Question about bridging code

[Robert Watson]

On Thu, 16 Oct 2003, Barney Wolff wrote:

> On Thu, Oct 16, 2003 at 05:18:24PM -0400, Robert Watson wrote:
> > 
> > My conclusion from my BPF bridge experience was that prototyping in
> > userspace made it a lot easier to experiment with changes, and
> > dramatically reduced the development time.  On the other hand, it did
> > terrible things to performance on high bandwidth tests, and because we
> > weren't using mbufs in userspace, made it harder to port to the kernel. 
> > One nice benefit, though, was that we had TCP/IP people programming TCP/IP
> > stuff without having to teach them about mbuf semantics or kernel
> > debugging :-).
> 
> It's actually not so hard to get kernel mbuf-oriented code running in
> userspace.  I did a userspace PPP implementation in 1994, and when it
> came time to do VJ compression I took the BSD kernel VJ code (from
> lbl.gov, if I recall correctly), defined some of the mbuf fields in my
> own structs, and it compiled and worked correctly without changing a
> single line of the VJ code. 
> 
> That project would never have survived if every bug had caused a kernel
> panic.  The code is still running in commercial service today. 

There are also at least a couple of implementations of a full BSD network
stack in userspace (Alpine being one, I believe), as well as the Linux
network stack.  A number of pretty large product companies also ship
network appliances and related products based on userspace network stacks
optimized for what they're doing, using the OS basically as a device
driver and development framework.

Robert N M Watson             FreeBSD Core Team, TrustedBSD Projects
robert at fledge.watson.org      Network Associates Laboratories