CVE-2026-43444 Linux CVE debrief

Who should care

Linux kernel maintainers, distribution security teams, and operators running AMD GPU/KFD workloads on affected kernels should prioritize this fix, especially on systems where local users can interact with amdkfd.

Technical summary

The vulnerable path is in amdkfd queue-update failure handling. The NVD description says the buffer object (bo) must be unreserved on failure, implying the flaw was a missing cleanup step in an error path. NVD maps the issue to CWE-476 and assigns CVSS 3.1 AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H, which aligns with a local privilege requirement and a primary availability impact rather than confidentiality or integrity compromise.

Defensive priority

Medium. The issue is local and requires limited privileges, but it can affect availability on impacted AMD GPU/KFD systems. Patch priority is higher for multi-user hosts, compute nodes, and any environment that depends on amdkfd stability.

Recommended defensive actions

• Apply the kernel fixes referenced by NVD for your branch or backport the stable patch set.
• Confirm whether your deployed kernels fall within the affected ranges: 6.12 before 6.12.78, 6.13 before 6.18.19, 6.19 before 6.19.9, or 7.0-rc1 through 7.0-rc3.
• Prioritize systems that use AMDGPU/KFD or other drm/amdkfd functionality.
• If you maintain downstream kernels, verify the cleanup path in the queue-update failure handling is present in your build.
• Track local-user activity and kernel error logs on affected hosts while remediation is rolled out.

Evidence notes

Source evidence is limited to the official NVD record and its referenced kernel patches. NVD states: "drm/amdkfd: Unreserve bo if queue update failed" and notes the error-handling path should unreserve the buffer object before returning failure. NVD also lists the CVSS vector AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H, CWE-476, vuln status Analyzed, and the affected Linux kernel version ranges.

Official resources