Hacker News Reader

Comment by eriknstr | Hacker News Reader

You're pretty determined when you decide to dissolve the packaging of an IC :)

Decapping an IC is simpler than it looks. Heck, anyone with access to a university cleanroom can (with practice) easily reverse a commercial IC. Understanding how it works is a different matter entirely though -- that's where microscopy and circuits knowledge come in. Analog circuits are significantly more complex as you need to characterize (measure resistance, capacitance, etc.) individual components to be able to accurately model their behavior.

Curious what your thoughts are on this:

https://reverseengineering.stackexchange.com/questions/5878/...

A modern Intel CPU is a different beast entirely. Even relatively simple ICs become exceedingly difficult to reverse at lower process nodes, so I agree with the answer. My comment was more geared towards simpler ICs at a fairly accessible process node (65 nm and older). Also, my use of the word "reverse" in no way includes the process of figuring out what each part of the circuit does in detail.

Nonetheless, at least in my field of research, the assumption is that there exist nation states with "unlimited" resources that can in fact reverse complex ICs. In reality, I doubt even the US DoD has that level of resources available, so I like to think of the assumption as an analogue to the oracle model in cryptography.

In reality, I doubt even the US DoD has that level of resources available, so I like to think of the assumption as an analogue to the oracle model in cryptography.

Actually I would look East, to China and Russia, for the latest technologies in IC reverse-engineering. I don't think nation-state levels of resources are necessary even to RE something like an Intel CPU. To actually understand how it all works is a different matter, but if you're talking about just deriving a transistor-level or gate-level schematic, or turning that into HDL, from my understanding it's actually quite affordable now. You can already recover code from most microcontrollers with <$10K USD (and I personally know others who have used those services before); I'd estimate getting a schematic to cost maybe 10x that.

Modern intel CPUs are very different beasts from your average microcontroller.

I think we can assume that if the potential gain is large enough, the US or other high tech economies (China/Japan/Korea/Russia/Singapore/France/Italy/etc) are definitely capable and willing to expend significant sums.

For example, the fragments of recent NorK missiles. If they had any computer or radar fragments left, they would be analysed, even if it cost $100M.

Also, military cryptographic processors, especially for strategic or space comms.

You might find this interesting:

https://www.sciencenews.org/article/what-donkey-kong-can-tel...

It was a post on HN a few months back

After decapping an IC, an experienced eye can usually identify patterns and make educated guesses at what each module in a circuit does, at least at a very high level. This can be helpful in and of itself, without diving into an exact understanding of how everything in the chip works.

you are determined when you unleash a technician with a FIB on the exposed chip.

And you can go to a lab and get this done for a few $100. It was exciting seeing the technician "fly" a electron microscope or something over the die and reach the exact spot we asked for and make modifications.

Reply on news.ycombinator.com