I moved my disks from the lame i386-20 to a i486-50 and the problem of
vm_faults and panics still occurs under stressful NFS use.  This is true
both with NQFS and std NFS.  I can't claim that ONLY nfs makes this happen,
but compiling with /usr/src on one machine and /usr/obj on another machine
eventually crashes one machine.  This machine runs slip and has GATEWAY and
NMBUF... = 1024.  To determine if nfs is the cause, I will leave the machine
running without NFS mounts for the next week.

Here is the server side nfsstat

Server Info:
  Getattr   Setattr    Lookup  Readlink      Read     Write    Create    Remove
     6568         0     33549       468     12222         0         0         0
   Rename      Link   Symlink     Mkdir     Rmdir   Readdir    Statfs  RdirLook
        0         0         0         0         0      1535      1975       118
   GLease    Vacate     Evict
       28         0         0
Server Ret-Failed
            22205
Server Faults
            0
Server Cache Stats:
   Inprog      Idem  Non-idem    Misses
       11        37         0     55637
Server Lease Stats:
   Leases     PeakL   GLeases
        0       112       863


I get errors like this in the client /var/log/messages just before the crash
Jul 13 18:13:44 atrium /netbsd: Data modified on freelist: word 0 of object 0xf8
6edc80 size 128 previous type free (0x0 != 0xdeadbeef)
Jul 13 18:13:44 atrium /netbsd: Data modified on freelist: word 0 of object 0xf8
6f0900 size 128 previous type free (0x0 != 0xdeadbeef)
Jul 13 18:13:44 atrium /netbsd: Data modified on freelist: word 0 of object 0xf8
6f0914 size 128 previous type hosttbl (invalid addr 0x6bb)
Jul 13 18:13:44 atrium /netbsd: Data modified on freelist: word 0 of object 0xf8
6f0914 size 128 previous type hosttbl (0x0 != 0xdeadbeef)

Or more often nothing at all in the log and a vm_fault on the console.
Jul 13 21:56:44 atrium savecore: reboot after panic: receive 3