Processes hung in 'inode' state ...

[David Schultz]

On Mon, Jun 23, 2003, Marc G. Fournier wrote:
> 
> Not sure what all to do, but doing a 'gdb -k kernel.debug /dev/mem', a
> backtrack on one of the processes shows ... server has been up 12 days
> now, and running a June 8th kernel ...
> 
> #0  0x20f4f0 in ?? ()
> (kgdb) proc 67258
> (kgdb) bt
> #0  mi_switch () at machine/globals.h:119
> #1  0x8014a1f9 in tsleep (ident=0x8a3d2200, priority=8, wmesg=0x80263d4a "inode", timo=0) at /usr/src/sys/kern/kern_synch.c:479
> #2  0x80141507 in acquire (lkp=0x8a3d2200, extflags=16777280, wanted=1536) at /usr/src/sys/kern/kern_lock.c:147
> #3  0x8014179c in lockmgr (lkp=0x8a3d2200, flags=16973826, interlkp=0xbdc27b6c, p=0xc26fa9c0) at /usr/src/sys/kern/kern_lock.c:355

Ooh, a deadlock.  I'm guessing unionfs is responsible again.  ;-)
Can you type in the following?

	print *(struct lock *)0x8a3d2200

Then get a backtrace of the process with pid equal to lk_lockholder.