5 ms·
"There is no practical way to 100% ensure that every piece of software, and every chip, in every piece of equipment is clean" If we really cared about security
by ClumsyPilot 5y ago
"There is no practical way to 100% ensure that every piece of software, and every chip, in every piece of equipment is clean"
If we really cared about security and reliabilty, all critical infrastructure would be open source, both software and hardware, so that multiple suppliers could produce the parts.
Instead we choose 'free market' with a thumb on the scale. This does nothing to protect from random hackers and criminals, ignorance, sloppyness and lazyness.
- titzer 5y agoI agree 100%. But we don't really care about security and reliability. The political system is both corrupt and poorly incentivized. It doesn't reward robustness, but saving money on paper while really, truly, funneling money to cronies[1]. [1] Corruption doesn't look like mobster movies. It's mostly backyard barbecues, buddies helping their buddies' business grow by throwing them a bone. At so many different scales.
- dexterous 5y agoLike I've always said, if there's something to be learned about society, you'll learn it in Star Trek. "Mr. Worf, villains who twirl their mustaches are easy to spot. Those who clothe themselves in good deeds are well-camouflaged." - Picard and Worf, discussing both the investigations and the misguidedness of Admiral Satie Season 4 Episode 21 The Drumhead https://memory-alpha.fandom.com/wiki/The_Drumhead_(episode)#Memorable_quotes https://memory-alpha.fandom.com/wiki/The_Drumhead_(episode)#...
- skinner927 5y agoThere’s really no way to ever be 100% sure of hardware. Sure we could use FPGAs for all hardware and then the hardware “source” is software, but who makes the FPGAs? There’s techniques even under X-ray to mask a rogue chip. https://youtu.be/Hzb37RyagCQ https://youtu.be/Hzb37RyagCQ
- ClumsyPilot 5y agoAre you 100% sure of anything? That there aren't any NSA employees bribed by the CPP? Just like Chelsea Manning and Snowden took gigabytes of confidential information right from under the US government nose, they could do the same without any elaborate techniques. Soviet spies were very successful and stole atomic bomb secrets. But let's suppose everyone working for the government or subcontracted by them is 100% loyal and CCP never manages to blackmail them. What does it take to bribe a DPD guy on minimum wage delivering your 'American produced' equipment, who can't afford a dentist and is in pain every day? Let's say you ask him to 'mistake' one package for another, and have millions at your disposal? You talk about "techniques even under X-ray to mask a rogue chip", but do network operators even bother opening the casing of the router to validate the motherboard has not been replaced entirely? What if it's networking equipment where China replaced a single chip, who is ever going to find out? Does the network operator validate firmware of every single chip? We have security holes all over the place, this talk of 100% certainty is basically fantasy.
- sekh60 5y agoI think we need to do defense in depth more. There was a funny pic in one of the sysadmin subs about buying firewalls from everyone and putting them in a chain, with each firewall labelled with which nation state compromised it. The joking idea being that sure, the U.S. has backdoors into your Cisco gear, but it you put that behind a Huawei firewall they may not have the backdoors to both, add in a Russian firewall and then an intruder needs all three. Funny but I think not so ridiculous these days if you are paranoid.
- ineedasername 5y agoSecurity folks recognize that 100% certainty is impossible. Instead, the problem is always trying to balance the acceptable risk for each application of tech against the increased costs for an extra "9" of confidence. Although I suppose the minimum should always make the adversary jump through enough hoops to deter opportunistic & medium effort attacks, along with close monitoring for rapid detection of anything that gets through. In this case, it's pretty low hanging fruit to make a policy w/ financial backing that says "we won't use hardware from companies controlled by slightly hostile countries with incompatible values".
- andreilys 5y agoDidn’t the heart bleed exploit originate from an open source library? I don’t think open source is the panacea you think it is, exploits will exist as long as they’re profitable and state actors have incentives to stock pile zero days. My guess is it’ll make it even easier to identify zero days if everything was open source.
- titzer 5y agoThey've been making this same argument for as long as I can remember, and it keeps being false. Open source means vulnerabilities get found and patched much quicker and people bring large-scale software bug-finding techniques to bear on the problem. For really robust software, we need better languages, better verification, and more and more testing and tools. All of which work best on source code.
- Bayart 5y agoThe immediate effect of Heartbleed was the OpenBSD folk [1] and Google [2] forking OpenSSL. There's a talk from Bob Beck of OpenBSD on pruning OpenSSL, it's pretty hilarious [3]. In that case open source was at least able to react appropriately, even if it didn't act preemptively. [1]: https://www.libressl.org https://www.libressl.org [2]: https://boringssl.googlesource.com/boringssl/ https://boringssl.googlesource.com/boringssl/ [3]: https://www.youtube.com/watch?v=GnBbhXBDmwU https://www.youtube.com/watch?v=GnBbhXBDmwU
- JohnBooty 5y ago"Open source" hardware would be a step forward, but how do you know that the hardware on the chip(s) matches the published design(s)? You'd have to randomly sample some statistically significant number of the chips in the distribution/retail/whatever pipeline and, I guess, scan them with the appropriate tools? And verify them that way?
- alfalfasprout 5y agoIt's worse than that... functionality testing wouldn't be enough. Short of putting every chip under an electron microscope you can't really guarantee it.
- worik 5y agoRandom sampling would be sufficient.
- hypertele-Xii 5y agoAnd the good news is you can sell each sampled chip at insane markup, because those few are guaranteed.
- JohnBooty 5y agoI'm sure somebody can correct me if I'm wrong, but I'm quite certain that scanning the chips with an electron microscope would destroy them. You'd need to decap the chips, and so on. But, perhaps there's a nondestructive way.
- ClumsyPilot 5y agoCan you guarantee that the router delivered to you actually came from CISCO and has not been modified in any way? What if some chip was replaced? Short of putting every chip under an electron microscope you can't really guarantee that either
- JohnBooty 5y ago
- eps 5y agoOpen hardware specs don't prevent spec-compliant chips from carrying some "extras".
- imoverclocked 5y agoEven if the hardware is only what is spec'd in the Open Source model, the model itself has a lot of complexity for bugs to hide in. Those bugs might be intentional, they might not be. The same obviously goes for software. Even really-high-profile open-source security suites have seen critical bugs that have been around for many years. Being open source is a good start but it's hardly sufficient to guarantee anything about the supply chain. The opposite may actually be true in some cases.