Re: trusting the processor chip

At 12:25 AM 4/27/96 -0500, Snow wrote:
>On Thu, 25 Apr 1996, jim bell wrote:
>> 
>> This analysis seems to assume that the entire production run of a standard 
>> product is subverted.  More likely,I think, an organization like the NSA 
>> might build a pin-compatible version of an existing, commonly-used product 
>> like a keyboard encoder chip that is designed to transmit (by RFI signals) 
>> the contents of what is typed at the keyboard.  It's simple, it's hard to 
>> detect, and it gets what they want.
>
>	I thought that most (all?) chips already radiated on the 
>electromagnetic spectrum? Isn't that what tempest is about?

There's a difference between trying to find a needle in a haystack, and 
finding a day-glo, red-hot needle that plays music at 110 decibels in that 
same haystack.  Digital logic chips do radiate EMI, but some radiate very 
little (because their are few logic transitions or they occur relatively 
infrequently) or are buried within other circuitry and they don't have a 
particularly good antenna.  The Trojan horse chip I'm hypothesizing would be 
specifically designed to radiate a fairly loud, continuous signal, on wires 
that are long enough to make a good antenna.  Ideally, the chip would have a 
crystal to produce a very constant frequency, so that other noise not on 
that frequency could be ignored.

The best place to put such a chip would be a location outside the computer's 
case, or at least it would have access to the outside.  I think that a 
keyboard controller would be optimum, because I suspect that there are a 
relatively small number of different designs.

Jim Bell
jimbell@pacifier.com

I realize that when one argues with a fool, no one can tell the 
difference, but as the dumbest person on the list, I figure I can learn 
from just about anyone here. Not that I am calling Mr. Bell a fool.

On Sat, 27 Apr 1996, jim bell wrote:
> At 12:25 AM 4/27/96 -0500, Snow wrote:
> >On Thu, 25 Apr 1996, jim bell wrote:
> >> product is subverted.  More likely,I think, an organization like the NSA 
> >	I thought that most (all?) chips already radiated on the 
> >electromagnetic spectrum? Isn't that what tempest is about?
> There's a difference between trying to find a needle in a haystack, and 
> finding a day-glo, red-hot needle that plays music at 110 decibels in that 
<snip>
> The best place to put such a chip would be a location outside the computer's 
> relatively small number of different designs.

	I still maintain that this would be less feasible than either:

	a) Tempest. Why bother resubverting each new processor (think 
about it, Which processor? Intel (all variants) Motorola (all variants), 
Digital (Alpha) etc. When it would be easier (It seems to me at least) to 
develop a system that _can_ find that needle in a hay stack, and simply 
develop translators for each kind of chip (which could be done in 
software I'd think) to show what the chip is doing. 

	b) physcailly compromising the work enviroment so that you see 
what the person is typing as well as what is on the screen. As well as 
get Voice etc.

	c) This I just thought of, and is kind of a hybrid of Mr. Bells 
idea and a tempest style attack, it isn't thought through real well, but 
I _think_ it would work. 

	Each processor would emit on a certain band, so you build a 
"repeater" that takes that band, encodes it, steps it to a different band 
and retrans it. This device probably could be made small enough to fit 
_easily_ inside a case, and draw very little power (the transmitting 
distance would not need to very far) and since most people never open 
their cases, it would be fairly safe from detection. It could even be 
designed to piggyback on common device interface cards (parallel/serial 
cards, Video cards) so that even if one _did_ open ones case you probably 
wouldn't notice.
	All that this would entail _after_ development would be a simple 
B&E. This wouldn't solve the problem of decoding, but it heats the 
needle, and makes it sound off at many times less cost than subverting 
the chip. 

Petro, Christopher C.
petro@suba.com <prefered>
snow@crash.suba.com

The promised reference:

"The Intel 80x86 Processor Architecture: 
Pitfalls for Secure Systems"

Olin Silbert, Oxford Systems Inc,
Phillip A Porras, The Aerospace Corp,
Robert Lindell, --- " ---

Abstract:
An in-depth analysis of the 80x86 processor families identifies
architectural properties that may have unexpected, and undesirable,
results in secure computer systems. In addition, reported
implementation errors in some processor versions render them 
undesirable for secure systems because of potential security and
reliability problems. In this paper, we discuss the imbalance in 
scrutiny for hardware protection mechanisms relative to software,
and why this imbalance is increasingly difficult to justify as
hardware complexity increases. We illustrate this difficulty with 
examples of architectural subtleties and reported implementation
errors.


My comments:
This is a high-security view paper, so they go on looking for
all possible covert channels etc. Not what we are discussing 
here, perhaps.

They note one problem with Page Access Control by the TCB through
the VERR and VERW instructions. In some cases it is possible
that these instructions leave "grant access" when they should have
said the opposite.

They note that the Timestamp Counter (TCS) in the pentium might
give out high-resolution timing information. This can be used 
attack sw RSA running in another task for example, I believe.

They have 102 flaw reports collected for 80386, 80486, Pentium.

There are 8 major security flaws reported. "7. The bits of the 
I/O Permission Bitmap (IOPB) correspond to individual byte addresses
in the I/O address space. The D0 step of the 386 permits
access to certain addresses prohibited by the I/O bitamap: if a 
4-byte access is performed, only 3 of the 4 relevant bytes are 
checked."

There were 9 denial-of-service as well, here's one "LAL, LSL, 
VERR, VERW for a null (zero) selector (A1 step) [Turl88]"

Quite fun reading, although I also recognizes that this kind of
attack is a bit down on the list of best cost/effort ratios.

-Christian