1 |
Frank Schafer wrote: |
2 |
> |
3 |
> Hi all, |
4 |
> |
5 |
> Somebody on b.g.o. suggests to try building with a vanilly kernel. Did |
6 |
> someone here make the same experience and if yes, will it run then on a |
7 |
> gentoo-kernel? |
8 |
|
9 |
Hi |
10 |
|
11 |
Not sure if you're talking in general about sig 11 on emerge, some |
12 |
specific problem with Xorg. |
13 |
|
14 |
Anyhow, I had this with various packages and found a few things helped: |
15 |
|
16 |
- Once I had a sig 11 a restart of the emerge would seg fault very |
17 |
soon. I suspected dodgy CPU or memory, so played with the cpuburn |
18 |
package. Curiously I found that running MMXburn (I think it was called) |
19 |
for a minute or so vastly extended how far into a build emerge would |
20 |
get. I have no idea what this program actually does, but I assumed it |
21 |
was heavily loading some memory or other so that when gcc started again |
22 |
it was given a different chumk to what it was getting previously. But |
23 |
that's all very hypothetical... |
24 |
|
25 |
- I used the FEATURES="keepwork" emerge... so that the work area was |
26 |
preserved and the emerge could continue from where it was. But this |
27 |
takes up a lot of disk space uless you manually delete them or use |
28 |
ebuild --clean once it's suceeded. |
29 |
|
30 |
- Install ccache to 'dynamically' cache/preserve the guts of whatever |
31 |
you're trying to build so that you don't need "keepwork". It's a bit |
32 |
slower, but easier to manage. |
33 |
|
34 |
- Do an emerge in a one line script something like this, so it'll loop |
35 |
around until it succeeds (best used with the FEATURES or ccache idea): |
36 |
|
37 |
while true; do emerge foo && break; done |
38 |
|
39 |
HTH |
40 |
IanC |
41 |
-- |
42 |
gentoo-user@g.o mailing list |