> I have a hunch that the allocation is failing because something else
> has used up all the available memory.

Not suprisingly, it looks like the bridge code is causing this.  After
running for 5 days with bridging, `vmstat -m` showed a total of 2292K
allocated (41 mbufs).  The numbers were increasing at a rate of about 30K
and 10 mbufs per day.  About 15 hours ago, I turned on ipf and the figures
have now jumped to 3124K and 3110 (!) mbufs.

J

-- 
  My other computer also runs NetBSD    /        Sailing at Newbiggin
        http://www.netbsd.org/        /   http://www.newbigginsailingclub.org/