Open rsidd120 opened 1 year ago
Sounds like a libinput bug.
Should it crash sway though? I understand that assertions help with debugging and also I was running a non-release sway version. Unfortunately this happened after an uptime of 2 months and I don't know how reproducible it is or whether it has been fixed already. I am now running sway version 1.8.1-68d620a8 (May 27 2023).
It happened again today, twice. I misunderstood the comment above: the assertion is in libinput code, not in sway. I have recompiled libinput changing that assertion to an "if" for the following line; I don't know whether that is correct and should I report it upstream to libinput folks?
Sway Version: sway version 1.8.1-68d620a8 (Mar 25 2023, branch 'v1.8')
Debug Log: https://gist.github.com/rsidd120/6219f5cbb5065b8667f8cc52cbf72229 I have included only the last 100 lines because the full file is 780MB after two months of uptime. (scroll down for the log file, sorry I included both config and log in same gist)
Configuration File: https://gist.github.com/rsidd120/6219f5cbb5065b8667f8cc52cbf72229 (sorry, same gist as above, includes both the config and the log file)
Stack Trace:
Unfortunately I restarted sway without obtaining this. If it recurs, I will update.
This was after two months of uptime (I suspend the laptop, I don't shut down or reboot except after system upgrades). I just clicked on a firefox window addressbar that I have clicked on many times before.
The relevant error in the log file seems to be 369:35:23.415 [DEBUG] [wlr] [types/wlr_xdg_activation_v1.c:47] Activation token 'b9eb30d3c1899db3744d773927e3a308' has expired sway: ../src/evdev.c:138: evdev_update_key_down_count: Assertion `device->key_count[code] > 0' failed. Gdk-Message: 06:32:27.036: Error reading events from display: Broken pipe