NFSv4 poops itself

[Jeff Garzik <jeff@garzik.org>]

Background:

Server: x86-64 dual core Intel, kernel 2.6.23-rc1 (my home fileserver)
	Exporting NFS/NFSv4 mounts.  Client count: 1  Uptime: 4 days

Client: x86-64 dual core Intel, kernel 2.6.23-rc1 (my main workstation)
	NFS mount setup:
	pretzel:/ on /g type nfs4 (rw,noatime,proto=tcp,addr=10.10.10.1)
	Uptime: 4 days

	Home directory mounted via NFSv4.

Problem:

My workstation has been happily talking to my file server for several 
days without incident.  An hour ago, my numeric keypad stopping working 
(unrelated problem... USB or X bug?).  The solution to the keypad 
problem is usually to log out of X and log back in, or worse case, reboot.

So, I log out, and log back in.  At first, a few shell windows open and 
successfully initialize themselves (read bash profile over NFS, etc.) 
Then, as more shell windows open, things start hanging.  I can easily 
switch to console and ssh to the fileserver, so it is clear this is an 
NFS hang.

No adverse messages at all on the client.

On the server, I see NFSv4 spamming dmesg with hundreds of thousands of 
messages:

Jul 27 08:20:53 pretzel kernel: NFSD: preprocess_seqid_op: old stateid!
Jul 27 08:21:24 pretzel last message repeated 167966 times
Jul 27 08:21:55 pretzel last message repeated 173628 times
Jul 27 08:21:55 pretzel kernel: NFSD: preprocess_seqid_op: old stateid!
Jul 27 08:22:26 pretzel last message repeated 171286 times
Jul 27 08:23:27 pretzel last message repeated 344461 times
Jul 27 08:23:30 pretzel last message repeated 18656 times

I rebooted the client, the problem disappeared, and everything is happy 
again...  but clearly NFSv4 shat itself.  And now I am worried this will 
happen again.

In all my quite-heavy use of NFSv4 I've never seen this behavior before, 
  so I would call this a regression.

I always run vanilla linux-2.6.git self-built kernels on both client and 
server.

	Jeff