CVE-2024-42268

What it actually does

Missing acquisition of the devlink lock before invoking devlink_remote_reload_actions_performed() during mlx5_sync_reset_reload_work triggers a lock assertion failure in devlink core, manifesting as a kernel warning and potential workqueue stall on affected network devices.

Why it matters to the business

Network interface resets on systems using mlx5 hardware can produce kernel warnings that interrupt operations, reduce availability of high-speed networking, and require restarts in environments with two or more downstream dependents relying on the affected kernel trees.

How an attack unfolds

An attacker with local access triggers a synchronous firmware reset on an mlx5 device, causing the reload work item to execute without the devlink lock; this leads to the assertion in devlink_notify when remote reload actions are reported, though no remote code execution or privilege escalation path is known.

How to fix it

Upgrade the Linux kernel to a version containing commit 5d07d1d40aabfd61bab21115639bd4f641db6002 or later; for the es5-ext package update to version 0.10.63 or newer. Apply the fix immediately given the 787 days since publication and presence in two downstream dependents.

How to confirm the fix worked

Verify the running kernel contains commit 5d07d1d40aabfd61bab21115639bd4f641db6002, confirm absence of the devl_assert_locked warning during mlx5 reset/reload sequences, and test devlink reload operations under load.

Typical attack path

1. Obtain local user or process execution on a host with mlx5 hardware
2. Trigger a firmware reset or devlink reload action that schedules mlx5_sync_reset_reload_work
3. Allow the work item to call devlink_remote_reload_actions_performed without holding the devlink lock
4. Observe the resulting WARNING from devl_assert_locked in the kernel log

How to detect it

Search kernel logs and dmesg for the string 'devl_assert_locked' combined with the mlx5_sync_reset_reload_work workqueue entry; inspect process trees for kworker threads invoking mlx5_fw_reset_events around the time of network device reloads.

• WARNING: CPU:.*devl_assert_locked+0x3e/0x50
• Workqueue: mlx5_fw_reset_events mlx5_sync_reset_reload_work [mlx5_core]
• devlink_notify+0x88/0x2b0

What your security team should do

• Prioritize kernel upgrades on all systems using mlx5 devices within the next maintenance window
• Monitor kernel logs for devl_assert_locked warnings on affected hosts
• Restrict local access to hosts with mlx5 hardware until patches are applied
• Track EPSS score stability and any new threat actor reports for this race condition