UBUNTU-CVE-2026-68237: Medium severity ubuntu/linux-hwe-edge vulnerability
In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu/userq: fix indefinite fence wait during GPU reset prereset only force-completes fences of MAPPED queues. A queue in any other state (e.g. mid-eviction) keeps its lastfence pending; after a GPU reset that fence never signals, so the eviction/suspend worker and process teardown (amdgpuevfmgrflushsuspend) wait on it forever and wedge the machine: INFO: task kworker/6:28 blocked for more than 120 seconds. Workqueue: events amdgpuevictionfencesuspendworker [amdgpu] Call Trace: dmafencewaittimeout+0x7e/0x130 amdgpuuserqevict+0x67/0x140 [amdgpu] amdgpuevictionfencesuspendworker+0xd8/0x160 [amdgpu] processscheduledworks+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)
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
ubuntu/linux-aws-7.0to a version that resolves this vulnerability.Fixed in 7.0.0-1014.14~24.04.1 - Upgrade
Upgrade
ubuntu/linux-azure-7.0to a version that resolves this vulnerability.Fixed in 7.0.0-1017.17~24.04.1 - Upgrade
Upgrade
ubuntu/linux-azure-fde-7.0to a version that resolves this vulnerability.Fixed in 7.0.0-1016.16~24.04.1 - Upgrade
Upgrade
ubuntu/linux-gcp-7.0to a version that resolves this vulnerability.Fixed in 7.0.0-1014.14~24.04.1 - Upgrade
Upgrade
ubuntu/linux-hwe-7.0to a version that resolves this vulnerability.Fixed in 7.0.0-38.38~24.04.4 - Upgrade
Upgrade
ubuntu/linux-nvidia-7.0to a version that resolves this vulnerability.Fixed in 7.0.0-1024.25~24.04.1 - Upgrade
Upgrade
ubuntu/linux-oracle-7.0to a version that resolves this vulnerability.Fixed in 7.0.0-1013.13~24.04.1 - Upgrade
Upgrade
ubuntu/linux-riscv-7.0to a version that resolves this vulnerability.Fixed in 7.0.0-38.38.1~24.04.1 - Upgrade
Upgrade
ubuntu/linuxto a version that resolves this vulnerability.Fixed in 7.0.0-38.38 - Upgrade
Upgrade
ubuntu/linux-awsto a version that resolves this vulnerability.Fixed in 7.0.0-1014.14 - Upgrade
Upgrade
ubuntu/linux-azureto a version that resolves this vulnerability.Fixed in 7.0.0-1017.17 - Upgrade
Upgrade
ubuntu/linux-azure-fdeto a version that resolves this vulnerability.Fixed in 7.0.0-1016.16 - Upgrade
Upgrade
ubuntu/linux-gcpto a version that resolves this vulnerability.Fixed in 7.0.0-1014.14 - Upgrade
Upgrade
ubuntu/linux-gketo a version that resolves this vulnerability.Fixed in 7.0.0-1008.9 - Upgrade
Upgrade
ubuntu/linux-ibmto a version that resolves this vulnerability.Fixed in 7.0.0-1015.15 - Upgrade
Upgrade
ubuntu/linux-nvidia-bosto a version that resolves this vulnerability.Fixed in 7.0.0-2022.23 - Upgrade
Upgrade
ubuntu/linux-oem-7.0to a version that resolves this vulnerability.Fixed in 7.0.0-1015.15 - Upgrade
Upgrade
ubuntu/linux-oracleto a version that resolves this vulnerability.Fixed in 7.0.0-1013.13 - Upgrade
Upgrade
ubuntu/linux-realtimeto a version that resolves this vulnerability.Fixed in 7.0.0-38.38.1 - Upgrade
Upgrade
ubuntu/linux-riscvto a version that resolves this vulnerability.Fixed in 7.0.0-38.38.1 - Upgrade
Upgrade
Linux kernel drm/amdgpu userqto a version that resolves this vulnerability.Patch 9102b39fa924dcc3dc75a3137bfa9633c40b88c0
Event History
Frequently Asked Questions
Which deployments are relevant to this issue?
The issue applies to the listed Ubuntu kernel packages when the AMDGPU userq functionality is in use. The affected package set includes linux-riscv, linux-realtime, linux-hwe-edge, linux-azure, linux-gcp, linux-azure-fde, linux-gke, and linux-aws-5.0.
What conditions can trigger the system hang?
A GPU reset must occur while a user queue is in a state other than MAPPED, such as during eviction. Its pending fence may then never signal, causing eviction, suspend, or process-teardown work to wait indefinitely.
How can I identify a system already experiencing this problem?
Kernel logs may report a kworker task blocked for more than 120 seconds. The reported workqueue is amdgpu_eviction_fence_suspend_worker, with a call trace involving dma_fence_wait_timeout, amdgpu_userq_evict, and amdgpu_eviction_fence_suspend_worker.