At 12:50 PM 12/13/95 +1100, lyalc@mail.mpx.com.au (lyal collins) wrote:

I meant that on-line certificate issuing, notary and similar services where data is submitted to a system for processing/RSA encryption are subject to this for of attack. Parts of the SEPP/STT protocols appear to require this of merchants and customers. I retract my comments about ecash/echeques - I'm not sure of the implications there yet. As for SEPP/STT - another nail in the coffin, me thinks.

For large environments like this, it's possible to work around the attack by methods like queueing up all the signature jobs and doing them serially; this makes it difficult for the Bad Guy to know whether the server is taking time doing his multiplications or Alice's or N other customers', so he can't control timing very well by picking otherwise-informative numbers. On the other hand, your smartcard or PC is still at risk, since it's _not_ doing a lot of them, unless it's doing them just sort of at random when it's got nothing better to do and throws the real work in the middle. #-- # Thanks; Bill # Bill Stewart, Freelance Information Architect, stewarts@ix.netcom.com # Phone +1-510-247-0663 Pager/Voicemail 1-408-787-1281