Hang when shutting down - seems umount and NFS related - kernel version 2.6.18.1

[Jesper Juhl]

I have a server that has the sole job of serving up a bunch of NFS
filesystems to webservers. The server is running kernel 2.6.18.1

A few days ago when attempting to reboot the server, the shutdown
process hung with a backtrace on the console. A photo was taken and is
available here:
 http://www.kernel.org/pub/linux/kernel/people/juhl/images/DSC00504.jpg

A quick (incomplete) text transcript of the photo is here:

EIP: [<c01a51a0>] nfs_clear_inode+0x61/0x6b SS:ESP 0068:eb476ee4
BUG: warning at kernel/exit.c:855/do_exit()
do_exit			do_trap
do_trap			do_invalid_op
do_invalid_op			nfs_clear_inode
scheduler_tick			radix_tree_gang_lookup_tag
nfs_wait_on_reques_locked	error_code
nfs3_forget_cached_acls	nfs_clear_inode
clear_inode			dispose_list
invalidate_inodes		generic_shutdown_super
kill_anon_super		nfs_kill_super
deactivate_super		sys_umount
sys_readlink			syscall_call
BUG: umount/20593, lock held at task exit time!
 {alloc_super}
.. held by:  umount:20593
... acquired at:     generic_shutdown_super
/etc/rc6.d/S01reboot: line4: 20593 Segmentation fault      $*
                                                         [FAILED]
Unmounting filesystems (retry):

The problem has only manifested itself this once, so unfortunately not
reproducible on demand, but I hope that maybe someone more
knowledgable of NFS internals than me can figure out what went wrong
and a way to fix it in the future (if it's not already fixed) :-)

--
Jesper Juhl <jesper.juhl@xxxxxxxxx>
Don't top-post  http://www.catb.org/~esr/jargon/html/T/top-post.html
Plain text mails only, please      http://www.expita.com/nomime.html
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/