You are not logged in.
Hi,
I am running arch on my vserver and installed yaourt for AUR Package installation. Running a system update using # yaourt -Syua got me a segfault on an AUR package update. I wanted to reinstall the yaourt and package-query packages by hand and even get segfaults trying that. Normal pacman -Syu updates work just fine.
The error I get building the package using makepkg is:
/usr/bin/fakeroot: line 181: 18123 Segmentation fault (core dumped) FAKEROOTKEY=$FAKEROOTKEY LD_LIBRARY_PATH="$PATHS" LD_PRELOAD="$LIB" "$@"
$ echo $?
139
A try without the usage of fakeroot (# makepkg --asroot):
Segmentation fault (core dumped)
# echo $?
139
Do you have any ideas what might causes this issue or what I can try to fix it? Where can I get the core dump to investigate the issue further?
Best regards and thanks for your help,
Watnuss
Offline
makepkg is just a bash script (which I wouldn't expect to segfault); try running it with `bash -vv /usr/bin/makepkg` to see what binary is segfaulting.
Offline
Check the health of your disk and filesystem, then reinstall fakeroot (assuming they're healthy).
As a sidenote: you don't need yaourt to install packages from the AUR, 'pacman -U' will install packages whether they're official or 'homemade'.
Sakura:-
Mobo: MSI MAG X570S TORPEDO MAX // Processor: AMD Ryzen 9 5950X @4.9GHz // GFX: AMD Radeon RX 5700 XT // RAM: 32GB (4x 8GB) Corsair DDR4 (@ 3000MHz) // Storage: 1x 3TB HDD, 6x 1TB SSD, 2x 120GB SSD, 1x 275GB M2 SSD
Making lemonade from lemons since 2015.
Offline
Check the health of your disk and filesystem, then reinstall fakeroot (assuming they're healthy).
The op states it fails without fakeroot too; and the reason fakeroot is in that message is probably because of what fakeroot actually executes otoh; it won't make things worse
Last edited by Spider.007 (2014-01-17 18:04:58)
Offline
Not a Sysadmin issue, moving to AUR Issues...
Offline
Hi,
I know, that I don't need to use yaourt but nevertheless I cannot install packages manually as 'makepkg' fails by itself. That was the reason why yaourt installations didn't work in the first place.
I tried to get any information by calling makepkg with bash -vv but the logfiles don't include any further information. (http://pastebin.com/3qe9ejwB if you wanna check by yourself. I called it with '$ bash -vv /usr/bin/makepkg >& logfile)
(see http://pastebin.com/s49YxRDx for calling with '# bash -vv /usr/bin/makepkg --asroot >& logfile'. The segfault message wasn't included to the log but given to the command line even though I used >& for redirection.)
Do you have any further ideas?
Best regards and thanks so far.
Watnuss
Last edited by Watnuss (2014-01-18 08:55:28)
Offline
From the log it seems du is responsible for the segfault; could you try `pacman -S coreutils` to make sure it's not corrupted? Also; by default the segfault should also end up in the systemd-journal; have a look at systemd-coredumpctl (as root) and let us know if any segfaults are logged there
Offline
Hi,
reinstalling coreutils didn't solve the issue. Makepkg still fails with the same error.
You can find the output of coredumpctl below (calls with uid 0 are makepkg --asroot and calls with uid 1000 are normal makepkg calls. Output with the parameter dump results in a blob. How can I get further information out of coredumpctl (didn't got the information from the manpages)
# systemd-coredumpctl
TIME PID UID GID SIG EXE
Tue 2014-01-07 19:52:57 CET 13537 0 0 7 /usr/bin/postconf
Tue 2014-01-07 19:53:13 CET 13714 0 0 11 /usr/bin/bash
Tue 2014-01-07 19:53:30 CET 14134 0 0 11 /usr/bin/bash
Tue 2014-01-07 19:56:37 CET 14856 0 0 11 /usr/bin/bash
Wed 2014-01-08 11:05:59 CET 551 0 0 11 /usr/bin/bash
Thu 2014-01-09 17:59:42 CET 17576 0 0 11 /usr/bin/bash
Thu 2014-01-09 18:00:14 CET 18052 0 0 11 /usr/bin/bash
Sat 2014-01-11 17:17:13 CET 5473 0 0 11 /usr/bin/bash
Fri 2014-01-17 15:50:15 CET 15003 1000 1000 11 /usr/bin/bash
Fri 2014-01-17 15:54:21 CET 24771 1000 1000 11 /usr/bin/bash
Fri 2014-01-17 15:59:18 CET 1238 0 0 11 /usr/bin/bash
Fri 2014-01-17 15:59:38 CET 4158 0 0 11 /usr/bin/bash
Fri 2014-01-17 16:03:37 CET 7847 1000 1000 11 /usr/bin/bash
Fri 2014-01-17 16:05:56 CET 11969 0 0 11 /usr/bin/bash
Fri 2014-01-17 16:06:41 CET 18123 1000 1000 11 /usr/bin/bash
Sat 2014-01-18 09:31:56 CET 32561 1000 1000 11 /usr/bin/bash
Sat 2014-01-18 09:32:21 CET 3217 1000 1000 11 /usr/bin/bash
Sat 2014-01-18 09:32:45 CET 6339 1000 1000 11 /usr/bin/bash
Sat 2014-01-18 09:33:54 CET 10745 1000 1000 11 /usr/bin/bash
Sat 2014-01-18 09:34:11 CET 13603 1000 1000 11 /usr/bin/bash
Sat 2014-01-18 09:35:22 CET 18052 1000 1000 11 /usr/bin/bash
Sat 2014-01-18 09:35:30 CET 3625 1000 1000 11 /usr/bin/bash
Sat 2014-01-18 09:36:38 CET 22721 1000 1000 11 /usr/bin/bash
Sat 2014-01-18 09:36:52 CET 25511 1000 1000 11 /usr/bin/bash
Sat 2014-01-18 09:38:48 CET 28919 0 0 11 /usr/bin/bash
Sat 2014-01-18 09:39:03 CET 31675 0 0 11 /usr/bin/bash
Sat 2014-01-18 09:39:58 CET 3213 0 0 11 /usr/bin/bash
Sat 2014-01-18 09:41:49 CET 11315 1000 1000 11 /usr/bin/bash
Sat 2014-01-18 09:43:45 CET 14733 0 0 11 /usr/bin/bash
Sun 2014-01-19 09:21:27 CET 18613 1000 1000 11 /usr/bin/bash
Sun 2014-01-19 09:22:23 CET 20312 0 0 11 /usr/bin/bash
Thanks for your help spider.
Watnuss
Offline
This is very interesting; it seems bash is the one segfaulting. You can try `systemd-coredumpctl gdb 13714`; or use `systemd-coredumpctl dump 13714` to dump the coredump to disk; then upload it somewhere for someone else to look at (maybe I can tell what is happening from the output)
Offline
Calling 'systemd-coredumpctl gdb 13714' gives me an error:
# systemd-coredumpctl gdb 13714
TIME PID UID GID SIG EXE
Tue 2014-01-07 19:53:13 CET 13714 0 0 11 /usr/bin/bash
Failed to invoke gdb: No such file or directory
# echo $?
1
Writing the dump out to disk (using the -o parameter) gives the following file: http://watnuss.de/public/bash_segfault.dump
Offline
If noone has any further ideas I would appreciate any hints on what I could do and where I could ask else to resolve this issue. I don't really feel like reinstalling the system from scratch.
Offline
To be honest, I don't know what to do with dump files, and Spider seemed to have a better grasp of the situation than me, but I loaded it up with gdb and it gave the following output:
Core was generated by `/usr/bin/bash /usr/bin/makepkg --asroot -s -f -p ./PKGBUILD'.
Program terminated with signal 11, Segmentation fault.
#0 0x00007fadf95a72c8 in __collidx_table_lookup () from /usr/lib/libc.so.6
So I'd try reinstalling glibc.
Sakura:-
Mobo: MSI MAG X570S TORPEDO MAX // Processor: AMD Ryzen 9 5950X @4.9GHz // GFX: AMD Radeon RX 5700 XT // RAM: 32GB (4x 8GB) Corsair DDR4 (@ 3000MHz) // Storage: 1x 3TB HDD, 6x 1TB SSD, 2x 120GB SSD, 1x 275GB M2 SSD
Making lemonade from lemons since 2015.
Offline
Thanks for your answer WorMzy. Sadly reinstalling glibc didn't change anything.
Offline
Well I found the same thing as WorMzy noted, glibc throwing strange errors. (it seems the collidx function does nothing special) Have you ever tried memtest86 on your machine? You could also have a look at mcelog to see if your processor has anything useful to tell.
Offline
Hi,
I was running memtest without any errors on a whole run using all test.
I installed and enabled the mcelog daemon. I checked outputs with 'mcelog --client' which didn't gave me any outputs. I hope that was the correct way to check MCEs.
Maybe it's time to forget about finding that rare error and reinstall the system. What do you think? (Don't want to waste your time any further.)
Best regards and thanks for your help.
Last edited by Watnuss (2014-01-27 14:27:47)
Offline
Well; you're not wasting any time, it is an intriguing problem; but it's hard to debug.
For future reference; here is the backtrace from gdb:
#0 0x00007fadf95a72c8 in __collidx_table_lookup () from /usr/lib/libc.so.6
#1 0x00007fadf961d245 in get_next_seq () from /usr/lib/libc.so.6
#2 0x00007fadf961def7 in wcscoll_l () from /usr/lib/libc.so.6
#3 0x000000000047aa38 in ?? ()
#4 0x000000000043e77d in ?? ()
#5 0x00000000004486a4 in ?? ()
#6 0x00000000004496ef in ?? ()
#7 0x000000000044ac9c in ?? ()
#8 0x000000000044b1fe in ?? ()
#9 0x0000000000446035 in expand_string ()
#10 0x0000000000444f7e in ?? ()
#11 0x0000000000447c4b in ?? ()
#12 0x00000000004496ef in ?? ()
#13 0x000000000044a3c2 in ?? ()
#14 0x000000000044ac9c in ?? ()
#15 0x000000000044b0fa in expand_string_assignment ()
#16 0x0000000000444f7e in ?? ()
#17 0x000000000044536f in ?? ()
#18 0x000000000044bd26 in ?? ()
#19 0x000000000042b0c0 in ?? ()
#20 0x000000000042cbaa in execute_command_internal ()
#21 0x000000000042c9e9 in execute_command_internal ()
#22 0x000000000042e12e in execute_command ()
.....
#138 0x000000000042e12e in execute_command ()
#139 0x00000000004192b3 in reader_loop ()
#140 0x000000000041793e in main ()
Offline
My machine (VM) also segfaults on bash. I couldnt even login. I had to change default shell to zsh.
Offline
It also happens for me, I'm not sure why, but it seems to be connected to the locale. My default locale is de_DE.utf8, when I switch to en_US.utf8 or C, i.e. "LANG=C makepkg", it works.1
Offline
Thanks tobias_,
[edit: double thanks actually as I was just making a backup to reinstall the server. This won't be necessary now I guess. So really HUGE thanks as you save me a lot of time.]
I was able to build the packages setting the LANG variable to 'C'. Setting it to 'en_US.UTF-8', which I originally had, still results in the segfault above.
Can anyone explain this behaviour?
Last edited by Watnuss (2014-02-08 14:00:40)
Offline
This is probably a bug. You should report this on the Arch bugtracker.
Offline
I found the source of the issue: I did not generate locales with locale-gen for de-DE.UTF-8
I'm not sure if segfaulting applications is a wanted or buggy behaviour then…
Offline
Someone did find a similar locale bug with pacman/makepkg a while back. I'll see if I can find the related topic.
I'll also say the archivist's name three times into a mirror in three minutes time (midnight here) and see if he appears with the link.
EDIT: Er, I think it worked.. https://bbs.archlinux.org/viewtopic.php?id=173500
Last edited by WorMzy (2014-02-09 00:01:08)
Sakura:-
Mobo: MSI MAG X570S TORPEDO MAX // Processor: AMD Ryzen 9 5950X @4.9GHz // GFX: AMD Radeon RX 5700 XT // RAM: 32GB (4x 8GB) Corsair DDR4 (@ 3000MHz) // Storage: 1x 3TB HDD, 6x 1TB SSD, 2x 120GB SSD, 1x 275GB M2 SSD
Making lemonade from lemons since 2015.
Offline