---
id: "CVE-2026-99149"
kind: "vulnerability"
aliases: ["CVE-2026-99149"]
title: "Linux kernel: AMD GPU userptr ioctl leaks memory objects on allocation failure"
severity: "unknown"
cvss: null
epss: null
exploited: false
status: "open"
category: "infrastructure"
products: []
ecosystems: []
tags: ["amd-gpu", "amdgpu", "infrastructure", "linux-kernel", "self-hosted"]
not_affected_if: ["amdgpu-not-loaded", "no-amd-gpu"]
published_at: "2026-10-28T11:17:34+00:00"
first_seen_at: "2026-10-28T11:32:35+00:00"
updated_at: "2026-10-28T11:35:58+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-99149", "first_seen_at": "2026-10-28T11:32:35+00:00"}]
---

# Linux kernel: AMD GPU userptr ioctl leaks memory objects on allocation failure

When the AMD GPU driver's userptr ioctl runs out of memory, it returns early and skips cleanup, leaving a GEM object, a VM root reference and an MMU notifier registration behind. The practical effect is a resource leak in the kernel driver, which can degrade a system under memory pressure.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- unknown

## Fix

- Update to a Linux kernel release that includes upstream commit bbe739e407a542073ec1a7872b14843566b4919e (the fix is cherry-picked into stable trees; check your distribution's kernel changelog for it).

## Timeline

- 2026-10-28T11:32:35+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-99149

## From the sources

In the Linux kernel, the following vulnerability has been resolved:

drm/amdgpu: fix GEM object leak on ENOMEM in userptr ioctl

When amdgpu_hmm_range_alloc() fails amdgpu_gem_userptr_ioctl() returns
-ENOMEM right away and skips the cleanup every other error path goes
through. The GEM object, the bo->parent reference on the VM root and the
MMU notifier registration are all left behind.

Use release_object here, not user_pages_done, range is still NULL at
this point.

(cherry picked from commit bbe739e407a542073ec1a7872b14843566b4919e)

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_