Thanks! I was wondering if there was a good way to do that!
I’ve attached the full journal from the time of the crash up to when I restarted:
Two weeks without a crash is good but I can’t confirm that the improvement came from the boot parameter because it just so happens that we finished raiding at the start of this month so my time played in WoW has reduced significantly too, so less chance of catching crashes.
I’ve now swapped to boot parameter amdgpu.runpm=0 only and will monitor for further crashes.
I don’t know if it is relevant to the crashes I’ve been having but I just bought Doom The Dark Ages and I was tinkering with the settings. Turning on Path Tracing caused an instant crash very similar to what I’ve been having.
After a reboot, the game crashed instantly as soon as it loaded. I had to delete the config file and re-do my settings without Path Tracing to continue.
Do either of these journals offer any further insight? And are they the same crash?
First journal is me changing the setting:
Second journal is me starting the game on a fresh boot
(Quick edit to say that in 70 minutes of play so far with Path Tracing off the game has run fine with no crashes. It crashed instantly when turning on Path Tracing)
I’ve just had a crash in Claire Obscure. Been playing for the last few weeks an hour or two a night with no issues. Tonight have had the first crash with this game (And the first crash since Doom path tracing). The error is back to what I was seeing at the start with lots of page flip timeout messages and amdgpu reset.
Journalctl below
If anyone can offer any suggestions or if the journalctl offers any more insight I’d love to hear it. In the meantime I’ll continue to the last boot parameter that I haven’t tried yet.
I’ve experienced this bug myself and have not yet resolved it or found a workaround. I think I triggered it last when invalid shaders loaded, but that’s not fully inside user control in unmodded games. Nepti has linked the issue I have been watching.
There was a commit to fix, but it seems like the fix targets context loss on hibernation. When I experienced this problem, it was entirely unrelated to hibernation. This new commit may not actually fix the problem
Potentially also related, especially regarding VRR / Adaptive Sync:
A lot of people seem to have this crash just during desktop use, but I’ve only experienced it in games so far. It doesn’t necessarily seem to happen at a reliable trigger though, I’ve taken to keeping LACT historical data open and it’s not like I’m seeing spikes around the crash time.
I did strongly suspect it was VRR / Freesync at the start because of the pageflip timeout part of the errors. However in my early testing I do think I experienced the crash even with VRR turned off in KDE settings.
Someone in those threads does mention seeing crashes stop after disabling Freesync at the monitor level, so I may have to test that (Though I’d really rather not lose VRR)!
