Avoid generating build time noise so that real errors stand out.
https://en.wikipedia.org/wiki/Unix_philosophy
Rule of Silence
Developers should design programs so that they do not print
unnecessary output. This rule aims to allow other programs
and developers to pick out the information they need from a
program's output without having to parse verbosity.
Test: Info messages no longer show up at build time.
Bug: 115998215
Change-Id: I33c18e2c7d77ed1bb4132debe13de2ae0907c34c
This scripts checks for common problems with SELinux policy,
including:
- Declared types that are not assigned to any files
- Files that don't exist on a running device
- Rules defined in the wrong file
- Using the wrong version of _file_perms/_dir_perms
These are heuristics, mainly because it does not fully parse regular
expressions and because policy might still be needed even if the
relevant file does not exist on a single device. But it hopefully is
a start at helping cleanup policy.
Bug: 30003114
Bug: 70702017
Test: Run script on core and device-specific policy.
Test: Verify that most of its results are correct.
Change-Id: I1ded4e9b18816841198dcbf72da65f046441d626
The tags property is (and has always been) unused by Soong. The property has
been defined as a list of strings, and the `androidmk` converted any
LOCAL_MODULE_TAGS entries over to it, but we've never done anything with it.
In preparation for removing the definition from Soong, I'm removing it from all
Android.bp files in the tree.
Since this has never done anything, this is a no-op, but if you really did want
the Android.mk behavior, the proper way to define a module to be installed in
userdebug / eng builds is to use PRODUCT_PACKAGES_DEBUG or PRODUCT_PACKAGES_ENG
in the appropriate product makefile.
Change-Id: Ia9a9b1c35533e02047cbb183b317ab93f1eeec6b
Exempt-From-Owner-Approval: global no-op build change
Test: remove `tags` from Soong, see errors go away.
This script will build the SELinux policy for multiple targets in parallel.
To use it, run:
./build_policies.sh <Android root directory> <output directory> [specific targets to build]
If you do not pass any individual targets, it will build all targets it can find.
It will print out the list of failing targets. You can open up the corresponding log file in the output directory to see the exact errors.
This script is still a work in progress. It currently cannot discover all build targets (it misses ones "lunch" does not list).
Bug: 33463570
Test: Ran script to build multiple targets with and without failures.
Change-Id: Iee8ccf4da38e5eb7ce2034431613fe10c65696ab
This CL lists all the exported platform properties in
private/exported_property_contexts.
Additionally accessing core_property_type from vendor components is
restricted.
Instead public_readable_property_type is used to allow vendor components
to read exported platform properties, and accessibility from
vendor_init is also specified explicitly.
Note that whitelisting would be applied only if
PRODUCT_COMPATIBLE_PROPERTY is set on.
Bug: 38146102
Test: tested on walleye with PRODUCT_COMPATIBLE_PROPERTY=true
Change-Id: I304ba428cc4ca82668fec2ddeb17c971e7ec065e
Code review of:
- https://android-review.googlesource.com/#/c/platform/system/sepolicy/+/512420/
had some comments. These were addressed and upstreamed here:
- 65620e0f94
Bring these changes back into the AOSP tree.
Test: verify that output sorted device files did not change hashes when built.
Change-Id: I7f07d3f74923cf731e853629034469784fc669f7
Signed-off-by: William Roberts <william.c.roberts@intel.com>
Update to commit:
- 5490639ac9
This solves all reported clang analyzer issues and is inline with upstream.
Test: veerify that md5sum of output files do not change.
Change-Id: I942145b8f9748c8ecd185f730c94d57cb77f5acc
Signed-off-by: William Roberts <william.c.roberts@intel.com>
Static analyzer complains the memory pointed by list in bcurrent is not
deallocated before returning. But since this code is in "main" routine,
we don't care about the deallocation. Just ignore the warning.
Bug: b/27101951
Test: Verified warning is gone.
Change-Id: I58d784b61a5ad30d1406bd7c6b28c5713abf2b34
Fix the following warnings:
system/sepolicy/tools/sepolicy-analyze/neverallow.c:346:9: warning:
Potential leak of memory pointed to by '__s1'
system/sepolicy/tools/sepolicy-analyze/neverallow.c:346:9: warning:
Potential leak of memory pointed to by 'id'
system/sepolicy/tools/sepolicy-analyze/neverallow.c:364:13: warning:
Potential leak of memory pointed to by 'classperms'
system/sepolicy/tools/sepolicy-analyze/neverallow.c:364:13: warning:
Potential leak of memory pointed to by 'node'
Bug: b/27101951
Test:Warnings are gone.
Change-Id: Ib9b2e0b9f19950b4b764d438ee58340e6c022ef5
Use the getline API correctly: keep a single buffer as long as
possible, and let the callee handle re-allocation. Move the final
free out of the loop.
Release the head of the linked list.
Bug: 37757586
Test: ASAN_OPTIONS= SANITIZE_HOST=address mmma system/sepolicy
Change-Id: I42424acba7cd68c1b9a7a43e916a421ac3e253f7
Empty typeset is not an issue in neverallow rules. The reason is that
it's completly normal for scontext or tcontext of neverallow rules to
evaluate to an empty type set. For example, there are neverallow rules
whose purpose is to test that all types with particular powers are
associated with a particular attribute:
neverallow {
untrusted_app_all
-untrusted_app
-untrusted_app_25
} domain:process fork;
Test: sepolicy-analyze neverallow -w -n \
'neverallow {} {}:binder call;'
produces empty output instead of "Warning! Empty type set"
Bug: 37357742
Change-Id: Id61b4fe22fafaf0522d8769dd4e23dfde6cd9f45
This could be useful in diffs between policy versions.
Bug: 37357742
Test: sepolicy-analyze lists all attributes in precompiled_policy.
Change-Id: I6532a93d4102cf9cb12b73ee8ed86ece368f9131
The sepolicy version takes SDK_INT.<minor> format. Make sure our
'current' policy version reflects the format and make it '100000.0'.
This ensures any vendor.img compiled with this will never work with
a production framework image either.
Make version_policy replace the '.' in version by '_' so secilc is
happy too.
This unblocks libvintf from giving out a runtme API to check vendor's
sepolicy version. The PLAT_PUBLIC_SEPOLICY_CURRENT_VERSION will
eventually be picked up from the build system.
(cherry-pick of commit 42f95984b5)
Bug: 35217573
Test: Build and boot sailfish.
Boot sailfish with sepolicy compilation on device.
Signed-off-by: Sandeep Patil <sspatil@google.com>
Change-Id: Ic8b6687c4e71227bf9090018999149cd9e11d63b
The sepolicy version takes SDK_INT.<minor> format. Make sure our
'current' policy version reflects the format and make it '100000.0'.
This ensures any vendor.img compiled with this will never work with
a production framework image either.
Make version_policy replace the '.' in version by '_' so secilc is
happy too.
This unblocks libvintf from giving out a runtme API to check vendor's
sepolicy version. The PLAT_PUBLIC_SEPOLICY_CURRENT_VERSION will
eventually be picked up from the build system.
Bug: 35217573
Test: Build and boot sailfish.
Boot sailfish with sepolicy compilation on device.
Signed-off-by: Sandeep Patil <sspatil@google.com>
Change-Id: Ic8b6687c4e71227bf9090018999149cd9e11d63b
added checkfc options 'l' and 'v' to verify hwservice_manager_type
and vndservice_manager_type on service context files, respectively.
The checkfc call to verify the new hwservice_contexts files will
be added together with hwservicemanager ACL CLs later.
Bug: 34454312
Bug: 36052864
Test: device boots, works
Change-Id: Ie3b56da30be47c95a6b05d1bc5e5805acb809783
sepolicy-analyze allows users to see all types that have a given
attribute, but not the reverse case: all attributes of a given type.
Add a '--reverse' option which enables this, but keeps the previous
interface.
Usage: sepolicy-analyze sepolicy attribute -r init
Bug: 36508258
Test: Build and run against current policy.
(cherry picked from commit d444ebedac)
Change-Id: I9813ebf61d50fb5abbc8e52be4cf62751979bbd4
sepolicy-analyze allows users to see all types that have a given
attribute, but not the reverse case: all attributes of a given type.
Add a '--reverse' option which enables this, but keeps the previous
interface.
Usage: sepolicy-analyze sepolicy attribute -r init
Bug: 36508258
Test: Build and run against current policy.
Change-Id: Ice6893cf7aa2ec4706a7411645a8e0a8a3ad01eb
untrusted_v2_app is basically a refinement of untrusted_app with legacy
capabilities removed and potentially backwards incompatible changes.
This is not currently hooked up to anything.
Bug: 33350220
Test: builds
Change-Id: Ic9fad57476bc2b6022b1eaca8667bf6d844753c2
This new input selector allows phasing in new security policies by
giving app developers an opportunity to make any needed compatibility
changes before updating each app's targetSdkVersion.
When all else is equal, matching entries with higher
minTargetSdkVersion= values are preferred over entries with lower
minTargetSdkVersion= values.
Test: Marlin builds and boots. Apps targeting targetSdkVersion<=25
run in untrusted_app_25 domain. Apps targeting the current development
build >=26 run in the untrusted_app domain with fewer permissions. No
new denials observed during testing.
Bug: 34115651
Change-Id: I14bf4f51dbe26cb9bd3f62ad0b281085441d9806
