1 |
On Fri, Jan 18, 2013 at 6:13 AM, Paul Arthur |
2 |
<junk+usenet@×××××××××××.com> wrote: |
3 |
> Yes. This is the exact same issue secure-delete has, since it uses |
4 |
> the same approach. shred is just as useful as srm (in fact it's more |
5 |
> useful, since it doesn't mandate the full, useless run of 38 passes |
6 |
> that srm does.) |
7 |
|
8 |
"srm" doesn't mandate rewrites either. |
9 |
|
10 |
Anyway, I actually forgot about "shred", so I remove my objection. |
11 |
Other utilities in secure-delete are either simple wrappers of |
12 |
rarely-used functionality ("sfill", "sswap"), or essentially useless |
13 |
for modern kernels ("smem" — good luck clearing free RAM in userspace, |
14 |
been there, tried that). |
15 |
|
16 |
Some comments on replies in this thread: |
17 |
|
18 |
1. Multiple rewrites are indeed useless for modern media, see [1]. |
19 |
2. So journal metadata is not cleared. BFD. If you need 100% |
20 |
guarantees, drop media in acid. |
21 |
3. Wear leveling on flash media is rarer than you think, and most |
22 |
likely doesn't do what you think, see [2]. |
23 |
4. Wear leveling is irrelevant for the usual attack vectors, which is |
24 |
a technician copying your naked gf photos. You need special hardware |
25 |
to access hidden sectors. If you are worried about that, see (2). |
26 |
|
27 |
[1] C. Wright et al., “Overwriting Hard Drive Data: The Great Wiping |
28 |
Controversy”, http://dx.doi.org/10.1007/978-3-540-89862-7_21 |
29 |
[2] E. Gal and S. Toledo, “Algorithms and Data Structures for Flash |
30 |
Memories”, http://dx.doi.org/10.1145/1089733.1089735 |
31 |
|
32 |
-- |
33 |
Maxim Kammerer |
34 |
Liberté Linux: http://dee.su/liberte |