virtuoso-t constantly segfaulting

Duncan 1i5t5.duncan at cox.net
Sun Jan 26 23:33:42 GMT 2014


Kevin Krammer posted on Sun, 26 Jan 2014 23:48:02 +0100 as excerpted:

> On Sunday, 2014-01-26, 23:42:12, Thomas Tanghus wrote:
>> The past week or so I get this every 10-20 seconds in/var/log/syslog:
>> 
>> kernel: [168098.800123] virtuoso-t[13981]: segfault at ffffffff ip
>> 0851efbb sp a7143870 error 7 in virtuoso-t[8048000+aa0000]
>> 
>> I get several hits when searching for it, but no solution to how to fix
>> it, and it makes the log almost unusable.
>> 
>> Any idea what is going on?
>> 
>> I run KDE Platform 4.12 on Kubuntu 13.10.
> 
> Since this is Kubuntu it could be AppArmor interfering, e.g. not
> allowing virtuoso to access its files or something.

Good point.

> It is kind of strange though that a user process would log to syslog.

Note that it's actually the kernel doing the logging to syslog, logging 
the segfault of the app in question!

Barring a simple fix for virtoso, my solution to the syslog noise problem 
would be a syslog filter to filter out that noise.  I've done that a 
number of times here.

What (k)ubuntu uses for syslog I have no idea.  On (my distro of choice) 
gentoo there are several options and the handbook actually gives you the 
choice of which one you want to install.  I picked syslog-ng years ago 
when I first installed, and while it has gotten more complex and advanced 
since then such that I'm now running something rather more complex than 
necessary for the job I give it to do, it's actually quite easy to deal 
with something like this on syslog-ng by just adding a filter to the 
syslog config. =:^)

Since it's likely that kubuntu uses a logger with a quite different 
configuration I'll omit the explanation here, but if the OP does happen 
to be using syslog-ng, just ask, and I can post my config and a 
"short"[1] walk-thru of what it does and why I configured it that way.


As for trying to fix virtuoso... I should first mention as a caveat that 
I took advantage of gentoo's USE flags to set USE=-semantic-desktop and 
related flags for kde some time ago (early kde 4.7 era, so 2.5+ years 
ago), and after rebuilding kde, entirely purged all the related semantic-
desktop junk from my system, including virtuoso, nepomuk, etc.  So I've 
not run it or even had it installed in some years.

That said, the possibility that immediately comes to mind here is that 
its database is corrupted perhaps by an unclean shutdown aka system crash 
just as virtuoso was writing something, so that now, when it (apparently 
repeatedly) tries to load, it hits that corrupt database and segfaults.

If that's correct, the fix should in theory be pretty simple: Find that 
database and delete it, allowing nepomuk/virtuoso to rebuild it as 
necessary (or turning it off at runtime at least, if you don't use it, so 
it doesn't rebuild the database).

The problem is finding and deleting that database.  Since I'm not running 
it here and long ago deleted it, I can't tell you for sure where to 
look.  Perhaps Kevin can help with that.

---
[1] Short:  Caveat, just so readers know and as I'm sure Kevin can 
attest, my "short" could well end up being several hundred lines of 
example and instructions! =:^)

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman

___________________________________________________
This message is from the kde mailing list.
Account management:  https://mail.kde.org/mailman/listinfo/kde.
Archives: http://lists.kde.org/.
More info: http://www.kde.org/faq.html.




More information about the kde mailing list