Systemd uses a number of UNIX sockets for communication (notify
socket [1], journald socket). These sockets are normally created at
start-up after the SELinux policy is loaded, which means that the
kernel socket objects have proper security contexts of the creating
processes.
Unfortunately things look different when the system is started with an
initrd that is also running systemd (e.g. dracut). In such case the
sockets are created in the initrd systemd environment before the
SELinux policy is loaded and therefore the socket object is assigned
the default kernel context (system_u:system_r:kernel_t). When the
initrd systemd transfers control to the main systemd the notify socket
descriptors are passed to the main systemd process [2]. This means
that when the main system is running the sockets will use the default
kernel securint context until they are recreated, which for some
sockets (notify socket) never happens.
Until there is a way to change the context of an already open socket
object all processes, that wish to use systemd sockets need to be
able to send datagrams to system_u:system_r:kernel_t sockets.
Parts of this workaround were earlier hidden behind RedHat-specific
rules, since this distribution is the prime user of systemd+dracut
combo. Since other distros may want to use similar configuration it
makes sense to enable this globally.
[1] sd_notify(3)
[2] https://github.com/systemd/systemd/issues/16714
Signed-off-by: Krzysztof Nowicki <krissn@op.pl>
tmp
I have been working to support IMA/EVM on a system. It
requires having keys added to the kernel keyring. Keys
added with keyctl and evmctl. I am creating keys in the
ima_key_t type. Once the keys are created, many domains
then need search permission on the type of the key. The
following changes are needed to get things to work.
Need to add keys to the kernel keyring (keyctl).
type=AVC msg=audit(1585420717.704:1868): avc: denied { write } for pid=8622 comm="keyctl" scontext=system_u:system_r:cleanup_t:s0 tcontext=system_u:system_r:kernel_t:s0 tclass=key permissive=1
Allow all domains to search key
type=AVC msg=audit(1587936822.802:556): avc: denied { search } for pid=5963 comm="kworker/u16:6" scontext=system_u:system_r:kernel_t:s0 tcontext=system_u:object_r:ima_key_t:s0 tclass=key permissive=1
type=AVC msg=audit(1587936822.804:559): avc: denied { search } for pid=5963 comm="systemd-cgroups" scontext=system_u:system_r:systemd_cgroups_t:s0 tcontext=system_u:object_r:ima_key_t:s0 tclass=key permissive=1
type=AVC msg=audit(1587936822.809:560): avc: denied { search } for pid=5964 comm="(sysctl)" scontext=system_u:system_r:init_t:s0 tcontext=system_u:object_r:ima_key_t:s0 tclass=key permissive=1
type=AVC msg=audit(1587936822.813:562): avc: denied { search } for pid=5964 comm="sysctl" scontext=system_u:system_r:initrc_t:s0 tcontext=system_u:object_r:ima_key_t:s0 tclass=key permissive=1
type=AVC msg=audit(1587936823.149:604): avc: denied { search } for pid=5987 comm="setsebool" scontext=system_u:system_r:semanage_t:s0 tcontext=system_u:object_r:ima_key_t:s0 tclass=key permissive=1
Signed-off-by: Dave Sugar <dsugar@tresys.com>
CRIU can influence the PID of the threads it wants to create.
CRIU uses /proc/sys/kernel/ns_last_pidto tell the kernel which
PID it wants for the next clone().
So it has to write to that file. This feels like a problematic as
it opens up the container writing to all sysctl_kernel_t.
Using new label container_t will just write to
sysctl_kernel_ns_last_pid_t instad writing to more generic
sysctl_kernel_t files.
Every Infiniband network will have a default pkey, so that is labeled.
The rest of the pkey configuration is network specific. The policy allows
access to the default and unlabeled pkeys for sysadm and staff users.
kernel_t is allowed access to all pkeys, which it needs to process and
route management datagrams.
Endports are all unlabeled by default, sysadm users are allowed to
manage the subnet on unlabeled endports. kernel_t is allowed to manage
the subnet on all ibendports, which is required for configuring the HCA.
This patch requires selinux series: "SELinux user space support for
Infiniband RDMA", due to the new ipkeycon labeling mechanism.
Signed-off-by: Daniel Jurgens <danielj@mellanox.com>
This patch doesn't do everything that is needed to have systemd-nspawn work.
But it does everything that is needed and which I have written in a clear and
uncontroversial way. I think it's best to get this upstream now and then
either have a separate discussion about the more difficult issues, or wait
until I devise a way of solving those problems that's not too hacky.
Who knows, maybe someone else will devise a brilliant solution to the remaining
issues after this is accepted upstream.
Also there's a tiny patch for systemd_machined_t that is required by
systemd_nspawn_t.
Description: systemd-nspawn
Author: Russell Coker <russell@coker.com.au>
Last-Update: 2017-03-29
This patch adds missing permissions in the kernel module that prevent
to run it without the unconfined module.
This second version improves the comment section of new interfaces:
"Domain" is replaced by "Domain allowed access".
Signed-off-by: Guido Trentalancia <guido@trentalancia.net>
Interface fs_register_binary_executable_type allow registering
interpreters using a filesystem monted on /proc/sys/fs/binfmt_misc. In
order to access this filesystem, the process needs to search every
parent directory of the mountpoint.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
The kernel_read_unix_sysctls() and kernel_rw_unix_sysctls() currenly
don't allow listing the /proc/sys/net/unix directory, contrary to the
other sysctl interfaces.
The kernel_delete_unlabeled_chr_files interface is called by the
(deprecated) files_delete_isid_type_chr_files interface in
kernel/files.if.
Signed-off-by: Sven Vermeulen <sven.vermeulen@siphos.be>
The kernel_delete_unlabeled_blk_files interface is called by the
(deprecated) files_delete_isid_type_blk_files in kernel/files.if.
Signed-off-by: Sven Vermeulen <sven.vermeulen@siphos.be>
The kernel_delete_unlabeled_sockets interface is called by the
(deprecated) files_delete_isid_type_sock_files interface in
kernel/files.if.
Signed-off-by: Sven Vermeulen <sven.vermeulen@siphos.be>
The kernel_delete_unlabeled_pipes interface is called by the
(deprecated) files_delete_isid_type_fifo_files interface in
kernel/files.if.
Signed-off-by: Sven Vermeulen <sven.vermeulen@siphos.be>
The kernel_delete_unlabeled_symlinks interface is called by the
files_delete_isid_type_symlinks interface (in kernel/files.if). This
interface is deprecated (and calls kernel_delete_unlabeled_symlinks).
Signed-off-by: Sven Vermeulen <sven.vermeulen@siphos.be>
Back when the SECMARK implementation was new, the packet class was always
checked. Because of that, unlabeled_t packet rules proliferated refpolicy
since the common case was to have no SECMARK rules. Since then, the kernel
has been modified to only enforce the packet class if there are SECMARK
rules. Remove the unlabeled_t packet rules, since users of SECMARK will
likely want no unlabeled_t packet rules, and the common case users will
have no impact since the packet class isn't enforced on their systems.
To have partial SECMARK confinement, the following rule applies:
allow { domain -type_i_want_to_constrain_t } unlabeled_t:packet { send recv };
It seems like over-allowing, but if you have no SECMARK rules, it's the equivalent of:
allow * unlabeled_t:packet { send recv };
Signed-off-by: Chris PeBenito <cpebenito@tresys.com>
The attached patch adds a few database object classes, as follows:
* db_schema
------------
A schema object performs as a namespace in database; similar to
directories in filesystem.
It seems some of (but not all) database objects are stored within
a certain schema logically. We can qualify these objects using
schema name. For example, a table: "my_tbl" within a schema: "my_scm"
is identified by "my_scm.my_tbl". This table is completely different
from "your_scm.my_tbl" that it a table within a schema: "your_scm".
Its characteristics is similar to a directory in filesystem, so
it has similar permissions.
The 'search' controls to resolve object name within a schema.
The 'add_name' and 'remove_name' controls to add/remove an object
to/from a schema.
See also,
http://developer.postgresql.org/pgdocs/postgres/sql-createschema.html
In the past discussion, a rubix folks concerned about no object
class definition for schema and catalog which is an upper level
namespace. Since I'm not certain whether we have a disadvantage
when 'db_schema' class is applied on catalog class, I don't add
this definition yet.
Default security context of 'db_table' and 'db_procedure' classes
get being computed using type_transition with 'db_schema' class,
instead of 'db_database' class. It reflects logical hierarchy of
database object more correctly.
* db_view
----------
A view object performs as a virtual table. We can run SELECT
statement on views, although it has no physical entities.
The definition of views are expanded in run-time, so it allows
us to describe complex queries with keeping readability.
This object class uniquely provides 'expand' permission that
controls whether user can expand this view, or not.
The default security context shall be computed by type transition
rule with a schema object that owning the view.
See also,
http://developer.postgresql.org/pgdocs/postgres/sql-createview.html
* db_sequence
--------------
A sequence object is a sequential number generator.
This object class uniquely provides 'get_value', 'next_value' and
'set_value' permissions. The 'get_value' controls to reference the
sequence object. The 'next_value' controls to fetch and increment
the value of sequence object. The 'set_value' controls to set
an arbitrary value.
The default security context shall be computed by type transition
rule with a schema object that owning the sequence.
See also,
http://developer.postgresql.org/pgdocs/postgres/sql-createsequence.html
* db_language
--------------
A language object is an installed engine to execute procedures.
PostgreSQL supports to define SQL procedures using regular script
languages; such as Perl, Tcl, not only SQL or binary modules.
In addition, v9.0 or later supports DO statement. It allows us to
execute a script statement on server side without defining a SQL
procedure. It requires to control whether user can execute DO
statement on this language, or not.
This object class uniquely provides 'implement' and 'execute'
permissions. The 'implement' controls whether a procedure can
be implemented with this language, or not. So, it takes security
context of the procedure as subject. The 'execute' controls to
execute code block using DO statement.
The default security context shall be computed by type transition
rule with a database object, because it is not owned by a certain
schema.
In the default policy, we provide two types: 'sepgsql_lang_t' and
'sepgsql_safe_lang_t' that allows unpriv users to execute DO
statement. The default is 'sepgsql_leng_t'.
We assume newly installed language may be harm, so DBA has to relabel
it explicitly, if he want user defined procedures using the language.
See also,
http://developer.postgresql.org/pgdocs/postgres/sql-createlanguage.htmlhttp://developer.postgresql.org/pgdocs/postgres/sql-do.html
P.S)
I found a bug in MCS. It didn't constraint 'relabelfrom' permission
of 'db_procedure' class. IIRC, I fixed it before, but it might be
only MLS side. Sorry.
Thanks,
--
KaiGai Kohei <kaigai@ak.jp.nec.com>
policy/flask/access_vectors | 29 ++++++++
policy/flask/security_classes | 6 ++
policy/mcs | 16 ++++-
policy/mls | 58 ++++++++++++++-
policy/modules/kernel/kernel.if | 8 ++
policy/modules/services/postgresql.if | 125 +++++++++++++++++++++++++++++++--
policy/modules/services/postgresql.te | 116 +++++++++++++++++++++++++++++-
7 files changed, 342 insertions(+), 16 deletions(-)
Add ability to dontaudit requiests to load kernel modules. If you
disable ipv6 every confined app that does ip, tries to get the kernel to
load the module.
Better handling of unlabeled files by the kernel interfaces
