[109491] in Cypherpunks
Re: About Alpha radiation... (fwd)
daemon@ATHENA.MIT.EDU (Tim May)
Thu Mar 25 21:28:51 1999
In-Reply-To:
<Pine.BSF.3.96.990326014120.11414B-100000@pakastelohi.cypherpunks.to>
Date: Thu, 25 Mar 1999 18:06:18 -0800
To: cypherpunks@cyberpass.net
From: Tim May <tcmay@got.net>
Reply-To: Tim May <tcmay@got.net>
At 4:48 PM -0800 3/25/99, Lucky Green wrote:
>On Thu, 25 Mar 1999, Jim Choate wrote:
>> One could say that since your original statement was incomplete and left out
>> important boundary conditions would imply you didn't understand it any
>> better than the original poster or the textbook authors.
>
>Here is my original statement:
>
>> One of my favorite examples is DRAM. Yes, your DRAM may have a 60ns
>> refresh cycle. Does this mean pulling the plug on your computer will
>> permanently erase the key that was stored in the same DRAM segment for
>> days? Not even close. Sure, the OS can't read out the key. But serious
>> destructive analysis is able recover the key for weeks or months after you
>> pushed that "emergency" button.
>
>I'd love to know what's unclear about this statement. Am talking
>*anywhere* in here about long-term charge retention being the mechanism
>that allows recovery of the key? Of course not
>
Lucky is right. Sandia and Intel and others did a lot of work on recovering
the contents of RAM, both dynamic and static, by various means.
It turns out that there are various subtle "memory effects" in RAM
cells...charge trapping in the potential wells, for example. Subtle shifts
in threshold voltages.
(This is one of the reasons that "good cryptographic hygiene" has included
arranging for keys loaded into RAM to use method to ensure that the same
RAM cells aren't constantly being loaded with the same bits.)
OK, what are some of the methods looked at?
1. Freezing. it turns out that freezing some RAMs, esp. DRAMs, could render
an unplugged chip readable. Not perfectly readable, but we're talking about
reading enough to degrade the entropy, e.g., to cut the search time way
down.
2. Flash irradiation. Co-60 irradiation can also "freeze" the state of a
RAM. I talked to Bob Gregory of Sandia about this back in 1978. There are
probably some unclassified papers on this, but much of the best stuff done
by the government is probably still classified...and I know the work we did
at Intel was never published.
And so on.
(Am I saying _all_ RAMs? No. _Any_ RAM? No. Which RAMs? This would need
detailed characterization. Maybe the spooks have done this characterization
recently, maybe not.)
Lucky's original point about how folks ought not be confused by simplistic
analyses is valid. Textbooks are a _starting_ point. Researchers always
have to get beyond the simple summaries of how things work.
I could write for hours about my alpha particle experiences, including
hilarious encounters with some Japanese who had absolutely no common sense
when it came to this, but this list is not about alpha particles and memory
chips.
(BTW, the Japanese eventually got past the "textbook understanding" and
eventually published many good, if not especially original, papers in this
area.)
--Tim May
Y2K: Hope for the best, prepare for the worst, enjoy the ride.
---------:---------:---------:---------:---------:---------:---------:----
Timothy C. May | Crypto Anarchy: encryption, digital money,
ComSec 3DES: 831-728-0152 | anonymous networks, digital pseudonyms, zero
W.A.S.T.E.: Corralitos, CA | knowledge, reputations, information markets,
Licensed Ontologist | black markets, collapse of governments.
