As documented in art/runtime/jni/jni_internal.cc, ART has deviated
from the RI by using a 4-byte encoding instead of the 3-byte encoding
required by the JNI specification.
Some users are okay with this 4-byte encoding (where they control
both the reading and writing logic) but other users require
compatibility with the DataOutput/DataInput API contract, so this
change lets users request either behavior.
This change now exercises all tests in both 4-byte and 3-byte modes,
and exhaustively confirms that all valid code-points match the
DataOutput/DataInput contract when in 3-byte mode.
Benchmark results still show significant performance benefits when
using this 3-byte encoding over the upstream RI:
timeRead_Upstream_mean (ns): 5090068
timeRead_LocalUsing3ByteSequences_mean (ns): 1996032
timeRead_LocalUsing4ByteSequences_mean (ns): 1813250
timeWrite_Upstream_mean (ns): 3856276
timeWrite_LocalUsing3ByteSequences_mean (ns): 1632697
timeWrite_LocalUsing4ByteSequences_mean (ns): 886503
Bug: 236923096
Test: atest FrameworksCoreTests:CharsetUtilsTest
Test: atest FrameworksCoreTests:FastDataTest
Test: atest FrameworksCoreTests:XmlTest
Test: atest FrameworksCoreTests:BinaryXmlTest
Test: ./frameworks/base/libs/hwui/tests/scripts/prep_generic.sh little && atest CorePerfTests:FastDataPerfTest
Change-Id: Ibddd36410a0d4a909522de011f23a337b53d6889
The test performed a lot of repetitions of the same append call with a buffer that kept being extended which resulted in OOM errors, but now we use a new buffer each time.
Test: atest CharsetUtf8PerfTest
Bug: 235455275
Change-Id: I10427d315a6d0b45394c74d0c8642f0c085ceb07
JarFilePerfTest was incorrectly submitted and not working yet, and
SignaturePerfTest had a failing test case.
Test: atest SignaturePerfTest
Change-Id: I739a87c02fdb3dd8c82cd63c8187fa9a8bf05ca7
Dump last interfaces of the network information in dumpsys of
netpolicy.
Output:
Network to interfaces:
100: {rmnet_data1}]
101: {v4-wlan0, wlan0}]
Bug: 237640200
Test: FrameworksNetTests
Change-Id: I258f5565e62666cd71222073c692e45e2e5828e2
When the last item of sRunningAttributionSources is removed, the
current check condition will cause the process of removing
sRunningAttributionSources stopped and return directly, bring in weak
global reference table overflow.
Change the order for smooth removal.
Bug: 218794347
Test: run audio stress test pass
Change-Id: I39072f3b72f0be6d6debbcbc948d7a940ad92785
Extend the support of the exception list for SC-V2 and apps targeting
S and SC-V2.
Test: atest ActivityRecordTests
Test: com.android.server.wm.SplashScreenExceptionListTest
Bug: 231708538
Merged-In: I5412e81f70cbc9aac3861d13d85e199e949bedc7
Change-Id: I70a2aa4684c1267fe98e0e2260c61042db9c2e36
In that file, we can list methods which we want to compile but that our
boot image profiling implementation did not cover.
Test: m
Bug: 235557326
Change-Id: Id5241dac826b5d4f581573d8b9b556bde2dabdeb
RemoteAnimationController
Sometimes mPendingAnimations.remove will be called more than once in one cycle (line 255 - 270) of
onAnimationFinished, e.g., mPendingAnimations.remove being called again in
adapters.mAdapter.mCapturedFinishCallback.onAnimationFinished, resulting in
IndexOutOfBoundsException.
Bug:237989368
Change-Id: Ica771f1ae4cb4ddddb7684bd964504ca9f31ea12
Test: Monkey test
* changes:
Update method visibility for testing
Stop VPN profiles by exiting VpnRunner instead of prepareInternal
Update the prefix in keystore for app exclusion
RPC Parcels need to be interpreted in the context of an RPC session.
Otherwise, they don't make sense.
Bug: N/A
Test: N/A
Change-Id: I7a7b83341372faae357927f1b23be181b2f907af
A kernel binder object for libbinder can either be a binder object
or an FD.
Bug: N/A (from email thread)
Test: N/A
Change-Id: Iaa02595c5859879749f776222eccca52e782cbfb
