| 1 |
On Wed, 2003-10-29 at 08:41, Mike Williams wrote: |
| 2 |
> For about the last week I have had an SELinux install running, it all appeared |
| 3 |
> to be working OK. |
| 4 |
> I was getting loads of access denied warnings from selinux about portage |
| 5 |
> access it's cache and things, the best I could do was to unset 'userpriv'. |
| 6 |
|
| 7 |
I don't use any of the usersandbox/userpriv stuff, so I can't test this |
| 8 |
right now. Are you in permissive, like I suggested? :) |
| 9 |
|
| 10 |
> Then I had problems upgrading portage itself. The compile and merge would go |
| 11 |
> without a hitch, but it would seg fault at exactly the same point everytime |
| 12 |
> when doing the qmerge (to the filesystem). |
| 13 |
> I *think* it worked ok this morning when I upgraded to 2.0.49-r15. |
| 14 |
|
| 15 |
I think there was a segfault bug with an older portage version. |
| 16 |
|
| 17 |
> My major problem now is with glibc (and zlib), both when merging to the live |
| 18 |
> filesystem will get to a point (--- /lib/, --- /usr/lib), and truncate the |
| 19 |
> file (libc-2.3.2.so, libz.blah) |
| 20 |
|
| 21 |
> If I boot off a CD and mount the partitions /lib/libc-2.3.2 is 0 bytes (can't |
| 22 |
> remember zlib), the file waiting in the sandboxed image is 1.5M, and simply |
| 23 |
> copying it over and rebooting fixes it. |
| 24 |
|
| 25 |
This is probably some weird problem with the old python-selinux. |
| 26 |
Unmerge that, and it will most likely work. We've done a lot more |
| 27 |
testing on python-selinux-2.7 for the new API, and hopefully this kind |
| 28 |
of weird bug will not happen. |
| 29 |
|
| 30 |
|
| 31 |
-- |
| 32 |
Chris PeBenito |
| 33 |
<pebenito@g.o> |
| 34 |
Developer, SELinux |
| 35 |
Hardened Gentoo Linux |
| 36 |
|
| 37 |
Public Key: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE6AF9243 |
| 38 |
Key fingerprint = B0E6 877A 883F A57A 8E6A CB00 BC8E E42D E6AF 9243 |
Archives