John Gilmore <[EMAIL PROTECTED]> writes: >Despite a bunch of PC graphics chips and boards having announced HDCP >support, according to the above article, it turns out that none of them will >actually work. It looks like something slipped somewhere, and an extra >crypto-key chip needed to be added to every existing board -- at >manufacturing time.
The extra item was just a little serial EEPROM with per-device keying/entitlement info. All the HDCP stuff (the challenge-response handshake over the DVI link and the actual content en/decryption and MAC'ing) are performed by the DVI controller chip using data from the external EEPROM. Since adding the EEPROM would have added a few tens of cents to the cost of the hardware (the hardware itself, the manufacturing cost, the cost of personalising and testing each EEPROM, and the fact all portions of the manufacturing process that come into contact with keys require special security measures), there was no clear idea when anyone could actually use it, and there would no doubt be interop problems once devices did hit the market, leading to product returns from dissatisfied customers, manufacturers skipped the cost and overhead of adding it. It makes sense really, why add something that's both completely useless to users and a potential liability to manufacturers? If I were creating the devices, I'd have done the same thing. It's like region coding in DVD players, you reduce cost and add value by *not* including it. Peter. --------------------------------------------------------------------- The Cryptography Mailing List Unsubscribe by sending "unsubscribe cryptography" to [EMAIL PROTECTED]