Yes, I agree that this would be a good idea. Building and verifying a separate path to the CRL issuer is a less known requirement of RFC 3280. Thanks, Steve "David A. Cooper" wrote: > > All, > > In some discussions that I have seen having with Santosh Chokani, he > asked me whether PKITS included any tests that verified that a relying > party would only use a CRL if the signature on the CRL could be verified > using an authenticated public key. > > PKITS already includes some such tests in the Basic Certificate > Revocation Tests section, including tests where the CRL has a bad > signature and the CRL has the wrong issuer name. However, there are no > tests to verify that a CRL will only be used if there is a valid > certification path to the verification key, since in almost every test > the certification path is the same for the CRL and the certificate being > verified. > > I have attached a figure depicting three new tests that could be added > to the Basic Certificate Revocation Tests section that (1) would verify > that relying parties can use CRLs even if the certification path leading > to the CRL verification key is different than the one leading to the > certificate being verified and (2) would verify that relying parties > will not use a CRL if the certification path for the verification key is > invalid. > > In the first two tests, the trust anchor has issued two certificates to > the intermediate CA: one certifying the CA's certificate signing key and > one certifying the CA's CRL signing key. In one test, the end entity's > certificate is valid and in the other the end entity's certificate has > been revoked. > > In the third test, the trust anchor has again issued two certificates to > the intermediate CA: one certifying the CA's certificate signing key, > which is valid, and one certifying the CA's CRL signing key, which has > been revoked. Since the certificate issued to the intermediate CA's CRL > signing key has been revoked, the signature on the CRL can not be > validated and so the status of the end entity certificate can not be > determined. > > Do people believe that these would be worthwhile tests to add? > > Thanks, > > Dave > > ------------------------------------------------------------------------ > Name: new_tests.pdf > new_tests.pdf Type: Acrobat (application/pdf) > Encoding: base64
S/MIME Cryptographic Signature