Open mb720 opened 1 year ago
Probably related:
00:00:34.986 [ERROR] [wlr] [backend/drm/atomic.c:72] connector DP-2: Atomic commit failed: Device or resource busy
00:00:34.986 [DEBUG] [wlr] [backend/drm/atomic.c:76] (Atomic commit flags: PAGE_FLIP_EVENT | ATOMIC_NONBLOCK)
Can you capture a dmesg log?
Hm nothing interesting in dmesg.
I am having the same issues, both at my work and personal computer. For me, it's mostly around 1 second and it's not that frequent (Maby once or twice in two hours). Nothing interesting in dmesg for me aswell.
I am using Alpine Linux and Intel graphics.
me too!
Glad to see I'm not the only one.
I observed that lags occur more frequently in kitty than in wezterm.
I use Sway on Arch Vanilla (LTS) with all tools mentioned here, especially Kitty and Neovim. I do not have any of the new issues, all responses are lightning fast.
@mb720, I see you mentioning Mesa 22.3.6, while I also run Arch and the Aur gives me Mesa 23.0.3
@jstdk I upgraded to Mesa 23.0.0-3 in the meantime (package is from Arch's "extra" repository) but it didn't help with the lags.
I can provoke lags to happen more often in Neovim when running a video in the browser simultaneously.
Probably the same as https://gitlab.freedesktop.org/wlroots/wlroots/-/issues/3507
I'm having the same issue, especially in foot and neovim.
This might actually be a problem with neovim rather than sway. I don't have problems anymore since I switched to vis a few weeks ago.
This might actually be a problem with neovim rather than sway. I don't have problems anymore since I switched to vis a few weeks ago.
It also happens randomly outside of neovim. For example, when typing a command like "ls" and pressing Enter, sometimes the screen doesn't update until you press another key. Also, i see in the logs a bunch of "Atomic commit failed" errors like those mentioned by @emersion
I upgraded to Fedora 38 and encountered this bug. I was informed of a workaround via IRC.
From wlroots documentation: WLR_DRM_NO_ATOMIC: set to 1 to use legacy DRM interface instead of atomic mode setting
Setting this variable appears to help. No more errors printed in the logs and I haven't noticed the lag. I put it in /usr/share/wayland-sessions/sway.desktop
, replacing the Exec
line with Exec=env WLR_DRM_NO_ATOMIC=1 sway
Thanks at @adamthiede. I started Sway using WLR_DRM_NO_ATOMIC=1 sway
yesterday and had no lag issues since. Looks like an effective workaround. 👍
@adamthiede legend.
Just wanted to report that this still happens on latest master after the scene graph rewrite, with the same [ERROR] [wlr] [backend/drm/atomic.c:74] connector eDP-1: Atomic commit failed: Device or resource busy
error message.
I faced the same thing right after reboot, but with segfault in dmesg.
sway log: https://pastebin.com/8DRq8vDG
dmesg log: https://pastebin.com/ChqeHjGf
Sorry for the low resolution. This is my output:
I do not know what is causing it. But I use foot terminal.
Hi and thanks a lot for Sway!
Recently, I noticed that there is sometimes a delay in updating the screen of up to about five seconds.
Most commonly these delays occur in command line applications:
One of the symptoms in Neovim is that when moving the cursor, for example with I, it sometimes takes a couple of seconds till the cursor has actually moved.
I've experienced this issue both with kitty and wezterm as the terminal emulators.
The reason I think it's an issue with Sway is that there's sometimes a delay when switching workspaces: The workspace I switched to is only highlighted as the current one after a few seconds.
I noticed these delays after updating Sway to version 1.8.1.
Has anybody else seen lags like these?
Arch Linux 6.2.5
kitty 0.27.1 wezterm 20221119-145034-49b9839f
glxinfo | grep "OpenGL version"
: OpenGL version string: 3.3 (Compatibility Profile) Mesa 22.3.6Excerpt from
lshw -C display
:Output of
sway -d 2> ~/sway.log
is here.