I'm seeing problems with the new NFS code that I didn't
have before the NFS3 stuff from lite2 was integrated.
(Interestingly enough, I have the same similar problem
with BSD/OS 2.0 and 2.01).  In a nutshell, using 1024 for
rsize and wsize causes NFS-mounts to hang.  With an 8K
blocksize everything is just dandy.  This is with -current
as of 10am Feb 24 on the iX86 port.  Is anyone else seeing
this (mis)behaviour?

NB: The reason for using a 1K block size is that we were
getting poor performance using large block sizes on NFS
mounts on various platforms that went through our router.

 - todd
---
Todd C. Miller   Sysadmin--University of Colorado   Todd.Miller@cs.colorado.edu