Song from 1989 declared a cybersecurity vulnerability for crashing hard drives | Hacker News Reader

Song from 1989 declared a cybersecurity vulnerability for crashing hard drives

techspot.com

123 pointsquyleanh53 comments

The vulnerability and the fix remind me of a chapter in Gödel, Escher Bach where Achilles gives the tortoise (or maybe the other way around) a record that, when played, destroys his record player. So fun to see that IRL.

That actually illustrates why it’s a bad analogy for communicating the concept: it’s not intuitive at all why record player could be destroyed by playing the right record — most have volume limits that make this very difficult in practice, and most people never see one even getting close enough to see the dynamics that would cause it to happen.

And so it definitely doesn’t help communicate the broader idea that all formal systems (meeting some minimal criteria) should have a corresponding flaw.

(I would have gone with a helicopter as the system but that would require domain familiarity.)

> I would have gone with a helicopter as the system but that would require domain familiarity

Would you give it a try?

Well, there isn’t much to add beyond the fact them being more prone to acoustic breakage. Helicopters are a nightmare for vibrations because you have to spin the rotor really fast and forcibly, risking over-stressing critical structural components if you excite them at their resonant frequency. They’re designed to operate at only one rotor speed because that’s enough of a headache in terms of preventing it from hitting resonant modes that the structure can’t handle.

So for helicopters it’s much easier to imagine a pattern of rotor movement that will break the system — though again, getting the intuition down would require having modeled vibratory systems to see when they go out of control.

But at least it’s more intuitive than record players, which work with much smaller vibrations and generally operate in moderate volume limits.

I wonder why they didn't have Achilles give out a special rotor for a helicopter that would have caused the entire helicopter to break. It seems like a great analogy for people who have the required domain knowledge as well as think it would be reasonable for someone to gift someone else a helicopter rotor.

This reminds me of the Donald Duck cartoon in which for some reason some of the Duck family have acquired a steam calliope (kind of like a somewhat portable pipe organ), and are trying to figure out what to do with it. Whereupon they meet a strange hermit who lives in a cave, or something, and who says that he has absolutely everything he needs in life, and there is nothing they can offer him that would be of any value.

Except for one thing.

Of course, it turns out to be a steam calliope.

Similarly, I suppose that, sometimes, when you really need a helicopter rotor, you really need a helicopter rotor!

Did you miss that the analogy would have “rotor input” instead of a replacement rotor, and that I agreed it would require some domain knowledge, or is this just you adding toxic snark to look clever and discourage substantive contributions to these threads?

In any case, the record player analogy is still unhelpful because it conveys no practical intuition about why a record might break its player — that doesn’t happen in practice. Unless you want to make a substantive remark by showing an everyday case where a normal record might come close to breaking a normal player.

Giving someone a rotor input seems unrealistic. I've never heard of someone gifting one and who would swap out a rotor input with one that would cause damage to the helicopter? I think readers would find that example too farfetched.

First of all, I already agreed in the initial post that it would require some experience with helicopters. Second, in this case, "giving it a new input" could be as simple as "changing the standing rotor speed -- 5500 RPM instead of 6500 RPM". They're so finicky, you don't need to specify a complex time history to hit a dangerous resonant mode-- though it certainly helps!

Third, the fact that it's a gift is not at all required for an analogy. Of course you don't have to keep that part, and would have to reframe it so that the details of how the helicopter is provided a new input are different. You're not even trying to be charitable at this point, but make it sound as stupid as you possibly can. Please work on your curiosity.

It does explain it in the book in a fairly tortured way - its a perfect record player than plays things perfectly and can play anything and that makes it vulnerable. A worse record player would not be vulnerable.

Zip drives had something like this. A malfunctioning drive could damage the removable discs in such a way that they would then damage other drives they were inserted into, then the cycle repeats...

https://en.m.wikipedia.org/wiki/Click_of_death#Iomega_Zip_dr...

OMG recall having a zip and 1 gig IOMEGA drive in your tower? and how cool you thought you were

Especially when that tower was a licensed mac clone SuperMac. I don't think our Jaz drives ever really worked.

I kinda always thought Zip was a kludge technology, and I thought I was way cooler because I used magneto-optical instead. (I kinda still do, because those discs are still working, 937 years later.)

The mechanism of damage is different. The faulty disks aren't vibrating the drive to death, the drive is head-banging[0]. Iomega cost-reduced the drives to a fault; removing a tiny piece of foam that is there specifically so that head-banging doesn't damage the drive head.

Head-banging is an intentional feature on disk drives[1] to correct for read errors and get back to a known good state; so a corrupted disk would cause it to head-bang every time you tried to read from it. And since there was no cushioning on the drive rails each head-bang would misalign the drive... making it corrupt more disks!

I'm not sure what the record player equivalent to this would even be. A very heavy stylus could damage the disc and carve into it, but there's no way you could have a disc permanently change the tracking weight of that stylus.

[0] When a disk drive intentionally crashes its head against a mechanical limit to ensure position of that head in lieu of having an actual sensor

[1] The most egregious case being the Apple ]['s 5 1/2" drives, which intentionally do this on every power-up.

> I'm not sure what the record player equivalent to this would even be. A very heavy stylus could damage the disc and carve into it, but there's no way you could have a disc permanently change the tracking weight of that stylus.

Conceptually, you could have a record made of "tar", which got scraped off onto the stylus and was then impossible to clean off.

I mentioned this when this was posted last week.

https://news.ycombinator.com/item?id=32507898

The Tortoise gave the phonograph record to the Crab.

(The Crab had a record player which was claimed to reproduce any sounds.)

Thank you for the correction and expansion. I didn't have the book with me at the time so I was just going from memory. This must have been the section titled "Crab Cannon".

Quite a few submissions on this recently.

"Janet Jackson had the power to crash laptop computers" https://news.ycombinator.com/item?id=32483211

(This is based on the original Raymond Chen blog post.)

Of course, I'm reminded of Brendan Gregg shouting at hard drives:

https://www.youtube.com/watch?v=tDacjrSCeq4

Hard disk drives are not dissimilar from other high-precision acoustic systems. The actuator is even called a "voice coil."

That we are able to transmit a range of frequencies with sufficient accuracy and precision to induce a magnet to move a near-microscopic coil of wire at the end of armature flying microns above the surface of a platter spinning at a rotational velocity of thousands of RPMs and a linear velocity of inches per millisecond, land at exact locations, and detect the polarity of a magnetic field… I can't even. I don't even know what to compare it to.

Last time someone posted this he showed up in the comments lol

https://news.ycombinator.com/item?id=32513240

We have a word for that.

>MAGICK

A lot of signal processing/information theory stuff crosses over pretty seamlessly between hard drives and radio transmission, too: https://www.schrankmonster.de/wp-content/uploads/2019/12/Scr...

The presentation that (deleted) tweet screenshots: https://www.iaria.org/conferences2015/filesAICT15/AnIntroduc...

Compare it to … Havana syndrome?

or Apple: https://www.youtube.com/watch?v=C5d151lqJsA&t=150s

I was setting up a trade show in Zurich when Janet Jackson was soundchecking Rhythm Nation in the next building. None of our hard drives crashed but that was in 88 or 89 so I guess maybe the vulnerability showed up much much later.

The live version may not have that exact frequency. Or the acoustics of the hall and the audio system may have acted as a filter at that frequency.

But it might be more likely that the hard disks just weren't vulnerable, yeah. (5400rpm in '89, was that common?) Otherwise, if they did play the original song, then the loudness could have made the effect even worse.

All true. What I remember was just how damn loud it was. We were quite a bit away and you could feel it in your bones.

There's a story (not confirmed) about a Black Team member using resonance to crash (physically) a tape cabinet: http://www.penzba.co.uk/GreybeardStories/TheBlackTeam.html

I've been so disappointed in the reaction of tech sites to this story. It's a third-hand anecdote about an unspecified model of hard drive from fifteen years ago which might be complete bullshit but I'm seeing links to articles about it one or times a day for a straight week.

I ran across it randomly the other day (or so I thought) and invited the submitter to repost it. If I'd known it was a repetitive thing I might have hesitated. Not their fault of course.

Edit: looks like it had at least 2 recent threads:

A laptop hard drive model could crash when exposed to a certain audio frequency - https://news.ycombinator.com/item?id=32507147 - Aug 2022 (65 comments)

Janet Jackson had the power to crash laptop computers - https://news.ycombinator.com/item?id=32483211 - Aug 2022 (9 comments)

Yeah, and that's just HN. Google News, suggested stories, my Firefox startup suggestions, Youtube suggestions, I cannot get away from this nothingburger.

Definitely sorry for inviting the nothingburger.

This reminds me of a similar experience I have from a time when I was briefly working in the field on machinery containing huge (multiple megawatts) industrial motors. We had finished doing some work on the control electronics, and the plant operator warned us to turn off our laptops before he started up the main motor because he had previously noticed they would always BSoD and occasionally completely fail to boot afterwards due to disk corruption. This was in a room far from the motor, and it was not very loud even without hearing protection, but the vibration was still enough to hit a resonance in the HDD.

Of course, that was so long ago that this "CVE-chaser" culture of paranoia had not yet appeared.

FTA, originally from https://www.theregister.com/2022/08/18/janet_jackson_video_c...

Yeah, link should probably be changed to that.

I initially wondered why this is not just pointing to Raymond Chen's excellent blog (https://devblogs.microsoft.com/oldnewthing/20220816-00/?p=10... ), which is also linked by both The Register and Techspot, but the The Register article adds that it's not officially a CVE.

I had the "vibration" problem a few times in me IT career, train, construction work and once elevators.

It is the kind of issue that are really hard to diagnose the first time but after the first time it's part of the standard "diagnose suite".

I love to show this video about the effect of noise on HDD (be sure to put the volume down) https://www.youtube.com/watch?v=tDacjrSCeq4

Has someone demonstrated this? It was immediately my favorite CVE but without poc and based on a single story I'm quite skeptical.

Whilst not precisely the same, you do have the famous "Screaming in the Data Centre" [0] from over a decade ago.

[0] https://youtube.com/watch?v=tDacjrSCeq4

So if you stand in Times Square with a boombox and play this song at full volume, does that constitute a cyber attack?

I'd consider it social engineering in this day and age.

Who _wouldn't_ want to tell you all their encryption keys and passwords after listening to Janet Jackson? ;-)

I suppose in the same way throwing a USB stick full of malware at someone would be, yeah.

Disagree, I guess. Throwing a USB stick of malware at someone doesn't realistically serve the purpose of infecting their stuff with that malware. But playing this song in Times Square very much does have a realistic chance of breaking their hard drives, if said drives are old enough, etc.

I think it's actually more likely that someone will pick up a USB stick and infect themselves than they will be using a '88 5400 RPM hard drive in the vicinity close enough to be affected by the resonance.

There is a repeated note (or sound effect) in the song which has a pitch which sweeps through 7200 Hz. In fact, it is reminiscent of a kind of sound that can occasionally emanate from a failing drive which spins at that rate. I wonder if they got the "5400 RPM" detail wrong?

I don't think the rotational speed of the drive is that relevant; the resonance that was being excited is certainly different as otherwise the drive would've excited it via its own rotation.

We don't actually know that this phenomenon is due to resonance.

Previously shared here, via the cve, which is a less interesting read of course.

Link: https://exploit.report/cve/cve-2022-38392/

Ah yes, Janet Jackson's infamous 'hard drive malfunction'.

> special malware that is able to encode the data to be transmitted through direct manipulation of the fan speed

So in addition to being more pleasant to use, fanless laptops (computers) are also more secure!

Not more pleasant --- to be honest, it's unsettling how little feedback there is from systems these days. Back then, it was natural to feel how the system was doing from the sounds of hard drives, the fans, etc., and malware trying to do something in the background would often result in a very noticeable change.

Reply on news.ycombinator.com