Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 27 Mar 2018 08:30:08 -0700
From:      Conrad Meyer <cem@freebsd.org>
To:        Benjamin Kaduk <bjkfbsd@gmail.com>
Cc:        svn-src-head@freebsd.org, svn-src-all@freebsd.org,  src-committers <src-committers@freebsd.org>
Subject:   Re: svn commit: r331618 - head/share/man/man7
Message-ID:  <CAG6CVpWf5Vkz_ACsAOrzuPR9-4z8hR6ATxnePKpMuP_jLkvVRA@mail.gmail.com>
In-Reply-To: <CAJ5_RoCNDvr5mh7%2B0Hp44zOOjJFqATNandSbomp8nmcSTGLbgQ@mail.gmail.com>
References:  <201803271451.w2REpJP9078197@repo.freebsd.org> <201803271457.w2REv6tH052497@pdx.rh.CN85.dnsmgr.net> <CAJ5_RoCNDvr5mh7%2B0Hp44zOOjJFqATNandSbomp8nmcSTGLbgQ@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
Thinking of the network as attacker-controlled is fine, but without
the CA certificate database in ports, TLS provides neither data
integrity nor confidentiality.[0]

Even with certificate validation, it's unlikely that TLS provides
meaningful confidentiality for svn.freebsd.org =E2=80=94 IP still exposes t=
he
server's address:

$ host 8.8.178.107
107.178.8.8.in-addr.arpa domain name pointer svnmir.ysv.freebsd.org

Even a naive network attacker can determine that you are interacting
with a FreeBSD source mirror, and can determine the direction of the
flow of information based on simple count of upload / download bytes.

Best,
Conrad

P.S., we should probably ship a CA database in base.  Maybe with an
override version in ports to match our release model.  But, base
should be able to authenticate certificates out of the box.

[0]: https://github.com/moxie0/sslsniff

On Tue, Mar 27, 2018 at 8:01 AM, Benjamin Kaduk <bjkfbsd@gmail.com> wrote:
> On Tue, Mar 27, 2018 at 9:57 AM, Rodney W. Grimes
> <freebsd@pdx.rh.cn85.dnsmgr.net> wrote:
>>
>> > Author: trasz
>> > Date: Tue Mar 27 14:51:19 2018
>> > New Revision: 331618
>> > URL: https://svnweb.freebsd.org/changeset/base/331618
>> >
>> > Log:
>> >   Use https:// instead of http://.
>> >
>> >   MFC after:  2 weeks
>> >
>> > Modified:
>> >   head/share/man/man7/development.7
>> >
>> > Modified: head/share/man/man7/development.7
>> >
>> > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D
>> > --- head/share/man/man7/development.7 Tue Mar 27 14:50:12 2018
>> > (r331617)
>> > +++ head/share/man/man7/development.7 Tue Mar 27 14:51:19 2018
>> > (r331618)
>> > @@ -57,7 +57,7 @@ can be found at:
>> >  FreeBSD src development takes place in the CURRENT branch in
>> > Subversion,
>> >  located at:
>> >  .Pp
>> > -.Lk http://svn.FreeBSD.org/base/head
>> > +.Lk https://svn.FreeBSD.org/base/head
>> >  .Pp
>> >  There is also a read-only GitHub mirror at:
>> >  .Pp
>>
>> Why do we want to run the load of TLS for what are public bits?
>> And fyi a default install of FreeBSD can not use https, you have
>> to install certs from ports before any of these https links
>> can even work, and that can be a royal pita in some situations.
>
>
> Many of us are used to thinking of the network as controlled by an attack=
er.
> Running http-not-s to fetch the sources lets "the attacker" supply an
> arbitrary
> collection of bits under the name FreeBSD without a good way for the user=
 to
> check that the bits on their disk match what the FreeBSD Project expects
> them to be.
> TLS provides data integrity as well as confidentiality...
>
> -Ben



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAG6CVpWf5Vkz_ACsAOrzuPR9-4z8hR6ATxnePKpMuP_jLkvVRA>