certmonger post-save scripts & certmonger_unconfined_t domain
by Sam Morris
Certmonger allows for the configuration of a post-save command to be run after it has obtained new certificates. This can be used to copy the key & certificates out of wherever certmonger is allowed to put them, and save them elsewhere with a particular owner/group, combine the certificate & chain into a single file as required by some software, etc.
The problem comes with SELinux which prevents my post-save scripts from being able to do all of that. I thought the solution was to give the scripts the context of certmonger_unconfined_exec_t, which would cause a transition to the certmonger_unconfined_t domain which is as its name suggests unconfined; but I can't get this to work.
I'm trying to use runcon to simulate certmonger executing a fake script:
# cat /tmp/fakescript
#!/bin/bash
set -eu
id -Z
# /tmp/fakescript
unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023
# ls -Z /tmp/fakescript
unconfined_u:object_r:certmonger_unconfined_exec_t:s0 /tmp/fakescript
# runcon system_u:system_r:certmonger_t:s0 /tmp/fakescript
runcon: ‘/tmp/fakescript’: Permission denied
Here is the avc denial:
----
type=PROCTITLE msg=audit(27/04/21 16:16:47.156:153492) : proctitle=runcon system_u:system_r:certmonger_t:s0 /tmp/fakescript
type=SYSCALL msg=audit(27/04/21 16:16:47.156:153492) : arch=x86_64 syscall=execve success=no exit=EACCES(Permission denied) a0=0x7ffd8aa768ab a1=0x7ffd8aa75888 a2=0x7ffd8aa75898 a3=0x0 items=0 ppid=177795 pid=177796 auid=sam.admin uid=root gid=root euid=root suid=root fsuid=root egid=root sgid=root fsgid=root tty=pts5 ses=103 comm=runcon exe=/usr/bin/runcon subj=unconfined_u:unconfined_r:unconfined_t:s0-s0:c0.c1023 key=(null)
type=AVC msg=audit(27/04/21 16:16:47.156:153492) : avc: denied { entrypoint } for pid=177796 comm=runcon path=/tmp/fakescript dev="dm-0" ino=33563064 scontext=system_u:system_r:certmonger_t:s0 tcontext=unconfined_u:object_r:certmonger_unconfined_exec_t:s0 tclass=file permissive=0
Even though:
# sepolicy transition -s certmonger_t -t certmonger_unconfined_t
certmonger_t @ certmonger_unconfined_exec_t --> certmonger_unconfined_t
Diving in a little deeper, I can see that certmonger can execute the file:
# sesearch -s certmonger_t -t certmonger_unconfined_exec_t -c file -p execute -A
allow certmonger_t certmonger_unconfined_exec_t:file { execute execute_no_trans getattr ioctl map open read };
... and that the file type is an entrypoint for the certmonger_unconfined_t domain:
# sesearch -s certmonger_unconfined_t -t certmonger_unconfined_exec_t -c file -p entrypoint -A
allow certmonger_unconfined_t certmonger_unconfined_exec_t:file { entrypoint execute getattr ioctl lock map open read };
... and that transition is permitted from certmonger_t:
# sesearch -s certmonger_t -t certmonger_unconfined_t -c process -p transition -A
allow certmonger_t certmonger_unconfined_t:process transition;
Which leaves me scratching my head, unsure why it doesn't work in practice...
--
Sam Morris <https://robots.org.uk/>
PGP: rsa4096/CAAA AA1A CA69 A83A 892B 1855 D20B 4202 5CDA 27B9
2 years, 12 months
iscsi.service: Unit cannot be reloaded because it is inactive.
by Jason Long
Hello,
I'm using Fedora Server as an iSCSI Shared Storage. When I rebooted my server then the "iscsi.service" couldn't load:
[root@node3 ~]# systemctl status iscsi.service
● iscsi.service - Login and scanning of iSCSI devices
Loaded: loaded (/usr/lib/systemd/system/iscsi.service; enabled; vendor preset: enabled)
Active: inactive (dead)
Condition: start condition failed at Sat 2021-04-03 18:49:08 +0430; 2s ago
└─ ConditionDirectoryNotEmpty=/var/lib/iscsi/nodes was not met
Docs: man:iscsiadm(8)
man:iscsid(8)
Apr 03 18:39:17 node3.localhost.localdomain systemd[1]: Condition check resulted in Login and scanning of iSCSI devices being skipped.
Apr 03 18:39:17 node3.localhost.localdomain systemd[1]: iscsi.service: Unit cannot be reloaded because it is inactive.
Apr 03 18:39:17 node3.localhost.localdomain systemd[1]: iscsi.service: Unit cannot be reloaded because it is inactive.
Apr 03 18:49:08 node3.localhost.localdomain systemd[1]: Condition check resulted in Login and scanning of iSCSI devices being skipped.
SELinux is enabled on my Fedora Server:
# sestatus
SELinux status: enabled
SELinuxfs mount: /sys/fs/selinux
SELinux root directory: /etc/selinux
Loaded policy name: targeted
Current mode: enforcing
Mode from config file: enforcing
Policy MLS status: enabled
Policy deny_unknown status: allowed
Memory protection checking: actual (secure)
Max kernel policy version: 33
[root@node3 ~]# ps -eZ | grep iscsid_t
[root@node3 ~]#
And when I looked at the log, then I saw below errors:
# dmesg -H -l err
[Apr 4 15:05] [drm:vmw_host_log [vmwgfx]] *ERROR* Failed to send host log message.
[ +0.000009] [drm:vmw_host_log [vmwgfx]] *ERROR* Failed to send host log message.
[ +9.037994] dev[000000004a7f146c]: Unable to change SE Device alua_support: alua_support has fixed value
[ +0.000014] dev[000000004a7f146c]: Unable to change SE Device alua_support: alua_support has fixed value
[ +0.000798] dev[000000004a7f146c]: Unable to change SE Device pgr_support: pgr_support has fixed value
[ +0.000004] dev[000000004a7f146c]: Unable to change SE Device pgr_support: pgr_support has fixed value
How can I configure SELinux for an iSCSI Shared Storage?
Thank you.
3 years
Looking for users of userfaultfd(2) syscall in Fedora
by Ondrej Mosnacek
Hi all,
Kernel 5.12 added support to SELinux for controlling access to the
userfaultfd interface [1][2] and we'd like to implement this in
Fedora's selinux-policy. However, once we add the corresponding class
to the policy, all SELinux domains for which we don't add the
appropriate rules will have any usage of userfaultfd(2) denied.
Therefore, we would like to identify as many users of this syscall as
possible before we make that change, so that we can add and test all
the needed rules in one go, minimizing the amount of denials found
after the fact. My understanding is that userfaultfd(2) doesn't have
many users among system services, so it should be possible to catch
most/all of them in advance.
So if you know that your (or any other) Fedora component uses
userfaultfd(2), please let us know. AFAIK, at least QEMU most likely
uses it, so we'll have that one on our radar, but we'd like to know if
there are any other programs/services we need to cover.
Thanks!
[1] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit...
[2] https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit...
--
Ondrej Mosnacek
Software Engineer, Linux Security - SELinux kernel
Red Hat, Inc.
3 years
Re: Looking for users of userfaultfd(2) syscall in Fedora
by Ondrej Mosnacek
On Tue, Apr 6, 2021 at 7:33 PM Zbigniew Jędrzejewski-Szmek
<zbyszek(a)in.waw.pl> wrote:
> On Tue, Apr 06, 2021 at 06:57:27PM +0200, Ondrej Mosnacek wrote:
> > Hi all,
> >
> > Kernel 5.12 added support to SELinux for controlling access to the
> > userfaultfd interface [1][2] and we'd like to implement this in
> > Fedora's selinux-policy. However, once we add the corresponding class
> > to the policy, all SELinux domains for which we don't add the
> > appropriate rules will have any usage of userfaultfd(2) denied.
>
> https://codesearch.debian.net/search?q=userfaultfd(&literal=1
> lists a few candidates…
Thanks, that's a nice tool!
Filtering out false-positives, the kernel itself, and user programs
that would normally run under unconfined_t, packages dead in Fedora,
..., the only relevant one seems to be 'criu' (already mentioned in
this thread). Strange that it didn't find QEMU... maybe needs a more
generic search...
--
Ondrej Mosnacek
Software Engineer, Linux Security - SELinux kernel
Red Hat, Inc.
3 years
Re: Looking for users of userfaultfd(2) syscall in Fedora
by Florian Weimer
* Zbigniew Jędrzejewski-Szmek:
> The code is available. From what I remember, they had a fairly beefy
> server dedicated to the indexing... But if somebody provides that, it
> should be fairly easy to duplicate.
Michael even expressed interest about setting up an instance, if I
recall correctly, but that was quite some time ago.
Thanks,
Florian
3 years
