[tor-talk] Dutch CA issues fake *.torproject.org cert (among many others)

Lee ler762 at gmail.com
Sat Sep 3 18:36:54 UTC 2011


On 9/3/11, Julian Yon <julian at yon.org.uk> wrote:
> On 03/09/11 15:59, Jim wrote:
>> I don't have a solution to this problem but I am raising it in case
>> somebody else does.  It's great that you not only sign your packages but
>> that the page above also lists the fingerprints of the signing keys.
>> But in case of a man-in-the-middle attack (or a compromised website) the
>> attacker could provide his own signatures for trojaned packages and then
>> display a page that shows the signature for *his* signing key(s) in
>> place of those for the real keys.
>
> There's no general solution as this is a bootstrapping problem.

Is there a solution for this specific case?  Someone claiming to be
Roger Dingledine included a PGP signature block in the msg that
started this thread.  Nobody's responded "Hey! That wasn't me!!" or
"That's not my PGP sig!" so it seems safe enough to trust that sig.

Is there a secure way to get from that PGP sig to whatever's necessary
for verifying a TOR package one just downloaded?

Lee


> However
> anyone experienced enough to be responsible for signing releases of a
> project such as Tor will undoubtedly have left traces elsewhere on the
> net (unless they're working anonymously). For instance, they may be a
> Debian developer, or have a Twitter account. By comparing fingerprints
> or contact details published on different websites you can confirm that
> they belong to the same person and therefore haven't been tampered with.
> Deciding whether that person *should* be signing packages is a separate
> problem, but which can be approached in a similar manner. Once you've
> verified a key to your satisfaction, then add it to your keyring and
> lsign it (that's "lsign", not "sign"). If ever a future download fails
> verification because you don't trust the key, and you haven't been
> notified of a change in signing key, then you know something is amiss.
>
> The real problem is in educating people that it's a good idea to go
> through this rigmarole. I'd love to see a solution to that.


More information about the tor-talk mailing list