You are not logged in.
Hi all,
for a some months my firefox instances crash, when I update my computers (pacman -Syyu, laptop & pc).
I wonder, if this is normal for arch/firefox or if this might be a problem of my installation - and where to start to look for the reason.
% inxi -SI
System:
Host: hog4 Kernel: 6.7.0-arch3-1 arch: x86_64 bits: 64 Desktop: KDE Plasma
v: 5.27.10 Distro: Arch Linux
Info:
Processes: 401 Uptime: 1d 17h 24m Memory: total: 16 GiB note: est.
available: 14.97 GiB used: 9.97 GiB (66.6%) Shell: Zsh inxi: 3.3.31
Any idea anyone?
Thanks & kind regards,
Peer
Offline
pacman -Syyu
Don't use double y by default, it is only needed in very special cases.
I wonder, if this is normal for arch/firefox or if this might be a problem of my installation
The second one .
Are you running pacman -Syu while firefox is active ?
Last edited by Lone_Wolf (2024-01-20 13:59:53)
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline
More of https://bbs.archlinux.org/viewtopic.php … 2#p2126702 ?
Do you have a backtrace for FF?
Offline
It's been the same for me since a couple of months. It seems to be triggered by one of the libalpm hooks. It could be 30-update-mime-database.hook but I'm not sure. I was never able to reproduce the crash manually running this or other hooks.
Offline
Did you see the link and attempt to
sudo touch /etc/systemd/do-not-udevadm-trigger-on-update
?
Offline
Did you see the link and attempt to
sudo touch /etc/systemd/do-not-udevadm-trigger-on-update
?
I cannot reproduce the crash calling `/usr/share/libalpm/scripts/systemd-hook udev-reload` but it's still worth a try. I'll report back but it may take weeks or even months to determine if this fixes the issue or not.
Offline
Sorry, was out of business for some days ..
pacman -Syyu
Don't use double y by default, it is only needed in very special cases.
Usually I used it for command-recall: "ctrl-r" yy
I am going to change it now, thank you for the hint!
I wonder, if this is normal for arch/firefox or if this might be a problem of my installation
The second one .
Are you running pacman -Syu while firefox is active ?
Yes: In the past, I have not observed an interaction between pacman and other running programs - except pacman ;-)
Kind regards,
Peer
Offline
More of https://bbs.archlinux.org/viewtopic.php … 2#p2126702 ?
Do you have a backtrace for FF?
Actually, I don't know, how to generate a backtrace. Do you know a handy How-To for the creation?
Kind regards,
Peer
Last edited by PeerK (2024-01-27 12:25:49)
Offline
seth wrote:Did you see the link and attempt to
sudo touch /etc/systemd/do-not-udevadm-trigger-on-update
?
I cannot reproduce the crash calling `/usr/share/libalpm/scripts/systemd-hook udev-reload` but it's still worth a try. I'll report back but it may take weeks or even months to determine if this fixes the issue or not.
Same here:
sudo /usr/share/libalpm/scripts/systemd-hook udev-reload
does not crash FF.
Kind regards,
Peer
Offline
https://wiki.archlinux.org/title/Core_d … _core_dump
You could still just put the file there and see whether FF still crashes on updates - it's a race condition and by nature not guaranteed to be triggered by a synthetic reload.
Offline
I'm having the same issue. It's crashing not only on updates, but also when a package is removed. For example, today I uninstalled some packages, and Firefox crashed. This time the only hook that was executed was the update-mime-database one, so I think it's safe to say that this is the culprit. As far as I can tell, the hook is only triggered when the contents of the mime database actually change, so presumably Firefox is crashing due to the cache changing. Manually running the hook without the changes to the mime database does not trigger a crash.
Offline
Do you get a coredump?
If this is half-wise reliable, what if you remove the PKGSYSTEM_ENABLE_FSYNC=0 environment from the hook?
(This will most likely slow the process down considerably, though)
Offline
Do you get a coredump?
It's doesn't appear in
coredumpctl list
, so I guess not.
It's not 100% reproducible. I couldn't reproduce it just now. But I have to say that I rarely turn off my computer, I tend to use suspend-to-disk or suspend-to-ram, and only do a full reboot after a system upgrade (usually once every 1-3 months). I'll try to reproduce after waking from suspend later, in case that has something to do with it.
Offline
If the update touches something firefox needs, then it is normal it complains, happens here too, but never bothered.
Doesn't just restart the browser after the update finished work?
Help me to improve ssh-rdp !
Retroarch User? Try my koko-aio shader !
Online
The package I removed was completely unrelated to Firefox (it was openmpi). And yes, one can restart Firefox after the update, but it is very annoying to have it crash when installing/removing unrelated packages.
Offline
openmpi doesn't seem to have any mime entries, though?
Offline
True, looks like I was mistaken, it's the update-desktop-database.hook, which gets triggered by hwloc (a dependency of openmpi). I was confused because the command line output reads "Updating the desktop file MIME type cache...". This is the pacman.log:
[2024-03-13T15:45:31+0100] [PACMAN] Running 'pacman -R -s -c -n --config /etc/pacman.conf -- openmpi'
[2024-03-13T15:45:32+0100] [ALPM] transaction started
[2024-03-13T15:45:32+0100] [ALPM] removed openmpi (5.0.2-6)
[2024-03-13T15:45:32+0100] [ALPM] removed prrte (3.0.4-3)
[2024-03-13T15:45:32+0100] [ALPM] removed openucx (1.15.0-2)
[2024-03-13T15:45:32+0100] [ALPM] removed openpmix (4.2.9-1)
[2024-03-13T15:45:32+0100] [ALPM] removed libfabric (1.20.0-1)
[2024-03-13T15:45:32+0100] [ALPM] removed hwloc (2.10.0-1)
[2024-03-13T15:45:33+0100] [ALPM] transaction completed
[2024-03-13T15:45:33+0100] [ALPM] running '30-systemd-update.hook'...
[2024-03-13T15:45:33+0100] [ALPM] running 'update-desktop-database.hook'...
I could not reproduce the crash right after waking from suspend, I will observe further.
Offline
Did you already https://bbs.archlinux.org/viewtopic.php … 9#p2145689 - because that's notorious for crashing stuff™ during the update and what you see in the terminal when your FF window disappears is only mildly indicative of the cause.
Offline
Not yet. But that relates to reloading udev rules, which is done by 30-systemd-udev-reload.hook. As far as I can tell, the only thing that 30-systemd-update.hook does is execute touch -c /usr.
Offline
Yes, I simply don't put any trust the log to be complete there in some form of heisencrash.
You could try to use "--verbose" instead of "--quiet" in /usr/share/libalpm/hooks/update-desktop-database.hook and see whether that narrows down the trigger.
Offline
...
for a some months my firefox instances crash, when I update my computers (pacman -Syyu, laptop & pc).
...
1. At a minimum, when updating, I would not have any GUI open. If you want to go to the next level, only do an update on a tty.
2. Check the system log.
# Look for when you began the install and then firefox.
journalctl -b
# To narrow it down a bit:
journalctl --since '2024-03-18 18:00:00' --until '2024-03-18 19:30:00'
3. Look for posts in this forum that mention **do-not-udevadm-trigger-on-update**. It resides in /usr/share/libalpm/scripts/systemd-hook. Perhaps there was something updated or some action occurred and udev took action that affected systemd, all bets are off --- end result, terminate applications.
On a side note, I always reboot after an upgrade because system files were updated and it just takes a minute. Just prior to the reboot I do some of the following:
- Review the pacman output. I capture the output of the command
- sudo pacman -Fy
- sudo DIFFPROG=meld pacdiff
Last edited by mountaintrek (2024-03-18 23:35:59)
Offline
Hi all,
I don't know, if the bug was fixed or if a dependency, that might have caused the error, has been removed.
But as far as I remember, I do not observe any FF-crashes during update processes for many weeks now.
So I am wondering to close the thread. I will wait for a week or so. Maybe someone knows more and could possible write a line, if it was fixed.
Kind regards,
Peer
Offline
Did you ever add /etc/systemd/do-not-udevadm-trigger-on-update ?
Offline