disk with high frequency noise only on FreeBSD

[Michael Powell]

James Phillips wrote:

> 
>> Date: Fri, 11 Dec
> 2009 23:52:50 +0200 > From: ly4uk Root <ly41k at ukr.net>
>> Subject: disk with high frequency noise only on FreeBSD
[snip] 
> 
> Now, this post is interesting. I'm sure many people with a software
> background may be tempted to write this report off as completely
> implausible. The truth is even "non-moving" parts such as inductors
> and possibly capacitors can move in response to an applied signal.
> For example, my ADSL modem with no moving parts makes an audible
> hissing noise louder than the (80mm) fan noise of my BSD server.

It is possible for one or more electronic components to emit a noise, even 
an integrated circuit. This is usually the result of an abnormal operating 
condition which has established a self sustaining oscillation, which 
requires some form of feedback loop to operate.

The quintessential example is the horizontal output transistor in an analog 
television or CRT style monitor. These normally operate at about 15KHz but 
not audibly. When the components around them have altered value enough to 
change bias voltages they will oscillate and produce a loud high pitched 
whine. Failure is what eventually occurs in this situation.

I had one machine that the memory would "sing" only when a make buildworld 
was run in FreeBSD. I have an old British Airways movie headphone set from 
back when their system was acoustic with air tubes. This works really well 
for examining where a sound is coming from.
 
> I have no idea what would be causing this in 8.0-RC2, but I can
> suggest what to look for: anything polling the drive in the audible
> frequency range (20 to 20 thousand times per second). Another
> possiblity is any action the repeats at that rate, but was not
> present in ealier versions. The timer interrupt is in that range, but
> other systems like GNU/Linux (before the tickless kernel) and Windows
> use a similar timer.

There are now 3 timers to choose from, and I think the default changed to 
the acpi fast timer. Interesting analysis, but very well could be related.

> To the original poster: you say this is a laptop. How do you know the
> noise is coming from the hard drive and not some other component like
> the speakers/Network card/fan?
> Regards,
> 

Yes, the noise could be coming from elsewhere. And if it is indeed coming 
from a VLSI type of chip it does not bode well. This indicates some form of 
abnormal operation which is most often eventually destructive in nature. 
There is actually probably very little you can actually do about it, so just 
live with it and "if it ain't broke don't fix it". When it is broke hit 
repeatedly with progressively larger hammers until it's in pieces and you 
now need a new one.

-Mike