RE: Cached OCSP responses vs. single entry CRLs

["Santosh Chokhani" <chokhani@xxxxxxxxxxxx>]

Title: Message

Ryan:

 

Assuming the CA and the client software are working properly, there will be one DP (URL) in the CDP and that will point to the proper partition.  The client will get the CRL from that URL.  The DP in the IDP of the CRL will be matched to the DP in the CDP of certificate and all will be well.

 

If there is a glitch on the network or active adversary, then these won't match, but in those circumstances there are innumerable (albeit finite) way to cause denial of service, CRL partitioning notwithstanding.

-----Original Message-----
From: owner-ietf-pkix@xxxxxxxxxxxx [mailto:owner-ietf-pkix@xxxxxxxxxxxx] On Behalf Of Ryan M. Hurst
Sent: Tuesday, December 09, 2003 10:02 PM
To: Jean-Marc Desperrier; ietf-pkix@xxxxxxx
Subject: RE: Cached OCSP responses vs. single entry CRLs

I have seen engines that do not deal with extension criticality in CRLs, some of the java implementations I have seen in-fact; these would trust these partial CRLs as a full CRL. Now I know folks will say thats a bad implementation, and I am not arguing that what these clients have done is right but its yet another example of how CRLs fail in this area due to the history of implementations out there.

 

Another example is where the client does properly deal with critical extensions in the CRL, in this case almost all implementations I have seen would show the CRL as bad and fail validation (as it should), I dont know of one that would try the next URL in the CDP to see if it was the full CRL. And even if it did thats extra data to be downloaded to discover its not the rigth CRL.

 

Ryan

---

From: Jean-Marc Desperrier
Sent: Tue 12/9/2003 12:19 PM
To: ietf-pkix@xxxxxxx
Subject: Re: Cached OCSP responses vs. single entry CRLs

Ryan M. Hurst wrote:

>[...] there is not way to tell from a CDP if the
>data on the other end represents a partitioned CRL or a full one.
>  
>
But then clean client should check the content of the IDP extension 
after getting the CRL and check it's distribution point matches the 
distribution field of the CDP, and if not refuse the CRL, and supposedly 
try the next crldp inside the cert.
RFC3280 6.3.3 (b)(2)(i)

I'm a bit sceptic it really works, but if the CRL's IDP is marked 
critical as it should, maybe a good number of client will duly reject 
the CRL when they can not support that case.