David Brown wrote:
> On Thu, Jun 19, 2008 at 12:37:17PM -0700, James G. Sack (jim) wrote:
> 
>>> I read some years ago that self-signed certs are more secure as the data
>>> encrypted by certified keys are accessible to those with access to the
>>> key data from the CA (e.g. - the feds).
>>
>> I don't really see that argument. The overwhelming use of ssl certs is
>> to establish credible identity, I believe. If the top level root CA
>> certs have gu'ment fingers in them, then the only consequence is
>> questionable authenticity of root-signed certs or, I suppose, bogus
>> revocations.
>>
>> Am I mistaken?
> 
> I would be concerned as to how a key is generated.  If I generate my own
> key, and ask the root CA to sign it, then I'm only using them to
> authenticate my certificate.  If their procedure requires them to generate
> my key first, it could be a weak key in a way that wouldn't be easy for me
> to detect.  It would at least be a good sanity test to test the things that
> are supposed to be prime for primality and probably that the factors are
> safe primes and stuff like that.
> 
> Anybody ever gotten an official CA key?  Did they generate the key, or do
> they let you generate the key, and they then sign it as the service?
> 

Oh wow, I never considered that a CA would generate a key -- isn't that
a violation of the whole concept of private keys.

I suppose key-signing authorities might provide key-generating utilities
to naive clients, but I would be inclined not to use any such tool, for
the reason you raise.

I'd like to know the answer to your question!


Regards,
..jim


-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list

Reply via email to