Hi All -

I assuming that utmp can only hold 'maxusers' as defined in the
kernel. After roughly 9 days of uptime with 100+ users on at any
given time, I decided to load a kernel with maxusers set to '256'
(was 64).  Thats about all I changed.  After about 6 hours of
running, there was a kernel panic:

Aug 21 13:23:18 sdf /netbsd: fatal kernel trap:
Aug 21 13:23:18 sdf /netbsd:
Aug 21 13:23:18 sdf /netbsd:     trap entry = 0x2 (memory management fault)
Aug 21 13:23:18 sdf /netbsd:     a0         = 0x20
Aug 21 13:23:18 sdf /netbsd:     a1         = 0x1
Aug 21 13:23:18 sdf /netbsd:     a2         = 0x0
Aug 21 13:23:18 sdf /netbsd:     pc         = 0xfffffc0000379da0
Aug 21 13:23:18 sdf /netbsd:     ra         = 0xfffffc0000379650
Aug 21 13:23:18 sdf /netbsd:     curproc    = 0xfffffc00606d6a90
Aug 21 13:23:18 sdf /netbsd:         pid = 14376, comm = mc
Aug 21 13:23:18 sdf /netbsd:
Aug 21 13:23:18 sdf /netbsd: panic: trap
Aug 21 13:23:18 sdf /netbsd: syncing disks... 20 16 16 13 12 12 12 11 11 9 9 9 8 8 8 6 6 5 5 4 3 3 3 3 3 2 2 1 1 1 done
Aug 21 13:23:18 sdf /netbsd:
Aug 21 13:23:18 sdf /netbsd: dump to dev 8,1 not possible
Aug 21 13:23:18 sdf /netbsd: rebooting...

I realise that there are other parameters affected by 'maxusers' and 
that this panic could be totally unrelated (comm = mc and MM fault).

config will complain if maxusers is > 64 .. are there other things I
should be looking at? (also, I didn't find it, but is this something
that can be tuned with sysctl ?)