Yousuf said:
I wonder if revoking the keys is going to work at all anymore? It looks
like with this technique, all anyone has to do is wait for the
processing key to appear in memory, and they have it again (albeit,
easier said than done). No matter how many keys they revoke and reissue,
they will always lose it.
Yousuf Khan
Of cause this shows a fundamental limit of security that can be implemented
in a PC. PC does not have a secure memory, period. That itself says that
NO protection scheme in the PC can be secure even theoretically.
To have a secure protection scheme in a device, it requires a memory space
that is not accessible by any processes outside the CPU micro-code, e.g. secure
memory. That means micro-code itself would have to do all the decrypting,
authentication etc. Obviously this memory space has to be factory programmed, or at least
a seal code has to be factory programmed which allows access to programming
the secure memory.
None of these exists in a PC architecture, so as long as something is
PC-playable, it is not going to be secure (until this architecture is changed).
To change this architecture is not going to be easy of cause, it would conceptually
obsolete most of existing software.
Most important, every software installation would require to have a
hardware component with its own secure memory that has to be securely(*)
attached to the processing unit to program secure
memory with the new description code specific for this new software.
"Securely attached" means excluding external monitoring of communication
lines to steal the software decryption key, which is fundamentally possible only with QM
cryptography. Good enough approximation (which does not need a hardware
component) might be public-key cryptography with the software
decryption key being encrypted with the public key of the "internal"
recipient (e.g. micro-code in CPU) while secret key is factory programmed
in the secure memory of CPU and allows the micro-code to decrypt the software
key.
Unfortunately this "internal" secret key would become
a holly grail of the whole system which is now more difficult to compromise
because it is in secure memory, but considering its uniqueness to entire
architecture of the world PCs, it could still be eventually compromised by
hardware attack or social engineering attack. It is just not right to
center security of the whole system around one key. Which brings
us back to QM-cryptography and hardware modules coming with each software.
Anyway, I think copy-right concept that was initially intended for protecting book
publishers from competitors copying their products have been incorrectly extended on
private activities people themselves involve with. This resulted in
creating an artificial monopoly on content distribution (which should actually
be competitive and as wide open as possible) instead of solving
actual problem of compensating content creators. So, if the means on
maintaining this unproductive state-sponsored monopoly are being
cracked, why should I care. But the problem of securing the content
in a PC still remains interesting topic in itself.
Regards,
Evgenij
