Linux Kernel AMDGPU GFX9 CP ECC Error IRQ Handling Vulnerability

A vulnerability exists in the Linux kernel's handling of the Graphics Processing Unit (GPU) error interrupts for AMD's GFX9 architecture. Specifically, the issue arises in the AMDGPU driver when the 'cp_ecc_error_irq' interrupt is disabled without the proper conditions being met. This can lead to a call trace error, as the interrupt disablement should only occur when the legacy GFX Reliability, Availability, and Serviceability (RAS) is asserted. The vulnerability is present in the Linux kernel's stable releases that include this faulty interrupt handling.

Impact

The vulnerability can cause a call trace error due to an improper handling of the 'cp_ecc_error_irq' interrupt, which is related to error correction codes for the GPU's command processor. This mismanagement can disrupt the normal operation of the GPU, potentially leading to issues in graphics processing or computation tasks that rely on the GPU.

Reproduction

The vulnerability can be reproduced by suspending a device that has the 'cp_ecc_error_irq' interrupt enabled, without first ensuring that the legacy GFX RAS is asserted. This can be done by manually suspending the device through the PCI power management freeze process, which does not check the RAS condition before disabling the interrupt. The improper handling will trigger the call trace error, indicating that the 'amdgpu_irq_put' function was called without the necessary conditions being met.

Remediation

Users can upgrade to the latest version of the Linux kernel where this vulnerability has been addressed. The specific commit that resolves this issue is available in the Linux kernel stable tree.