witness_checkorder panic

Kostik Belousov kostikbel at gmail.com
Mon Dec 4 00:29:54 PST 2006


On Sun, Dec 03, 2006 at 09:12:08PM +0100, Arno J. Klaassen wrote:
> Hello,
> 
> > On Sat, Dec 02, 2006 at 07:15:37PM +0100, Arno J. Klaassen wrote:
> > > 
> > > Hello,
> > > 
> > > I just got this on a box I'm testing before installation.
> > > It has clean RELENG_6 from about two weeks ago with only
> > > some small if_bge.c-patches Bruce Evans sent me for testing
> > > performance/hang problems.
> > > Since I doubt this panic is related to that, I just post
> > > it here in case someone is interested in more info :
> > > 
> > > [sorry, no serial console attached ... just copy-paste from
> > > screen, but I will leave the box in the debugger for the
> > > WE ]
> > > 
> > >   struct mount mtx (struct mount mtx) @ /files/bsd/src6/sys/ufs/ufs/ufs_vnops.c:138
> > >   KDB: stack backtrace :
> > >   witness_checkorder()
> > >   _mtx_lock_flags()
> > >   ufs_itimes()
> > >   ufs_getattr()
> > >   VOP_GETATTR_APV()
> > >   filt_vfsread()
> > >   knote()
> > >   VOP_WRITE_APV()
> > >   vn_write()
> > >   dofilewrite()
> > >   kern_writev()
> > >   write()
> > >   syscall()
> > >   Xfast_syscall()
> > >   --- syscall (4, FreeBSD ELF64, write), rip = 0x4363dc, rsp = 0X7fffffffdd78, rbp = 0x2f6 ---
> > >   KDB: enter: witness_checkorder
> > >   [thread pid 3987 tid 100133 ]
> > > 
> > > Kernel config is stripped GENERIC +
> > > 
> > >   options AHC_ALLOW_MEMIO
> > >   options TCP_DROP_SYNFIN
> > >   options         KDB
> > >   options         KDB_TRACE
> > >   options         DDB
> > >   options         KTRACE
> > >   options         INVARIANTS
> > >   options         INVARIANT_SUPPORT
> > >   options         DDB_NUMSYM
> > >   options         BREAK_TO_DEBUGGER
> > >   options         INVARIANTS
> > >   options         INVARIANT_SUPPORT
> > >   options         WITNESS
> > >   options         WITNESS_KDB
> > >   options         DEBUG_LOCKS
> > >   options         DEBUG_VFS_LOCKS
> > >   options         DIAGNOSTIC
> > >   options         MUTEX_PROFILING
> > >   options         MUTEX_DEBUG
> > >   options         SLEEPQUEUE_PROFILING
> > >   options         TURNSTILE_PROFILING
> > >   options         DEBUG_MEMGUARD
> > > 
> > > The box was doing (/usr/src nfs-mounted):
> > > 
> > >   nohup time make -j 2 -DNO_CLEAN buildworld > /tmp/bw_alone.log 2>&1 &
> > > 
> > > It paniced shortly after I started 'tail -f /tmp/bw_alone.log' in another
> > > window, and /tmp is mfs.
> > 
> > Please, try the patch at
> > http://people.freebsd.org/~kib/kqueue-lor.1.patch
> > 
> 
> for now this prevents the LOR-panic. Thank you very much!

Thank you for testing. Do you have a procedure to reliably reproduce the
LOR ?

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 187 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-stable/attachments/20061204/7890654e/attachment.pgp


More information about the freebsd-stable mailing list