'Severe' OpenSSL vuln busts public key crypto | Hacker News Reader

'Severe' OpenSSL vuln busts public key crypto

theregister.co.uk

20 pointsbensummers19 comments

So you need physical access to the server/device, specialized power supply equipment, and then 104 hours on an 81-machine cluster of 2.4 GHz Pentium-4's to do the exploit. All of which can be solved by simply salting an error checking algorithm with a little randomness.

Their definition of 'Severe' is severely lacking.

Well, it is the register after all.

Wish they still had the about page that said "We are the corruptibles,...", good times.

Yeah, those were fun times when they were on 64kbps! I have a "Integrity, we've heard of it" t-shirt :)

Well you obviously missed this part:

"The scientists are also experimenting with the possibility of exploiting the bug using lasers or natural radiation sources, they said."

Where once I could bring my laser or a large piece of radioactive rock to work, now I will be met with suspicion, even outright hostility :-(

el'reg love to try and make out interesting exploit stories as "we're all going to die".

Which is a shame because it is an interesting story.

Dan Goodin is one of the good guys. No tech pub, Ars included, is staffed with people who can independently validate stories about crypto. Goodin at least tracks down multiple sources for his stories.

True; but they do have a tendency to big up these vulnerabilities more than other sites (in this case only in the title, it seems, but still...)

When you've (a) got a bunch of UMich researchers telling you it's big, (b) got the OpenSSL team telling you it's important enough to merit an immediate fix, (c) got Karsten Nohl telling you it's big, and (d) it's in SSL, then you're going to run a story saying it's big.

Look, it's not big. We probably all get that. But there's nothing negative you can really say about the Register for how they ran this story. I think they do a better job with security than almost any other tech pub.

Well, SSL is based off of the concept of a web of trust, which means that if you obtain a root SSL cert (like those held by Verisign, etc) you can perform man-in-the-middle attacks on countless sites. So while the requirements for successfully mounting the attack are very high, so are the potential rewards.

SSL is based on trust anchors (PKI), not web of trust (web of trust is PGP).

"on the" not "off of the"

I assume that you're talking about obtaining a root SSL cert private key.

In order to do this using this exploit, you'd need physical access to a machine that has that private key stored on it already (presumably a few very high security servers at Verisign, etc).

I'm pretty sure that you'd find it easier to compromise such a server remotely using some other attack not discussed here. I'm also sure that companies holding such a key take their security very seriously.

In short, I don't think that this side-channel attack is even close to being applied to anything but a DRM system; certainly not to a root SSL cert.

This kind of attack is not useful for taking on a server, but, for example, it could be used to completely break HDCP. Get a couple of Blu-Ray players, use this to sniff out their keys and distribute those. Do it a couple of times and key revocation becomes infeasible because of the cost of replacing hardware.

> Do it a couple of times and key revocation becomes infeasible because of the cost of replacing hardware.

Who bears the cost of replacing hardware when a Blu-Ray key is revoked?

That's a serious question. If the key used by my Blu-Ray player is revoked, who pays to fix things so I can continue to use it to watch new movies?

Not to be cynical or anything, but I'll bet that I'm the one on the hook after the warrantee period, and maybe even earlier.

If I'm correct, key revokation is a public relations nightmare waiting to happen.

My sense is that revocation targets the software players more than the set-top boxes, but I note also that the set-top boxes are firmware-updatable.

HDCP is also not where the battle's taking place; the rippers are targeting BD+, Blu-Ray's built-in DRM scheme. BD+ is considerably more sophisticated than a simple key you can rip out of memory.

I'm getting sick of these articles. Waiting for Brother Bruce to weigh in and silence this alarmist nonsense.

So what you're saying is that you already knew that OpenSSL had a glitching-style side channel attack, and that you're waiting for someone who has never assessed OpenSSL's code to shut down news stories about it?

No, what I'm saying is that this title is completely alarmist, because the OpenSSL vulnerability that has been discovered is not vulnerable to attack past the data center from which it originates. The phrase "busts public key crypto" is completely misleading.

It busts public key crypto --- in one of its most popular implementations in both closed and open products, using a class of attacks which has devastated many other implementations --- in settings where power is under the control of attackers, which isn't even a remotely uncommon situation. You think you're debunking a story, but really what you're doing is making it sound like the only exposure you have to cryptography comes from Schneier's blog.

Reply on news.ycombinator.com