I'm having some problems running ipnat.

Kernel version is 1.4.1.

I have a single ipnat rule:

    rdr fxp0 0.0.0.0/0 port www -> 127.0.0.1 port 8000

I'm doing transparent proxying, so this simply routes www traffic into
my squid server.

Everything seems to work very well, but if I do an ipnat -l there are
often a couple of occurrences of:

    kmemcpy:read: Bad address

in amongst all the regular

    RDR 127.0.0.1    8000  <- -> a.b.c.d    80    [e.f.g.h y]

lines.

Anyone have any thoughts on what might cause this?  Will it cause me
any grief?

The theory that occurs to me (keeping in mind that I know nothing
about the internals of the NetBSD ipnat system) is that ipnat is
trying to read entries that have already been flushed from the nat
cache and failing.


Peter Yandell
Chief Technology Officer
Freeonline.com.au Pty Ltd
peter_yandell@freeonline.com.au