1 |
Brian Kroth wrote: |
2 |
> I had also seen this in the bug. Do you know if there's a way to |
3 |
> specify per package CFLAG customizations in portage? |
4 |
> |
5 |
> Also, does anyone know if this is being fixed on bind's end of things? |
6 |
> |
7 |
> Thanks, |
8 |
> Brian |
9 |
> |
10 |
> Akos Szalkai wrote: |
11 |
>>> I recently upgraded to bind-9.3.4 via portage, and it now keeps core |
12 |
>>> dumping on me. Anyone running chrooted |
13 |
>>> bind-9.3.4+hardended+grsecurity having this problem? |
14 |
>> Yes, it also happens to non-chrooted bind. This is bug #158664. |
15 |
>> Basically, using -O instead of -O2 (gcc optimization) solved it for me. |
16 |
>> |
17 |
>> Akos |
18 |
>> |
19 |
I've done this on my laptop a few times, and am not too familiar. But I |
20 |
have the file: |
21 |
/etc/portage/env/dev-libs/beecrypt-3.1.0-r2 |
22 |
and the contents are: |
23 |
CFLAGS="-O2 -march=pentium3 -pipe -fomit-frame-pointer" |
24 |
This is on a Pentium-M laptop, but it won't compiler unless I set |
25 |
"march=pentium3". With this structure I've changed march= for just the |
26 |
one version of the package. I've experimented with having an unversioned |
27 |
package and it works, too. In this particular case, I now have |
28 |
beecrypt-4.1.2-r1 and apparently it has compiled just fine with my |
29 |
default "march=pentiumm". I believe it overrides the whole CFLAGS |
30 |
environment variable, not piece by piece. |
31 |
|
32 |
Dale Pontius |
33 |
-- |
34 |
gentoo-hardened@g.o mailing list |