CVE-2024-36026

What it actually does

During S4 suspend operations on AMD SMU v13.0.4/11, the GC, RLC, and PMFW components can enter an invalid state leading to hard system hangs.

Why it matters to the business

Repeated hibernation operations on affected systems can cause unexpected freezes and downtime, reducing reliability for users relying on suspend-to-disk functionality.

How an attack unfolds

An attacker with local access repeatedly triggers S4 suspend cycles to force the power management firmware into an invalid state, resulting in system-wide hard hangs.

How to fix it

Apply the kernel fix that adds a GFX reset immediately before sending the MP1_UNLOAD message; update to a patched Linux kernel version containing the resolution for this issue.

How to confirm the fix worked

Run repeated S4 stress test cycles after applying the update and confirm that GC/RLC/PMFW components no longer enter invalid states or cause hangs.

Typical attack path

1. Obtain local access to the target system
2. Initiate multiple S4 suspend-to-disk operations in succession
3. Trigger the race condition in SMU power management before the MP1_UNLOAD message

How to detect it

Review kernel logs for entries related to SMU, GC, RLC or PMFW invalid states occurring around S4 suspend events.

What your security team should do

• Patch affected Linux kernels promptly to resolve the power management race condition
• Test hibernation functionality thoroughly in staging environments before broad deployment
• Monitor system logs for GPU power management errors during suspend operations