KernelScan.io

HIGH Introduced in 4.5

vc4 Hangcheck UAF

CVE-2026-99095

CVSS 7.0 / 10.0 KernelScan AI

CVSS:3.1/AV:L/AC:L/PR:H/UI:N/S:U/C:H/I:H/A:H

01

In the Linux kernel, the following vulnerability has been resolved: drm/vc4: drain the hangcheck timer and works on V3D unbind Nothing stops the hangcheck timer that submitted jobs arm, or the job_done_work that the render-done interrupt queues, at teardown: vc4_irq_disable() cancels only overflow_mem_work, and vc4_gem_destroy() runs from the drm-managed release, after vc4_v3d_unbind() has already uninstalled the V3D interrupt and cleared vc4->v3d. A timer still armed by then reads V3D registers through the NULL vc4->v3d pointer, and the late callbacks run on the vc4_dev embedding them after it has been freed. Drain them in vc4_v3d_unbind(): shut the timer down and cancel reset_work before the interrupt is taken down, because its vc4_irq_reset() re-enables it, then flush job_done_work so that completions queued at teardown still release their jobs. This issue was found by an in-house static analysis tool.

02

Engine v0.9.0

Risk summary

A use-after-free in the VC4 GPU driver's V3D unbind path allows a local attacker with root privileges to trigger memory corruption during driver teardown. The hangcheck timer and job completion work items are not properly drained before the V3D interrupt is uninstalled and the device structure is freed, leading to NULL pointer dereferences and callbacks executing on freed memory. This primarily affects Raspberry Pi and other VideoCore IV/VI platforms where the VC4 driver is loaded.

Affecteddrivers/gpu/drm/vc4/vc4_v3d.c (vc4 DRM driver)

Vulnerability analysis

During V3D driver unbind, the hangcheck timer (armed by submitted GPU jobs) and the job_done_work workqueue item (queued by the render-done interrupt) remain active after the V3D interrupt is uninstalled and the vc4->v3d pointer is cleared. The timer callback subsequently reads V3D registers through the NULL pointer, and the work callbacks execute on the vc4_dev structure after it has been freed. The fix adds proper cleanup in vc4_v3d_unbind(): the hangcheck timer is shut down synchronously and the reset work is cancelled before the interrupt is uninstalled (since vc4_irq_reset() would re-enable it), and the job_done_work is flushed rather than cancelled so that any queued completions can still release their jobs cleanly. The vulnerability is reachable only during driver teardown — module unload, device removal, or explicit driver unbind via sysfs — all of which require root privileges or physical access to the hardware.

03

BranchIntroducedFixed inPatch commit
5.154.55.15.224839eebfc95f8
6.14.56.1.191a1e7c7789aef
6.124.56.12.113a046cd60ff3f
6.184.56.18.57e61465d3e029
6.64.56.6.16017c5fa371de2
7.24.57.2.11ade226dfd28c
mainline4.57.394cf5971b1d0
cip/6.124.56.12.114-cip34a046cd60ff3f

04

State of the newest release of each Civil Infrastructure Platform SLTS tree, from CIP's own fix records.

CIP treeState
cip/6.12Fixed in 6.12.114-cip34
cip/6.1Affected — no CIP fix yet (latest 6.1.190-cip62)
cip/6.1-rtAffected — no CIP fix yet (latest 6.1.189-cip61-rt32)
cip/5.10Affected — no CIP fix yet (latest 5.10.272-cip79)
cip/5.10-rtAffected — no CIP fix yet (latest 5.10.272-cip79-rt35)
cip/4.19Affected — no CIP fix yet (latest 4.19.325-cip137)
cip/4.19-rtAffected — no CIP fix yet (latest 4.19.325-cip137-rt51)