NIAP certification requires that all cryptographic functions
undergo a self-test during startup to demonstrate correct
operation. init now performs this check during startup.
The self-test is forked from init. For the child process
to be able to request a reboot it needs permissions to
set the sys.powerctl property.
Bug: 119826244
Test: Built for walleye. When the BoringSSL self test was forced
to fail the device rebooted into the bootloader, as
expected.
Change-Id: I4171b1dd0a5e393252ae5c002171ac51c9cbb3e6
The SELinux policy language supports an expandattribute statement.
Similar to the C "inline" declaration, this expands the permissions
associated with types, instead of using the attribute directly. Please
see
1089665e31
for more detail on this language option.
Expansion of attributes causes consistency problems with CTS. If a
neverallow rule exists which refers to an expanded attribute, the CTS
neverallow test will fail, because the policy does not have the
attribute embedded in it. Examples:
* b/119783042 (fixed in 536d3413b8)
* b/67296580 (fixed in 6f7e8609f9)
* b/63809360 (fixed in 89f215e6a0)
etc...
Instead of waiting for the CTS test to fail, modify the Android.mk file
so that we do checks similar to CTS. This allows us to fail at compile
time instead of waiting for a CTS bug. For example, for b/119783042,
instead of the compile succeeding, it will now fail with the following
error message:
[ 70% 190/268] build out/target/product/crosshatch/obj/ETC/sepolicy_neverallows_intermediates/sepolicy_neverallows
FAILED: out/target/product/crosshatch/obj/ETC/sepolicy_neverallows_intermediates/sepolicy_neverallows
/bin/bash -c "(ASAN_OPTIONS=detect_leaks=0 out/host/linux-x86/bin/checkpolicy -M -c
30 -o out/target/product/crosshatch/obj/ETC/sepolicy_neverallows_intermediates/sepolicy_neverallows.tmp
out/target/product/crosshatch/obj/ETC/sepolicy_neverallows_intermediates/policy.conf ) &&
(out/host/linux-x86/bin/sepolicy-analyze
out/target/product/crosshatch/obj/ETC/sepolicy_neverallows_intermediates/sepolicy_neverallows.tmp
neverallow -w -f out/target/product/crosshatch/obj/ETC/sepolicy_neverallows_intermediates/policy_2.conf
|| ( echo \"\" 1>&2; echo \"sepolicy-analyze failed. This is most likely due to the use\" 1>&2;
echo \"of an expanded attribute in a neverallow assertion. Please fix\" 1>&2;
echo \"the policy.\" 1>&2; exit 1 ) ) &&
(touch out/target/product/crosshatch/obj/ETC/sepolicy_neverallows_intermediates/sepolicy_neverallows.tmp )
&& (mv out/target/product/crosshatch/obj/ETC/sepolicy_neverallows_intermediates/sepolicy_neverallows.tmp
out/target/product/crosshatch/obj/ETC/sepolicy_neverallows_intermediates/sepolicy_neverallows )"
libsepol.report_failure: neverallow violated by allow vold hal_bootctl_default:binder { call };
libsepol.check_assertions: 1 neverallow failures occurred
sepolicy-analyze failed. This is most likely due to the use
of an expanded attribute in a neverallow assertion. Please fix
the policy.
15:44:27 ninja failed with: exit status 1
Test: Revert 536d3413b8 and verify compile
fails as above.
Test: Compile succeeds
Bug: 119783042
Change-Id: I5df405b337bb744b838dadf53a2234d8ed94bf39
server_configurable_flags_data_file is used for storing server
configurable flags which have been reset during current booting.
system_server needs to read the data to perform related disaster
recovery actions.
For how the data is read, see SettingsToPropertiesMapper.java.
Test: build succeeds & manual on device
Change-Id: Ifa22aecc13af2c574579299d28433622abbe6b85
Hals have 3 attributes associated with them, the attribute itself, the
_client attribute, and the _server attribute. Only the server attribute
isn't expanded using the expandattribute keyword, and as a result, is
the only attribute which can be used in neverallow rules.
Fix neverallow rule to use hal_bootctl_server, which is not expanded,
instead of hal_bootctl.
Introduced in: https://android-review.googlesource.com/c/platform/system/sepolicy/+/777178
Test: policy compiles
Bug: 119500144
Change-Id: I8cff9cc03f4c30704175afb203c68f237fbd61ca
During the build process, use a temporary file until we've determined
that every step of the build process has completed. Failure to do this
may cause subsequent invocations of the make command to improperly
assume that this step ran to completion when it didn't.
Test: code compiles.
Change-Id: I9a28e653e33b61446a87278975789376769bcc6a
There is no real need to access the manifest.json (which is being
renamed in other CLs anyway). So remove the access to it.
Bug: 119672727
Test: m, installed on device, boots.
Change-Id: I2d82062031da36f871b2a64d97a50a6f1e6fc3dd
We introduced a new API to allow Device Owner to install an OTA file on disk.
This in turn requires system_server to be able to copy the OTA file to a known
OTA file location, call into update_engine to start the installation and let
update_engine to call back to the system_server to deliver any error conditions
asynchronously. This CL modifies the SELinux policy to allow these interaction.
Test: manual in TestDPC, CTS tests for negative cases: atest com.android.cts.devicepolicy.DeviceOwnerTest#testInstallUpdate
Change-Id: Id1fbea9111f753c5c80f270c269ecb9ef141cd79
Bug: 111173669
In recovery, everything is labeled rootfs, including
/system/bin/*. Allow postinstall to execute them in recovery.
Test: sideload
Bug: 116608795
Fixes: 119877813
Change-Id: I5682bdecd0df1cb9ff3bc968ea29449b0b8588f4
Commit ebc3a1a34c ("Move to ioctl
whitelisting for plain files / directories", Oct 10th), enabled ioctl
filtering on all files, including functionfs files. However, recovery
performs the ioctl FUNCTIONFS_ENDPOINT_DESC on functionfs files, so
allow it.
Addresses the following denial:
audit: type=1400 audit(673009.476:507811): avc: denied { ioctl } for pid=731 comm="recovery" path="/dev/usb-ffs/adb/ep1" dev="functionfs" ino=473 ioctlcmd=0x6782 scontext=u:r:recovery:s0 tcontext=u:object_r:functionfs:s0 tclass=file permissive=1
Test: policy compiles.
Bug: 119877813
Change-Id: I09715acc16ab319b8d8b1f233cefaec23a358962
Gathering file contexts for all APEXes there for easier auditing.
Test: m com.android.resolv
Bug: 119527674
Change-Id: I0f06c21c77f4b537e7c7d590204569f4531b5302
grant rw_dir_perms of dir
/data/server_configurable_flags to flags_health_check.te, in order to
enable flags_health_check to record reset flags data as file under this
dir for later use. See function:
server_configurable_flags::ServerConfigurableFlagsReset for how the
permission is used.
Test: manual on device
Change-Id: I1df7b8cadfbe279f26bf828e9e725ce170a376f7
By convention, auditallow statements are always placed in
userdebug_or_eng() blocks. This ensures that we don't inadvertently ship
audit rules on production devices, which could result in device logspam,
and in pathological situations, impact device performance (generating
audit messages is much more expensive than a standard SELinux check).
Bug: 117606664
Test: policy compiles.
Change-Id: I681ed73c83683e8fdbef9cf662488115f6e7a490
Commit b4f17069b3 ("sepolicy: Drop
BOARD_SEPOLICY_IGNORE/REPLACE support.", Mar 2015) made it a compile
time failure to use BOARD_SEPOLICY_REPLACE or BOARD_SEPOLICY_IGNORE.
As these restrictions have been in place since 2015, we can safely
assume all usages of this have been cleaned up, and there is no further
need to check for this.
8 lines deleted from Android.mk, 1720 lines to go.
Test: compiles.
Change-Id: I23249e4b2e9ec83cb6356a6c5a6e187ae1fc9744
Remove the permission to execute dex2oat from apps targetSdkVersion>28.
This has been historically used by ART to compile secondary dex files
but that functionality has been removed in Q and the permission is
therefore not needed.
Some legacy apps do invoke dex2oat directly. Hence allow (with audit) for
targetSdkVersion<= 28.
Test: atest CtsSelinuxTargetSdk25TestCases
Test: atest CtsSelinuxTargetSdk27TestCases
Test: atest CtsSelinuxTargetSdkCurrentTestCases
Bug: 117606664
Change-Id: I2ea9cd56861fcf280cab388a251aa53e618160e5
This reverts commit 92bde4b941.
Reason for revert: Rebooting after OTA fails due to the
filesystem still seeing the old label on the device.
Bug: 116528212
Bug: 119747564
Change-Id: Ib5f920f85c7e305e89c377369dca038d2c6c738c
Test: rollback change
This is used for querying the installed packages, as well as
coordinating the installations of packages.
Test: ran an app that queries PM, that queries apexd.
Bug: 117589375
Change-Id: I38203ffe6d0d312d6cc38e131a29c14ace0ba10c
This is world-readable so it can be checked in libc's process init.
Test: m
Test: flash sailfish
Bug: 117821125
Change-Id: Iac7317ceb75b5ad9cfb9adabdf16929263fa8a9d
system server reads this property to keep track of whether server
configurable flags have been reset during current boot.
system server needs this information to decide whether to perform
following disaster recovery actions on framework level.
the get_prop added in this cl in system_server.te is not grouped
in the same place as the set_prop in system_server.te in another
cl (https://android-review.googlesource.com/c/platform/system/sepolicy/+/828284).
This is because these 2 properties are serving for different purposes:
device_config_flags_health_check_prop is used to control features(so will be
all the future set_prop added by other feature teams under "# server configurable flags properties"),
while device_config_reset_performed_prop is used by our API's internal implementation.
So I feel like it might be clearer if I put this get_prop in a different place rather than
appending to "# server configurable flags properties".
Test: build suceeded.
Change-Id: I64379aa8f0bbe093969b98d62093696a32aabe59
I added ro.bionic.(2nd_)?_(arch|cpu_variant) to vendor system
properties. And have init to write them to files under dev/.
This change set SELinux rules for these properties and files.
For the system properties: vendor/default.prop will set them. init will
read them.
For the files /dev/cpu_variant:.*: init will write them. bionic libc
will read them. (Basically world readable).
This is to allow libc select the right optimized routine at runtime.
Like memcpy / strcmp etc.
Test: getprop to make sure the properties are set.
Test: ls -laZ to make sure /dev/cpu_variant:.* are correctly labeled.
Change-Id: I41662493dce30eae6d41bf0985709045c44247d3
The dynamic linker calls realpath(3) on paths found in the linker config
script. Since realpath() calls lstat() on the parent paths, not allowing
getattr on /apex and its subdirectories will cause selinux denial spam
whenever something is executed from APEXes.
Silence the spam by allowing getattr on apex_mnt_dir.
Bug: 117403679
Bug: 115787633
Test: m apex.test; m; device is bootable
Change-Id: Ic659582760a3ae146e73770266bc64332b36a97c
