Successfully triggered it (drained half the battery, plugged it in with a couple of infinite loops going and a bit of hotspot usage), and alas, attempts at mode switching aren’t solving it: `swaymsg output DSI-1 disable && swaymsg output DSI-1 enable`, `swaymsg -- output DIS-1 mode --custom 360x720`, all this kind of stuff causes the screen to turn off and back on again, but not start working again.
Actually, though, I’ve found something useful in dmesg, which is making me suspect that the other time I went looking I forgot to look there and only looked at journald.
[Mar14 21:17] [drm:lima_sched_timedout_job] *ERROR* lima job timeout
[ +3.422897] lima 1c40000.gpu: fail to save task state from sway pid 3227: error task list is full
[ +0.000115] lima 1c40000.gpu: gp task error int_state=0 status=0
Then a trace, including pc and lr being inside lima_devfreq_record_idle, which might point towards something about it not happening when the device is in active use. And yeah, lima is GPU driver.
Well, it’s something. Hmm, wonder if I can be bothered delving further. It doesn’t bother me as much as it possibly should, I really don’t interact with my phone directly all that much.
Actually, though, I’ve found something useful in dmesg, which is making me suspect that the other time I went looking I forgot to look there and only looked at journald.
Then a trace, including pc and lr being inside lima_devfreq_record_idle, which might point towards something about it not happening when the device is in active use. And yeah, lima is GPU driver.Well, it’s something. Hmm, wonder if I can be bothered delving further. It doesn’t bother me as much as it possibly should, I really don’t interact with my phone directly all that much.
(Seems to be reported as https://gitlab.freedesktop.org/mesa/mesa/-/issues/6036.)