[NV126] Random unrecoverable hangs with "reason 02 [PTE]" channel faults in dmesg
Software:
Arch Linux
Linux 5.10.7
Mesa 20.3.3
Sway 1.5.1
This issue manifests as a frozen screen, requiring a reboot to recover from. I have attached 4 separate kernel logs of this issue happening. The first two were taken on a tainted kernel, but the second two are from an untainted kernel and demonstrate the same issue. Excerpt of relevant lines from dmesg-2
:
1220:Dec 09 21:45:19 glados kernel: nouveau 0000:01:00.0: fifo: fault 01 [WRITE] at 0000000002cc0000 engine 15 [CE0] client 01 [HUB/CE0] reason 02 [PTE] on channel 1 [007f8df000 DRM]
1221:Dec 09 21:45:19 glados kernel: nouveau 0000:01:00.0: fifo: channel 1: killed
1222:Dec 09 21:45:19 glados kernel: nouveau 0000:01:00.0: fifo: runlist 0: scheduled for recovery
1223:Dec 09 21:45:19 glados kernel: nouveau 0000:01:00.0: fifo: runlist 4: scheduled for recovery
1224:Dec 09 21:45:19 glados kernel: nouveau 0000:01:00.0: fifo: engine 4: scheduled for recovery
1225:Dec 09 21:45:45 glados kernel: nouveau 0000:01:00.0: DRM: DDC responded, but no EDID for DVI-I-1
1226:Dec 09 21:45:46 glados kernel: nouveau 0000:01:00.0: DRM: DDC responded, but no EDID for DVI-I-1
1237:Dec 09 21:46:09 glados kernel: nouveau 0000:01:00.0: Renderer[3209]: failed to idle channel 5 [Renderer[3209]]
1238:Dec 09 21:46:24 glados kernel: nouveau 0000:01:00.0: Renderer[3209]: failed to idle channel 5 [Renderer[3209]]
1239:Dec 09 21:46:24 glados kernel: nouveau 0000:01:00.0: fifo: fault 00 [READ] at 0000000000013000 engine 07 [HOST0] client 07 [HUB/HOST_CPU] reason 02 [PTE] on channel 5 [007eeb1000 Renderer[3209]]
1240:Dec 09 21:46:24 glados kernel: nouveau 0000:01:00.0: fifo: channel 5: killed
1241:Dec 09 21:46:24 glados kernel: nouveau 0000:01:00.0: fifo: runlist 0: scheduled for recovery
1242:Dec 09 21:46:24 glados kernel: nouveau 0000:01:00.0: Renderer[3209]: channel 5 killed!
1254:Dec 09 21:46:39 glados kernel: nouveau 0000:01:00.0: systemd-logind[447]: failed to idle channel 2 [systemd-logind[447]]
1255:Dec 09 21:46:54 glados kernel: nouveau 0000:01:00.0: systemd-logind[447]: failed to idle channel 2 [systemd-logind[447]]
1256:Dec 09 21:46:54 glados kernel: nouveau 0000:01:00.0: fifo: fault 00 [READ] at 0000000000013000 engine 07 [HOST0] client 07 [HUB/HOST_CPU] reason 02 [PTE] on channel 2 [007f8d6000 systemd-logind[447]]
1257:Dec 09 21:46:54 glados kernel: nouveau 0000:01:00.0: fifo: channel 2: killed
1258:Dec 09 21:46:54 glados kernel: nouveau 0000:01:00.0: fifo: runlist 0: scheduled for recovery
1259:Dec 09 21:46:54 glados kernel: nouveau 0000:01:00.0: fifo: engine 7: scheduled for recovery
1260:Dec 09 21:46:54 glados kernel: nouveau 0000:01:00.0: systemd-logind[447]: channel 2 killed!
Edited by Tom Hebb