibernation on Physical Function (PF) passthrough setups. The root cause is that during hibernation, a Mode1 reset occurs, but the partition mode is not restored upon resume. This leaves registers mmCP_HYP_XCP_CTL and mmCP_PSP_XCP_CTL in an incorrect state, causing the Command Processor (CP) to use a wrong stride size when accessing the Memory Queue Descriptor (MQD) buffer object, leading to an out-of-bounds access and a page fault.
Exploitation requires a system with an AMD GPU configured in PF passthrough mode, where hibernation and resume occur. No authentication or network access is needed; the vulnerability manifests during the resume process. The bug specifically affects the gfx_v9_4_3 component and is triggered when the partition mode fails to be restored after a hibernation cycle.
The impact of this vulnerability is a GPU page fault, which can result in system instability, application crashes, or denial of service. In a PF passthrough environment, this could disrupt virtualized workloads relying on GPU acceleration.
The fix is already merged into the Linux kernel (commit 5d1b32cfe4a676fe552416cb5ae847b215463a1a) and backported. The patch ensures that gfx_v9_4_3_switch_compute_partition() is called during resume from hibernation to correctly restore the partition mode [1]. Users are advised to update their kernel to include this fix.
AI Insight generated on May 19, 2026. Synthesized from this CVE's description and the cited reference URLs; citations are validated against the source bundle.
Affected products
2Patches
0No patches discovered yet.
Vulnerability mechanics
AI mechanics synthesis has not run for this CVE yet.
References
3News mentions
0No linked articles in our index yet.