[Linux] Frequent crashes when starting application, X-related

[the-drunk-coder]

Hi,

recently I'm having plenty of (non-deterministic) crashes when starting my egui-based application.

[xcb] Unknown sequence number while processing queue
[xcb] Most likely this is a multi-threaded client and XInitThreads has not been called
[xcb] Aborting, sorry about that.
megra_rs: xcb_io.c:269: poll_for_event: Assertion `!xcb_xlib_threads_sequence_lost' failed.

Not really sure what this means. Usually I have to try two or three times, then the application starts. Any hints ?

I'm on Arch, Xorg, AwesomeWM

Best,
N

[slve]

I see get the same message, not just frequently, but any time I try to use the lib through epick.
Arch, xorg, dwm here.

[jhimcras]

I've got exactly same error message above.
I'm on Arch(Manjaro), i3wm

[the-drunk-coder]

There was something in the underlying winit library, supposedly fixed:

rust-windowing/winit#458

[d4h0]

I get this error as well (also on Arch Linux with i3, running the egui_template or the demo app).

Does anybody know how to overcome this?

The web backend works, fortunately, but it would be nice to run apps without a browser and the limitations of WASM.

[d4h0]

Because this was tagged with egui_glium I tried to run glium's examples. 80% percent fail with the same error that everybody here gets (the other 20% fail in different ways).

I cloned egui_template again, and tried to execute the example without modifying the code in any way. The app opened once, but not again after.

Edit: Actually, these are glitum's examples, not glium's (but glium seems to depend on glitum)

[emilk]

Similar issue, reported closed: rust-windowing/winit#458

Because this was tagged with egui_glium I tried to run glium's examples. 80% percent fail with the same error that everybody here gets (the other 20% fail in different ways).

Let's figure out if this is a problem with glium or winit. There is a new glium-free backend in the works at #685 - can you check it out and try to run it?

I cloned egui_template again, and tried to execute the example without modifying the code in any way. The app opened once, but not again after.

try removing the state at /home/UserName/.local/share/egui_template - does it work now? If so, see if you can figure out what part of the config file is causing the problem (maybe restoring the position or the size of the window?)

[AZMCode]

After some testing with the glium-free backend, it seems this is not a problem with glium. I'd advise to remove the egui_glium tag,

[AZMCode]

Since I'm running Gentoo after all I've been compiling a lot of stuff. And I don't know if this is a lead or not, but there seem to be less crashes when my computer is busy doing something.

[the-drunk-coder]

After some testing with the glium-free backend, it seems this is not a problem with glium. I'd advise to remove the egui_glium tag,

Yes, just tried with the glow backend, same problem.

[smorokin]

I get also the same error message using the eframe_template on Kubuntu 21.10:

[xcb] Unknown sequence number while processing queue
[xcb] Most likely this is a multi-threaded client and XInitThreads has not been called
[xcb] Aborting, sorry about that.
egui_test: ../../src/xcb_io.c:269: poll_for_event: Assertion `!xcb_xlib_threads_sequence_lost' failed.

Using the glow backend produces the same error.

[LongHairedHacker]

I ran into the same !xcb_xlib_threads_sequence_lost assertion a few days ago.
For me any egui example I try to run crashes 100% of the time, so my system might be a good environment to reproduce this problem.

Since rust-windowing/glutin#1034 (comment) suggests that it might be window manager related, I'd like to add that I also use i3 (also on Arch Linux).

I did not yet test the glow backend, but I will give it a try if can free up some time later today.

If there is anything else I should give a try just point me at piece of code and I'll try it out.
Likewise if there is any additional information about my system (software versions, graphics hardware...) that might be helpful, just ask.

Also I have one casual observation I'd like to add: Most of the earlier reports for this problem mention Arch Linux, and only the later one other distros like Ubuntu.
So it might be change Xorg/XCB/Mesa, that got adopted early by Arch and has now made its way into Ubuntu that causes the issue.

[LongHairedHacker]

Nope ... unfortunately the glow_example #685 runs into the same error:

[sebastian@laptop egui_glow] (master)$ cargo run --example pure
    Finished dev [unoptimized + debuginfo] target(s) in 0.16s
     Running `/tmp/egui/target/debug/examples/pure`
[xcb] Unknown sequence number while processing queue
[xcb] Most likely this is a multi-threaded client and XInitThreads has not been called
[xcb] Aborting, sorry about that.
pure: xcb_io.c:269: poll_for_event: Assertion `!xcb_xlib_threads_sequence_lost' failed.
Aborted (core dumped)

[LongHairedHacker]

Just for kicks I patched my mesa library using https://bugs.freedesktop.org/attachment.cgi?id=143767 and it actually seems to solve the problem. All examples I tested work now.
So if this is actually a mesa bug creating an error where there shouldn't be one, is there anything that can be done to work around it on the application side?

[NiceneNerd]

I can confirm the same issue on Arch under both i3 and Plasma. Patching mesa does indeed fix it.

[the-drunk-coder]

Hmm I just updated to egui 0.17.0, using glow instead of glium. Seems like it doesn't happen anymore ? But there's other things that changed in my system configuration as well, so I can't really tell what fixed it.

[emilk]

Should we perhaps close this in favor of rust-windowing/glutin#1034 ?

[the-drunk-coder]

Not opposed to that ...

[emilk]

Closed in favor of rust-windowing/glutin#1034