1, /metadata/aconfig is the directory that stores aconfig storage
related protobuf files and flag value files boot copy. Grant read
access to everybody. But limit the write access only to init and
aconfig storage service process (to be created later)
2, /metadata/aconfig/flags is the sub directory that stores persistent
aconfig value files.Initially set it up to be accessible by
system_server process only . When aconfig storage service process is
created, will add another permission to storage service process.
Context to why we are hosting flag data on /metadata partition:
Android is adopting trunk stable workflow, flagging and A/B testing is
essential to every platform component. We need some place to host the
flag that are accessible to system processes that starts before /data
partition becomes available.
In addition, there has been a long discussion regarding utilizing
/metadata partition for some process data, another example is mainline
modules, we are trying to make them to be able to be mounted earlier,
but cannot due to /data availability.
Bug: 312444587
Test: m
Change-Id: I7e7dae5cf8c4268d71229c770af31b5e9f071428
Add neverallow reading these files because this came
up in a review recently, and they contain information
about processes which is important for security, so
we'd like to avoid accidentally granted these
permissions.
Fixes: 306036348
Test: build (is build time change)
Change-Id: I8b8917dacd2a65b809b7b6fb7c1869a3db94156b
The majority of code for media encoding and decoding occurs within the
context of client app processes via linking with libstagefright. This
code needs access to server-configurable flags to configure
codec-related features.
Bug: 234833109
Test: manual test with 'adb shell device_config' commands
Change-Id: I95aa6772a40599636d109d6960c2898e44648c9b
These were unnecessarily lax. Some additional places
additionally exclude only the generic proc type, but
we don't care about those places.
Bug: 281877578
Test: boot
Change-Id: I9ebf410c12a41888ab1f5ecc21c95c34fc36c0d0
Otherwise, we will encounter SELinux denials like:
W binder:6200_7: type=1400 audit(0.0:327): avc: denied { read } for name="PrebuiltGmsCoreNext_DynamiteLoader.apk" dev="dm-51" ino=2576 scontext=u:r:artd:s0 tcontext=u:object_r:privapp_data_file:s0:c512,c768 tclass=lnk_file permissive=0
Bug: 262230400
Test: No longer see such SELinux denials.
Change-Id: Iccb97b1973f8efbe859b59e729f7a0194d05ba5e
With AIDL fastboot, wiping partition will be handled by new service.
Allow hal_fastboot_server to the exception to neverallow rule.
Bug: 260140380
Test: th
Test: fastboot -w
Change-Id: Ic38ad715cb097ccd9c8936bb8e2a04e3e70b3245
Signed-off-by: Sandeep Dhavale <dhavale@google.com>
This way we can prevent private types (e.g., sdk_sandbox) from accessing
those properties.
Bug: 210811873
Test: m -j, boot device
Change-Id: Idbcc4928c8d0d433f819d8b114e84a5f09466ad0
Secondary dex files are in app data directories. In order to perform
secondary dex compilation, artd needs permissions to:
- Read secondary dex files
- Create "oat" dir
- Create a reference profile in "oat" dir
- Rename the reference profile
- Delete the reference profile
- Read the current profile in "oat" dir
- Delete the current profile
- Create compilation artifacts in "oat" dir
- Rename compilation artifacts
- Delete compilation artifacts
Bug: 249984283
Test: -
1. adb shell pm art optimize-package --secondary-dex -m speed-profile -f com.google.android.gms
2. See no SELinux denial.
Change-Id: I19a0ea7895a54c67959b22085de27d1d0ccc1efc
Reduce use of "exported_system_prop" by defining 2 new (currently
identical) "locale_prop" and "timezone_prop" types for the system
properties that are for "global system settings". See the comments in
private/property_contexts for details.
Initially the rights of the new types should be identical to
exported_system_prop but they will be reduced with a follow-up commit to
enable easier rollback / progress to be made on related work.
Bug: 236612872
Test: treehugger
Change-Id: I8d818342023bc462376c091b8a522532ccaf15d3
Manual testing protocol:
* Verify prng_seeder daemon is running and has the
correct label (via ps -Z)
* Verify prng_seeder socket present and has correct
label (via ls -Z)
* Verify no SELinux denials
* strace a libcrypto process and verify it reads seeding
data from prng_seeder (e.g. strace bssl rand -hex 1024)
* strace seeder daemon to observe incoming connections
(e.g. strace -f -p `pgrep prng_seeder`)
* Kill daemon, observe that init restarts it
* strace again and observe clients now seed from new instance
Bug: 243933553
Test: Manual - see above
Change-Id: I0a7e339115a2cf6b819730dcf5f8b189a339c57d
ro.secure and ro.debuggable system properties are not intended
to be visible via Android SDK. This change blocks untrusted
apps from reading these properties.
Test: android.security.SELinuxTargetSdkTest
Bug: 193912100
Change-Id: I40ac5d43da5778b5fa863b559c28e8d72961f831
- Adapt installd rules for app compilation.
- Add profman rules for checking the profile before compilation. This is new behavior compared to installd.
Bug: 229268202
Test: -
1. adb shell pm art optimize-package -m speed-profile -f \
com.google.android.youtube
2. See no SELinux denial.
Change-Id: Idfe1ccdb1b27fd275fdf912bc8d005551f89d4fc
If something starts before servicemanager does,
intelligently wait for servicemanager to start rather
than sleeping for 1s.
Bug: 239382640
Test: boot
Change-Id: If0380c3a1fce937b0939cd6137fcb25f3e47d14c
Removes all references to vendor_service in policy except the
definition of this type, which also needs to be removed by
clients.
We don't need this because interface type shouldn't be associated
with where they are served. We can serve HALs from anywhere if they are
implemented in software.
Bug: 237115222
Test: builds
Change-Id: If370a904af81e015e7e1f7a408c4bfde2ebff9a4
Currently, vendor_service is excluded from this neverallow
for the same reason. However, the current plan is to remove
vendor_service. Since some vendor HAL services are not
marked as hal_service_type, this part of the change needs
to be submitted independently in order to clean them up.
Bug: 237115222
Test: build
Change-Id: I7893184c4d1011881b721d0b851e07c17f73732b
The feature was superseded by tzdata mainline module(s).
Bug: 148144561
Test: see system/timezone
Test: m selinux_policy
Change-Id: I48d445ac723ae310b8a134371342fc4c0d202300
Merged-In: I48d445ac723ae310b8a134371342fc4c0d202300
Adding a new system property that will act as a toggle
enabling/disabling the framework changes that were submitted to prevent
leaked animators.
Bug: 233391022
Test: manual.
Merged-In: I57225feb50a3f3b4ac8c39998c47f263ae211b66
Change-Id: Ifc339efc1c3a5e19920b77d1f24bef19c39d5f44
Remove references in sepolicy. Leave a few of the types defined since
they're public and may be used in device-specific policy.
Bug: 211461392
Test: build/boot cuttlefish
Change-Id: I615137b92b82b744628ab9b7959ae5ff28001169
Creating a per-user encrypted directory such as /data/system_ce/0 and
the subdirectories in it too early has been a recurring bug. Typically,
individual services in system_server are to blame; system_server has
permission to create these directories, and it's easy to write
"mkdirs()" instead of "mkdir()". Such bugs are very bad, as they
prevent these directories from being encrypted, as encryption policies
can only be set on empty directories. Due to recent changes, a factory
reset is now forced in such cases, which helps detect these bugs;
however, it would be much better to prevent them in the first place.
This CL locks down the ability to create these directories to just vold
and init, or to just vold when possible. This is done by assigning new
types to the directories that contain these directories, and then only
allowing the needed domains to write to these parent directories. This
is similar to what https://r.android.com/1117297 did for /data itself.
Three new types are used instead of just one, since these directories
had three different types already (system_data_file, media_rw_data_file,
vendor_data_file), and this allows the policy to be a bit more precise.
A significant limitation is that /data/user/0 is currently being created
by init during early boot. Therefore, this CL doesn't help much for
/data/user/0, though it helps a lot for the other directories. As the
next step, I'll try to eliminate the /data/user/0 quirk. Anyway, this
CL is needed regardless of whether we're able to do that.
Test: Booted cuttlefish. Ran 'sm partition disk:253,32 private', then
created and deleted a user. Used 'ls -lZ' to check the relevant
SELinux labels on both internal and adoptable storage. Also did
similar tests on raven, with the addition of going through the
setup wizard and using an app that creates media files. No
relevant SELinux denials seen during any of this.
Bug: 156305599
Change-Id: I1fbdd180f56dd2fe4703763936f5850cef8ab0ba
The binder driver now advertises the features it supports through
individual files under /dev/binderfs/features/*. Let all domains have
access to these files to determine how to interact with the driver.
Bug: 191910201
Tested: clients are able to read feature files via libbinder
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Change-Id: Ice5de9efee74e571ef0a23ce093af162fc3b276e
Used for *TS testing to ensure that user devices do not multi-install
APEXes.
Bug: 216852347
Test: (non root) getprop | grep ro.boot.vendor.apex
Change-Id: Ibc670fefbf89c4a4c1fa5d2ab9d7784c04946690
Because mtectrl is a system internal domain, and we don't need to expose
the type to vendor.
Test: build and boot
Change-Id: Idb5c4a4c6f175e338722971944bf08ba99835476
As we need to create new sysprops for Bluetooth mainline
configs, we need to have a property context available to
vendors and be able to access configs from other packages.
Tag: #feature
Bug: 211570675
Test: Added overlays and logs
Change-Id: If9c61f251578b61c070619069519e0aa563a9573
Stop using these attributes since these will be removed soon.
Bug: 202520796
Test: source build/envsetup.sh && lunch aosp_x86_64 && m && launch_cvd
Change-Id: I61dffb482f4e952299156f34be642ae52fcbfeb3
Signed-off-by: Bart Van Assche <bvanassche@google.com>
Without this change, any crash from an executable in /data/local/tests
is incomplete. Specifically, function names are missing which makes
the crash nearly useless for debugging.
Bug: 197229540
Test: Used the crasher executable and copied it to /data/local/tests
Test: and verified that running it as root and shell results in
Test: tombstones that have full unwinds with function names.
Change-Id: Ic4862ca6ee9b02132a593ccd5fe26508ed5c8510
