On 04/11/2013 3:24 AM, grarpamp wrote:
HS -> Guard -> Middle -> Middle -> Rendpoint <- Middle <- Guard <- ClientI hate to say it that gives me no confidence at all. Only the randomly chosen guard at the HS end needs to be compromised and the whole chain becomes worthless. A timing attack will reveal all.I'd thought it trivial to time/packet/byte count and find the: - HS, IF you ran both its guard and a client. - Client, IF you ran both its guard and its destination of choice (HS/[clearnet/exit]).
Yes
It depends on whether the guard has knowledge that some of the streams going through its relay function are in fact destined for a HS/client and not some other relay. I never properly followed up on that question...
This knowledge is trivial to obtain, a timing attack isn't even necessary. Whether that represents a true risk I haven't yet determined. I need to consider what surety one can have that the guard is not compromised.
DC -- tor-talk mailing list - tor-talk@xxxxxxxxxxxxxxxxxxxx To unsubscribe or change other settings go to https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-talk