Deleting files

Question about which tools to use, bugs, the best way to implement a function, etc should go here. Don't forget to see if your question is answered in the wiki first! When in doubt post here.
Post Reply
ManOfSteel

Deleting files

Post by ManOfSteel »

Since I will begin my file system driver soon, I thought about this: why does some "file wiping utilities" overwrite a file more than once before deleting it, I mean, if you fill the entire file with zeros and than delete it, the data that the file contained before that is totally erased, right?
Thanks.
User avatar
Pype.Clicker
Member
Member
Posts: 5964
Joined: Wed Oct 18, 2006 2:31 am
Location: In a galaxy, far, far away
Contact:

Re:Deleting files

Post by Pype.Clicker »

well, yes and no.
If you extract the disk platters and put them in a more sensible device, you may still find traces of the old values because some magnets haven't been reset correctly while the majority of them were ... So the previous signal is still there but with a very smaller amplitude.
By repeating the process of zeroing, you reduce the chances non-zero bytes remain and by writing garbage after that zeroing, you avoid that a quick-scan reveals large zeroes zones (where previous data may have been written and then erased ... hum hum, let's see if we can recover something out of it }:*D )

That's a bit too technical for me to explain it comprehensively, but looking at "man shred" may give you a better idea...
Imho, it's something you shouldn't worry about since it can easily be implemented at the application level...
Schol-R-LEA

Re:Deleting files

Post by Schol-R-LEA »

This is also why most 'strong deletion' functions use more than one fill pattern (e.g., an all 1 fill, then an all 0 fill, then alternating byte all 1 and all 0 fills, then an alternating bit 1 and 0 fill followed by the same pattern rotated by one bit, then a quasirandom-bitpattern fill, then finally another all 0 fill), to ensure that every bit has been changed repeatedly relative to it's neighboring bits, which should in theory completely randomize any remaining resonance.

Of course, the people who worry about this sort of thing are usually the sort who put faraday cages around their workspaces and don't read any unencrypted e-mails... which isn't to say that such paranoia is never warranted, just that's it's a bit excessive in the general case. OTOH, we all have things to hide, and in those cases, these kind of precautions are perfectly undertandable. Having worked in a secure facility for a major insurance company once I know that there are times when security really does call for a certain amount of paranoia. On the gripping hand, they laid out several million US dollars for a sealed inner computer room with a faraday cage in the walls, shielded cabling, and no line of sight from any exterior windows or doors, only to shut the project down a few months later over some bureaucratic in-fighting, so perhaps the effort might have been better spend elsewhere.
Post Reply