Gentoo Archives: gentoo-hardened

From: Petre Rodan <petre.rodan@××××××××××××.com>
To: Chris PeBenito <pebenito@g.o>
Cc: Petre Rodan <petre.rodan@××××××××××××.com>, Hardened Gentoo Mail List <gentoo-hardened@g.o>
Subject: Re: [gentoo-hardened] Re: selinux and ~x86
Date: Tue, 04 Nov 2003 16:09:02
Message-Id: 20031104160326.GA9858@peter.rav.local
In Reply to: Re: [gentoo-hardened] Re: selinux and ~x86 by Chris PeBenito
1 On Tue, Nov 04, 2003 at 09:39:14AM -0600, Chris PeBenito wrote:
2 > On Tue, 2003-11-04 at 07:45, Petre Rodan wrote:
3 > > one thing about
4 > > http://www.gentoo.org/proj/en/hardened/selinux/selinux-quickstart.xml:
5 > >
6 > > you do warn people that there is a bug with grub, but there are a few
7 > > implications with that which are not quite obvious:
8 > >
9 > > if the user blindly removes the /boot/boot symlink and modifies
10 > > grub.conf there is the possibility that on next reboot he'll see 'grub
11 > > error 15'
12 > > he sould have re-installed grub (I think grub still expects to find
13 > > the stages in (hd0)/boot/grub/stageX (instead of (hd0)/grub/stageX)
14 > > and hence the error)
15 >
16 > Yes, the grub bug is really nasty and irritating. I'll add to the doc
17 > to reinstall grub.
18
19 you can also add the fact that the menu.lst symlink can be removed, since it can only make matters worse.
20
21 > > and I use monolithic kernels, so tweaking mkinitrd was also needed
22 >
23 > In what way? If you run 'make initrd' from the policy directory, it
24 > doesn't try to install any modules.
25
26 erm
27 /lib/modules/2.4.22-hardened is not a directory.
28 No dep file found for kernel 2.4.22-hardened
29 and so on ...
30
31 I've only made an ugly hack to make it create my initrd.gz, if you are interested I think I could do a cleaner diff on this issue.
32
33 > > BTW,
34 > > this idea of having the policy in two places (in
35 > > /etc/security/selinux/.. and inside initrd.gz) is temporary, right? =)
36 >
37 > Well the old way of having the kernel read the policy was rejected from
38 > the 2.6 kernel for several reasons. So the initrd (or initramfs) is
39 > currently the only way to load the policy before running init. The
40 > policy needs to be loaded before init, so init can transition to the
41 > right domain. There are some ugly hacks and games that RedHat is
42 > trying, to skip the initrd. But they are ugly :\ So right now the
43 > initrd is just a fact of life.
44
45 there are interesting ideas on the selinux list, and as I see it a lot of people is against initrd
46 for compatibility reasons (the Sun/Cobalt Qube/RaQ hardware which has a Linux kernel in the BIOS which can read an Ext2/Ext3 file system and load a kernel - they never wrote code for loading an initrd) to quote Russel)
47
48 maybe a good solution will pop up in the near future.
49
50
51 bye,
52 peter
53
54
55 > --
56 > Chris PeBenito
57 > <pebenito@g.o>
58 > Developer, SELinux
59 > Hardened Gentoo Linux
60 >
61 > Public Key: http://pgp.mit.edu:11371/pks/lookup?op=get&search=0xE6AF9243
62 > Key fingerprint = B0E6 877A 883F A57A 8E6A CB00 BC8E E42D E6AF 9243
63
64
65
66 --
67
68 Petre Rodan
69 Senior Network Engineer
70 GeCAD Software - RAV Division
71
72 ----------------------------------------------------------------------
73 Tel/Fax: +40-21-321-7803
74 Hotline: +40-21-321-7859
75
76 This message is confidential. It may also be privileged or otherwise
77 protected by work product immunity or other legal rules. It may contain
78 personal views which are not the views of the GeCAD unless specifically
79 stated.
80 If you have received it in error, please delete it from your system.
81 Do not use, copy or disclose the information in any way nor act in
82 reliance on it and notify the sender immediately.