Retiring portsnap [was MITM attacks against portsnap and freebsd-update]
Ian Smith
smithi at nimnet.asn.au
Mon Apr 14 05:09:43 UTC 2014
On Sun, 13 Apr 2014 10:33:53 -0400, Lowell Gilbert wrote:
> David Noel <david.i.noel at gmail.com> writes:
>
> > My main point was that if you don't trust Subversion it makes no sense
> > to say you trust portsnap. Portsnap pulls the ports tree from
> > Subversion. Using Subversion! The portsnap system relies on the trust
> > of both svnadmin and svn. Just as it does when you run svn co and svn
> > up. If you say you don't trust Subversion, essentially what you're
> > saying is that you don't trust anything running on your computer.
>
> You were talking about MITM attacks. Portsnap uses secured access for
> getting updates out of Subversion, whereas doing "svn co" remotely
> generally does not. This is not a trivial point.
Indeed it is not. David's solution - which seems to amount to removing
portsnap and herding the cats at home to DTRT about using svn securely -
relies on other cats being as smart and aware of the ramifications as he
is - a highly questionable proposition especially for the numerous more
naive users that portsnap renders the process of securely upgrading the
ports tree just about as simple and consistent as it can be.
David, perhaps your obvious talent for auditing the portsnap code and
its server-side configuration might be better applied to remedying any
perceived vulnerabilities in conjunction with present and past security
officers and teams?
cheers, Ian
More information about the freebsd-security
mailing list