CVE-2022-49012

What it actually does

Incorrect use of atomic_inc_return instead of atomic_read on a reference counter prevents proper decrement and cleanup of afs_server structures, causing a resource leak that blocks module unload.

Why it matters to the business

Kernel module removal operations hang indefinitely, disrupting system maintenance, updates, and potentially leading to prolonged resource consumption in environments relying on AFS filesystem support.

How an attack unfolds

An attacker with local access triggers repeated server reference operations to exacerbate the leak; combined with module unload attempts this can be used to cause denial of service during system administration, though remote exploitation is not indicated.

How to fix it

Upgrade to a fixed kernel version containing the afs_put_server correction; for affected Gradio environments apply the patch from commit bbf9ba7e997022960c621f72baa891185bd03732 or later.

How to confirm the fix worked

Confirm atomic_read is restored in afs_put_server, verify module unload completes without the 'Can't purge' warning, and test AFS server reference counting under load.

Typical attack path

1. Obtain local access to a system with the vulnerable AFS kernel code
2. Perform operations that increment server active counts
3. Attempt rmmod or module unload to trigger the hang warning

How to detect it

Search kernel logs for 'Can't purge s=' messages during module unload; review dmesg and syslog for afs_put_server related warnings on systems with AFS enabled.

What your security team should do

• Prioritize kernel updates on all systems using AFS
• Monitor for module unload failures in production logs
• Restrict local access to kernel module management interfaces