Due to an influx of spam, we have had to impose restrictions on new accounts. Please see this wiki page for instructions on how to get full permissions. Sorry for the inconvenience.
Admin message
Equinix is shutting down its operations with us on April 30, 2025. They have gracefully supported us for almost 5 years, but all good things come to an end.
Given the time frame, it's going to be hard to make a smooth transition of the cluster to somewhere else (TBD). Please expect in the next months some hiccups in the service and probably at least a full week of downtime to transfer gitlab to a different place.
All help is appreciated.
libx11 1.8.2 causes firefox to freeze when highlighting text
My platform is Arch Linux, running KDE, X11, Nvidia GPU.
I am getting intermittent freezes in firefox. It appears to happen when I highlight text. When it does happen, firefox becomes completely unresponsive. I haven't observed it happening in other applications, but I also don't highlight nearly as often in other applications.
I've narrowed down the package that causes the issue to Arch Linux's upgrade from libx11 1.8.1 to 1.8.2.
I have the same issue running Manjaro, KDE Plasma 5.26.3, Kernel 6.0.8, libX11 1.8.2, Nvidia 520.56.06 and similar segfaults as noted in the Firefox issue:
[ 7732.083412] firefox[13211]: segfault at d8 ip 00007f3451a04874 sp 00007ffc1e5b1e88 error 4 in libX11.so.6.4.0[7f34519f5000+8b000]
I will say though that the behavior is not consistent on my end with an RTX 3070 as well - I can highlight text usually so it doesn't seem like that is the problem for me but I'll comment again if I find a consistent way to make the behavior happen.
Thanks for the info, John. Right now I'm working on bisecting the commit.
A note about the freezes. I don't have a way to reproduce it on command either. I've just noticed that it seems like it always happens just as I highlight (or possibly attempt to u highlight) text.
Since it takes a while to trigger the bug, it's gonna be a few hours minimum before I know which commit has the bug.
Commit 79775575 is the culprit. It looks like a change in event handling and locking, seems reasonable that it could cause a lockup at first glance.
I've found I can reproduce it fairly reliably when firefox is restoring lots of windows and tabs. If I go to highlight some text while it is still loading/restoring, it almost always triggers the bug.
Hi, I also ran into this bug. Most likely it happens when I have 2 firefox windows open. One with a video and a second one where I highlight text. (I also recorded a video on how I reproduced it, crash at 26s https://www.youtube.com/watch?v=iOgGa4qTIJI )
As soon as I disable hardware acceleration in the firefox settings menu, I can't get it to crash anymore.
I'm running arch with nvidia on libx11 1.8.2-2. Hope this info helps to reproduce the bug.
It may have, just tested the changes and it have not been able to reproduce yet. Too early to tell for sure though. I retract what I said here, I am observing crashes as well. It seems to occur while highlighting.
@calcifer@seriousm4x@Littlejth can you test too? Just install libx11-git from the AUR.
I installed the git version and tried again. Firefox doesn't freeze anymore but now instantly crashes. The second time I tried, firefox gave me a crash log. I don't think it will help much as it only contains system infos.
Are there any x11 logs that might be helpful?
I did notice a crash on the first start of firefox, but after that it is working fine. Did you try launching it a third time?
Did you do a clean build of libx11-git?
When I tested it I only restarted sddm though, maybe something is lingering using the old x11 library on my system. Will try a full reboot later to see if the problem fully goes away.
@rockybulwinkle Yes, did a clean build and rebooted afterwards. Crash is as reproducible as before.
@crimist These are the only related logs in journalctl
Nov 22 21:08:30 max-arch firefox.desktop[17026]: Exiting due to channel error.Nov 22 21:08:30 max-arch firefox.desktop[14657]: Exiting due to channel error.Nov 22 21:08:30 max-arch firefox.desktop[17023]: Exiting due to channel error.Nov 22 21:08:30 max-arch firefox.desktop[14410]: Exiting due to channel error.Nov 22 21:08:30 max-arch firefox.desktop[14551]: Exiting due to channel error.Nov 22 21:08:30 max-arch firefox.desktop[16755]: Exiting due to channel error.Nov 22 21:08:30 max-arch firefox.desktop[17189]: Exiting due to channel error.Nov 22 21:08:30 max-arch firefox.desktop[14346]: Exiting due to channel error.Nov 22 21:08:30 max-arch firefox.desktop[14407]: Exiting due to channel error.Nov 22 21:08:30 max-arch firefox.desktop[16770]: Exiting due to channel error.Nov 22 21:08:30 max-arch firefox.desktop[14303]: Exiting due to channel error.Nov 22 21:08:30 max-arch systemd[1154]: app-gnome-firefox-14206.scope: Consumed 7min 21.903s CPU time.
I wanted to be sure I could still repro the crash on 1.8.2-2 - it took a few hours but finally occurred with logs similar to seriousm4x above. I have now installed libx11-git (1.8.2.r3.ga9e84580-1), I'll see if things crash.
Nov 22 23:17:35 matt-work1 plasmashell[9507]: [xcb] Unknown sequence number while processing queueNov 22 23:17:35 matt-work1 plasmashell[9507]: [xcb] You called XInitThreads, this is not your faultNov 22 23:17:35 matt-work1 plasmashell[9507]: [xcb] Aborting, sorry about that.Nov 22 23:17:35 matt-work1 plasmashell[9507]: firefox: xcb_io.c:278: poll_for_event: Assertion `!xcb_xlib_threads_sequence_lost' failed.Nov 22 23:17:35 matt-work1 plasmashell[9507]: ExceptionHandler::GenerateDump cloned child 57397Nov 22 23:17:35 matt-work1 plasmashell[9507]: ExceptionHandler::SendContinueSignalToChild sent continue signal to childNov 22 23:17:35 matt-work1 plasmashell[57397]: ExceptionHandler::WaitForContinueSignal waiting for continue signal...Nov 22 23:17:35 matt-work1 plasmashell[55493]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[57206]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[57289]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[29069]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[56895]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[57351]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[57110]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[48866]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[56892]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[21325]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[11459]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[9951]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[9949]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[43809]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[9762]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[21396]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[9754]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[14378]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[9757]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[9603]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[9657]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[11191]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[53262]: Exiting due to channel error.Nov 22 23:17:35 matt-work1 plasmashell[975]: file:///usr/share/plasma/plasmoids/org.kde.plasma.mediacontroller/contents/ui/ExpandedRepresentation.qml:649: TypeError: Value is undefined and could not be converted to an objectNov 22 23:17:35 matt-work1 plasmashell[975]: file:///usr/share/plasma/plasmoids/org.kde.plasma.mediacontroller/contents/ui/CompactRepresentation.qml:77: TypeError: Cannot read property 'implicitWidth' of nullNov 22 23:17:58 matt-work1 kwin_x11[875]: kwin_core: XCB error: 152 (BadDamage), sequence: 34636, resource id: 17607341, major code: 143 (DAMAGE), minor code: 3 (Subtract)Nov 22 23:18:00 matt-work1 plasmashell[57398]: Failed to open curl lib from binary, use libcurl.so insteadNov 22 23:18:00 matt-work1 plasmashell[57446]: [GFX1-]: glxtest: VA-API test failed: failed to initialise VAAPI connection.
Another crash, this time no crash report dialog. Happened when I tried to paste into a google sheet.
Click to expand
Nov 23 01:30:19 matt-work1 plasmashell[57446]: [xcb] Unknown sequence number while processing queueNov 23 01:30:19 matt-work1 plasmashell[57446]: [xcb] You called XInitThreads, this is not your faultNov 23 01:30:19 matt-work1 plasmashell[57446]: [xcb] Aborting, sorry about that.Nov 23 01:30:19 matt-work1 plasmashell[57446]: firefox: xcb_io.c:278: poll_for_event: Assertion `!xcb_xlib_threads_sequence_lost' failed.Nov 23 01:30:19 matt-work1 plasmashell[57446]: ExceptionHandler::GenerateDump cloned child 79197Nov 23 01:30:19 matt-work1 plasmashell[79197]: ExceptionHandler::WaitForContinueSignal waiting for continue signal...Nov 23 01:30:19 matt-work1 plasmashell[57446]: ExceptionHandler::SendContinueSignalToChild sent continue signal to childNov 23 01:30:19 matt-work1 plasmashell[78773]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[78770]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[77690]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[75402]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[67652]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[78227]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[78182]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[77312]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[57939]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[65117]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[58028]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[79033]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[58014]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[57684]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[59929]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[62600]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[78934]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[57676]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[59235]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[57886]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[57679]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[57590]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[57541]: Exiting due to channel error.Nov 23 01:30:19 matt-work1 plasmashell[975]: file:///usr/share/plasma/plasmoids/org.kde.plasma.mediacontroller/contents/ui/ExpandedRepresentation.qml:649: TypeError: Value is undefined and>Nov 23 01:30:19 matt-work1 kwin_x11[875]: kwin_core: Failed to focus 0xa0000b3 (error 3)Nov 23 01:30:19 matt-work1 plasmashell[975]: file:///usr/share/plasma/plasmoids/org.kde.plasma.mediacontroller/contents/ui/CompactRepresentation.qml:77: TypeError: Cannot read property 'im>Nov 23 01:30:19 matt-work1 systemd[810]: app-firefox\x2dnightly-f2e0616932a04b82a65b7ea53943be81.scope: Consumed 2h 11min 58.776s CPU time.
When I tested that commit, it caused firefox to crash instead of hanging. I haven't had time to investigate further, and have been holding my libx11 on 1.8.1 on my arch install.
Also downgraded the rest of my Arch system to November 17th, but with that pull request. So far, haven't seen the crash, but definitely need to test it for longer. saw a crash, see below.
It definitely crashes when I revert everything back to the 17th. Now trying with everything up to date, plus that pull request, I'll let you know if I see a crash with everything else up to date.
Something else to note, I haven't recognized any other crashes, it seems like firefox is the only one that is crashing. I'm starting to wonder if this really is a firefox bug? I don't know what the stance is on core libraries like libx11 is w.r.t. breaking other apps. Like, if it's anything like with the kernel, where if it breaks a userland app it's basically definitionally a kernel bug.
One last thing, my firefox is now on 107.0.1, before it was on 107.0, though they don't list anything in their release notes about fixing a copy/paste bug, maybe a fix on firefox's side snuck in? pretty sure the firefox version number isn't relevant.
Anyway I hope we can get this sorted out soon. Let me know if there's anything more I can do to help.
Something I forgot to say clearly, the crash happened just after highlighting some text and hit CTRL+C, whereas in the original presentation, it would freeze while just highlighting. If it matters, the crash happened in an entry box, not unlike the one I'm typing this message in now.
Using latest libX11 v1.8.2 on openSUSE Tumbleweed, vkquake also has this issue, crashing between 2s and 1min after starting a map:
[xcb] Unknown sequence number while processing queue[xcb] You called XInitThreads, this is not your fault[xcb] Aborting, sorry about that.vkquake: xcb_io.c:278: poll_for_event: Assertion `!xcb_xlib_threads_sequence_lost' failed.Aborted (core dumped)
On ba095967, the crashes I saw always seem to be when pasting into empty inputs, but no obvious pattern beyond that and most of the time it doesn't crash. Usually a crash report was created, I have 3 similar crash reports on this commit.
Edit: I was copying between a plain HTML page and google docs, 70+ copy/pastes in, and it crashed. Nothing obvious that was different.
Anything I could do to passively collect helpful info?
1.8.2 with a9e84580 crashes GZDoom, sometimes very quickly. I'm pretty sure I saw it happen with 1.8.3 too, but it didn't happen when I played for a few minutes just now.
I've also noticed this the past few weeks. It happened to me tonight with three windows open, one was a YouTube video and another with a lot more things queued up in tabs, the third with a duolingo.com lesson open.
It hung when I pressed "Check" on the Duolingo Firefox window which had no other tabs. I noticed the button "Check" was highlighted by accident as all three Firefox windows hung as a result and I could no longer click "Check" on the page to continue the lesson. Seems consistent enough an experience with this bug's title.
On the latest libx11, I no longer experience this issue. I haven't tested if !176 (merged) was really what solved it, but I assume so. I think the issue can be closed.
Ah, my mistake. For some reason I thought it made it into 1.8.3. I see now that it did not, and yes arch is in fact pulling in !176 (merged) in their PKGBUILD.
I pulled in the patch in the Arch package because things break quite a bit when it's not in there. I'd appreciate this to be fixed upstream, of course, but for the time being I don't want our users to get too annoyed.
1.8.4's been working well for me too. And I actually also had a severe Steam Link performance hit with 1.8.2. Now both Steam Link and Firefox have been normal since 1.8.4.