> Sigh, this is becoming quite annoying.
>
> panic: kkernel: page fault trap, code=0
> Stopped at      0x801b38a4:     kernel: page fault trap, code=0
> Stopped at      db_disarm+0x1b: movl    PTmap(%eax),%eax
>
> -
> That's exact wording, there was a slight typo in my last email (movl).
> kkernel is kkernel, not a typo.
>
> Still unsure how to debug it so all I can do is say I haven't changed my
> config recently.
>
> It's NetBSD 1.5 i386 Generic running on a pentium 100. Using rasppoe for a
> dsl connection and ipfilters/ipnat (forwarding = 1) to provide a
> gateway for
> lan net access.
>
> The system previously ran NT4 server for about 2-3 years without hicups
> (usually 7-10 day uptimes before being shut down on purpose or
> powerfailures).
>
> Also I just tried to reboot it and got this:
>
> swapctl: adding /dev/wd0b as swap device at priority 0
> Automatic boot in progress: starting file system checks./dev/rwd0a: 3634
> files, 27625 used, 56859 free (1155 frags, 6963 blocks, 1.4%
> fragmentation)
> /dev/rwd0a: MARKING FILE SYSTEM CLEAN
> kernel: page fault tap, code=0
> Stopped at if_slowtimo+0x24: movw 0x2e(%ebx),%ax

Thanks to all who replied. After a couple of crashes (and me not being able
to access email to see possible solutions or how to gather further info) the
machine started to fail it's boot up memory test, I think I've found the
real problem. The 'page fault' bit should have clued me in to this I guess.
Hopefully we've got some old spare ram around here :)

Julio Merino's suggestion backs this up too, thanks.

Strange though, it's been running for the past 10 hours without hicups again
*touch wood*

-Gwilym