Open gustafla opened 2 years ago
I get similar behaviour and logs when I swayidle
calls wlr-randr DP-1 --off
in a multi-monitor setup (note that I'm using river
, but this is a compositor that is also built upon wlroots, like sway
This happens to me constantly, and also when plugging/unplugging displays.
I switched to sway a few days a go, I got this same issue a few times when waking up from sleep or leaving idle mode after a dpms off
. I do not use kanshi or any dynamic screen manager.
I don't know if it's related but I have a lot of errors like:
Sep 16 10:07:35 yuki /usr/libexec/gdm-wayland-session[6188]: 01:56:46.890 [ERROR] [wlr] [backend/drm/atomic.c:34] connector DP-4: Atomic commit failed (pageflip): Device or resource busy
Sep 16 10:07:35 yuki /usr/libexec/gdm-wayland-session[6188]: 01:56:46.890 [ERROR] [sway/config/output.c:501] Failed to commit output DP-4
Sep 16 10:07:35 yuki /usr/libexec/gdm-wayland-session[6188]: 01:56:46.890 [ERROR] [wlr] [backend/drm/atomic.c:34] connector eDP-1: Atomic commit failed (pageflip): Device or resource busy
Sep 16 10:07:35 yuki /usr/libexec/gdm-wayland-session[6188]: 01:56:46.890 [ERROR] [sway/config/output.c:501] Failed to commit output eDP-1
The relevant logs before crash:
Sep 16 10:31:39 yuki /usr/libexec/gdm-wayland-session[42922]: 00:00:03.088 [ERROR] [common/ipc-client.c:87] Unable to receive IPC response
Sep 16 10:31:42 yuki /usr/libexec/gdm-wayland-session[42927]: 00:00:03.067 [ERROR] [common/ipc-client.c:87] Unable to receive IPC response
Sep 16 10:31:49 yuki /usr/libexec/gdm-wayland-session[42975]: 42199
Sep 16 10:31:49 yuki /usr/libexec/gdm-wayland-session[42975]: 42214
Sep 16 10:31:49 yuki /usr/libexec/gdm-wayland-session[22852]: 2022-09-16 10:31:49 - [main.c:392] Destroying output DP-4 (Iiyama North America PL2730H 1153621211304)
Sep 16 10:31:49 yuki waybar[22857]: Invalid id 8 passed to g_bus_unwatch_name()
Sep 16 10:31:49 yuki /usr/libexec/gdm-wayland-session[22857]: [2022-09-16 10:31:49.809] [info] Bar removed from output: DP-4
Sep 16 10:31:49 yuki waybar[22857]: Dropped dispatcher message as the dispatcher no longer exists
Sep 16 10:31:50 yuki /usr/libexec/gdm-wayland-session[6188]: 02:21:01.238 [ERROR] [wlr] [types/output/output.c:656] Basic output test failed for eDP-1
Sep 16 10:31:50 yuki /usr/libexec/gdm-wayland-session[6188]: 02:21:01.238 [ERROR] [sway/config/output.c:501] Failed to commit output eDP-1
Sep 16 10:31:50 yuki /usr/libexec/gdm-wayland-session[22852]: 2022-09-16 10:31:50 - [main.c:340] Found config Iiyama North America PL2730H 1153621211304 for output DP-4 (Iiyama North America PL2730H 1153621211304)
Sep 16 10:31:50 yuki /usr/libexec/gdm-wayland-session[6188]: 02:21:01.332 [ERROR] [wlr] [types/output/output.c:656] Basic output test failed for eDP-1
Sep 16 10:31:50 yuki /usr/libexec/gdm-wayland-session[6188]: sway: types/output/render.c:103: output_attach_back_buffer: Assertion `output->back_buffer == NULL' failed.
Perhaps it would be useful to have an option to disable output hotplug handling on sway's side if external tool like kanshi is used.
Yeah, that would be a nice feature.
Sway Version:
sway version 1.7
Debug Log:
Configuration File: My main sway config file And the included file
output
:Stack Trace:
Description:
When I launch sway, sometimes (<5% of the time?) it crashes very soon after startup. From reading the logs, to me it looks like it's trying to test/assert some properties about a display (eDP) which
kanshi
has already disabled. I'm not sure if this is sway's or kanshi's bug, but I believe sway should "lock" the displays from external influence when it's going to do assertions or configuration. Or buffer the API/IPC calls kanshi makes during the critical section. I don't know, I haven't read the sources of either program.