CVE-2026-68237
Received Received - Intake

GPU Reset Deadlock in AMDGPU Linux Kernel

Publication date: 2026-08-10

Last updated on: 2026-08-10

Assigner: kernel.org

Description
In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu/userq: fix indefinite fence wait during GPU reset pre_reset only force-completes fences of MAPPED queues. A queue in any other state (e.g. mid-eviction) keeps its last_fence pending; after a GPU reset that fence never signals, so the eviction/suspend worker and process teardown (amdgpu_evf_mgr_flush_suspend) wait on it forever and wedge the machine: INFO: task kworker/6:28 blocked for more than 120 seconds. Workqueue: events amdgpu_eviction_fence_suspend_worker [amdgpu] Call Trace: dma_fence_wait_timeout+0x7e/0x130 amdgpu_userq_evict+0x67/0x140 [amdgpu] amdgpu_eviction_fence_suspend_worker+0xd8/0x160 [amdgpu] process_scheduled_works+0xa6/0x420 Force-complete every queue's fence regardless of state. The unmap and mark-hung step stays gated on MAPPED, since unmapping a queue that is not mapped is invalid. (cherry picked from commit 9102b39fa924dcc3dc75a3137bfa9633c40b88c0)

CVSS Scores

EPSS Scores

Probability:
Percentile:

Meta Information

Published
2026-08-10
Last Modified
2026-08-10
Generated
2026-08-10
AI Q&A
2026-08-10
EPSS Evaluated
N/A
NVD
CVE-2026-68237
EUVD
EUVD-2026-55338

Affected Vendors & Products

Showing 1 associated CPE
Vendor Product Version / Range
amd amdgpu *

Helpful Resources

Exploitability

CWE
CWE Icon
KEV
KEV Icon
CWE ID Description
CWE-UNKNOWN

Attack-Flow Graph

AI Quick Actions

Instant insights powered by AI
Executive Summary

This vulnerability in the Linux kernel involves a flaw in the AMD GPU driver where fences for GPU queues are not properly handled during a GPU reset. Specifically, queues not in a mapped state retain their last fence, which never signals after a reset, causing indefinite waiting and system hangs.

Detection Guidance

This vulnerability is specific to the Linux kernel's AMDGPU driver and may cause system hangs during GPU operations. Detection involves checking kernel logs for blocked tasks or GPU-related errors. Use commands like dmesg | grep -i amdgpu or journalctl -k | grep -i blocked to identify hanging processes or fence-related issues.

Impact Analysis

This vulnerability can cause your system to hang or become unresponsive during GPU operations, particularly during suspend or eviction processes. It may lead to kernel tasks being blocked indefinitely, requiring a hard reboot.

Mitigation Strategies

Update the Linux kernel to a patched version that includes the fix for this AMDGPU fence issue. If immediate patching is not possible, avoid GPU-intensive tasks or suspend/resume operations that may trigger the bug. Monitor system logs for hanging processes as a temporary workaround.

Chat Assistant

Ask questions about this CVE
Hi! I’m here to help you understand CVE-2026-68237. Ask me anything about the vulnerability, its impact, or mitigation strategies.
0/70

EPSS Chart