Created attachment 295021 [details] relevant dmesg output after kernel recovered from freeze Occasional temporary kernel freeze when monitor wakes up. The freeze only happens sometimes (which makes it hard to reproduce). Most of the time, it wakes up just fine. The last time it happened, I decided to wait, rather than resetting the computer, and the system recovered after some time (less than a minute). My graphics card is a Radeon RX Vega 64. CPU is Intel i7-7700K. I have 2 connected monitors: - Gigabyte G27QC connected via DisplayPort, running at its highest mode: 2560x1440@165Hz. - Samsung SyncMaster 2243WM connected via HDMI through a DvI adapter, also at its highest mode: 1680x1050@60Hz I don't know if the issue will occur if only one of the two monitors is connected; as I said, the bug is hard to reproduce. I am using the `sway` wayland compositor. After the system recovered from the freeze, I obtained the relevant `dmesg` output. I am attaching it as `dmesg-post-freeze.txt` (with relative timestamps). I am also attaching `dmesg-boot.txt` with the kernel messages from the system booting, in case you need it for more info about my system.
Created attachment 295023 [details] dmesg output from the system booting (for extra info)
OK, it just happened again. More info: The secondary monitor lights up, but just shows a black screen with the mouse cursor, but the primary monitor does not. This makes me suspect that my primary monitor is the culprit. As before, the kernel freezes for about a minute or so in that state, and then recovers. The dmesg output is shorter this time: [Feb 3 19:20] [drm] perform_link_training_with_retries: Link training attempt 1 of 4 failed [ +0.407677] [drm] perform_link_training_with_retries: Link training attempt 2 of 4 failed [ +0.460615] [drm] perform_link_training_with_retries: Link training attempt 3 of 4 failed [ +0.420583] [drm] enabling link 0 failed: 15 [ +0.177390] [drm] perform_link_training_with_retries: Link training attempt 1 of 4 failed [ +0.060044] [drm] perform_link_training_with_retries: Link training attempt 2 of 4 failed [ +0.111304] [drm] perform_link_training_with_retries: Link training attempt 3 of 4 failed [ +0.159993] [drm] enabling link 0 failed: 15 [Feb 3 19:21] [drm] REG_WAIT timeout 10us * 3000 tries - dce110_stream_encoder_dp_blank line:955 There was no backtrace like last time. BTW, note that the kernel "thinks" that ~1 second passed, based on those timestamps, but in reality, it was about a minute.
It happens every time I turn off monitor or it blanks out (and monitor goes into sleep mode). Kernel 5.11.0 rx 5700 xt Mesa 20.3.2 Pop OS 20.10 I can get back to DE but nothing works besides cursor movement (no response from clicking or keyboard). Only option is to go to terminal only screen and reboot.
I suspect you might be experiencing a different bug from mine, because in my case the kernel recovers after some time and everything is back in a usable state. There could be many different causes of the kernel freezing up, and I suspect the different symptoms probably mean that these are different bugs. Without a response/assistance from a kernel/driver developer, IDK what to do to investigate these bugs further.
I have the same/similar problem with my 2 monitors - but only when waking them from sleep. However, switching to terminal (ctrl+alt+fX) allows me to regain display output in both monitors, after that I can switch back to X and everything works - until the sleep timer turns the monitors off again.