fcontext nightmare - Help please?

[Daniel J Walsh]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 09/16/2012 09:00 PM, Tom London wrote:
> On Mon, Aug 20, 2012 at 2:59 AM, Daniel J Walsh <dwalsh at redhat.com> wrote: 
> On 08/19/2012 04:24 PM, Tom London wrote:
>>>> On Tue, Aug 14, 2012 at 2:21 PM, Dominick Grift
>>>> <dominick.grift at gmail.com> wrote:
>>>>> You might want to check out the semanage --equiv option. (man
>>>>> semanage)
>>>>> 
>>>>> That basically allows you to alias existing file context
>>>>> structures:
>>>>> 
>>>>> heres an example from man semanage:
>>>>> 
>>>>> For home directories under top level directory, for example
>>>>> /disk6/home, execute the following commands. # semanage fcontext -a
>>>>> -t home_root_t "/disk6" # semanage fcontext -a -e /home /disk6/home
>>>>> # restorecon -R -v /disk6
>>>>> 
>>>>> so in your case you might want to make /data equivalent to / or 
>>>>> something
>>>>> 
>>>>> semanage fcontext -a -e / /data restorecon -R -v -F /data
>>>>> 
>>>>> That should label /data root_t, /data/var var_t, /data/var/lib
>>>>> var_lib_t etc.
>>>>> 
>>>>> just as if it was your main file system.
>>>>> 
>>>> 
>>>> So this sounds exactly what i would like to do with my Luks encrytped
>>>> USB back up drive.
>>>> 
>>>> Unfortunately, I'm stumbling across the fact that the drive is 
>>>> 'automagically' mounted (when I login or power it on), and it gets
>>>> mounted on /run/media/tbl/Backup1TB:
>>>> 
>>>> /dev/mapper/luks-94a9d7d7-f819-4c2c-b735-81bb28db0426 on 
>>>> /run/media/tbl/Backup1TB type ext4 
>>>> (rw,nosuid,nodev,relatime,seclabel,data=ordered,uhelper=udisks2)
>>>> 
>>>> The 'semanage -e' command spews:
>>>> 
>>>> [root at tlondon ~]# semanage fcontext -a -e /
>>>> /run/media/tbl/Backup1TB/X200 /sbin/semanage: File spec
>>>> /run/media/tbl/Backup1TB/X200 conflicts with equivalency rule '/run
>>>> /var/run'; Try adding '/var/run/media/tbl/Backup1TB/X200' instead
>>>> [root at tlondon ~]#
>>>> 
>>>> Appears that '/var/run/media' doesn't exist on my system (I guess
>>>> /run and /var/run are not really 'equivalent'?).
>>>> 
>>>> This an issue with my system (e.g., do I need an explicit entry in
>>>> fstab or some such)? With the scaffolding that deals with /run and
>>>> /var/run? Other? Should this work?
>>>> 
>>>> Thanks, tom
>>>> 
> Yes it is telling you about a double equivalence.  systemd guys have
> suggested that we reverse the equivalence. since /var/run does not really
> exist anymore, they suggested we move to /var/run -> /run rather then what
> we currently have /run -> /var/run.  My concern with this switch would be
> if users/package developers had already added file context for /var/run
> 
> So I tried this to work around the 'one-level equivalence detection':
> 
> [root at tlondon ~]# mount --bind /run/media/tbl/Backup1TB/X200/ /mnt 
> [root at tlondon ~]# semanage fcontext -a -t root_t /mnt [root at tlondon ~]#
> semanage fcontext -a -e / /mnt [root at tlondon ~]# restorecon -v -R /mnt 
> restorecon reset /mnt context 
> system_u:object_r:admin_home_t:s0->system_u:object_r:root_t:s0 restorecon
> reset /mnt/.tcshrc context 
> staff_u:object_r:admin_home_t:s0->staff_u:object_r:etc_runtime_t:s0 
> restorecon reset /mnt/run context 
> staff_u:object_r:admin_home_t:s0->staff_u:object_r:var_run_t:s0 restorecon
> reset /mnt/enable-unconfined context 
> unconfined_u:object_r:admin_home_t:s0->unconfined_u:object_r:etc_runtime_t:s0
>
> 
restorecon reset /mnt/.lesshst context
> staff_u:object_r:admin_home_t:s0->staff_u:object_r:etc_runtime_t:s0 
> <<<<<SNIP>>>>> <<<<<Lots of relabelling here>>>>> restorecon reset
> /mnt/var/cache/krb5rcache context 
> staff_u:object_r:var_t:s0->staff_u:object_r:krb5_host_rcache_t:s0 
> restorecon reset /mnt/var/cache/jetty context 
> system_u:object_r:var_t:s0->system_u:object_r:jetty_cache_t:s0 restorecon
> reset /mnt/var/cache/jetty/temp context 
> system_u:object_r:var_t:s0->system_u:object_r:jetty_cache_t:s0 restorecon
> reset /mnt/var/cache/httpd context 
> staff_u:object_r:var_t:s0->staff_u:object_r:httpd_cache_t:s0 restorecon
> reset /mnt/var/cache/httpd/proxy context 
> staff_u:object_r:var_t:s0->staff_u:object_r:httpd_cache_t:s0 [root at tlondon
> ~]#
> 
> I checked a few relabelled files, and the contexts seem correct, for
> example: restorecon reset /mnt/usr/share/jetty/bin/jetty.sh context 
> staff_u:object_r:bin_t:s0->staff_u:object_r:httpd_exec_t:s0
> 
> 
> I should have used something other than '/mnt', of course. And since the
> drive is not persistently mounted, I'm thinking of wrapping  the 'rsync'
> command with 'semanage' commands that temporarily add/delete the mappings.
> 
> Am I correct in assuming that the way to do this is (presuming bind mount
> the mounted path to '/backup'):
> 
> 
> semanage fcontext -a -t root_t /backup semanage fcontext -a -e / /backup
> 
> rsync ..... lots of options
> 
> semanage fcontext -d -e / /backup semanage fcontext -d -t root_t /backup
> 
> 
> That seem right?
> 
> Thanks! tom
> 

I would figure

/backup/run/blah or /backup/usr/lib64 might be labeled differently then /run
and /usr/lib64.

Since only one substitution would happen.  You would really need to do all of
the substitutions again.

/backup/run == /var/run
/backup/usr/lib64 == /usr/lib
...

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://www.enigmail.net/

iEYEARECAAYFAlBXKvoACgkQrlYvE4MpobNRdgCgvmALwQjwpB+oEB2l2a6akHF9
rwkAniexDIyfYtm4IUlvYeCTs7c9gIUu
=1igJ
-----END PGP SIGNATURE-----