100 likes | 312 Views
Certificate and CRL Profile and Public Key Algorithms. Russ Housley & Tim Polk Dec. 13, 2000. Status. New algorithms draft posted in November draft-ietf-pkix-ipki-pkalgs-01.txt Ready for WG Last Call now New son-of-RFC 2459 posted in November draft-ietf-pkix-new-part1-03.txt
E N D
Certificate and CRL ProfileandPublic Key Algorithms Russ Housley & Tim Polk Dec. 13, 2000
Status • New algorithms draft posted in November • draft-ietf-pkix-ipki-pkalgs-01.txt • Ready for WG Last Call now • New son-of-RFC 2459 posted in November • draft-ietf-pkix-new-part1-03.txt • Son-of-RFC 2459 will be ready for WG Last Call in Late January 2001
Algorithms • It is done… • However, some people think that it should specify a MUST implement certificate and CRL signature algorithm
Profile Modifications (1 of 2) • Recommended support for pseudonym to align with QC • Support for inhibit any-policy extension now required for conforming clients • Permit use of subject directory attributes extension • One new IETF extension • subject information access
Profile Modifications (2 of 2) • Removed public key validity period from the path validation algorithm inputs • Removed the issuer unique identifier from the trust anchor information • Aligned policy mapping processing with X.509-2000 rules • Other minor clarifications
Remaining Profile Work • More minor clarifications • relying party can supply constraints to the validation algorithm via inputs, not initial state variables • CRL DP extension explanation text • Finish removal of UIDs from path validation algorithm (missed one place) • Name constraints…
Name Constraints • The relying party may be interested in using an alternative name form, such as the rfc822name • The CA MUST be able to constrain the contents of alternative names without requiring that all certificates in the chain include the constrained name form
X.509 and Name Constraints • Semantics of excluded subtrees are clear • Semantics of permitted subtrees are subject of some dispute: • PKIX Folks: permitted subtrees are satisfied by a certificate if it contains one or more acceptable names • X.509 Folks: permitted subtrees are satisfied when every subsequent certificate contains the specified name form and that name is within the subtree
Name Constraints Example Cert #1Cert #2 Issuer c=US,cn=CA1 c=US,o=XX,cn=CA1 Subject c=US,o=XX,cn=CA1 c=US,o=XX,cn=CA2 Alt Name (none) (none) Constraints permit: c=US,o=XX (none) Cert #3 Issuer c=US,o=XX,cn=CA2 Subject (empty) Alt Name alice@xx.com
Way Forward • Two Alternatives • Clarify X.509 so it works for PKIX-style paths • Develop new IETF name constraints extension • Clearly, the first alternative is preferable • Not in our control! • How long should we wait?