Return to mroe.idea CVS log

Up to [CSRG BSD Unix] / 43BSDReno / contrib / isode-beta / others / X / ideas

Annotation of 43BSDReno/contrib/isode-beta/others/X/ideas/mroe.idea, revision 1.1

1.1     ! root        1: (Message inbox:23)
        !             2: To: Steve Kille <[email protected]>
        !             3: cc: Mike Roe <[email protected]>, Jon Crowcroft <[email protected]>
        !             4: Subject: Re: Moron X Window protocol on ISODE TS 
        !             5: In-reply-to: Steve Kille's message of Tue, 17 Oct 89 08:15:08 +0100.
        !             6:              <[email protected]> 
        !             7: Date: Tue, 17 Oct 89 12:41:13 +0100
        !             8: From: Mike Roe <[email protected]>
        !             9: 
        !            10: 
        !            11: > Forget transport layer authentication.  If you want to add in authentication,
        !            12: > do it where it fits "proerly".    I think that the aim is a simple
        !            13: > and reaonabvly efficient mapping of X.  For this, cutting at the 
        !            14: > transport layer makes a lot of sense.  But don't try hacking in bells
        !            15: > and whistles.   
        !            16: 
        !            17: Firstly, I agree "X over TS" is an orthogonal issue to authentication, and 
        !            18: should be tackled separately.
        !            19: 
        !            20: However, I'm now going to fall for it and ask "What's wrong with transport
        !            21: level authentication?".
        !            22: 
        !            23: Here, you have a stream of data between a sender and a receiver. (Ok, so the 
        !            24: stream consists of X protocol datagrams). All you want to do is convince the
        !            25: receiver that everything came from the sender --- there is no need for
        !            26: non-repudiation etc.
        !            27: 
        !            28: Clearly, you want to insert a checksum into the stream every so often, at
        !            29: least once per (application) datagram. Note that this ought to be a
        !            30: simple hash (eg DES in CBC mode). Sending a full authenticator (Certification
        !            31: path + RSA encrypted token) each time is unacceptably wasteful.
        !            32: 
        !            33: Claim: The end of a TSDU or SSDU is as good a place as any to put the checksum.
        !            34: 
        !            35: I can see some of the arguments against it :
        !            36: 
        !            37: 1. (Pragmatic) As I said before, no agreed way to set up the key.
        !            38: 
        !            39: 2. (Religious) It's the layer 7 entity you want to authenticate, not the T-layer
        !            40:    The T-layer should not have to know about application layer information.
        !            41: 
        !            42: 3. The only time the Rx needs to look at the checksum is at the end of a 
        !            43:    datagram, so it should be sent only at the end of a datgram. 
        !            44:    From the standpoint of the 7 layer model, only application/presentation can
        !            45:    do this.
        !            46:    From the standpoint of the actual protocols, session can also do this,
        !            47:    as a P-DATA.REQUEST maps directly onto an S-DATA.REQUEST.
        !            48: 
        !            49: Religious argument for it :
        !            50: 
        !            51:   The application layer doesn't know about the concrete encoding --- so it
        !            52:   can't calculate a checksum.
        !            53: 
        !            54:   While we're on this, the X.509 "SIGNED" and "ENCRYPTED" macros are somewhat
        !            55:   broken : They prevent you from negotiating a different transfer syntax!
        !            56: 
        !            57: THESIS: Only the lower layers know which encoding to sign.
        !            58: ANTITHESIS: Only application layer knows which key to use.
        !            59: SYNTHESIS: Sometimes the seven layer model creates imaginary problems.
        !            60: 
        !            61: Mike

This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.