On Wed, Jan 03, 2007 at 02:32:56PM -0500, chris@xxxxxxxxxxxx wrote: > I'm studying section 2 of the spec on connections > and just wanted to confirm some items with the good people of Tor.... > > * The 'short-term connection key' mentioned in third paragraph is an AES key > right? No; it's a short term RSA key. We don't say that, because you can't (sanely) stick symmetric keys in a certificate. I'll fix the spec so it's (hopefully) readable without assuming the reader knows so much about TLS. > * The 'identity key' is the RSA public key associated with a router right? Yes. > * Spec says this identity key is self-signed but did not say the 'short-term > connection key' is signed. The 'short-term connection key' is signed by > encrypting its hash with router's RSA private key right? The identity key certificate is self-signed. The certificate with the short term connection key is signed by the identity key. > * Spec introduced the terms digital signature and certs in section 2 without > mentioning all the boring details like what standard is used for these two > things...e.g. X509? X.509 is the only certificate standard supported by TLS (at least, in any implementation I've seen). > * Are all the aforementioned certs and keys mentioned above sent in > 'cells'? Which cell types? This was not specified. No. This is part of the TLS handshake. I'll try to make that clear if I can. > * It appears each onion router has a RSA public key that can be > acquired from a directory server or EXTEND cells. The begs the > question how do the Onion Routers safely get the public keys of > directory servers? I assume routers talk to them over HTTS / SSL > right? Directory servers are specified in dir-spec.txt. Directory _authority_ locations and keys ship with the tor source; clients learn about caches from the authorities. HTH, -- Nick Mathewson
Attachment:
pgpZdwQnMHVI1.pgp
Description: PGP signature