1 |
On 8/14/2011 5:25 AM, Sven Vermeulen wrote: |
2 |
> On Sat, Aug 13, 2011 at 8:33 PM, Mike Edenfield |
3 |
> <kutulu@××××××.org <mailto:kutulu@××××××.org>> wrote: |
4 |
> |
5 |
> My system-auth doesn't have anything about SELinux in |
6 |
> it. The pam_selinux.so |
7 |
> calls are in system-login. This looks like what pambase |
8 |
> is supposed to be |
9 |
> doing. system-login.in <http://system-login.in> has these: |
10 |
> |
11 |
> #if HAVE_SELINUX |
12 |
> session required pam_selinux.so close |
13 |
> #endif |
14 |
> |
15 |
> and system-auth.in <http://system-auth.in> doesn't. |
16 |
> |
17 |
> Which one should kdm/gdm be using? Right now |
18 |
> /etc/pam.d/kde pulls in system- |
19 |
> auth. Can I just move the pam_selinux calls? |
20 |
> |
21 |
> |
22 |
> If you do, does it break things (like logon through terminals)? |
23 |
> If not, does it fix the KDM logons? |
24 |
|
25 |
It fixed my KDM logins to be unconfined, but it appears to |
26 |
break a bunch of other things: |
27 |
|
28 |
kutulu@platypus ~ $ id -Z |
29 |
unconfined_u:unconfined_r:unconfined_t |
30 |
kutulu@platypus ~ $ sudo -s |
31 |
Password: |
32 |
platypus kutulu # id -Z |
33 |
unconfined_u:unconfined_r:bootloader_t |
34 |
|
35 |
bootloader_t seems pretty random so its possible I screwed |
36 |
up my policy in some unrelated way. I'm reinstalling all the |
37 |
policy packages and relabeling, we'll see what happens. |
38 |
|
39 |
--Mike |