System is opensuse, kernel 4.12.5
Recently I receive error messages by invoking journalctl -k
Nov 26 16:57:43 linux-zbez.bw-host.de kernel: multiqueue1:src: segfault at 7fcb240b5708 ip 00007fcb678173f9 sp 00007fcb3c800d50 error 6 in libc-2.25.so[7fcb6779b000+1
Nov 26 16:57:43 linux-zbez.bw-host.de kernel: multiqueue1:src: segfault at 7f3e680b5708 ip 00007f3eaec5c3f9 sp 00007f3e76dbad50 error 6 in libc-2.25.so[7f3eaebe0000+1
Nov 26 16:57:44 linux-zbez.bw-host.de kernel: multiqueue1:src: segfault at 7f433c0b5708 ip 00007f43814c03f9 sp 00007f4351757d50 error 6 in libc-2.25.so[7f4381444000+1
The glibc rpm package was not touched from the beginning and is verified to be uncompromised.
Whole system is in a very good condition.
I was able to track down the program being responsible for the error message, this is
"/usr/bin/dbus-daemon --session --address=systemd: --nofork --nopidfile --systemd-activation --syslog-only"
Killing the dbus-daemon process the error messages go away (of course X window system is unusable then).
Installation of dbus-daemon is 1.12.2, without any modification from the rpm source package.
some additional information:
glibc is version 2.25 but I will not touch this... This is a production machine.
some more information...
well, did some more investigations.
Seems to be that there is some connection to configuration files (in home
A newly created standard user with default home directory does not cause this error to happen. Well, I created a clean .config and .local and .bashrc and .profile for the affected user, however, this did not help.
What else can I do?
The easiest way to debug a crash like this is to get a backtrace. Since you’re using systemd, you can use the incredibly helpful systemd-coredump to get one.
See https://www.suse.com/support/kb/doc/?id=7017137 for how to enable and use coredumpd.
(In reply to Martin Stenzel from comment #0)
> I was able to track down the program being responsible for the error
> message, this is
> "/usr/bin/dbus-daemon --session --address=systemd: --nofork --nopidfile
> --systemd-activation --syslog-only"
I don't think the dbus-daemon is responsible for this. The kernel messages indicate that the segfaults happen in threads named "multiqueue1:src", but the dbus-daemon doesn't have a thread with that name. Also, your crash messages came from multiple different process IDs within a very short space of time. If the dbus-daemon was segfaulting, every time it did, it would have a larger effect on your system, equivalent to what happens when you kill it yourself.
As you pointed out, killing the dbus-daemon stops the entire X11 session. I think the program that is crashing is part of the X11 session, so stopping the session has the side-effect of stopping the program that is restarting the thread that repeatedly crashes with a segfault.
Philip's suggestion to use coredumpd is an excellent one. A thread named multiqueue1:src sounds like it might be related to GStreamer, so I think you'll probably find that it's some process that uses GStreamer for multimedia (music and video players, web browsers, voice-over-IP clients, that sort of thing).
If you have evidence that dbus-daemon is responsible for this crash, please quote it here. Otherwise, this is not a dbus bug - good luck tracking it down, but we can't help you further.
thank you all for your support.
After some more working with the system, including some software updates
but no further problem hunting the error message disappeared from the system logs (with rsyslog running...).
Unfortunately I cannot tell you what changed matters.
Stays somehow nebulous.