Chipwhisperer lite on kickstarter (Hackaday 2nd price winner)

[all_repair]

Just want to shout out to let you know that chipwhisperer having a kickstarter campaign now. 

https://www.kickstarter.com/projects/coflynn/chipwhisperer-lite-a-new-era-of-hardware-security

[Bud]

board is assembled right here in Nova Scotia, Canada. This means you are getting a high-quality product that was built ethically - no concern about supporting questionable labour standards or companies ignoring environmental rules.

Cmon, what a gimmic this statemeng is. Have not seen anyone would bang their head against the wall feeling guilty after sending Gerbers to China.

[zapta]

I like this picture. Mechanical strength of SMD USB connectors is often ignored and they get peeled with moderate force.

[coflynn]

Thanks for the kind comments (I'm behind the project), if you've got questions feel free to let me know!

Cmon, what a gimmic this statemeng is

At least I refrained from adding GMO-free to the description ;-)

But the PCB manufacturing is quite dirty (especially the electroplating process). I highly doubt companies are going to put as much effort into separating output waste without being forced to by government regulation - it would just cost a lot more money without being able to show anything to the customer. I know Apple's PCB supplier was hassled for example, so it's not just little fabs doing it.

I still use China for PCB manufacturing quite a bit too due to the much lower prices (so yeah I'm not a white knight here), but if possible for a lot of reasons prefer to use North-American companies. This was echoed by my assembly house, which tried using overseas PCBs a number of times, but had enough reliability issues there was no net savings.

The better control over environmental regulation is basically a side-effect of using the preferred fab for my manufacture. Of course it's entirely possible for a overseas-based fab to have better controls than a North-American one. But I would bet that two randomly picked fabs would see the overseas one with a worse track record.

Mechanical strength of SMD USB connectors is often ignored

Interestingly it seemed a lot of people don't use the through-hole/SMD combo parts. In this case I'm using one of those, which gives me a lot more mechanical strength. The prototype boards have worse strength than the production versions will have, since I haven't used proper plated slots for the legs (they are rectangular legs), just large holes that fit the legs, but with tons of extra room.

There's a slight cost increase in the production run by adding the through-hole aspect, but considering how annoying it would be to break off the micro-usb connector it's well worth the minimal extra cost!

EDIT: Edit to notify me of replies

[ferrix]

Hi Colin,

Neat project and I backed it after listening to your Amp Hour interview. Looking forward to having a play with some AES one-time passcode authentication devices.

[coflynn]

@ferrix: Thanks very much for your support!

Should be a fun project - I'm hoping it surprises a ton of engineers/designers with how easily they can break AES and other algorithms. And maybe people will finally take all these attacks seriously.

[snoopy]

Thanks for the kind comments (I'm behind the project), if you've got questions feel free to let me know!

Cmon, what a gimmic this statemeng is

At least I refrained from adding GMO-free to the description ;-)

But the PCB manufacturing is quite dirty (especially the electroplating process). I highly doubt companies are going to put as much effort into separating output waste without being forced to by government regulation - it would just cost a lot more money without being able to show anything to the customer. I know Apple's PCB supplier was hassled for example, so it's not just little fabs doing it.

I still use China for PCB manufacturing quite a bit too due to the much lower prices (so yeah I'm not a white knight here), but if possible for a lot of reasons prefer to use North-American companies. This was echoed by my assembly house, which tried using overseas PCBs a number of times, but had enough reliability issues there was no net savings.

The better control over environmental regulation is basically a side-effect of using the preferred fab for my manufacture. Of course it's entirely possible for a overseas-based fab to have better controls than a North-American one. But I would bet that two randomly picked fabs would see the overseas one with a worse track record.

I think the "bring back the jobs to the US" campaign is starting to raise eyebrows in other countries and rightly so. It's one way to help a flailing economy.

Will this device be able to read back configuration data in CPLD devices ? I sort of hope not 

cheers

[coflynn]

Will this device be able to read back configuration data in CPLD devices ?

It's main focus is really crypto, not fuses or anything. It's also setup as labs/training mostly, so learning how the attacks work for example.

There was an article posted a long time ago about fuses in micros at http://www.cl.cam.ac.uk/~sps32/mcu_lock.html . He mentions in the article CPLDs tend to have better security, so I've no idea if they might be harder to break? I've never tried any of that stuff so can't knowledgeablya comment.

[Kjelt]

Hi Colin,
great project but also pretty scary for companies that protect their IP with AES and PSKs. So I have seen your youtube movie about 120 second AES128 hack it looked very impressive as you meant it to be. Then I wonder in real life with a micro that does tons of other stuff and now and then uses AES128 to decrypt something, how easy is it for you to retrieve the key using your tools? Are you really saying that no AES PSK implementation is safe anymore from now on?

[coflynn]

Provided you can cause the decryption, it's not to difficult to pick out where AES is happening, especially if it's software-based. AES hardware accelerators are harder to break, but not by a significant amount (i.e. requires 4000 traces instead of 40). This is all for generic devices, there is specially protected devices which are much much harder to break but they are not normally used.

I've got an example of breaking an AES-256 bootloader (it's my own bootloader so I'm not picking on a specific implementation, but it's pretty close to what a lot of people do), here's a few versions of the description:

* Hackaday blog post: http://hackaday.io/project/956-chipwhisperer-security-research/log/10108-aes-256-is-not-enough-breaking-a-bootloader
* Academic Paper: https://eprint.iacr.org/2014/899.pdf
* Tutorial: http://www.newae.com/sidechannel/cwdocs/tutorialaes256boot.html

No AES PSK implementation has been safe for 14 years basically (side-channel power attacks were introduced in 1998, before AES was a standard even). Unfortunately very few designers know about these attacks, so are basically ignored during system development. I'm hoping to change that viewpoint, as they have *always* been possible for attackers.

As an example Xilinx bitstream encryption has been broken on most products several years ago (https://eprint.iacr.org/2011/391.pdf), and Altera bitstream encryption also has been broken (https://www.emsec.rub.de/media/crypto/veroeffentlichungen/2013/01/11/StratixIIDPA_2.pdf). But I doubt many people using those parts are aware of this, so trust the encryption to completely protect their IP.

The solution is to pressure the silicon vendors to use more secure implementations - Microsemi for example has countermeasures in their products now, after a paper came out breaking their secure product line.