CVE-2025-37854

In the Linux kernel, the following vulnerability has been resolved:

drm/amdkfd: Fix mode1 reset crash issue

If HW scheduler hangs and mode1 reset is used to recover GPU, KFD signal user space to abort the processes. After process abort exit, user queues still use the GPU to access system memory before h/w is reset while KFD cleanup worker free system memory and free VRAM.

There is use-after-free race bug that KFD allocate and reuse the freed system memory, and user queue write to the same system memory to corrupt the data structure and cause driver crash.

To fix this race, KFD cleanup worker terminate user queues, then flush reset_domain wq to wait for any GPU ongoing reset complete, and then free outstanding BOs.

This vulnerability poses minimal risk to Red Hat environments as the susceptible code path is not present in supported Red Hat Enterprise Linux kernels. The issue stems from a race condition during hardware recovery in the AMD compute driver, which requires specialized AMD GPU hardware and local execution privileges to trigger. Supported Red Hat enterprise products are not affected by this flaw.