Broadwell: Asynchronous wait on fence 0000:00:02.0:Xorg[707]:4c90 timed out (hint:intel_atomic_commit_ready [i915]), GPU HANG ecode 8:0:00000000
Rewriting the issue documented in #8687 to make it clearer: On laptops with the Broadwell IGP, the video chipset is subject to random freezing for around 10 or so seconds followed by a quick recovery, logging a dmesg error and GPU_HANG. In some cases the entire video output on the CF-RZ4 would freeze up. I have confirmed this issue in a Lenovo Thinkpad Yoga 12 (i5-5200u) and Panasonic Let's Note CF-RZ4 (m5-5y70) and both kernels 6.1 and 6.6. Attached are files from both the Yoga 12 and CF-RZ4.
[ 168.077810] Asynchronous wait on fence 0000:00:02.0:Xorg[707]:4c90 timed out (hint:intel_atomic_commit_ready [i915])
[ 172.174339] i915 0000:00:02.0: [drm] GPU HANG: ecode 8:0:00000000
[ 172.175347] i915 0000:00:02.0: [drm] Resetting rcs0 for stopped heartbeat on rcs0
To reproduce this error; download GPUTest 0.4.0 and run the furmark windowed over and over. Depending on the hardware, it might take more attempts to hang the GPU with a black window before it recovers and runs furmark. One of the users in #8687 made a script to automate this, I'm attaching it as an sh because the paste function of gitlab is kind of buggy.