1 |
On Sat, 2004-02-28 at 18:05, Michael Ihde wrote: |
2 |
[NIS stuff cut] |
3 |
|
4 |
I cant help since I'm not familiar with NIS. |
5 |
|
6 |
> On to NFS, mount needed |
7 |
> allow { mount_t } resolv_conf_t:file { read getattr }; |
8 |
> allow { mount_t } mount_t:tcp_socket { write read }; |
9 |
|
10 |
This should be ok. |
11 |
|
12 |
> Right now all the nfs files and directories and labeled (null). And a |
13 |
> relabel does not change this. Because of this they seem to all be in |
14 |
> the context nfs_t:dir and nfs_t:file. I don't want to give users full |
15 |
> access to nfs_t. How can I label these nfs file systems appropriatly |
16 |
> (btw...I believe they are ext2). |
17 |
|
18 |
Mounting a NFS export does not support labels, even if the server has |
19 |
labels. The label is given by genfs contexts, won't be shown by ls, and |
20 |
can't changed. There is work to have labeled NFS, so you can have |
21 |
labels work between two SELinux machines, but its still going to be some |
22 |
time before it's ready. So, for now, all NFS filesystems will be nfs_t, |
23 |
and you'll have to allow the access and make sure the regular DAC is set |
24 |
correctly :\ |
25 |
|
26 |
-- |
27 |
Chris PeBenito |
28 |
<pebenito@g.o> |
29 |
Developer, |
30 |
Hardened Gentoo Linux |
31 |
Embedded Gentoo Linux |
32 |
|
33 |
Public Key: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE6AF9243 |
34 |
Key fingerprint = B0E6 877A 883F A57A 8E6A CB00 BC8E E42D E6AF 9243 |