> On Oct 23, 2014, at 4:51 PM, Ryan Sleevi <ryan-mozdevsecpol...@sleevi.com> 
> wrote:
> 
> On Thu, October 23, 2014 1:08 pm, John Nagle wrote:
>> Examine the cert of "https://www.sevendays.co";.
>> 
>> Here's one of those certs with a huge number of unrelated hosts.
>> This seems to be a Cloudflare legacy setup from the pre-TLS era.
>> Unfortunately, this cert became valid on 10/09/2014. It's
>> not a legacy cert.
> 
> So? It's perfectly valid and conforming with all of the policies set out
> by Mozilla and the CA/Browser Forum.

And I suspect it is related to this:
http://blog.cloudflare.com/introducing-universal-ssl/


>> Should certs like this be rejected as misrepresenting the identity of
>> the organization?   Junk certs like this let any domain in the cert
>> impersonate any other domain, so they are a form of "wildcard" cert.
>> Now that all major browsers have TLS, they are unnecessary and can
>> be phased out, correct?
> 
> So what?
> 
> 1) It's not misrepresenting the organization. BR certs do not, have not,
> and have never been a means of asserting the organization. Thinking that
> TLS certs do that is a fabrication that some CAs promulgate, but has
> nothing to do with the purpose of TLS certificates (which assert a binding
> between a *domain* and a key)
> 
> 2) Calling them junk certs implies you think there's something wrong with
> them, but you'd need to spell that out. Nothing is wrong with them by any
> of the root store policies.
> 
> 3) Wildcard certs are both valid and useful. They are expressly permitted
> in the BRs.

And notably, multi-SAN certs are not wild-card certs; they are far more 
limited.  A wildcard can be used for an indefinite number of names, none of 
which are individually validated.  A multi-SAN cert has a prescribed list of 
domains, each of which must be individually validated according to the BRs.

--Richard

>> 
>> Firefox displays "You are connected to sevendays.co. Verified
>> by GlobalSign NV-SA" for this site. That misrepresents what the
>> cert really told Firefox.
> 
> No it doesn't. You connected to the IP address returned by dns for
> sevendays.co, and received a cert certified by GlobalSign NV-SA that
> asserts that an entity with operational control over sevendays.co has
> control over a private key presented in the certificate and the TLS chain.
> 
> This is all truthful and accurate information.
> 
>> The message should read "You are
>> connected to one of the following sites, and give the list.
>> Or perhaps "You are connected to ssl2910.cloudflare.com" which
>> appears to be authorized to host sevendays.co".  That would
>> assist in accelerating the phaseout of these old certs.
> 
> Accelerating the phaseout of these certs is a non-goal with no positive
> security value.
> 
>> 
>> There's a real risk here.  A break-in at any of those sites
>> allows impersonating all of them.  This creates a huge
>> attack surface.
> 
> This doesn't change at all if they're using separate certs, if the certs
> are still hosted by the same company on the same machine.
> 
> I'm sorry, but I believe you've been confused as to what the purpose is,
> but there's nothing wrong with these certs.
> 
> _______________________________________________
> dev-security-policy mailing list
> dev-security-policy@lists.mozilla.org
> https://lists.mozilla.org/listinfo/dev-security-policy

_______________________________________________
dev-security-policy mailing list
dev-security-policy@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-security-policy

Reply via email to