The pattern that matches /usr/include uses a dot, in order probably to
avoid calling m4's function include(). This also matches other paths
such as /usr/inclu/e. Such a side-effect can be avoided by inserting an
empty string which is removed by m4's preprocessing.
Mailing-list discussion: https://lore.kernel.org/selinux-refpolicy/CAJfZ7=krh_TaCBQzFxLM394Sc5-82ZO0DdcfvWON-RXu-wqBVw@mail.gmail.com/t/#u
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
There are two patterns that define file contexts for /run/rpc.statd.pid:
* in policy/modules/services/rpcbind.fc:
/run/rpc.statd\.pid -- gen_context(system_u:object_r:rpcbind_var_run_t,s0)
* in policy/modules/services/rpc.dc:
/run/rpc\.statd\.pid -- gen_context(system_u:object_r:rpcd_var_run_t,s0)
They coexist even though their labels differ because the first one uses
a unescaped dot. As it does not seem to exist other files matching the
first pattern, remove it in order to only keep the second one.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
Patterns using this have a small issue:
/opt/Adobe/Reader.?/Reader/intellinux
The issue is that the dot can also match a slash. A bettern pattern
would be:
/opt/Adobe/Reader[^/]?/Reader/intellinux
In this specific case, the intent is to match digits (like
/opt/Adobe/Reader9). Use [0-9] for this.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
The following pattern seems to match much more than intended:
/usr/(.*/)?bin/java[^-]*
According to the commit which introduced it
(0190325c18),
the aim was to match java1.4, java5, java6, and not java-config nor
java-check-environment. The issue is that the pattern also matches
sub-directories such as:
/usr/share/my-application/bin/java/myfile
Prevent this by adding / in the character blacklist of the pattern.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
This seems to be a misspelling, and there is no reason which would
explain why monit's init script would be labeled with a different
sensitivity while the main binary uses s0.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
haveged listens to commands using a Unix socket
(at path "\0/sys/entropy/haveged" according to
https://github.com/jirka-h/haveged/blob/1.9.4/src/havegecmd.h#L34).
The implementation of this server is available on
https://github.com/jirka-h/haveged/blob/1.9.4/src/havegecmd.c .
This fixes the following denial:
type=AVC msg=audit(1551002989.396:27): avc: denied { listen } for
pid=262 comm="haveged"
path=002F7379732F656E74726F70792F68617665676564
scontext=system_u:system_r:entropyd_t
tcontext=system_u:system_r:entropyd_t tclass=unix_stream_socket
permissive=1
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
When ulogd is run by systemd on Debian, it logs messages to the journal,
it used a PID file in /run/ulog/ulogd.pid, and logs packets to
/var/log/ulog/syslogemu.log. This last ones triggers a dac_read_search
capability check because the directory is configured as:
drwxrwx---. ulog adm /var/log/ulog
(root does not have an access to the directory without bypassing the DAC.)
Add a comment describing how to avoid allowing dac_read_search to ulogd_t.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
Debian uses /var/log/ulog/syslogemu.log by default to log network
packets sent through a netlink multicast group by the firewall.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
type=AVC msg=audit(1560944462.698:217): avc: denied { map } for pid=1265 comm="rpm" path="/sys/fs/selinux/status" dev="selinuxfs" ino=19 scontext=system_u:system_r:rpm_t:s0 tcontext=system_u:object_r:security_t:s0 tclass=file permissive=1
v2 - Create new interface to allow mapping security_t and use this interface by rpm_t
Signed-off-by: Dave Sugar <dsugar@tresys.com>
Messages like this are added to the audit log when an rpm is installed:
type=SOFTWARE_UPDATE msg=audit(1560913896.581:244): pid=1265 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:rpm_t:s0 msg='sw="ntpdate-4.2.6p5-25.el7_3.2.x86_64" sw_type=rpm key_enforce=0 gpg_res=0 root_dir="/" comm="rpm" exe="/usr/bin/rpm" hostname=? addr=? terminal=? res=success'
These are the denials that I'm seeing:
type=AVC msg=audit(1560913896.581:243): avc: denied { audit_write } for pid=1265 comm="rpm" capability=29 scontext=system_u:system_r:rpm_t:s0 tcontext=system_u:system_r:rpm_t:s0 tclass=capability permissive=1
type=AVC msg=audit(1561298132.446:240): avc: denied { create } for pid=1266 comm="rpm" scontext=system_u:system_r:rpm_t:s0 tcontext=system_u:system_r:rpm_t:s0 tclass=netlink_audit_socket permissive=1
type=AVC msg=audit(1561298132.446:241): avc: denied { write } for pid=1266 comm="rpm" scontext=system_u:system_r:rpm_t:s0 tcontext=system_u:system_r:rpm_t:s0 tclass=netlink_audit_socket permissive=1
type=AVC msg=audit(1561298132.446:241): avc: denied { nlmsg_relay } for pid=1266 comm="rpm" scontext=system_u:system_r:rpm_t:s0 tcontext=system_u:system_r:rpm_t:s0 tclass=netlink_audit_socket permissive=1
type=AVC msg=audit(1561298132.447:243): avc: denied { read } for pid=1266 comm="rpm" scontext=system_u:system_r:rpm_t:s0 tcontext=system_u:system_r:rpm_t:s0 tclass=netlink_audit_socket permissive=1
v2 - Use interface rather than adding permissions here - this change may
confuse subsequent patches in this set, if so let me know and I will
submit a pull request on github.
Signed-off-by: Dave Sugar <dsugar@tresys.com>
Nicolas Iooss
Chris PeBenito
Sugar, David
Laurent Bigonville
Chris PeBenito
Alexander Miroshnichenko