Bug #579

panic at reboot - assertion: vp->v_lock.lk_exclusivecount == 1 in vgone_ interlocked

Added by thomas.nikolajsen about 7 years ago. Updated over 6 years ago.

Status:ClosedStart date:
Priority:HighDue date:
Assignee:-% Done:

0%

Category:-
Target version:-

Description

FYI: I still see this panic on HEAD (today);
it seem it is only when using vn(4); panic not seen on every update.

New core dump avail. on request.

-thomas
<http://leaf.dragonflybsd.org/mailarchive/bugs/2007-01/msg00214.html>

History

#1 Updated by dillon about 7 years ago

:
:FYI: I still see this panic on HEAD (today);
:it seem it is only when using vn(4); panic not seen on every update.
:
:New core dump avail. on request.
:
: -thomas
:<http://leaf.dragonflybsd.org/mailarchive/bugs/2007-01/msg00214.html>

Could you give me the exact vnconfig lines you are using to configure
your VN? Including any partitioning and filesystem mounts you are doing.

Are you unmounting the filesystem using the VN device before rebooting?
Are you vnconfig -u'ing (turning off) the VN device before rebooting?

-Matt
Matthew Dillon
<>

#2 Updated by thomas.nikolajsen about 7 years ago

Commands used (script): inspired by vkernel(7):

vnconfig -cs labels vn0 FILE
fsck -y /dev/vn0a
mount /dev/vn0a /mnt
make installworld DESTDIR=/mnt
umount /mnt
vnconfig -u vn0
reboot

FILE on UFS.
It might seem like panic occurs if I fiddle with /mnt while vn0a mounted: df; df /mnt

-thomas

#3 Updated by thomas.nikolajsen about 7 years ago

Btw: I also saw this panic once without using vn(4):
using 1.8.0 SMP kernel om MP (HTT) machine.

Was upgrading from 1.8.0 to 1.8.1:
Did build/install world & kernel using nfs mounted /usr/src & /usr/obj;
at reboot it panic'ed, but core dump wasn't saved to file at next boot.

-thomas

..

#4 Updated by corecode almost 7 years ago

does this still happen on -DEVEL?

#5 Updated by thomas.nikolajsen almost 7 years ago

I haven't seen this problem since June 6th, so it may be resolved.
I normally update to HEAD a few times a week.

I did set up a stress test yesterday to reproduce this problem:
continously updating(*) system for 24h: no problem.

* cvs update; build&install world&kernel (mix of quick&full); reboot

-thomas

#6 Updated by thomas.nikolajsen over 6 years ago

Problem haven't shown up in several months.

-thomas

Also available in: Atom PDF