Modify create_file_perms and create_dir_perms so it doesn't have
the "link" permission. This permission controls whether hard links
are allowed or not on the given file label. Hard links are a common
source of security bugs, and isn't something we want to support by
default.
Get rid of link_file_perms and move the necessary permissions into
create_file_perms and create_dir_perms. Nobody is using this macro,
so it's pointless to keep it around.
Get rid of unlink on directories. It returns EISDIR if you attempt to
do it, independent of SELinux permissions.
SELinux domains which have a need for hard linking for a particular
file type can add it back to their permission set on an as-needed basis.
Add a compile time assertion (neverallow rule) for untrusted_app.
It's particularly dangerous for untrusted_app to ever have hard
link capabilities, and the neverallow rule will prevent regressions.
Bug: 19953790
Change-Id: I5e9493d2bf5da460d074f0bc5ad8ba7c14dec6e0
Add a compile time assertion that capabilities other than setuid
and setgid are never granted to run-as.
This is a compile time assertion only. No new capabilities are granted
or removed.
Change-Id: Ie86d651b539cdfb6f3eaafef0d5d3b716610a220
Add a compile time assertion that gpsd never has capabilities other
than block_suspend.
Bug: 19908228
Change-Id: Iaaf83191902ed04fe9df52c1ed44248fb1ce732d
Android has long enforced that code can't compile with text
relocations present. Add a compile time assertion to prevent
regressions.
Change-Id: Iab35267ce640c1fad9dc82b90d22e70e861321b7
Make sure we're not running fsck on block devices where it
doesn't make any sense. In particular, we should not be running
fsck on /system since it's mounted read-only, and any modification
to that block device will screw up verified boot.
Change-Id: Ic8dd4b0519b423bb5ceb814daeebef06a8f065b4
/odm has the same permissions as /system/... for devices with a
separate odm partition
Bug: 19609718
Change-Id: I6dd83d43c5fd8682248e79d11b0ca676030eadf0
Commit a191398812 added a new
SELinux label to /system/xbin/procrank, which had the effect of
preventing dumpstate from executing procrank. Allow dumpstate
to execute procrank.
Bug: 18342188
Change-Id: If5b781db0d3af34912f3c803b7fa73d53120f3ba
/system/xbin/procrank is a setuid program run by adb shell on
userdebug / eng devices. Allow it to work without running adb root.
Bug: 18342188
Change-Id: I18d9f743e5588c26661eaa26e1b7e6980b15caf7
This is causing more harm than good. We'll just make these all link
libc++ again and work out the CTS issues if they still exist.
Bug: 19778891
This reverts commit 3812cf58cb.
Change-Id: Iaea8f6acb147da4275633a760ccb32951db7f8b6
This is causing more harm than good. We'll just make these all link
libc++ again (another revert) and work out the CTS issues if they still
exist.
Bug: 19778891
This reverts commit a5113a1500.
Change-Id: I35a4c93dae4abb66e3525451d5ce01e33a540895
Address sanitizer requires using libc++ (apparently). We removed
libc++ from these projects since they were C and the SDK/CTS was not
able to find libc++.
If we're interested in continuing to use ASAN on these tools
(probably), we should turn libc++ back on once we're sure CTS won't
die.
Bug: 19778891
Change-Id: I3c1913171a15396ead73277ec1186fead730f66d
Addresses the following error when running CTS on master:
junit.framework.AssertionFailedError: The following errors were encountered when validating the SELinuxneverallow rule:
neverallow { appdomain -bluetooth } self:capability *;
/tmp/SELinuxHostTest5593810182495331783.tmp: error while loading shared libraries: libc++.so: cannot open shared object file: No such file or directory
Also indicate that none of the sepolicy tools need c++ std lib.
Bug: 19617220
Change-Id: I713b3cbd1220655413d399c7cd2b0b50459a5485
Needed since Iff1e601e1268d4d77f64788d733789a2d2cd18cc removed it
from appdomain.
Change-Id: I9fc08b525b9868f0fb703b99b0c0c17ca8b656f9
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
This was rendered obsolete when SELinuxDomainTest was ported
to SELinuxHostTest and only makes sense if allowing search
to domain:dir and { open read } to domain:file in order to
open the /proc/pid/attr/current files in the first place.
SELinux applies a further :process getattr check when
reading any of the /proc/pid/attr/* files for any process
other than self, which is no longer needed by app domains to
pass CTS.
Change-Id: Iff1e601e1268d4d77f64788d733789a2d2cd18cc
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Executing /system/xbin/su is only supported on userdebug builds
for a limited number of domains. On user builds, it should never
occur.
Add a compile time assertion (neverallow rule) that this is
always true.
Bug: 19647373
Change-Id: I231a438948ea2d47c1951207e117e0fb2728c532
Require equivalence for all write operations. We were already
doing this for app_data_file as a result of restricting open
rather than read/write, so this makes the model consistent across
all objects and operations. It also addresses the scenario where
we have mixed usage of levelFrom=all and levelFrom=user for
different apps on the same device where the dominated-by (domby)
relation may not be sufficiently restrictive.
Drop the System V IPC constraints since System V IPC is never allowed
by TE and thus these constraints are dead policy.
Change-Id: Ic06a35030c086e3978c02d501c380889af8d21e0
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Assigning mlstrustedsubject to untrusted_app would undermine
the per-user isolation model being enforced via levelFrom=user
in seapp_contexts and the mls constraints. There is no direct
way to specify a neverallow on attribute assignment, but this
makes use of a particular property of the fork permission to
prevent ever adding mlstrustedsubject to untrusted_app.
A similar restriction for app_data_file and mlstrustedobject
is also important for the same reason, but cannot be expressed
as a neverallow.
Change-Id: I5170cadc55cc614aef0cd5f6491de8f69a4fa2a0
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
There were a few instances where allow rules were appended
after the neverallow rules stanza in the .te file. Also
there were some regular allow rules inserted into the CTS-specific
rules section of app.te. Just move the rules as appropriate.
Should be no change in policy.
Change-Id: Iec76f32d4b531d245bbf5dd9f621a71ff5c71f3e
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Add an attribute command to sepolicy-analyze for displaying the list
of types associated with an attribute in a policy. This is for use
by CTS to check what domains and types are associated with certain
attributes such as mlstrustedsubject and mlstrustedobject.
Change-Id: Ie19361c02feb1ad14ce36862c6aace9e66c422bb
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
With changes I431c1ab22fc53749f623937154b9ec43469d9645 and
Ia54aa263f2245c7090f4b9d9703130c19f11bd28, it is no longer
legitimate to use BOARD_SEPOLICY_IGNORE or REPLACE with
any of the *_contexts files since the CTS requires the AOSP
entries to be present in the device files.
Further, these changes render BOARD_SEPOLICY_IGNORE unusable for
most policy files since all domains and types referenced within any
of the AOSP *_contexts entries must be defined in the kernel policy, so
you cannot use BOARD_SEPOLICY_IGNORE to exclude any .te file
that defines a type referenced in any of those *_contexts files.
There does not seem to be a significant need for such a facility,
as AOSP policy is small and only domains and types used by most
devices should be defined in external/sepolicy.
BOARD_SEPOLICY_REPLACE is commonly misused to eliminate neverallow rules
from AOSP policy, which will only lead to CTS failures, especially
since change Iefe508df265f62efa92f8eb74fc65542d39e3e74 introduced neverallow
checking on the entire policy via sepolicy-analyze. The only remaining
legitimate function of BOARD_SEPOLICY_REPLACE is to support overriding
AOSP .te files with more restrictive rule sets. However, the need for this
facility has been significantly reduced by the fact that AOSP policy
is now fully confined + enforcing for all domains, and further restrictions
beyond AOSP carry a compatibility risk.
Builders of custom policies and custom ROMs still have the freedom to
apply patches on top of external/sepolicy to tighten rule sets (which are
likely more maintainable than maintaining a completely separate copy of
the file via BOARD_SEPOLICY_REPLACE) and/or of using their own separate
policy build system as exemplified by
https://bitbucket.org/quarksecurity/build-policies
Change-Id: I2611e983f7cbfa15f9d45ec3ea301e94132b06fa
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Failed to include base_rules.mk, so this target was not being built.
Change-Id: I2414fa6c3e3e37c74f63c205e3694d1a811c956e
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Generate general forms of the remaining *_contexts files with only the
device-independent entries for use in CTS testing.
Change-Id: I2bf0e41db8a73c26754cedd92cbc3783ff03d6b5
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Periodically, SELinux denials of the form:
type=1400 audit(0.0:8574): avc: denied { setsched } for comm="system_server" scontext=u:r:system_server:s0 tcontext=u:r:kernel:s0 tclass=process permissive=0
are being generated. These denials come from system_server and other
processes. There's no reason why system_server should be calling
sched_setscheduler() on a kernel thread.
Current belief is that these SELinux denials are a bug in the kernel,
and are being inappropriately triggered.
Revert 2d1650f407. The original reason
for accepting this change was to see if it would fix bug 18085992.
Unfortunately, even after the commit, the bug was still present.
The change had no impact on the bug.
Don't inappropriately grant system_server the ability to minipulate
the scheduling priority of kernel threads.
This reverts commit 2d1650f407.
Change-Id: I59bdf26ad247a02b741af2fa58a18e7e83ef44d8
Generate a general_seapp_contexts file with only the
device-independent entries, similar to general_sepolicy.conf.
This is for use by CTS tests to compare with the prefix of
device seapp_contexts.
Change-Id: If8d1456afff5347adff7157411c6a160484e0b39
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Instead of displaying the boolean count, display a list of booleans
defined in the policy, if any. This makes sepolicy-analyze booleans
consistent with sepolicy-analyze permissive and allows automated tests
to simply check whether there was any output at all.
Change-Id: I221b60d94e6e7f6d80399bf0833887af3747fe83
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
- allow access for /data/system/packages.xml.
- deprecate access to /dev/logd_debug (can use /dev/kmsg for debugging)
- allow access to /dev/socket/logd for 'logd --reinit'
Bug: 19681572
Change-Id: Iac57fff1aabc3b061ad2cc27969017797f8bef54
Implement the booleans test in sepolicy-analyze so
that we can move the no-booleans check from the
SELinuxTest to the SELinuxHostTest along with the
other policy checks.
Change-Id: I95d7ad34da10c354470f43734d34a6ec631a7b4e
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
With the exception of the factory reset protection block device,
don't allow system_server to read or write to any other block
devices. This helps protect against a system->root escalation
when system_server has the ability to directly minipulate raw
block devices / partitions / partition tables.
This change adds a neverallow rule, which is a compile time
assertion that no SELinux policy is written which allows this
access. No new rules are added or removed.
Change-Id: I388408423097ef7cf4950197b79d4be9d666362c
Add neverallow rules to ensure that zygote commands are only taken from
system_server.
Also remove the zygote policy class which was removed as an object manager in
commit: ccb3424639821b5ef85264bc5836451590e8ade7
Bug: 19624279
Change-Id: I1c925d7facf19b3953b5deb85d992415344c4c9f
Allow system server to handle already open app unix_stream_sockets.
This is needed to support system_server receiving a socket
created using socketpair(AF_UNIX, SOCK_STREAM) and
socketpair(AF_UNIX, SOCK_SEQPACKET). Needed for future Android
functionality.
Addresses the following denial:
type=1400 audit(0.0:9): avc: denied { read write } for path="socket:[14911]" dev="sockfs" ino=14911 scontext=u:r:system_server:s0 tcontext=u:r:platform_app:s0:c512,c768 tclass=unix_stream_socket permissive=0
Bug: 19648474
Change-Id: I4644e318aa74ada4d98b7f49a41d13a9b9584f39
