Subject: Re: verified executable kernel modification committed
To: Seth Kurtzberg <seth@cql.com>
From: Gavan Fantom <gavan@coolfactor.org>
List: tech-security
Date: 11/03/2002 19:50:34
On 3 Nov 2002, Seth Kurtzberg wrote:

> That's a good point.  There are some situations, though, where this is
> undesirable.  Drives are inexpensive, but the cost is not zero.

I'm not sure such a device would have a large enough market to bring the
cost below an inexpensive drive though.

> More significantly, two drives consume significantly more power than one.

That's a much better reason. I can't profess to knowing how much overlap
there is between power-critical and security-critical applications out
there.

> You might argue that it takes a somewhat more sophisticated person to
> defeat the system (because you could check for the device in software
> and because any monkey can replace a cable), but it may be true that
> anyone who cares enough to open the box is probably capable of defeating
> it.

Snip off the write pin on the drive. It's much harder to solder a new pin
on than to just replace an IDE cable.

Of course, then it's also much harder to upgrade the OS.

CDROM springs to mind here. Sure, you may be able to replace the CDROM
with another disc, but people would start noticing when the whole set of
system binaries and configurations went missing.

-- 
Gillette - the best a man can forget