Documents from Peter Sewell on TCP stack...

[Peter Sewell]

On 14 October 2015 at 17:58, hiren panchasara <hiren at strugglingcoder.info>
wrote:

> On 10/10/15 at 04:30P, Peter Sewell wrote:
> > On 10 October 2015 at 16:27, George Neville-Neil <gnn at neville-neil.com>
> > wrote:
> >
> > > Howdy,
> > >
> > > These are a bit long but this is the work that Peter (who I think is
> now
> > > on this list) and his group
> > > did in 2005 on the FreeBSD TCP stack.
> > >
> > > http://www.cl.cam.ac.uk/~pes20/Netsem/tr.pdf
> > > http://www.cl.cam.ac.uk/~pes20/Netsem/alldoc.pdf
> > >
> > > Much of the overview material including the "Quick Introduction" which
> I
> > > recommend you read first,
> > > is here http://www.cl.cam.ac.uk/~pes20/Netsem/index.html
>
> Thanks George.
> > >
> > >
> > (probably the SIGCOMM paper [
> http://www.cl.cam.ac.uk/~pes20/Netsem/paper.ps]
> > comes after that)
>
> Thanks Peter for the link. I (very) briefly looked at it and have a few
> questions (apologies if I missed something obvious):
>

These are mostly questions for Hannes and David, who have picked the
project up again.


> 1) It refers to FreeBSD 4.6-R. Have you tried any newer version?
>

Not yet, though that's part of their plan.


> 2) How many total test-cases have you created for TCP? Section 5.
> RESULTS mentions 1095 total traces and 1004 turned out good bases on
> the test-cases. Where do I look at the test-cases in detail?
>

There's more detail about the test generation in Section 3 of "Volume 1".
We generated around 1000 TCP tests - they're mostly pretty short, oriented
towards testing connection setup, teardown, and suchlike, rather than long
connections that move significant data.  I don't think we put the
individual traces on the web anywhere, but perhaps Hannes can send you a
few.



> 3) Section 6. IMPLEMENTATION ANOMALIES lists a bunch of bugs. It'd be
> good to know if they are now fixed or not.
>

These are also detailed in "Volume 1".  We didn't chase them at the time,
as we ran out of available effort - I don't recall whether we reported
them. So they may or may not have been fixed.

best,
Peter



> I think this is a great piece of work. I'd be interested in knowing
> future plans wrt this project.
>
> Cheers,
> Hiren
>