CVE-2026-23034: drm/amdgpu/userq: Fix fence reference leak on queue teardown v2
In the Linux kernel, the following vulnerability has been resolved:
drm/amdgpu/userq: Fix fence reference leak on queue teardown v2
The user mode queue keeps a pointer to the most recent fence in
userq->last_fence. This pointer holds an extra dma_fence reference.
When the queue is destroyed, we free the fence driver and its xarray,
but we forgot to drop the last_fence reference.
Because of the missing dma_fence_put(), the last fence object can stay
alive when the driver unloads. This leaves an allocated object in the
amdgpu_userq_fence slab cache and triggers
This is visible during driver unload as:
BUG amdgpu_userq_fence: Objects remaining on __kmem_cache_shutdown()
kmem_cache_destroy amdgpu_userq_fence: Slab cache still has objects
Call Trace:
kmem_cache_destroy
amdgpu_userq_fence_slab_fini
amdgpu_exit
__do_sys_delete_module
Fix this by putting userq->last_fence and clearing the pointer during
amdgpu_userq_fence_driver_free().
This makes sure the fence reference is released and the slab cache is
empty when the module exits.
v2: Update to only release userq->last_fence with dma_fence_put()
(Christian)
(cherry picked from commit 8e051e38a8d45caf6a866d4ff842105b577953bb)
Security readout for executives and security teams
Plain-English summary
This Linux kernel AMDGPU issue is a cleanup bug. When a user-mode GPU queue is destroyed, the driver can keep one fence object referenced. The reported effect is a lingering slab object and warnings during driver unload, not disclosed code execution or data exposure.
Executive priority
Treat as routine kernel maintenance unless AMDGPU stability during unload is operationally important. There is no sourced evidence of active exploitation or severe security impact in the provided materials.
Technical view
In drm/amdgpu/userq, userq->last_fence holds an extra dma_fence reference. Queue teardown freed the fence driver and xarray without dma_fence_put() on last_fence. The fix releases and clears that pointer in amdgpu_userq_fence_driver_free().
Likely exposure
Exposure appears limited to Linux systems with affected AMDGPU kernel code and relevant user-mode queue teardown paths. The record lists Linux versions including 6.16, 6.18.7, and 6.19, but distro backports must be checked.
Exploitation context
The source bundle provides no KEV listing, no active exploitation evidence, and no exploitability details. The observable impact described is resource leakage during driver unload, shown by slab cache objects remaining at module exit.
Researcher notes
The issue is a dma_fence reference leak in AMDGPU user queue teardown. Current public evidence supports a lifecycle/resource-management impact, but does not establish privilege escalation, denial-of-service severity, or exploit preconditions.
Mitigation direction
Check your distribution kernel advisories for CVE-2026-23034 status.
Update to a kernel containing the referenced stable fixes.
Prioritize AMDGPU hosts where module unloads or GPU queue teardown occurs.
If no vendor package exists, follow vendor kernel guidance before custom backporting.
Validation and detection
Identify hosts running affected Linux kernel versions or AMDGPU backports.
Confirm whether the referenced stable commits are present in your kernel source.
Review kernel logs for amdgpu_userq_fence slab cache warnings during module unload.
Track vendor CVE metadata because the source lacks CVSS and CWE details.
Generated from the cited source records. This long-tail analysis has not been individually reviewed by a named human.
Potential ATT&CK relevance
Conservative CVE-to-ATT&CK context
These mappings and lookup hints may be relevant to the vulnerability behavior, CWE, affected product, or exposure path. Glexia-inferred context is not an official MITRE, ATT&CK, CWE, or CVE Program mapping.
ATT&CK lookup starting points
Use these exact CWE pages and searches to review the Glexia ATT&CK library from this CVE's weakness and description context.
cve · low confidence lookup
CVE-2026-23034 mapping review
Open the CVE-to-ATT&CK bridge for reviewed, inferred, or future official mappings tied to this CVE.
These fields come from the CVE record and ADP containers, not from Glexia's Take. They preserve time-varying source decisions such as CISA SSVC, KEV status, CVSS metrics, and provider references.
0CVSS vectors
3Timeline events
0ADP providers
3Source links
Vulnerability timeline
Timeline events are normalized from CVE metadata, CNA source timelines, ADP timelines, and KEV metadata when present.
CVE reservedCVE Program
The CVE ID was reserved by the assigning CNA.
CVE publishedCVE Program
The CVE record was published.
Jan 31, 2026, 11:42 UTC (UTC+00:00)
CVE updatedCVE Program
The CVE record metadata indicates this as the latest update time.