> >Only if you're right :)  I was really just trying to generate a little
> >interest in seeing if we could prove that fragmentation of the kernel
> >heap is bounded for any given kernel config. It would be a nice warm fuzzy.
> 
> I think that's orthogonal to the issue of interface detaching, yes? :-)

Depends on whether the interface detaching can cause a slow leak or not.

If the free'd regions are never subdivided, but only sit around waiting to
be reused (mbufs are like this), then we're golden. Otherwise it gets more
interesting.

Todd Whitesel
toddpw @ best.com