On 8/31/11 6:34 AM, Gervase Markham wrote:
There is a way to encode this in certificates, called basicConstraints,
although I suspect very few CAs do that. (Why limit your market?)
One possible answer to this last question could probably be "because it
will make it simpler to end up in our trusted CA list".
Hard to enforce on CAs that are already in it, of course, unless we're
willing to announce a policy change and have CAs either add such
constraints or go through whatever hoops we want CAs that can issue to
any hostname to go through.
Another possible answer is "reduces the chance of your CA being executed
like DigiNotar was". Maybe.
I guess NSS could have a feature to impose them on a CA from outside.
Or that, yes.
That would limit the extent to which a compromise of one of these
"specialist" CAs could be exploited (e.g. we'd notice that a Dutch CA is
being used to sign the Mossad's website and cry foul, without
pre-pinning the CA for the presumably rarely visited Mossad site).
Just because they are a Dutch CA doesn't mean they are necessarily only
working with Dutch sites. Verisign/Symantec is an American CA.
I didn't say this applies to all CAs. I said that there are CAs in our
cert store it applies to. The phrase "Dutch CA" above means "A CA that
told us it only issues certificates to .nl sites".
We don't want to put ourselves in a position of entrenching incumbents.
Yes, I agree. CAs should be free to not thus restrict themselves, but I
think that should imply a higher trust bar on our part....
In addition, even a CA focussed only on Dutch _companies_ would want to
issue certs for .com, because I'm sure a lot of Dutch companies have
..com sites.
This is focusing on the wrong thing. I said up front that not all CAs
are "specialist" CAs. The question is how many of the ones in our cert
store _are_. If it's too few, this is not really a proposal that flies.
But we need that data.
Given that the aim would be to prevent them issuing bogus
certs for mozilla.org, paypal.com, twitter.com etc., I'm not sure many
of them
Please continue? ;)
We did consider imposing such a restriction for government CAs, on the
basis that the audit rules relating to them are necessarily different.
OK. Considered and rejected? Or considered and limboed?
Has this been considered before? Is my assumption that a lot of the CAs
in our trust list would only issue to a small subset of possible
hostnames accurate?
It is, but perhaps not in a useful way which allows us to contain risk. :-|
OK, can we get data on this somehow? Asking the CAs themselves, say?
-Boris
_______________________________________________
dev-security mailing list
dev-security@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-security
