Hacker News new | past | comments | ask | show | jobs | submit login
A laptop hard drive model could crash when exposed to a certain audio frequency (microsoft.com)
256 points by drKarl on Aug 18, 2022 | hide | past | favorite | 65 comments



And don't even think about standing up from your office chair either when using Dell DisplayLink 3100

https://support.displaylink.com/knowledgebase/articles/73861...

https://www.youtube.com/watch?v=voW5kEI7JKE


I actually have this! It was driving me crazy WHY this happened. I do not have a DELL monitor but some AOC Gaming widescreen thing but to make it more weirder: it only occurs when I am using my laptop not my desktop. Both systems are connected using HDMI to the monitor.

Only when I am working on my laptop and I have audio over my HDMI cable using my speakers on my monitor and I walk away my monitor blanks and the sounds stops for 5 seconds...


Protip: You can just buy ferrite beads and stick them on cables. They aren't even super expensive. Don't put them all the way to one end but a bit back. Use one on both ends if need be.


I guess your desktop is earthed


We had that in the office around 2015-16 with one certain combo of chair model + screen model, but I don't remember the specifics.

You actually had to let yourself fall into the chair though, it did not trigger when sitting down or standing up.


Yes! For the longest time I could swear my display blinked every time a colleague stood up.... people looked at me like I was crazy....


That's interesting. Been getting the same behavior from my Gigabyte M32U monitor from things like disconnecting a different monitor's HDMI cable to my Switch, even though the Gigabyte is only connected to my PC. This seems like the most obvious explanation.


Recently I bought a chair and my Bluetooth dongle stops working sometimes when I stand up. Perhaps something like this happend for me..., but my new chair doesn't have elevating mechanism.


> When people stand or sit on gas lift chairs, they can generate an EMI spike which is picked up on the video cables, causing a loss of sync.

What is the mechanism here?


Just taking a guess, rubber gasket on metal cylinder are maybe like a poorly optimized metal brush/rings on rubber belt as in a Van de Graaff generator, used for generating static charges and discharge can be used for generating xrays.


So this is what those $800 dollar shielded hdmi cables are for.


> It turns out that the song contained one of the natural resonant frequencies for the model of 5400 rpm laptop hard drives that they and other manufacturers used.

> The manufacturer worked around the problem by adding a custom filter in the audio pipeline that detected and removed the offending frequencies during audio playback.

Does this mean that every machine with a hdd has the same problem? Is it just about finding the right song or do engineers include some kind of dampeners to prevent this?


In fact, almost everything has this problem.

There seems to be (someone more knowledgeable please correct me) no way to avoid having a resonant frequency. The only thing that can be done is to change it.

The resonant frequency of a glass can shatter it. The resonant frequency of a room can make it uncomfortable to be in. The resonant frequency of a bridge can tear it apart.

We even have to take special precautions between 2.4GHz wifi and processors running at a 2.4GHz clock speed because they are capable of hitting the exact same frequency.

As someone pointed out, the difference in manufacturers would have made the resonant frequency slightly differently between the specific drives.


One example of a precaution taken: Spread-spectrum clocking.

Perfectly clocked signals have really high EM spectral peaks (noise radiated at particular frequencies). You can intentionally make the clock a little worse, spreading that radiation over more of the spectrum, and it will reduce the chance that those peaks break other things.

https://www.microsemi.com/document-portal/doc_download/13543...


I had a DVR (digital video recorder) that was constantly recording the video feed into hard drive. It was in my bedroom. One day when I was cleaning I moved to box and didn't think much of it. Then in the evening I was getting ready to sleep and on the side of the bed where I sleep I could hear extremely loud boomy sound. It was driving me crazy, but as soon as I moved to find the source the sound "disappeared". Going back to bed and it's back again. I was going around the room like crazy trying to find the source, or maybe it was outside? Eventually I put my ear to DVR and noticed the sound was quite similar, but very faint. By doing this I moved the DVR and when I returned to bed the sound was gone. This was so frustrating. But I least I learned something how to find a source when something like this happens next time. Basically the unlikely positioning of DVR, my furniture and myself created loud resonance.


I'm curious if this is really the answer and not something like sound tunneling and amplification.

Like when you stand somewhere further away and there is a spot were you can very clearly hear the people talking across the street or on a balcony


Natural amplification (especially through materials, which is what I would guess happened here) is resonance. It’s wild how pervasive it is.


An excellent demonstration of this is The Whispering Arch in Liverpool Anglican Cathedral or The Whispering Gallery of St Paul's in London.


>The resonant frequency of a glass can shatter it. The resonant frequency of a room can make it uncomfortable to be in. The resonant frequency of a bridge can tear it apart.

I love the roller-coaster of magnitude changes in the consequences listed: "shatter, unpleasantness, tear apart"


> The only thing that can be done is to change it.

You can also dampen the resonance, to limit its response at the resonance frequency.


Yes, but that’s a mitigation technique and not about eliminating the resonance frequency itself.


A famous example is the old Tacoma Narrows Bridge[1].

[1] https://en.m.wikipedia.org/wiki/Tacoma_Narrows_Bridge_(1940)


That's just so cool.

And probably very annoying for a lot of people.


There’s a whole chapter in Godel Escher Bach specifically about this! (OK, so it’s supposed to be a metaphor.)



What are you referring to?


Audiophiles hate this trick


The weakness is in the specific size and shape of the drive hardware, so the exact resonant frequency would vary a bit.


But for the specific production run of hard drives combined with the specific production run of laptop casings etc, with today's manufacturing tolerances, it will be a problem.


Doesn’t that mean hearing the song from another source will still break the hdd ?


> And then they discovered something extremely weird: Playing the music video on one laptop caused a laptop sitting nearby to crash, even though that other laptop wasn’t playing the video!

Yes


Yes I read the article so I was clarifying whether or not they fixed that problem.


Seems like it would have to be vulnerable to it still, right? If their "fix" was to filter audio output. They can't filter everyone's laptops..



With the CVE date being a day after this blog post, and ~17 years after the actual vulnerability.

Filing a CVE for this seems like a bit of a cheap gag (and i enjoy cheap gags). Is there any real practical value in filing a CVE with this little detail, this long after the fact?


Maybe manufacturers will check to see if they are still vulnerable or enterprises using old technology. If it were just a blog post, nobody would probably know about it.


  Historical records should never be underestimated just by assuming they don't look too "practical" nowadays.


Moral: Computers are hard



_electronics_ are hard. I always remember back to one of my electrical engineering labs... The oscilloscopes were sensitive enough that someone walking past them was visible on the scope. Really, any motion near it. It was especially funny when we were trying to figure out why a particular experiment wasn't working (we couldn't pick up any signal from our circuit at all), yet we could pick up noise from movement around the scope. Turns out the scope had a faulty BNC connector... It was very frustrating at the time, but many humorous memories in the decades since.


> And I’m sure they put a digital version of a “Do not remove” sticker on that audio filter. (Though I’m worried that in the many years since the workaround was added, nobody remembers why it’s there. Hopefully, their laptops are not still carrying this audio filter to protect against damage to a model of hard drive they are no longer using.)

The audio filter is Chesterton's fence

https://en.wikipedia.org/wiki/G._K._Chesterton#Chesterton's_...


Somewhat frustratingly, the developers could have avoided it becoming a Chesterton's Fence by putting why the (metaphorical) sticker shouldn't be removed on it.

I feel like there's a lot of times when people make labels/verbal advice/laws/comments/whatever saying "don't do X" without saying why you shouldn't do X when there was no reason not to include that information.


I wonder if the filter was embedded in the firmware of the CODEC's internal DSP (as it should) or simply done via software driver. For the latter it would mean the user would have to download the custom driver from the laptop manufacturer's website to get the fix as a generic driver would not have it. This would also mean that people using Linux distros on that laptop would not get the fix as well.

Though in this case the filter would not prevent crashes if the song was played externally, like from another device or over a PA system.


Specifically Rhythm Nation.

You might be wondering "does it really matter which Janet Jackson song it is? Is that relevant to HN, or even the CVE?".

...

You're welcome.


It’s considered irresponsible disclosure to publish the hit track proof-of-concept prematurely. You should have waited another 5-10 years.


It's already been 208 years gimme a break!

https://en.wikipedia.org/wiki/Janet_Jackson%27s_Rhythm_Natio...


In my opinion, if a disclosure by a person takes place within their lifetime then it’s too soon.


Well, the specific model of HDD hasn't been mentioned yet as far as I can tell.


A phone we were developing could be made crash by taking a photo of it with a Xenon flash camera


The photoelectric effect (where metals emit electrons when hit by light); see https://www.raspberrypi.com/news/xenon-death-flash-a-free-ph...


You don't even need that. Xenon flash tubes require quite a powerful surge of high voltage electricity, and it's common for them to emit considerable amounts of radio-frequency interference.


EEVBlog made a video demonstrating that and explaining it generally: https://youtu.be/SrDfRCi1UV0


This happened to an earlier version of Raspberry Pi


Yes the phone used some of the same Broadcom chipset as the (original, I think) Raspberry Pi


Nikola Tesla had claimed that a resonance oscillator that he had built was capable of causing earthquakes.

See: https://en.wikipedia.org/wiki/Tesla%27s_oscillator

Even though MythBusters team tried to replicate this in one of their episodes and concluded the claim to be false, the OP shows that resonance is a real problem with spinning disks.


Shouting at disks is also a bad idea: https://m.youtube.com/watch?v=tDacjrSCeq4


This is much more credible than saying that the laptop would "crash".

During XP era all HDDs already had some builtin error correction. Unless the "song" was affecting the electronics rather than the mechanical parts (which is not impossible but much more unlikely), the only effect the user of the laptop would see is an increase in the correctable read error rate (since it's also highly unlikely that the song would prevent the head from ever reading the correct value 100% of the time), and perhaps a delay due to a retry (which would then always read the correct value, again because of the previous reason).


Technically every laptop will crash if you play a sound loud enough


For solid state machines I think nearby humans would crash first!


I REALLY appreciate the format of this post. Clear enough for a non-technical individual to understand and brief enough to nearly guarantee being worth the time.


Reminds me of the self-destructing record for the "perfect" record player from Gödel, Escher, Bach.


Wow, did dang really change the title AND the url of this post? Didn't know that was possible!! This post was pointing to mitre.org, now it points to Raymond Chen's blog...


A mod changed the URL and title as the CVE (https://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2022-3839...) seems to be largely a gag, and the title formulated by the submitter (as the CVE doesn't have another title) was baity.

Btw, this is in the site guidelines:

"Please submit the original source. If a post reports on something found on another site, submit the latter."

https://news.ycombinator.com/newsguidelines.html


The reason I think that matters is that I had originally read the blog post, but I thought it was interesting on its own that there's an actual CVE published!!


checks date

No, it's not April 1st.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: