data remanence

Ringo Kamens 2600denver at
Mon May 15 21:21:14 UTC 2006

Perhaps servers should generate new private keys upon starting and store
them in ram?

On 5/15/06, Lionel Elie Mamane <lionel at> wrote:
> On Mon, May 15, 2006 at 08:30:49AM -0400, Michael Holstein wrote:
> >> Flash is writable, so can be tampered. The critical secret (the
> >> server's key) is in the ro part. So what good is it?
> > There are a variety of hardware devices that attempt to address this
> > ..  IBM's secure crypto cards (PCI) come to mind -- these are used
> > in a variety of ATMs for example, and store the key in volatile
> > memory, protected against side-channel and physical attack.
> My reasoning was that the server key is long-lived, and that from the
> key and a communication transcript, one can decrypt everything. I make
> this error more often. When using DH, you can have perfect forward
> secrecy, which is exactly avoiding this attack.
> --
> Lionel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the tor-talk mailing list