| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
| |
Bug: 154796679
Test: m droid
Change-Id: I72fd2b0541693fdc2608e744a46229ce6914994b
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The main goal here is performance by avoiding JNI and especially
NativeAllocationRegistry overhead. We gain a factor of 10 or so
on small BigInteger arithmetic. For large computations, we gain
substantially in a few cases where OpenJDK seems to use better
algorithms. AFAIK, with this version we never lose by integral
factors relative to what we had before.
A secondary goal is to clean out open BigInteger bugs.
The base version is 8u252, which for this part of the tree is
identical to June 15 2020 ToT.
Note that this means we included the java.math part of
https://hg.openjdk.java.net/jdk8u/jdk8u/jdk/rev/a5f5d7fd9be6 (29 Oct 2019)
That appears to be a separable fix that makes no interface changes.
I re-added @NonNull annotations. We also removed some code to write
certain backwards-compatibility fields during serialization, since we
never had those.
This also adds a the <X>ValueExact Java 8 BigInteger methods that had
not been previously supported.
This removes a Harmony test for compatibility with the RI implementation
that the RI implementation itself apparently fails. And, IMO, the
observed behavior is better than what we previously tested for.
Fixed a bunch of comment spelling errors. All of these have now
been fixed in upstream ToT.
Benchmarking showed that a straight move to the RI version slowed down
large multiplications, divisions, and modular exponentiation enough to
make it problematic. Thus I reintroduced NativeBN to allow those to fall
back to boringssl, at least for sufficiently large inputs. It was moved
to a hopefully more appropriate location. The fallback is tuned for
64 bits; for 32 bits it's probably much less useful; much of the
boringssl performance advantage comes from 64-bit "digits".
The boringssl fallbacks are not completely free, since we need extra
conversions on each operation. But since we only do this for large
asymptotically expensive computations, we see at mosts tens of percents
regressions, which probably qualifies as "in the noise" here. If we
find additional performance issues, we can add more boringssl fallbacks;
the required code is now fairly straightforward.
Unlike the old version, this no longer uses NativeAllocationRegistry or
similar mechanisms at all. Native memory is only used on a short-term
basis, with explicit deallocation. We no longer use boringssl for
simple linear-time operations like addition.
Microbenchmark results for the newly added benchmark, and for a
close-to-final BigInteger version, listed as "combined":
Msecs/iteration
Digits: 5/10 50/100 500/1000 5000/10000
Inner product, 1000 terms, factors of larger indicated size:
current 2.1 2.5 6.4 168
RI 0.11 0.66 11.0 486
combined 0.12 0.67 9.1 189
Harmonic series, uses smaller indicated size
current 3.7 3.2 4.3 17.6
RI 0.16 0.34 1.4 14.3
combined 0.17 0.34 1.23 14.2
(1+1/100000)^100000, larger size
current 0.07 0.073 0.33 15.8
RI 0.011 0.049 1.553 63.6
combined 0.011 0.049 0.48 13.6
Single modPow() call, smaller size
current 0.005 0.011 1.1 583
RI 0.006 0.038 7.2 5580
combined 0.011 0.012 1.1 541
Single modInverse call, larger size
current 0.003 0.014 0.375 27.8
RI 0.003 0.003 0.026 1.6
combined 0.002 0.002 0.008 0.4
Bug: 160641142
Bug: 136887041
Bug: 119491938
Bug: 28214673
Bug: 28251030
Bug: 2950143
Test: AOSP Boots. Ran some manual Calculator tests on Cuttlefish.
Change-Id: Id577d99328013b1e3c389973dcb0195fa7f52b0b
|
| |\ |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This CL moves Flow.java to openjdk_javadoc_files which has the effect of
adding the interfaces contained in it to the public API surface.
The file derives from OpenJDK 9+181 and is identical to the upstream
version except that a piece of documentation referring to the
SubmissionPublisher class (not present on Android) has been removed.
The file contains only interfaces with no default methods, so there are no tests.
Fixes: 129875045
Test: Treehugger
Change-Id: Ie44055c927b2feb86f064526eca32851bfc34b76
|
| |/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This covers
- integration of all changes from OpenJDK 9+181 into
java.util.{AbstractList,List,Map,Objects,Set}.
- Map, List, Set gain static of() / ofEntries() factory methods.
- Objects gains misc static checker methods.
- When implementing RandomAccess, AbstractList.spliterator() as
well as subList.spliterator() is now built on top of random-access
List.get(int) rather than List.iterator().
- addition of various new files that are needed by those.
- addition of Nullablity annotations for the new APIs.
Changes to reference upstream versions
--------------------------------------
While the other files were previously derived from OpenJDK 8u121-b13,
Map.java was previously derived from OpenJDK 9b113+. After this CL,
all of the touched files derive from OpenJDK 9+181 with no further
changes waiting to be integrated; Android-changed and similar markers
note where Android carries local patches.
Notes on test coverage
----------------------
This CL adds test coverage for the new APIs/behavior.
The test coverage for Set.of() duplicates a bunch of logic from
the List.of() coverage. I experimented with ways to reuse code
but that reduced clarity and I didn't want to spend a lot of effort
on a major rearchitecture of the tests.
AbstractListTest asserts that iterator() is not called
while exercising basic existing Android spliterator, and checks
the fail-fast behavior guaranteed by the documentation.
Bug: 147483640
Test: atest CtsLibcoreTestCases:libcore.java.util.{Map,List,Set}OfTest
Test: atest CtsLibcoreTestCases:libcore.java.util.{AbstractList,Objects}Test
Test: atest CtsLibcoreTestCases
Test: Checked how the added @implSpec appear in generated docs: they
appear under headings "Implementation Requirements:".
Change-Id: I362ec405b270ba00fe3176dc19f08943b7d350d5
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This file is exactly identical to the copy in OpenJDK9+181,
checked out per instructions from:
https://android.googlesource.com/platform/libcore/+/7a23698de2d48fcc806413ff758e17232c3739aa/tools/upstream/src/main/java/libcore/CompareUpstreams.java#62
A future CL might add this class to the public API but for now, it is
listed in openjdk_internal_files which has the same effect as @hide
but allowed me to keep the file exactly identical to upstream.
Since this purely a collection of interfaces with no default methods,
no tests were added. Editing the documentation to remove references
to classes (SubmissionPublisher) that Android does not carry is left
to a follow-on CL.
Test: Treehugger
Bug: 129875045
Change-Id: I634b9bd479f7984d5263520f6922abb0207cf7bf
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
A change to add system_modules to droidstubs build rules allows the
same system modules to be used to create the stubs as will be used to
compile them. That avoids the need to duplicate the libs from the
system modules on the droidstubs and allows their visibility to be
restricted.
When sdk_version: "none" is specified it is required that the
system_modules property is also set; consistent with how it behaves on
java_library. That requires system_modules: "none" to be added to
droidstubs rules.
Bug: 142534789
Test: m checkbuild
built offline-sdk-docs without my changes and took base copy
repeated with my changes and compared with base copy
verified that only the timestamp.js file had changed.
Change-Id: Id62970a91e86253c122358c26891c9d9c8765507
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Extracts art_module_api_files filegroup from core_api_files to just
contain the files provided by the ART module.
Adds droidstubs and java_library definitions to generate the stubs
sources and compile them for each of the art, conscrypt and i18n
modules. Also adds a java_system_modules definition for art module
intra core api stubs to allow it to be used to build the i18n and
conscrypt libraries.
Removes generation of core-intra-stubs and instead builds
core.intra.stubs from the module specific libraries.
The art module reuses the API specification files (e.g.
mmodules/intracoreapi/api/intra/current-api.txt, the other modules
have their own directory.
The changes have been made in here to make reviewing easier. However,
the intent is that the i18n and conscrypt build rules and API
specification files will be moved to external/icu and
external/conscrypt respectively in a follow up change.
The core-libart-oj.depscheck build module has been modified to be
more consistent with core-all and consequently has to use the
i18n stubs source due to limitations with javac. A follow up change
will switch core-all over to build against i18n stubs source at which
point this will become unnecessary and will be removed.
Test: m checkbuild
Bug: 141935400
Change-Id: I8d820c0303614939ff8a042ba4f75511bebcdbe7
|
| |
|
|
|
|
|
|
|
|
| |
Use filegroup to specify sources files with base directory via 'path'
property. This allows the build system to filter Java source files by
their packages.
Bug: 140764681
Test: m
Change-Id: I52760b3cbe3c99f5c48473f667322d6102cb7f53
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The openjdk_internal_files and openjdk_mmodule_extra_files should be
disjoint but they actually overlap, with the following duplicate files:
ojluni/src/main/java/sun/misc/Cleaner.java
ojluni/src/main/java/sun/nio/ch/DirectBuffer.java
ojluni/src/main/java/sun/security/jca/Providers.java
ojluni/src/main/java/sun/security/pkcs/PKCS7.java
ojluni/src/main/java/sun/security/pkcs/SignerInfo.java
This removes the duplicates.
Bug: 138207002
Test: m droid
Change-Id: Ie20cb23302f8762c7683f60d28d88fe9829804a8
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Takes advantage of a new mechanism for specifying the default visibility
of all modules in a package to restrict the visibility of libcore
modules as much as possible:
package {default_visibility: ["//visibility:private"]}
Bug: 133290645
Bug: 134379140
Test: m droid
Change-Id: I2ba1a6131993d218fd65cd4a8f0ad29a7a83df14
|
| |
|
|
|
| |
Test: m droid
Change-Id: Ia3835306970445b6f224b1c329dc4e63b845e544
|
| |
|
|
|
|
| |
Bug: 134566750
Test: m
Change-Id: I154a1324deef7fdb9dabf51f2924f5efdec72aa5
|
| |
|
|
|
|
| |
Bug: 132757397
Test: TH
Change-Id: I1201b3e6116ba5f1578e800fd65ecdd2f13a6645
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Some code generation tools, notably dagger, produce output which
includes a @Generated annotation when built with javac -source 9
-target 9. This annotation is not part of any Android API, so
compilation fails.
As a workaround for this dagger behaviour, this change introduces a
stub for that annotations, available at compile time only, to keep
javac happy. (This topic should be reverted if dagger is fixed not to
use the annotation in its output.)
The upstream source for Generated.java is OpenJDK 9+181.
N.B. The problem this is fixing does not manifest in AOSP, only in
downstream branches.
Bug: 123891440
Test: EXPERIMENTAL_USE_OPENJDK9=true make droid
(cherry picked from commit 81dd1d9418f62f6888063aeec53e15d123948f0c)
Merged-In: I06ff179384160a598b70959a0b0cd82d439490ce
Change-Id: I55971ce65c74bac75b32f1258a96de7368ac97c0
|
| |
|
|
|
|
|
| |
These are used by the framework's protobuf implementation.
Test: compiles
Change-Id: I8a8aebbdd94c0012768b7a16732e708ee35714e6
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
1) Delete unused Docs.mk.
This removes the last reference to non_openjdk_java_files.bp and
openjdk_java_files.bp filegroups from .mk files.
This means that export_to_make_var can be removed in those files.
2) Introduce a filegroup, core_public_api_files, that can be used
from frameworks/base/ rather than "internal" variables defined in
non_openjdk_java_files.bp, openjdk_java_files.bp and conscrypt.
Bug: 126332291
Test: build only
Merged-In: I3c85d17dd7405e65a6be9e60831b1a07e4f08abf
Change-Id: I3c85d17dd7405e65a6be9e60831b1a07e4f08abf
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Adds stubs for ojluni classes some of whose hidden members are used by
applications. Each such member is annotated with the UnsupportedAppUsage
annotation.
The stubs are built as part of the core-oj-hiddenapi target from which
the build will extract information about the hidden class members that
are used by applications.
This was tested by making and then manually checking that the generated
out/target/common/obj/PACKAGING/hiddenapi-light-greylist.txt was the
same (after sorting) before and after this change.
Test: see above
Bug: 117818301
Change-Id: Icfe773e00ba1f96a379b705f26ef631ec34fee5e
|
| |
|
|
|
|
|
|
|
| |
Internal builds have more dependencies on core libs.
This adds required CorePlatformApi elements.
Bug: 113148576
Test: build (internal)
Change-Id: I48942689a4d0e7cb9cbd0a4254a0a7af3a23b261
|
| |
|
|
|
|
|
|
| |
These annotations are required for further targets to build.
Bug: 113148576
Test: build
Change-Id: I695c6b973835b5de2311e0ed99cb2fb55b69bc94
|
| |
|
|
|
|
|
|
|
| |
Tested by adding a dependency from framework to core.platform.api.stubs
and modifying the build to disable link type checking.
Test: make checkbuild
Bug: 113148576
Change-Id: I618ade912f8cb4429b441a610ea16a9aa16c3398
|
| |
|
|
|
|
|
|
|
|
|
| |
These were found while investigating targets besides
"framework" needed by targets that are
closely tied to framework. e.g.
framework/core/services, framework/opt/telephony.
Bug: 113148576
Test: build
Change-Id: Ic45624d1404e8ee31ddf6c975b4bd6a8d1f8710a
|
| |\ |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This change adds two new make targets which create empty stub files suitable
for copying into ojluni/annotations, which the developer can then annotate.
The instructions (and a description of the previously undocumented
mmodule subdirectory) are included in the README.
Bug: 64930165
Bug: 115746226
Test: Followed the steps added to the README
Change-Id: Iea070ffb97cd93731f24ba379fd3a38d0b70b9c6
|
| |/
|
|
|
|
|
|
|
|
|
|
|
| |
non-lambda filegroups were accidentally referring to the lambda path.
This was a harmless mistake; it should only affect internal paths used
during builds.
Since openjdk_java_files contains files in different source directories,
and since path is optional, this CL drops the attribute altogether.
Bug: 116082167
Test: Treehugger
Change-Id: I484384c63a94cd139523b74cc80f500a373a21a5
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
An initial set of methods that need to be visible in
the intra-core API surface for conscrypt to compile.
There's a fix in JavaLibrary.bp to the core-current-stubs-gen
target to stop files in openjdk_mmodule_extra_files showing
up in the core.current.stubs.
Bug: 110404540
Bug: 113148576
Test: build
Change-Id: I4bade4dac925cfcb14df9335918319be897ac50d
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This change prepares for adding mainline mode API annotations
(@CorePlatformApi etc.) supplied in external .annotated.java files so
they are not carried as patches in ojluni source files.
No real annotations are introduced in this change, and it has no
effect on the generated stubs for any of the affected targets (except,
seemingly, a few trivial whitespace changes).
Specifically, this change does the following:
- Adds a placeholder .annotated.java file (to be removed as soon as we
have any real ones).
- Adds a @libcore.api.Hide annotation we can use as an alternative to
the @hide javadoc tag.
- Changes the relevant mainline module API droidstubs rules to merge
in the .annotated.java annotations and to respect the @Hide
annotation.
- Refactors the openjdk_java_files targets to create a new category of
source file, those which are annotated for the mainline module APIs
but are not part of the public API. (The alternative to doing this
would be to move them into the public category and change all the
public API droidstubs to merge in the .annotated.java annotations
and to respect @Hide, which would be significantly more invasive.)
- Changes the relevant mainline module API droidstubs rules use the
new source categories.
Bug: 115746226
Test: `make core-platform-api-stubs core-intra-stubs core-libart+oj-intra-stubs`
Change-Id: I33e80603686112d3eb5a05debb58f13764b819b6
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The MethodHandleInfo class has a number of patches to inline constants
and methods from the upstream MethodHandleNatives class instead of just
including the MethodHandleNatives class. This change adds the
MethodHandleNatives class and reverts those patches. The
MethodHandleNatives class does require some patches in order to remove
code that does not compile or which is otherwise unused. However,
patches that remove code are generally less risky and easier to
maintain than patches that change the code so this seems like a better
trade off.
The refKind...() methods in MethodHandleInfo could not be removed
because they were accidentally added to the public API in API level 26.
The removal of those methods can be tracked under bug 114175595.
Bug: 74379469
Test: make checkbuild, flash and start up device
Change-Id: I843707564e9cf5c7fbf52bfa7ce4b6a2b10c0836
|
| |
|
|
|
|
|
|
| |
This was inherited from upstream but is unused in android due to
changes in java.lang.String et al.
Test: make checkbuild
Change-Id: Id6f9db95139d39c1c7dbc818ef757ed2f67028b2
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
These are present in preparation for OpenJDK 9 to allow VM support for
VarHandle accessors and fences to be developed. For this initial
import the VarHandle APIs are hidden and unimplemented.
The implementation path will follow the approach taken with the
MethodHandle API: public API compatability and alternative
implementation strategy based on VM functionality.
The files in this CL under ojluni/ define the APIs and are based on
the corresponding files in upstream OpenJDK 9+181.
Bug: 65872996
Test: m -j32
Change-Id: I678ab2239507b349a9701a4a6f5bbc92bf18b40d
|
| |
|
|
|
|
|
|
|
|
|
| |
This reverts commit 9762264100d81d3dd653cbca2996b288c52f9546.
Jack is disabled everywhere now. This reapplies
I20a0f0327172bc51db47c7c56ccc806e20cf1244 with updates for changes
to Soong and to add sources to the core-lambda-stubs jar.
Test: m -j checkbuild
Change-Id: I77c9dc558bc6c0f3833b1b79a0707c498e560628
|
| |
|
|
|
|
|
|
| |
This reverts commit 70d863fdc17969b4638caf830d0567adc20a6100.
Reason for revert: Broke the few remaining builds that still use jack
Change-Id: I437d583cf33d68e6d8fa3367e1ea54848b56e05b
|
|
|
See build/soong/README.md for more information.
Test: m -j checkbuild
Test: classes.dex for core-oj is identical
Test: classes.dex for core-libart only has extra bridge methods (b/65645120)
Change-Id: I0f81937dcc6efba87e7f61b2d5840b18068cd587
|
