[TLS] order of evaluation when multiple client hello extensions are specified

[Rob Dugal <rdugal at certicom.com>]

How should a TLS server determine what is the best matching negotiation paramaters when a client sends several client hello extensions?
Suppose the client sends signature_algorithms, ec_pt_formats, and elliptic_curves extensions, containing multiple supported algorithms/formats/curves.
Which of these extensions should the server give the highest preference to?

For example, suppose client sends these:
elliptic_curves extension { secp384r1, secp256r1 } 
ec_pt_formats extension { ansiX962_compressed_prime, uncompressed } 
signature_algorithms extension { ecdsa/sha256, ecdsa/sha384, ecdsa/sha1 }


Without draft-rescorla-tls-suiteb being enforced, would secp384r1/sha384/uncompressed be considered a better match 
than secp256r1/sha256/ansiX962_compressed_prime?
 
Would the search order be this?
secp384r1/sha256/ansiX962_compressed_prime
secp384r1/sha384/ansiX962_compressed_prime
secp384r1/sha1/ansiX962_compressed_prime
secp384r1/sha256/uncompressed
secp384r1/sha384/uncompressed
secp384r1/sha1/uncompressed
secp256r1/sha256/ansiX962_compressed_prime
secp256r1/sha384/ansiX962_compressed_prime
secp256r1/sha1/ansiX962_compressed_prime
secp256r1/sha256/uncompressed
secp256r1/sha384/uncompressed
secp256r1/sha1/uncompressed


or maybe this?
secp384r1/sha256/ansiX962_compressed_prime
secp256r1/sha256/ansiX962_compressed_prime
secp384r1/sha256/uncompressed
secp256r1/sha256/uncompressed
secp384r1/sha384/ansiX962_compressed_prime
secp256r1/sha384/ansiX962_compressed_prime
secp384r1/sha384/uncompressed
secp256r1/sha384/uncompressed
secp384r1/sha1/ansiX962_compressed_prime
secp256r1/sha1/ansiX962_compressed_prime
secp384r1/sha1/uncompressed
secp256r1/sha1/uncompressed

-----------------------------------------------
Robert Dugal
Member of Development Group
Certicom Corp.
EMAIL: rdugal at certicom.com
PHONE: (905) 501-3848
FAX  : (905) 507-4230
WEBSITE: www.certicom.com


_______________________________________________
TLS mailing list
TLS at ietf.org
https://www.ietf.org/mailman/listinfo/tls