Re: Possibility of malicious CPUs
- From: Carlos Baiget <carlos.baiget@xxxxx>
- Date: Tue, 20 Jan 2009 17:33:26 +0100
Here is an article about this problem:
http://www.schneier.com/blog/archives/2008/10/designing_a_mal.html
Regards,
Carlos Baiget
Vesa-Matti Kari wrote:
Hello,
First of all, sorry for the poor newsgroup choice, but I
really could not find a better alternative.
I have recently given some thought to computer security.
This means a big dose of paranoia, but please bear
with me.
Suppose that we have succeeded to build an OS with perfect
security. By "perfect security" I mean the following:
1) There are no errors in the design of the security model
2) There are no errors in the implementation of the security model
3) There are no errors in the entire code development chain (e.g
build utilities, compilers, linkers) so we get perfect
executables from our perfectly designed perfect implementation
How safe would such a system be?
Descending to the hardware level, we could next investigate
the CPU. For example, suppose that we would examine a
modern Intel or AMD processor, with the protected mode and the
privilege rings 0-3.
Could it be possible that the CPU itself were malicious?
By "malicious" I mean that for those-in-the-know it would be
possible to bypass the CPU's security measures because the
CPU was designed that way from the beginning.
Suppose a CPU contains a mechanism similar to the one in
combination locks. Using machine code, you could feed the
CPU a secret sequence of undocumented OP-codes and bypass the
security at will. After "unlocking" the CPU, you could run
privileged instructions, have free access to all the memory
and so on. Game over.
My question is: How trustworthy are the CPUs that we
constantly use and trust? Any thoughts?
I know next to nothing about the low-level details of the CPU
implementations, so I cannot assess the feasibility of the
malicious CPUs I have described above.
Best regards,
vmk
.
- Follow-Ups:
- Re: Possibility of malicious CPUs
- From: Vesa-Matti Kari
- Re: Possibility of malicious CPUs
- References:
- Possibility of malicious CPUs
- From: Vesa-Matti Kari
- Possibility of malicious CPUs
- Prev by Date: Re: Possibility of malicious CPUs
- Next by Date: Re: Possibility of malicious CPUs
- Previous by thread: Re: Possibility of malicious CPUs
- Next by thread: Re: Possibility of malicious CPUs
- Index(es):
Relevant Pages
|