Regenerated sparse.apk with latest aapt2 and updated resource qualifier
to orientation from sdk version because config sdk version was fully
dropped in aapt2. Also added tests with runtime apk generated by Soong
build target to make these tests more comprehensive. Applied
parameterized test to avoid duplicate code.
Bug: b/249370927
Test: Added and verified affected atests pass
Change-Id: I94f6758d7862375fe4053f42fe89953d4afa48fd
The native code that underlies `Resources#getIdentifier()` did not take
sparse encoding into account when calculating offsets, resulting in
either garbage or SEGV_MAPERR whenever a resource is first encountered
in a sparsely encoded configuration.
Bug: 197976367
Test: atest libandroidfw_tests --host
Change-Id: Ib7550fe2e05005550f59129a06be5712b74bc9c8
Creates ApkAssets creation methods that allow an AssetsProvider to be
specified.
During idmap verification and creation, idmap2 currently opens the
target package and overlay package several times:
1) When the crc of the package is calculated in idmap2 verify
2) When the manifest of an overlay is parsed
3) When an ApkAssets is opened.
Opening large zip files (like framework-res.apk) is slow. If we opened
the zip one time as an ApkAssets, the resources.arsc would have to be
parsed (which means mmaping/unmapping and touching a lot of
resources.arsc pages). This would cause idmap2 to preform unnecessary
work just to check the crc of some files.
This change allows a ZipAssetsProvider to be created and then moved
for the creation of an ApkAssets. The zip file only needs to be opened
once and the resources.arsc is not parsed until reading resources is
actually necessary.
Bug: 172471315
Test: libandroidfw_tests
Test: CtsResourcesLoaderTests
Change-Id: I940bb2c13844c7f028776a623a9ecef45a4813bf
ResTable_config of every ResTable_type is read from device every time
AssetManager::RebuildFilterList is invoked. For large APKs (like
framework-res.apk), this causes a large number of page faults
when accessing the config from disk. The configs are also used in the
slow path of AssetManager::FindEntryInternal, which makes it even
slower. Instead cache the config on the TypeSpec of its ApkAsset.
Bug: 177247024
Test: libandroidfw_tests
Change-Id: I66d507c4eeb2399f7558f3d9dfc53c157129ada0
To make it easier to add the actor policy in a follow up CL,
move most of the policy handling to a central location.
The strings and transformation between strings and flags is
now handled in libidmap2policies, with libandroidfw
containing the single source of policy flags.
This also extracts all the test resource IDs into an R.h
so they can be swapped without having to edit a dozen files
each time.
Bug: 130563563
Test: m aapt2_tests idmapt2_tests and run from host test output
Test: atest libandroidfw_tests
Change-Id: Ie533c9cebf938215df7586f00c38763ae467e606
Now that RROs are loaded as shared libraries,
LoadedApk#makeApplication is attempting to find the onResourcesLoaded
method of the overlays. This is a performance hit and causes more
memory than necessary to be allocated during application start up.
Bug: 143314947
Test: com.android.performance.tests.HermeticMemoryTest
Change-Id: I3b8cd22dae83e0164d6678c80279f9fffceb34e6
This change allows RROs to reference their own internal resources as
expected.
Overlays are loaded as shared libraries so they can have their own
resource id space that does not conflict with the resource id space of
the target or other overlays.
References to overlay resources that override target resources now
appear as references to the target resources.
Overlay values that are inlined into the xml file specified using
android:overlayResources are now able to be used at runtime.
See go/rro-references for more information.
Bug: 135943783
Test: idmap2_tests
Test: libandroidfw_tests
Change-Id: Ie349c56d7fd3f7d94b7d595ed6d01dc6b59b6178
ResourceLoaders allow inserting another .apk/.arsc into AssetManager's
resource resolution search. The effect is similar to overlays,
where a entry of >= config later in the path list will return that
ApkAsset's resource value instead.
Because loading from an .arsc is supported, which doesn't contain
any actual files, ResourceLoader exposes loadDrawable and
loadXmlResourceParser to allow an application load those files from
anywhere or create them in code.
The data being loaded is either pushed into an .apk or .arsc that
mocks itself as the package being "overlaid" and is passed in
through ResourcesProvider, an interface with static methods that
supports loading from a readable path on disk or a FileDescriptor.
The APIs are accessed through a Context's getResources(), which
has been changed to be unique per "Context-scope", which is usually
the lifetime of the Java object. The exception is that Activities
who get their Resources object persisted across recreations
maintain that logic for persisting ResourceLoaders.
Bug: 135270223
Test: atest FrameworksResourceLoaderTests
Change-Id: I6929f0828629ad39a21fa155e7fec73bd75eec7d
Add a new, hidden method to AssetManager to extract a mapping
overlayable name -> overlayable actor for all <overlayable> blocks in a
package. [This will eventually be used to check if the caller of the OMS
AIDL API is the registered actor for a given overlay.]
Also, teach AssetManager2 to not accept packages that re-use the same
overlayable name. [Such packages have always been ill-formed.]
Bug: 123894537
Test: make libandroidfw_tests
Change-Id: I1117fd3503f04fe4c73eb7114901e022508f4d9e
Since Q will no longer have a product_services partition, remove instances of
the product_services policy across aapt2, androidfw, and idmap2.
Bug:122745343
Test: aapt2_tests and libandroidfw_tests
Change-Id: I97c223a0bf5a2eab95811e5f738b44af6335e0ea
Encode the actor and name attributes of <overlayable> elements in the
overlayable chunks. Two chunks cannot have the same name.
Bug: 110869880
Bug: 119390855
Test: aapt2_tests and libandroidfw_tests
Change-Id: I0ed95efef640b3927046b6155d1840f0d96030dd
This change defines two new chunks for encoding overlayable information.
RES_TABLE_OVERLAYABLE_POLICY_TYPE contains flags that represent
restrictions enforced on overlays that try to overlay a specific set of
resource ids. The chunk header is followed by ResTable_ref for each id
that belongs to the policy type. A policy chunk will be created for
every unique combination of policies that are defined in overlayable
declarations.
RES_TABLE_OVERLAYABLE_TYPE holds policy blocks. Since <overlayable>
does not currently have any attributes, only one overlayable block is
encoded in an APK.
This change also removes the SPEC_OVERLAYABLE flag because the runtime
does not use the flag, and the overlayable chunk encoding renders it
obsolete.
Bug: 110869880
Bug: 117545186
Test: libandroidfw_tests and aapt2_tests
Change-Id: I45ae9bf4176699f14c85e2b7a2e8560185d8a0b8
Add an iterator to LoadedPackage which allows the caller to iterate over
the resource IDs in the package. This will be used by idmap when
constructing the idmap file.
Bug: 78815803
Test: make libandroidfw_tests
Change-Id: Ia47daa21390d67ea2ef3665e88eb407837c4764f
AssetManager2 assumes that RES_TABLE_TYPE_SPEC_TYPEs must immediately
precede their associated RES_TABLE_TYPE_TYPEs. This is not correct.
RES_TABLE_TYPE_SPEC_TYPEs must precede their associated
RES_TABLE_TYPE_TYPEs, but they do not need to immediately precede them.
For example, this is what we currently expect:
RES_TABLE_TYPE_SPEC_TYPE id=1
RES_TABLE_TYPE_TYPE id=1
RES_TABLE_TYPE_SPEC_TYPE id=2
RES_TABLE_TYPE_TYPE id=2
but this is also valid:
RES_TABLE_TYPE_SPEC_TYPE id=1
RES_TABLE_TYPE_SPEC_TYPE id=2
RES_TABLE_TYPE_TYPE id=1
RES_TABLE_TYPE_TYPE id=2
Bug: 73052092
Test: make libandroidfw_tests
Change-Id: I1f3c43760f8108eee24c2c6ed7bc16f70e951c2b
AssetManager2 relied on creating a list of configurations
present in the resource table so as to avoid copying and
converting ResTable_config's from the APK on every
resource retrieval.
ResTable, however, had a better optimization that pruned
the configurations that didn't match the currently set
configuration. This vastly reduced the number of ResTable_configs
to test.
In this CL, AssetManager2 follows suite with this optimization
and only maintains the filtered ResTable_configs, falling back
to the slow path when the configuration is overridden.
Test: mma frameworks/base/libs/androidfw
Test: adb sync system data
Test: adb shell /data/benchmarktest64/libandroidfw_benchmarks/libandroidfw_benchmarks
Change-Id: Ib57b75fbb32e1d310eec146e5a12dfc6de4385f9
AssetManager2 relied on creating a list of configurations
present in the resource table so as to avoid copying and
converting ResTable_config's from the APK on every
resource retrieval.
ResTable, however, had a better optimization that pruned
the configurations that didn't match the currently set
configuration. This vastly reduced the number of ResTable_configs
to test.
In this CL, AssetManager2 follows suite with this optimization
and only maintains the filtered ResTable_configs, falling back
to the slow path when the configuration is overridden.
Test: mma frameworks/base/libs/androidfw
Test: adb sync system data
Test: adb shell /data/benchmarktest64/libandroidfw_benchmarks/libandroidfw_benchmarks
Change-Id: I5d46f8b005a37b72750d00bd75f090e7b5a36f60
AssetManager2 relied on creating a list of configurations
present in the resource table so as to avoid copying and
converting ResTable_config's from the APK on every
resource retrieval.
ResTable, however, had a better optimization that pruned
the configurations that didn't match the currently set
configuration. This vastly reduced the number of ResTable_configs
to test.
In this CL, AssetManager2 follows suite with this optimization
and only maintains the filtered ResTable_configs, falling back
to the slow path when the configuration is overridden.
Test: mma frameworks/base/libs/androidfw
Test: adb sync system data
Test: adb shell /data/benchmarktest64/libandroidfw_benchmarks/libandroidfw_benchmarks
Change-Id: I42d53f703623cf7c3f7cbcf194f5026ff540faf4
ApkAssets who have failed verification should still run for
compatibility. Not all resources are accessed, and therefore
errors in the APK are not necessarily fatal. However, this means
we must do bounds checks when retrieving resources, which is
slower.
Test: make libandroidfw_tests && $ANDROID_BUILD_TOP/out/host/<host>/nativetest64/libandroidfw_tests/libandroidfw_tests
Test: make libandroidfw_benchmarks && adb sync system && adb sync data && /data/benchmarktest64/libandroidfw_benchmarks/libandroidfw_benchmarks
Change-Id: I4cc926c064bca0491785d82cdac0419d74d7d9b0
This adds support for shared resource libraries in the new
ResTable/AssetManager implementation.
The dynamic package map encoded in resources.arsc is parsed
and stored with LoadedArsc, and combined to form a resolved table
in AssetManager2.
Benchmarks show that this implementation is an order of magnitude
faster on angler-userdebug (make libandroidfw_benchmarks).
Test: libandroidfw_tests
Change-Id: I57c80248728b63b162bf8269ac9495b53c3e7fa0
The multiwindow model and Resources-per-activity
model that came in N puts greater demands on AssetManagers.
They are created whenever window dimensions change, which
can be frequently. There is a need to be able to cheaply
create a new AssetManager for each Activity, which shares
a lot of underlying state.
In order to make the creation of AssetManagers cheap,
we need a new implementation of the native AssetManager
and ResTable to support immutable representations of
APKs. This new data structure/class is ApkAssets.
ApkAssets have the same functionality of an AssetManager, except
that they operate on a single APK, and they do not do any caching.
Once loaded, they are immutable.
ApkAssets will be exposed as a Java object, with its implementation in
native code. The existing Java StringBlock will be owned by ApkAssets,
which means that Strings can be shared across AssetManagers.
ApkAssets can be cached by the ResourcesManager. Creating an AssetManager
requires only a list of ApkAssets and a configuration.
AssetManager2 (named with the suffix '2' for now while transitioning
to the new implementation) caches bags that are accessed.
Since ApkAssets are expected to be kept around longer, they do more validation
of the resource table, which cause slower load times. Measured on an angler-userdebug,
loading the framework assets takes 11ms with ApkAssets, and 2ms with the old
AssetManager implementation.
The tradeoff is that there does not need to be any security checks once an ApkAssets
is loaded, and regular resource retrieval is faster. Measured on an angler-userdebug,
accessing resource (android:string/ok) with many locales takes 18us with AssetManager2,
and 19us with AssetManager (this is per resource, so these add up).
Test: make libandroidfw_tests
Change-Id: Id0e57ee828f17008891fe3741935a9be8830b01d
