CVE-2023-54115

What it actually does

Memory leak in nonstatic_release_resource_db() that fails to free socket_data structure allocated during nonstatic_init for PCMCIA sockets, leaving unreferenced kernel memory objects.

Why it matters to the business

Gradual kernel memory exhaustion on systems with PCMCIA/CardBus hardware when drivers are repeatedly loaded or sockets are registered, potentially causing performance degradation or instability over extended uptime.

How an attack unfolds

An attacker with local access repeatedly triggers PCMCIA socket registration and release paths (for example by binding and unbinding yenta_socket or similar drivers) to accumulate leaked socket_data allocations until memory pressure leads to denial of service.

How to fix it

Apply the upstream fix that adds the missing kfree of socket_data in nonstatic_release_resource_db, available in commit bde0b6da7bd893c37afaee3555cc3ac3be582313; update to a Linux kernel containing this commit.

How to confirm the fix worked

After patching, reproduce the bind/unbind sequence and confirm with kmemleak that no unreferenced pcmcia socket_data objects remain; verify the presence of the socket_data free call in the release function.

Typical attack path

1. Obtain local access to a system exposing PCMCIA sockets
2. Repeatedly load and unload PCMCIA resource drivers or manually bind/unbind yenta_cardbus
3. Trigger nonstatic_init followed by nonstatic_release_resource_db without the socket_data free
4. Accumulate unreferenced 64-byte kernel objects until kmemleak or OOM conditions appear

How to detect it

Scan kmemleak output and kernel logs for unreferenced objects sized 64 bytes with backtraces containing nonstatic_init and pcmcia_register_socket; correlate with yenta_socket or pcmcia_rsrc module activity.

• unreferenced object.*size 64.*nonstatic_init
• pcmcia_rsrc
• yenta_probe
• nonstatic_release_resource_db

What your security team should do

• Prioritize kernel updates containing bde0b6da7bd893c37afaee3555cc3ac3be582313 on any systems that may expose PCMCIA hardware
• Disable unused PCMCIA/CardBus drivers in kernel configuration where hardware is absent
• Monitor kmemleak reports on long-running systems for pcmcia-related leaks until patches are deployed