Open ghost opened 3 years ago
Can you try again with Sway master? See https://github.com/swaywm/sway/wiki/Development-Setup#compiling-as-a-subproject
So I compiled a binary , renamed it sway_master, made SUID and moved to /usr/local/bin After the next dpms event crashed wayland again I rebooted and tried to use it, it wouldn't even start. System became completely unresponsive.
If you are suggesting that master fixes this issue, I guess I will have to wait to try it.
Worth pointing out that the crash happened after far fewer dpms events than last time (I killed swayidle until I really needed it), there was probably something in the background preparing the crash.
Hopefully manually turning the monitors off doesn't trigger it too.
Looks like this issue is still here, don't know if it's a duplicate of #5757 but I'm linking to my comment there https://github.com/swaywm/sway/issues/5757#issuecomment-830917121
@emersion While I am not the reporter - I can see this exact behaviour (too many open files, "frozen windows that won't go away" & all) in current sway 1.6.1 & wlroots 0.14.1.
It also happens around the time of swaylock/while idling. I run exec swayidle -w
in my sway config and swayidle config is:
timeout 300 'swaylock -f -i <path to image>'
timeout 600 'swaymsg "output * dpms off"' resume 'swaymsg "output * dpms on"'
before-sleep 'swaylock -f -i <path to image>'
Given timing of your comment & Sway release - I suspect the answer to your question is that this is not fixed. But if current master might be better - I can retry.
cpu is Ivy Bridge,gpu radeon r5 230 linux void glibc, no elogind, no seatd, sway SUID.
I had a swayidle script that would liberally spam swaymsg "output * dpms on" (it had a 30 second idle window, intended to only trigger dpms off when lockscreen is engaged). This would cause a split second cursor stutter accompanied with a cpu spike wheneven the computer was idle for 30 seconds, even when swaylock was never used.
I have since fixed it to only work with swaylock running.
Nevertheless this seems like a bug: There were no windows utilizing xwayland at all.
sway debug log is littered with:
which seems to have at one point culminated in a wayland crash (all windows had a frozen image and would only be closed when the process in them was killed). sway would allow new windows.
as best as i can tell crash culminated with: