fsck on FAT32 filesystem?

Wojciech Puchar wojtek at wojtek.tensor.gdynia.pl
Thu Jul 19 11:10:32 UTC 2012

>> for very old drives it may not
> Would you be so kind as to point out the proof of that statement?

sorry but i didn't save that article on hard drive. So no proof if you 
don't believe me i've actually read it.

The main point is that you have

- track
- intra-track gap
- finite precision of writing head positioning.

When you write on track second time, head isn't positioned exactly as 
before so very thin stripe of previous recording remain.

With sophisticated enough tools you may recover it, requiring like 10 
times smaller head than normal.

With modern drives size of magnetic domains are larger than this 
imperfection. If drive record properly this "stripes" of leftover 
recording are just too small to be stable.

Even if it would, no hardware exist to do this, except maybe scanning 
electron microscope which would take years to scan whole surface of disk 

Not sure if it can "see" surface magnetization as i don't precisely know 
how such microscope works.

But i know it needs some time to scan even tiny thing.

> Please provide an address or location where the documentation
> supporting that statement can be found. By the way, "NOT READABLE" is
> not equal to "UNRECOVERABLE".
yes i know the difference.

Finally i am not sure if "bulk erases" can actually erase drives, for sure 
they can destroy disk electronics so disk appears cleared.

The field needed to clear modern magnetic media are just enormous. They 
are enormous under normal operation of disk, but power is small as track 
width is defined in nanometers.

If it would be my data to be erased i would just do
dd if=/dev/zero of=disk bs=..

or if really paranoid then after this

dd if=/dev/urandom of=disk bs=..

or if very paranoid then will just put that drives into fireplace, 
which would heat them over curie point which would definitely 
demagnetize whole media.

more sure than bulk eraser, and definitely secure, for free.

More information about the freebsd-questions mailing list