RAM encryption and key storing in CPU

RAM encryption and key storing in CPU

[someone called]

Hello,

==========
Problem:

Everything is stored in plaintext in the Memory.

So if although full disc encryption is used on an ex.: Linux Desktop, it is possible to copy the content of the memory, while the notebook was on suspend or it was running:

https://citp.princeton.edu/research/memory/media/

==========
Solution:

Can we (optionally*) encrypt the content of the memory and store the key for decryption in the CPU to avoid these kind of attacks in general?

Example patches (I am NOT related to them): 

https://www1.informatik.uni-erlangen.de/tresor

Is this solution already in the Linux kernel? If yes, how can a Linux enduser turn it on? If no, how can we get the code/idea in the mainline? What are the arguments against it?

*if someone would want to harden it's Linux Desktop (since notebooks could be stolen..) it could turn on this feature to avoid a policy to always turn off the notebook while not using it.

Thank you for your comments.

Re: RAM encryption and key storing in CPU

[Richard Weinberger]

On Tue, May 26, 2015 at 5:21 PM, someone called
<someonecalled@safe-mail.net> wrote:
> https://www1.informatik.uni-erlangen.de/tresor
>
> Is this solution already in the Linux kernel? If yes, how can a Linux enduser turn it on? If no, how can we get the code/idea in the mainline? What are the arguments against it?

No, these patches are not mainline.
IMHO such solutions paper over the real problem. Adding new layers won't help.
If we store keys in CPU registers, secret agencies will (or have
already) develop tools to dump them too.

-- 
Thanks,
//richard