1 |
On 03/03/2016 23:00, Grant Edwards wrote: |
2 |
> I'm sure I'm just being stupid, but I don't understand the lists of |
3 |
> affected and unaffected version numbers in Gentoo security advisories. |
4 |
> |
5 |
> For example: |
6 |
> |
7 |
> Package dev-libs/openssl on all architectures |
8 |
> Affected versions < 1.0.2f |
9 |
> |
10 |
> Unaffected versions >= 1.0.2f, revision >= 1.0.1r, revision >= 1.0.1s, |
11 |
> revision >= 1.0.1t, revision >= 0.9.8z_p8, |
12 |
> revision >= 0.9.8z_p9, revision >= 0.9.8z_p10, |
13 |
> revision >= 0.9.8z_p11, revision >= 0.9.8z_p12, |
14 |
> revision >= 0.9.8z_p13, revision >= 0.9.8z_p14, |
15 |
> revision >= 0.9.8z_p15 |
16 |
> |
17 |
> If it's true that versions >= 0.9.8z_p8 are unaffected, |
18 |
> why is there a need to list that versions >= 0.9.8z_p[9-15] are |
19 |
> unaffected? Are <> relationships betwen version numbers within the |
20 |
> 0.9.8z_pNNN seriels not transitive? |
21 |
> |
22 |
|
23 |
|
24 |
Easiest possible answer, and highly likely to be the correct one: |
25 |
|
26 |
Someone rushed out a notice. They made typos. Or the extract script has |
27 |
bad logic and printed the wrong compare symbols. Or the transitiveness |
28 |
of the subject matter simply never occurred to the author. |
29 |
|
30 |
No need to get all technical on this. Yes strictly speaking it is |
31 |
incorrect. But if you've been following DROWN the intent is crystal |
32 |
clear and you can put it down to yet another bug. |
33 |
|
34 |
Pity we can't take the same attitude with what openssl upstream did with |
35 |
1.0.2g |
36 |
|
37 |
|
38 |
-- |
39 |
Alan McKinnon |
40 |
alan.mckinnon@×××××.com |