Messages in this thread | | | Date | Fri, 17 Aug 2001 10:20:24 -0400 | From | Kent Borg <> | Subject | Re: ext2 not NULLing deleted files? |
| |
Andi Kleen <freitag@alancoxonachip.com> writes: >Just NULLing alone is quite useless anyways; just 0ed data can be >easily recovered in a special laboratory by using old traces of >magnetism on the surfaces. If you care about real data deletion you >should probably use an utility like wipe which does about 20-30 >passes with random data.
The services of such a laboratory are quite expensive, and invasive (they need the disk). An unerase utility is quite cheap (free) and can be quietly run from the other side of the planet.
It seems to me there be room for something simple that raises the cost of recovering deleted files to a price significantly above the current sale price of "free". Simple NULLing would do that and it could be done cheaply by a low priority daemon that goes around sweeping up deleted bits when nothing much else is happening.
Yes, there would still be a window when files will not have been NULLed, and some machines are too busy to allow such a daemon to run (are those machines also too busy to do encryption?), but it would be much better than the case now where nearly all of us have tons of deleted stuff just sitting there. (Do you?)
In the physical world documents are sometimes shredded. Yes, there is a window between when a document is designated to be shredded and when it can be shredded, and, yes, most shredders leave big enough pieces to reassemble the original. But shredders significantly lower one's exposure and they significantly raise the cost of recovering that data. Just because they are not perfect doesn't mean they are "quite useless". The same is true of NULLing deleted files.
Also, I note that such a userland daemon is not a kernel issue.
-kb, the Kent who doesn't consider seatbelts "quite useless" just because there are accidents for which they will not save his life.
P.S. We still don't know what was in the 18-1/2 minute gap. Maybe, after all these years, we will soon find out, but most of the folks involved are now dead. (How many on this list were not yet born then? Do they even know what I am talking about? That's pretty good security for daily use.) Simple erasure is not perfect security, but it is pretty damn good, and all it took was Rosemary Woods stretching to reach that "record" button. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |