Scott G. Miller wrote on 4/13/01 7:03 am:

<>

>We can't make this 
>assurance in Java.  

I know.  The suggestion (only store the value of the key when
it's needed, then immediately garbage collect it) was to reduce
the chance of it hitting swap space.  I'm quite aware that we the
chance of it being on swap will never hit 0 (unless the user shuts
off swap entirely), but we can at least come close.  When the program
needs the key again, it can read it back from the file (which
the paranoid will have put on a RAM disk). 

Note that the paranoid people will also make a script which, once
every day (ajust time for your level of paranoia) will copy
the encryption key to a second RAM disk, remounting this
second disk to the mount point of the first one, then wiping the 
first one.  It is possible (though very difficult) for a determend
opponent to recover data that is on RAM, and their job is made
much easier if a piece of data is sitting on the same address in
RAM for a long time.  This rotation script helps prevent this.


Timm Murray

-----------
Theory is when you know how it works, but fails.  Practice is when something 
works, but you don't know why.  Here, Theory and Practice come together.
Nothing works, and nobody knows why.

_______________________________________________
Devl mailing list
Devl at freenetproject.org
http://lists.freenetproject.org/mailman/listinfo/devl
>From - Sun Apr 15 14:51:11 2001

Reply via email to