CVE-2024-46821

What it actually does

The vulnerability permits negative clk_index values to be used as an array index into pptable->DpmDescriptor in the AMD power management code path, resulting in out-of-bounds memory read due to improper input validation.

Why it matters to the business

Local attackers could achieve kernel-level memory corruption on Linux systems running AMD hardware, enabling privilege escalation, system instability, or sensitive data exposure across servers and endpoints.

How an attack unfolds

An attacker with local access supplies crafted inputs that produce a negative clk_index; the value bypasses validation and indexes into the DPM descriptor array, allowing memory corruption that can be leveraged for further exploitation.

How to fix it

Update the Linux kernel to commit befd1dc693c98bad69a701ede3a298698f0f9436 or later versions that enforce non-negative clk_index checks before array access.

How to confirm the fix worked

Confirm the patch is applied by inspecting the clk_index return-value validation logic in amd/pm source; recompile and boot-test with negative index inputs to verify safe rejection.

Typical attack path

1. Obtain local shell or process execution on a vulnerable Linux kernel with AMD PM enabled
2. Trigger the drm/amd/pm code path with input producing negative clk_index
3. Exploit resulting out-of-bounds read for memory disclosure or write primitives
4. Escalate to root or achieve persistence

How to detect it

Inspect kernel logs and dmesg for AMD PM or DPM descriptor warnings and crashes; correlate with unusual local process activity around power management syscalls.

What your security team should do

• Prioritize kernel upgrades to the fixed commit on all AMD Linux systems
• Restrict untrusted local code execution until patched
• Review and harden AMD PM driver usage in production workloads