> > This is pretty naive. Every modern hardware contains software stored > > for example in EEPROMs or flash. If a malicious software is good > > enough it can do a lot of possible things that needs not only > > local attacks. > > I would consider these local attacks. You do not? No, it could be part of a site channel attack that works remote. > > There are a lot of possible attacks for malicious implementations. > > It can for example store the AES key in a flash memeory on the hardware > > and reveal on special commands or it can be exploited by site channel > > attacks. > > The flash again is a local attack. If you use the flash only to store a key you are right. If you use the flash to buffer the key for a site channel attack it is a remote attack. > As to side-channels, this is disk encryption. Sure, if in was > something going over the network, leaking key material via timing > information would be easy. However serving, e.g., websites from > an encrypted disk is a very bad idea in the first place. If you > do such a thing, there could be a side channel. If you don't use network for your computer, nearly all attack vectors go to local attacks. If you like to encrypt data from your server you have to do this. You could also see your browser as service because it interprets Javascript. I see no architectural problem on that. Yes, there could always be a site channel. But its much better than no encryption. Also you could use different keys for the disk space used from the service and the other one. > The objections actually go one step further. After all > the design documents would be doctored and, unlike software > available as source code, hardware is not easily alanysed or > checked to match its suppodes documantation. I also have this objections. Its not easy to verify correctness of software and for hardware this is much harder. greetings wof -- GMX FreeMail: 1 GB Postfach, 5 E-Mail-Adressen, 10 Free SMS. Alle Infos und kostenlose Anmeldung: http://www.gmx.net/de/go/freemail --------------------------------------------------------------------- dm-crypt mailing list - http://www.saout.de/misc/dm-crypt/ To unsubscribe, e-mail: dm-crypt-unsubscribe@xxxxxxxx For additional commands, e-mail: dm-crypt-help@xxxxxxxx
