Re: 2.1.70 knfsd oops

[Jonathan Corbet]

> From the decoded oops it appears a dentry had a NULL parent pointer,
> which should never happen. So it seems that either a dentry got trashed
> somehow, or else possibly there's an SMP locking problem. Is the server
> in question running on an SMP machine?

It is a uniprocessor machine.  I built the kernel with the SMP option
turned off as well.

> Is this a one-shor oops, or has something similar happened before?

It happened three times over the course of an hour.  This after the server
had been stable for a few days.  After that I put the old user-mode NFS
back, so, of course, the problem stopped...

Thanks for looking into this,

jon

Jonathan Corbet
National Center for Atmospheric Research, Atmospheric Technology Division
[e-mail]		     http://www.atd.ucar.edu/rdp/jmc.html