CVE-2026-46311: drm/amdgpu/userq: fix access to stale wptr mapping
In the Linux kernel, the following vulnerability has been resolved: drm/amdgpu/userq: fix access to stale wptr mapping Use drm_exec to take both locks i.e vm root bo and wptr_obj bo to access the mapping data properly. This fixes the security issue of unmap the wptr_obj while a queue creation is in progress and passing other bo at same address. (cherry picked from commit 1fc6c8ab45dbee096469c08c13f6099d57a52d6c)
Metrics
- CVSS v3.1
- 7.8
- Severity
- HIGH
- Fixed in
- 0
- Affected Products
- 2
HarborGuard Analysis
Synopsis
A stale-pointer vulnerability in the Linux kernel's AMD GPU driver (drm/amdgpu) allows a local attacker with a low-privilege account to exploit a race condition between queue creation and wptr_obj unmapping. By substituting a different buffer object at the same address during queue creation, an attacker can access stale mapping data in the userq subsystem. Successful exploitation gives the attacker full read, write, and crash capabilities over the affected system. A patched-image rebuild at the fix commits and version 7.0.9 is available on HarborGuard for environments running an affected kernel.
HarborGuard Coverage
Detection of CVE-2026-46311 is available across every HarborGuard environment; the CVE is ingested from upstream Linux kernel security feeds within minutes of publication and matched against customer images, including custom-built images that package an affected kernel version.
AvailableTriage is available with a CVSS v3.1 score of 7.8 (HIGH), weighted against each customer organization's compliance policy, and routed to the appropriate team inbox within the customer's HarborGuard environment.
AvailableA patched-image rebuild targeting the fix commits (336a9186f3a4b65bbd865d93936605ac8a1a3991, 6da7b1242da4455b11c24ce667d1cab1a348c8ea) and version 7.0.9 is available on HarborGuard for environments running an affected kernel version. For customers who opt into auto-remediation, HarborGuard performs the rebuild, runs a regression test suite, and opens a PR against affected workloads automatically.
AvailableExploit Conditions
- Network reachabilityNot required
The attacker needs an existing shell or process on the host; no network access is required to reach the vulnerable code path.
- AuthenticationRequired
Any low-privilege local account is sufficient to trigger the vulnerable queue creation path in the amdgpu driver.
- Victim interactionNot required
The attacker can carry out the exploit entirely on their own without requiring any action from another user.
- Attack complexityDetail
The exploit is reliable and condition-free once local access is established; no race-window tuning or specific memory layout is required beyond the inherent race in queue creation.
Blast Radius
- A successful attacker reads arbitrary kernel memory, including stored credentials, session tokens, and other sensitive in-memory data.
- A successful attacker writes to kernel memory structures, enabling privilege escalation or persistent modification of security-relevant kernel state.
- A successful attacker can crash the affected host by corrupting kernel data, causing a denial of service for all workloads on that node.
- Container workloads sharing the host kernel are exposed to the same impact, since kernel memory is not isolated between containers on the same node.
How HarborGuard Handles This
Available on HarborGuard: detection fires within minutes of CVE publication against all customer images that include an affected Linux kernel build, covering both upstream base images and internally built images. For environments where the fix version is available, a patched-image rebuild is offered; customers with auto-remediation enabled receive a rebuilt image, a regression-test run, and a PR opened against affected workloads automatically. For high-severity issues, the median time from CVE publication to a merged patch PR is around 90 minutes in environments with auto-remediation enabled. Because the vulnerability requires local shell access, compensating controls include tightening pod security admission to block privilege escalation, restricting access to AMD GPU device nodes via device plugin policy, and auditing which workloads mount GPU resources until the kernel image is updated.
Fix available
- Linux / Linux< 336a9186f3a4b65bbd865d93936605ac8a1a3991 (from 5fb2f7fc21a3668e5794cc0d153641b9719713e1) · < 6da7b1242da4455b11c24ce667d1cab1a348c8ea (from 5fb2f7fc21a3668e5794cc0d153641b9719713e1)
- Linux / Linux6.16Fixed in 0, 7.0.9, 7.1-rc3
CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H