CVE-2025-38319

What it actually does

A NULL pointer returned by smu_atom_get_data_table() is dereferenced without validation inside atomctrl_initialize_mc_reg_table() and atomctrl_initialize_mc_reg_table_v2_2(), producing a kernel NULL pointer dereference in the AMD powerplay subsystem.

Why it matters to the business

Affected Linux systems using AMD GPUs may experience kernel crashes or instability during graphics initialization or power management operations, leading to downtime, lost work, and the need for immediate kernel updates across fleets with high downstream package prevalence.

How an attack unfolds

An attacker with local access on an affected system triggers the AMD display driver code path that calls atomctrl_initialize_mc_reg_table(), supplying conditions where smu_atom_get_data_table() returns NULL so the subsequent dereference occurs.

How to fix it

Update linux-lts packages to the fixed versions 6.1.54-r0 or 6.6.58-r0, or apply commit 85cdcb834fb490731ff2d123f87ca799c57dacf2 in the stable kernel tree; rebuild and reboot with the patched kernel.

How to confirm the fix worked

Verify the running kernel version matches or exceeds the fixed releases, then inspect the atomctrl_initialize_mc_reg_table() source to confirm an explicit NULL check exists before using the vram_info pointer.

Typical attack path

1. Obtain local user access on a Linux system with an AMD GPU
2. Trigger graphics or power-management initialization that reaches atomctrl_initialize_mc_reg_table()
3. Cause smu_atom_get_data_table() to fail and return NULL
4. Dereference the NULL pointer to induce kernel panic or attempt memory corruption exploitation

How to detect it

Search kernel logs and crash dumps for oops or panic messages referencing atomctrl_initialize_mc_reg_table or drm/amd/pp; correlate with processes that initialize AMD display or powerplay drivers.

What your security team should do

• Prioritize kernel upgrades on all AMD GPU hosts within the next maintenance window
• Add the fixed package versions to internal build and deployment pipelines
• Monitor EPSS trend and any new exploit maturity signals for this memory corruption issue