Re: panic: ffs_valloc: dup alloc

["Steven M. Bellovin" <smb%cs.columbia.edu@localhost>]

On Sun, 18 Jan 2009 21:44:53 +0100
Lars Nordlund <lars.nordlund%hem.utfors.se@localhost> wrote:

> On Sun, 18 Jan 2009 16:37:20 +0100
> Tobias Nygren <tnn%NetBSD.org@localhost> wrote:
> 
> > On Sat, 10 Jan 2009 12:20:12 +0100
> > Lars Nordlund <lars.nordlund%hem.utfors.se@localhost> wrote:
> > 
> > > Hi
> > > 
> > > Hand-copied from screen:
> > > 
> > > dmode 8000 mode 8000 dgen 504d64d6 gen 504d64d6
> > > size 0 blocks 0
> > > ino 3 ipref 8
> > > panic: ffs_valloc: dup alloc
> > > fatal breakpoint trap in supervisor mode
> > > trap type 1 code 8 eip c02369c4 cs eflags 246 cr2 bbb99660 ilevel
> > > 8 Stopped in pid 5416.1 (mkdir) at   netbsd:breakpoint+0x4: popl
> > > %ebp
> > > 
> > > This happened on NetBSD-current from 2009-01-08 on my i386 UP
> > > machine while building some stuff from pkgsrc.
> > > 
> > > For some reason NetBSD can not boot with a PS2 keyboard attached,
> > > so I can not do anything at the moment at the ddb-prompt. I
> > > suppose crashdump is the only way forward.
> > > 
> > > Anyone else seeing this?
> > 
> > I saw this consistently today while trying to untar files on a newly
> > created ffs -O2 filesystem after booting a current NetBSD/alpha
> > sysinst kernel. Note I am *not* using WAPBL. The fix for me was to
> > mount the filesystem with -o sync while untarring. I've not been
> > able to reproduce the panic so far after rebooting into the newly
> > installed system, although I did see signs of fs corruption (a
> > corrupt directory). Will try to krank up debugging options.
> 
> I never got the previous installation to fly. I had to reinstall.
> 
> I had created separate filesystems (/, /var, /usr and /home) and /usr
> was constantly giving me problems. Trying to access some files
> under /usr/share would sometimes reboot the machine. The panic message
> in the core dump had a message 'bad dir' (or something).
> 
> In single-user mode, I copied away everything from /usr and newfs:ed
> (FFSv2), mounted (tried with and without WAPBL), and copied files
> back. Then I could not umount! The umount command would just freeze..
> When looking in ddb I could not say if something was really broken or
> just very inefficient/slow. I let it run a long time, but it never
> finished the umount. Hard reset was necessary.
> 
> One theory I had was that something previously written on the hard
> drive confused the kernel in some way.. I tried zeroing the partition
> before newfs:ing, but that did not help. The same, hanging umount kept
> coming back like a bad breath.
> 
> So I thought.. Everyone is using OneBigRoot(TM) these days.. Lets try
> that. I had done very little on the problematic installation anyway. I
> reinstalled, did the same as before, cvs checkout pkgsrc, start some
> builds, configure X, get things configured and running .. and
> everything seems to work fine now. No panics.
> 
> At the moment I am a bit short on spare computers, so I can not test
> installations for some time.
> 
While I'm not seeing exactly that problem, I do see problems that
result in duplicate blocks after a crash.  I think it's timing- and
disk-layout dependent, but I do think there's a serious file system
problem in -current.  I'm seriously contemplating wiping my disk and
installing a system from a couple of months ago.


                --Steve Bellovin, http://www.cs.columbia.edu/~smb