0

Huang: IRIS (Infra-Red, in situ) Project Updates

 6 months ago
source link: https://lwn.net/Articles/964956/
Go to the source link to view the article. You can view the picture content, updated content and better typesetting reading experience. If the link is broken, please click the button below to view the snapshot at that time.
neoserver,ios ssh client
User: Password: | |

Huang: IRIS (Infra-Red, in situ) Project Updates

[Posted March 10, 2024 by corbet]
Andrew 'bunnie' Huang provides an update on his IRIS infrared chip-scanning project as the starting point for a detailed summary on how chip customers can detect forgeries and modifications in general.
The technique works because although silicon looks opaque at visible light, it is transparent starting at near-infrared wavelengths (roughly 1000 nm and longer). Today's commodity optics and CMOS cameras are actually capable of working with lights at this wavelength; thus, IRIS is a low-cost and effective technique for confirming the construction of chips down to block level. For example, IRIS can readily help determine if a chip has the correct amount of RAM, number of CPU cores, peripherals, bond pads, etc. This level of verification would be sufficient to deter most counterfeits or substitutions.

(Log in to post comments)

Huang: IRIS (Infra-Red, in situ) Project Updates

Posted Mar 10, 2024 14:14 UTC (Sun) by dskoll (subscriber, #1630) [Link]

This is really cool. I suspect, however, that a determined attacker could still hide some malicious circuitry on the chip. For example, if a chip contains memory, it usually contains spare rows that can be activated by blowing fuses in case some cells are defective; this improves yield. It's not beyond the realm of possibility that a "spare row" could actually be something malicious, though it'd be tough to make it look as regular as an actual spare row.

Back in 1999, I worked for a company that reverse-engineered ICs. I created a system called Integrated Circuit Imaging System (ICIS... in retrospect, an unfortunate name) with a microscope, camera, and precision computer-controlled stage that could image an entire chip and let people reverse-engineer it at the transistor level. That was back when feature sizes were significantly larger than visible light wavelength.

Looking for your place in open source? Set up job alerts and get started today.

About Joyk


Aggregate valuable and interesting links.
Joyk means Joy of geeK