Re: stacked unionfs freeze and crash FreeBSD

On Wednesday 20 January 2010 18:05:54 David Naylor wrote:
> On Wednesday 20 January 2010 17:28:21 Gary Jennejohn wrote:
> > On Wed, 20 Jan 2010 15:43:12 +0200 David Naylor wrote:
> > > Hi,
> > >
> > > The attached script, that uses stacked unionfs, causes FreeBSD-9 (cvsup
> > > yesterday) to freeze and FreeBSD-8 (cvsup two days ago) to crash:
> > >
> > > Fatal double fault
> > > rip-0xffffffff81e4c1
> > > rsp=0xffffff80b133ef50
> > > rbp=0xffffff80b133f150
> > > cupid = 2; apic id = 02
> > > panic: double fault
> > > cpuid = 2
> > > uptime: 1h44m35s
> > > cannot dump.  Device not defined or unavailable
> > >
> > >
> > > Both systems use the stock GENERIC kernel, i.e. -9 had full diagnostics
> > > built in (and was run under VirtualBox) and -8 had no diagnostics (and
> > > was run native).
> > >
> > > A LOR is produces prior to freezing under -9 (quiet a time prior).  See
> > > kern/141950.
> > >
> > > The script uses unionfs to build ports (in an attempt to create a
> > > tinderbox without the need to delete and/or extract packages).  To use
> > > the script to:
> > >
> > > # mkdir /tmp/localbase /tmp/builddir
> > > # env LOCALBASE=/tmp/localbase BUILDDIR=/tmp/builddir
> > > ./ports-union-builder.sh
> >
> > Is your /tmp big enough?
> 
> I was actually using /home, didn't think people wanted to contaminate that
> directory.  Plenty of space.
> 
> > > This will try build everything for x11/xorg.
> > >
> > > Is it possible that VirtualBox is interfering in getting usable
> > > diagnostics for -9
> >
> > Who knows?  You might try posting your shell script so others can give
> > it a whirl on a "real" 9-current system.
> 
> I did attach the script.  Guess mailer ate it, again.  I'll try again...
> 
> > > and how can I setup a dump device for -8.  Currently I have:
> > >
> > > # swapinfo
> > > Device          1K-blocks     Used    Avail Capacity
> > > /dev/ad4s1b       8388608        0  8388608     0%
> > > /dev/ad8s1b       8388608        0  8388608     0%
> > > Total            16777216        0 16777216     0%
> >
> > Do you have dumpdev defined in /etc/rc.conf and/or do you see
> > /dev/dumpdev?  Having /dev/dumpdev indicates that dumpon succeeded.
> 
> No I haven't and no I don't.  I thought it automatically used swap?
> 
> I'll 'activate' it and post the results.

Here it is.  If anyone ones the full core.txt just shout.  

#0  doadump () at pcpu.h:223
223     pcpu.h: No such file or directory.
        in pcpu.h
(kgdb) #0  doadump () at pcpu.h:223
#1  0xffffffff80589ba9 in boot (howto=260)
    at /usr/src/sys/kern/kern_shutdown.c:416
#2  0xffffffff80589fdc in panic (fmt=0xffffffff8098bd81 "double fault")
    at /usr/src/sys/kern/kern_shutdown.c:579
#3  0xffffffff8086e546 in dblfault_handler (frame=Variable "frame" is not 
available.)
    at /usr/src/sys/amd64/amd64/trap.c:884
#4  0xffffffff808557fc in Xdblfault ()
    at /usr/src/sys/amd64/amd64/exception.S:278
#5  0xffffffff81e464c1 in unionfs_statfs (mp=Variable "mp" is not available.)
    at /usr/src/sys/modules/unionfs/../../fs/unionfs/union_vfsops.c:428
Previous frame inner to this frame (corrupt stack?)
(kgdb)