1 |
Il 16/04/2018 14:31, Anthony G. Basile ha scritto: |
2 |
> On 4/16/18 5:14 AM, Hanno Böck wrote: |
3 |
[snip] |
4 |
> |
5 |
>> |
6 |
>> There's also another question related to this: What's the future for |
7 |
>> Gentoo hardened? |
8 |
>> From what I can tell hardened consists of: |
9 |
>> * the things that try to make it compatible with grsec/pax |
10 |
>> (more or less obsolete). |
11 |
>> * things that are now in default profiles anyway (aslr, stack |
12 |
>> protector). |
13 |
>> * things that probably should be in default profiles (relro, now linker |
14 |
>> flags) |
15 |
>> * -fstack-check, which should eventually be replaced with |
16 |
>> -fstack-clash-protection (only available in future gcc's) and that |
17 |
>> should probably also go into default profiles. |
18 |
>> * Furthermore hardened disables some useful features due to their |
19 |
>> incompatibility with pax (e.g. sanitizers). |
20 |
>> |
21 |
>> So it's stuff that either is obsolete or probably should be a candidate |
22 |
>> for main profiles. Maybe we should strive for "hardened-by-default". |
23 |
>> |
24 |
> You're forgetting selinux. Most of Zorry's work has made it into gcc |
25 |
> and is now being enabled by our default toolchain. Some kernel features |
26 |
> have also been improved upstream. With upstream carrying a lot of the |
27 |
> work we did, I think 'hardened-by-default' minus selinux should be the |
28 |
> goal of Gentoo. |
29 |
> |
30 |
Hardened had strong impact in some workflows, surpassing 10%. |
31 |
Overhead could be acceptable in some situation but unwanted in others, |
32 |
main profiles are obscure and difficult to change for most. |
33 |
For this reason I'd like to ask to carefully evaluate if a security |
34 |
feature can be enabled without suddently change the behaviour (worse |
35 |
performances) of a machine running Gentoo. |
36 |
Instead it would be good to have a guide on how to further harden any |
37 |
profile. |
38 |
If the hardening at any cost argument wins however we MUST have a guide |
39 |
on ho to disable at least the most impactful options. |