| 1 |
Peter Humphrey <peter@××××××××××××.uk> wrote: |
| 2 |
> On Friday 07 Jul 2017 07:53:01 Martin Vaeth wrote: |
| 3 |
> |
| 4 |
>> ... my original text was arguing against the claim that the primary |
| 5 |
>> purpose of hardened kernels was to protect against untrusted users |
| 6 |
>> sitting in front of the keyboard. |
| 7 |
> |
| 8 |
> It wasn't a claim, just an impression |
| 9 |
|
| 10 |
Sorry that my formulation was unfortunate. |
| 11 |
My intention had been to explain why that impression is wrong IMHO. |
| 12 |
|
| 13 |
Anyway, this discussion is meanwhile almost pointless since |
| 14 |
hardened-sources are pratically no longer available for "normal" |
| 15 |
users, and so also the hardened profile has become almost pointless. |
| 16 |
|
| 17 |
As a small substitute, I would recommend to follow the recommendations |
| 18 |
of the kernel self projection project and to use the |
| 19 |
|
| 20 |
default/linux/amd64/17.0/desktop |
| 21 |
|
| 22 |
profile or - if you are limited to x86 - to combine |
| 23 |
|
| 24 |
default/linux/x86 |
| 25 |
releases/17.0 |
| 26 |
targets/desktop |
| 27 |
|
| 28 |
which enables the current compilers with some default-enabled |
| 29 |
security relevant CFLAGS. |
| 30 |
In addition you can also add -fstack-check=specific |
| 31 |
to CFLAGS and -Wl,-z,now -Wl,-z,relro to LDFLAGS. |
| 32 |
|
| 33 |
All this is not a complete substitute for TPE and friends but |
| 34 |
better than nothing. |
Archives