OSDN Git Service
Narayan Kamath [Thu, 20 Oct 2016 14:39:10 +0000 (14:39 +0000)]
Merge "Revert "Revert "Interpreter: Add support for direct handle invokes on methods."""
Roland Levillain [Thu, 20 Oct 2016 11:59:42 +0000 (11:59 +0000)]
Merge "ARM: VIXL32: Pass initial ART tests with new code generator."
Treehugger Robot [Thu, 20 Oct 2016 11:59:27 +0000 (11:59 +0000)]
Merge "DexCache: Add support for updating MethodType references during GC."
Treehugger Robot [Thu, 20 Oct 2016 11:45:05 +0000 (11:45 +0000)]
Merge "Return error message if IndirectReferenceTable construction fails."
Narayan Kamath [Thu, 20 Oct 2016 09:57:45 +0000 (10:57 +0100)]
Revert "Revert "Interpreter: Add support for direct handle invokes on methods.""
This reverts commit
cfa61ad52077df66a448b84c360b12bc6f0e3f51.
Test flakiness in 956-methodhandles is fixed by
d08e39b6f02368aaa668b5aae6b6077b3eb44d9c.
Test: make test-art-host
Change-Id: I56e02e9a5bbc2b992cf746a92fd95ea77d32456c
Narayan Kamath [Wed, 19 Oct 2016 13:16:35 +0000 (14:16 +0100)]
DexCache: Add support for updating MethodType references during GC.
Fixes flakiness in 956-methodhandles
Test: make test-art-host
bug:
30550796
Change-Id: Ia5359390d59e65fc6efc16cc55c11bc65029104d
Vladimir Marko [Thu, 20 Oct 2016 09:24:03 +0000 (09:24 +0000)]
Merge "Change pResolveString entrypoint to kSaveEverything."
Narayan Kamath [Thu, 20 Oct 2016 07:49:22 +0000 (07:49 +0000)]
Merge "Build: Avoid using ppid to generate temporary directory name."
Treehugger Robot [Thu, 20 Oct 2016 03:26:21 +0000 (03:26 +0000)]
Merge "ART: Harden dex-file verifier wrt/ LEB128"
Mathieu Chartier [Thu, 20 Oct 2016 00:25:39 +0000 (00:25 +0000)]
Merge "Persistent bitmap for region space"
Andreas Gampe [Sat, 3 Sep 2016 01:12:00 +0000 (18:12 -0700)]
ART: Harden dex-file verifier wrt/ LEB128
Use out-of-bounds checks when decoding (U)LEB128.
Bug:
31252966
Test: m test-art-host
Change-Id: Id6dbbceb37c4d8d698e7685c61bf31912e3bf69e
Mathieu Chartier [Fri, 7 Oct 2016 22:24:46 +0000 (15:24 -0700)]
Persistent bitmap for region space
Fix bug where region space was not safely walkable due to holes
in the unevac regions possibly having dangling class pointers.
No preformance change, RAM overhead 1.1-1.2% .Heap (non LOS).
Test: test-art-host
Test: https://android-review.googlesource.com/#/c/288907/
Bug:
31522820
Change-Id: Ic4f8b7175e117689cb1ce3e28b082cf63f1f7b5a
Mathieu Chartier [Wed, 19 Oct 2016 19:35:54 +0000 (19:35 +0000)]
Merge "Move most of class linker to ObjPtr"
Treehugger Robot [Wed, 19 Oct 2016 18:53:06 +0000 (18:53 +0000)]
Merge "Fix bug in dynamic BCE. With regression test."
Aart Bik [Tue, 18 Oct 2016 18:35:22 +0000 (11:35 -0700)]
Fix bug in dynamic BCE. With regression test.
Rationale:
We implicitly assumed anything on a fixed offset to a suitable
candidate could be picked for codegen. Although this is usually
true, the scanner found an exception (on types). So, it is best
to make the test explicit. Note: some type implications are TBD.
Bug:
32193474
Test: test-art-host and offending APKs
Change-Id: I664a2caa97a7f231509f8832b940dd72da11a372
Mathieu Chartier [Tue, 18 Oct 2016 23:27:40 +0000 (16:27 -0700)]
Move most of class linker to ObjPtr
Return values are still mirror pointer. Fix some failing asserts in
compiler driver and tests.
Bug:
31113334
Test: test-art-host
Change-Id: I4450bf9dfb2541749496b8388616e8aae8488919
Vladimir Marko [Fri, 30 Sep 2016 13:50:51 +0000 (14:50 +0100)]
Change pResolveString entrypoint to kSaveEverything.
Test: Run ART test suite including gcstress on host and Nexus 9.
Test: Run ART test suite including gcstress with baker CC on host and Nexus 9.
Bug:
20323084
Change-Id: I63c21a7d3be8ff7a5765b5003c85b5317635efe6
Mathieu Chartier [Wed, 19 Oct 2016 17:01:02 +0000 (17:01 +0000)]
Merge "Delete unused blocked_register_pairs_ in code generators"
David Sehr [Wed, 19 Oct 2016 16:18:50 +0000 (16:18 +0000)]
Merge "Remove mirror:: and ArtMethod deps in utils.{h,cc}"
Richard Uhler [Tue, 11 Oct 2016 14:06:38 +0000 (15:06 +0100)]
Return error message if IndirectReferenceTable construction fails.
Previously if there was an error when constructing the
IndirectReferenceTable, the error message was lost. Now expose and
include the error message when throwing an exception related to
failures to construct the IndirectReferenceTable.
The error message is propagated through JVMEnvExt, JavaVMExt, and
Runtime::Init as well.
Bug:
32013594
Test: Added new 151-OpenFileLimit runtest.
Test: m test-art-host, m test-art-target
Change-Id: I3692f6928c9570358571bce634569d6f14cdeb05
Treehugger Robot [Wed, 19 Oct 2016 14:55:34 +0000 (14:55 +0000)]
Merge "Revert "Interpreter: Add support for direct handle invokes on methods.""
Narayan Kamath [Wed, 19 Oct 2016 10:05:04 +0000 (11:05 +0100)]
Build: Avoid using ppid to generate temporary directory name.
When TMPDIR is left unset, use the BSD checksum of
$(ANDROID_BUILD_TOP)-$(USER) instead of the PPID to construct a root
directory for all test runs. This workaround lets us run tests from
multiple repos at the same time, and will also avoid having to deal
with path length issues on the build bots.
This is a hacky workaround, but the only proper fix is stop using a
test runner built on top of make.
Bug:
32165422
Test: test-art-host
Change-Id: I61bc1eb3cf57d498f910573926d826a36020067a
Narayan Kamath [Wed, 19 Oct 2016 13:21:59 +0000 (13:21 +0000)]
Revert "Interpreter: Add support for direct handle invokes on methods."
This reverts commit
e5cd2cd7d0f8e1332f25edfd2798d84fec871f10.
Reverting due to flakiness in 956-methodhandles. Flakiness will be
fixed by
67f0929710a4489e3603383e57c9d236920654aa but might take time
to submit.
Change-Id: Ifdda5d3caca62d16d21cefaab7742fb2e8ca6969
Roland Levillain [Wed, 19 Oct 2016 12:19:03 +0000 (12:19 +0000)]
Merge "Remove libcore tests failures expectations for the concurrent collector."
Treehugger Robot [Wed, 19 Oct 2016 11:32:59 +0000 (11:32 +0000)]
Merge "Fix string compression, add tests."
Narayan Kamath [Wed, 19 Oct 2016 10:30:57 +0000 (10:30 +0000)]
Merge "Interpreter: Add support for direct handle invokes on methods."
Treehugger Robot [Wed, 19 Oct 2016 10:04:50 +0000 (10:04 +0000)]
Merge "Blacklist test that has been flaky for days."
Vladimir Marko [Mon, 17 Oct 2016 13:45:29 +0000 (14:45 +0100)]
Fix string compression, add tests.
Fix String.compareTo() for interpreter; memcmp() does not
return the required result (only the right sign).
Fix x86-64 stub where the assembler silently accepted and
generated bad code for out-of-range JECXZ.
Add extensive tests for String.equals(), String.compareTo()
and String.indexOf().
Bug:
31040547
Test: Run ART test suite including interpreter tests on host and Nexus 9.
Test: Ditto with string compression enabled.
Change-Id: I21b7a74da8a577c8fbaf8d9225f048550236d414
Treehugger Robot [Wed, 19 Oct 2016 09:29:36 +0000 (09:29 +0000)]
Merge "Release version 0.8 of ahat."
Nicolas Geoffray [Wed, 19 Oct 2016 08:32:10 +0000 (09:32 +0100)]
Blacklist test that has been flaky for days.
Change-Id: Ic3f4a10568b03f02c2173fd18e7642bb9ae4e2b3
Treehugger Robot [Wed, 19 Oct 2016 02:00:08 +0000 (02:00 +0000)]
Merge "ART: Remove IRT cruft"
David Sehr [Thu, 13 Oct 2016 16:12:37 +0000 (09:12 -0700)]
Remove mirror:: and ArtMethod deps in utils.{h,cc}
The latest chapter in the ongoing saga of attempting to dump a DEX
file without having to start a whole runtime instance. This episode
finds us removing references to ArtMethod/ArtField/mirror.
One aspect of this change that I would like to call out specfically
is that the utils versions of the "Pretty*" functions all were written
to accept nullptr as an argument. I have split these functions up as
follows:
1) an instance method, such as PrettyClass that obviously requires
this != nullptr.
2) a static method, that behaves the same way as the util method, but
calls the instance method if p != nullptr.
This requires using a full class qualifier for the static methods,
which isn't exactly beautiful. I have tried to remove as many cases
as possible where it was clear p != nullptr.
Bug:
22322814
Test: test-art-host
Change-Id: I21adee3614aa697aa580cd1b86b72d9206e1cb24
Mathieu Chartier [Tue, 18 Oct 2016 20:24:05 +0000 (13:24 -0700)]
Delete unused blocked_register_pairs_ in code generators
Legacy code for compatibility with quick?
Test: test-art-host CC
Change-Id: I9de261daea67dfd9bd3df89826ba9d10f135e29e
Mathieu Chartier [Tue, 18 Oct 2016 20:39:28 +0000 (20:39 +0000)]
Merge "Update the Begin() of LargeObjectMapSpace"
Roland Levillain [Tue, 18 Oct 2016 18:07:49 +0000 (19:07 +0100)]
Remove libcore tests failures expectations for the concurrent collector.
Test: art/tools/run-libcore-tests.sh --mode=host --variant=X32 --debug (with ART_USE_READ_BARRIER=true)
Bug:
12687968
Change-Id: I470f11d0bc6e38011c5b8f3797f2c0e9ec1e56cc
Treehugger Robot [Tue, 18 Oct 2016 18:02:37 +0000 (18:02 +0000)]
Merge "Enable last value generation of periodic sequence."
Mathieu Chartier [Tue, 18 Oct 2016 17:40:52 +0000 (17:40 +0000)]
Merge "Move art/native to ObjPtr"
Mingyao Yang [Tue, 18 Oct 2016 17:32:46 +0000 (17:32 +0000)]
Merge "More store/allocation elimination for singletons in case of loops"
Treehugger Robot [Tue, 18 Oct 2016 17:32:14 +0000 (17:32 +0000)]
Merge changes If0d5eeeb,Id927f2b5
* changes:
ART: Fix and extend test 904
ART: Better handle multiple event enables
Andreas Gampe [Tue, 18 Oct 2016 17:13:02 +0000 (17:13 +0000)]
Merge "ART: Expect a null self in SetEntrypointsInstrumented"
Andreas Gampe [Tue, 18 Oct 2016 17:03:42 +0000 (10:03 -0700)]
ART: Expect a null self in SetEntrypointsInstrumented
Follow-up to commit
157c77eb9a5d6a7fe6ebcc29afa7f1913a877c87.
Bug:
31684277
Test: m ART_TEST_GC_STRESS=true test-art-host
Change-Id: I9de848ea6a9484798b75a594f3e390ef27394743
Narayan Kamath [Fri, 14 Oct 2016 17:49:12 +0000 (18:49 +0100)]
Interpreter: Add support for direct handle invokes on methods.
Constructors are still not supported and will be implemented in a
follow up change.
Also adds a new run-test that compiles from source using an alpha
version of Jack.
Test: make test-art-host
bug:
30550796
Change-Id: I3212c032a5885854f691156c29f12613e5194667
Aart Bik [Fri, 14 Oct 2016 16:49:42 +0000 (09:49 -0700)]
Enable last value generation of periodic sequence.
Rationale:
This helps to eliminate more dead induction. For example,
CaffeineLogic when compiled with latest Jack improves with
a 1.3 speedup (2900us -> 2200us) due to eliminating first
loop (second loop can be removed also, but for a later
case). The currently benchmarks.dex has a different construct
for the periodics, however, still to be recognized.
Test: test-art-host
Change-Id: Ia81649a207a2b1f03ead0855436862ed4e4f45e0
Richard Uhler [Tue, 18 Oct 2016 15:26:50 +0000 (16:26 +0100)]
Release version 0.8 of ahat.
Test: m ahat-test
Change-Id: I964f86fc66f901bff7ddeda81dd42a6e658fda9d
Andreas Gampe [Tue, 18 Oct 2016 01:05:19 +0000 (18:05 -0700)]
ART: Fix and extend test 904
Fix using the wrong thread when unregistering. The inner thread
should not use Thread.currentThread (copy-n-paste mistake).
Add test for registering in OnLoad.
Add test for keeping registered on exit.
Bug:
31684277
Bug:
31684920
Test: m test-art-host-run-test-904-object-allocation
Change-Id: If0d5eeeb864684ab42eeaf62698d1e639f2071d6
Andreas Gampe [Tue, 18 Oct 2016 15:15:56 +0000 (15:15 +0000)]
Merge changes I7061c450,Ie8f8f24b,Ia6cee2cb
* changes:
ART: Do not free allocation listener in heap destructor
ART: Fix allocation entrypoint instrumentation
ART: Fix lock order issue in allocation tracking
Narayan Kamath [Tue, 18 Oct 2016 09:49:15 +0000 (09:49 +0000)]
Merge "Revert "Revert "Revert "Temporarily blacklist 955-method-handles-smali.""""
Narayan Kamath [Tue, 18 Oct 2016 09:23:04 +0000 (09:23 +0000)]
Merge "Fix handling of dex cache arrays for method types."
tony.ys_liu [Mon, 17 Oct 2016 03:18:09 +0000 (11:18 +0800)]
Update the Begin() of LargeObjectMapSpace
Root cause:
LargeObjectMapSpace's begin_ is initialized by nullptr.
so, it's always 0 after std::min
Solution:
Consider the nullptr case when update begin_
Notes:
Remove the nullptr check for end_ since it's not needed
Test: make test-art-host-gtest-large_object_space_test
Change-Id: I9918c30916e5d359f0ab23ec2bdb0fc9f093a85f
Andreas Gampe [Tue, 18 Oct 2016 03:12:52 +0000 (20:12 -0700)]
ART: Remove IRT cruft
Remove dead code for initial vs maximum table size.
Bug:
32125344
Test: m test-art-host
Change-Id: Ie5806da7f5f3238483da918deca5982f01764466
Andreas Gampe [Tue, 18 Oct 2016 00:49:59 +0000 (17:49 -0700)]
ART: Better handle multiple event enables
Only do extra work like allocation listener registration when the
global state changes.
Bug:
31684920
Test: m test-art-host
Change-Id: Id927f2b504e02f2d68bc6e4af2658a6017920d8a
Andreas Gampe [Tue, 18 Oct 2016 00:46:15 +0000 (17:46 -0700)]
ART: Do not free allocation listener in heap destructor
Do not deallocate on exit. Libraries are freed only after the heap
has already been destroyed.
Bug:
31684277
Bug:
31684920
Test: m test-art-host
Change-Id: I7061c4500a9ad314e2c00d7d722bb3372dff4810
Andreas Gampe [Tue, 18 Oct 2016 00:44:41 +0000 (17:44 -0700)]
ART: Fix allocation entrypoint instrumentation
Fix the unstarted-runtime case. Special-case the reset call to
the current thread, only.
Bug:
31684277
Test: m test-art-host
Change-Id: Ie8f8f24b2a6041344266a4e6b6bf8ddd9a0352a5
Andreas Gampe [Tue, 18 Oct 2016 00:40:27 +0000 (17:40 -0700)]
ART: Fix lock order issue in allocation tracking
(Un)instrumenting requires the instrumenting lock, which must be
acquired before the mutator lock. As the plugin can be called both
with and without the lock held, be careful. For simplicity, acquire
the lock (potentially) and immediately suspend.
Bug:
31684277
Test: m test-art-host
Change-Id: Ia6cee2cbe90f13f5543bdfea815d469b28a0f8f4
Mathieu Chartier [Mon, 17 Oct 2016 22:46:31 +0000 (15:46 -0700)]
Move art/native to ObjPtr
Bug:
31113334
Test: test-art-host
Change-Id: I67eb89cf042c762c6dcd5eb8b008b9a28e9b3319
Treehugger Robot [Mon, 17 Oct 2016 23:08:41 +0000 (23:08 +0000)]
Merge "event interface going private"
Treehugger Robot [Mon, 17 Oct 2016 22:47:01 +0000 (22:47 +0000)]
Merge "Remove logging dependency on runtime"
Treehugger Robot [Mon, 17 Oct 2016 21:56:12 +0000 (21:56 +0000)]
Merge "Fix invoke-virtual not throwing ICCE in some cases"
Mark Salyzyn [Mon, 17 Oct 2016 21:11:20 +0000 (14:11 -0700)]
event interface going private
Test: compile
Bug:
30465923
Change-Id: I704e0e6ae886207a9625fcb59be4876307d21e5f
David Sehr [Mon, 17 Oct 2016 17:09:33 +0000 (10:09 -0700)]
Remove logging dependency on runtime
Moved the abort backtracing function to runtime, forcing callers to
supply the aborter at InitLogging. This makes runtime properly layer
on top of logging by removing the cyclic dependency.
Bug:
22322814
Test: test-art-host
Change-Id: I8b2e72174e937bb88fe1bddd6d04b564cfb011a9
Mathieu Chartier [Mon, 17 Oct 2016 20:29:37 +0000 (20:29 +0000)]
Merge "Move most mirror:: args to ObjPtr"
Alex Light [Mon, 17 Oct 2016 20:02:39 +0000 (13:02 -0700)]
Fix invoke-virtual not throwing ICCE in some cases
Due to an oversight invoke-virtual on an interface method would
not cause an ICCE to be thrown if the target method is default. This
could potentially cause incorrect methods to be called at runtime.
Bug:
32201623
Test: mma test-art-host-run-test-978-virtual-interface
Change-Id: Ie565cf2fbe8602b17be0fb051e21d221a17b518f
Mingyao Yang [Fri, 14 Oct 2016 23:26:08 +0000 (16:26 -0700)]
More store/allocation elimination for singletons in case of loops
For a store into a singleton's location, if it happens inside a loop, it
means the singleton's location value may be killed by loop side effects.
However if the singleton is defined inside that loop, that loop should
be skipped since its loop side effects kill values at loop header where
the singleton's location doesn't exist yet.
Test: test-art-host
Bug:
31716107
Change-Id: Iae2494ea93295977f90d1463ee136a7e2e09ba9b
Mathieu Chartier [Sat, 15 Oct 2016 01:43:19 +0000 (18:43 -0700)]
Move most mirror:: args to ObjPtr
Fixed possible moving GC bugs in ClinitImageUpdate class.
Bug:
31113334
Test: test-art-host
Change-Id: I0bf6578553d58b944aaa17665f1350bdf5ed15ec
Vladimir Marko [Mon, 17 Oct 2016 16:43:30 +0000 (16:43 +0000)]
Merge "Use FdFile::Copy() in dex2oat for better error checking."
Narayan Kamath [Mon, 17 Oct 2016 15:13:46 +0000 (16:13 +0100)]
Revert "Revert "Revert "Temporarily blacklist 955-method-handles-smali."""
This reverts commit
5665f6a561262b78d85afc975eca5a3138f4efa2.
Test failures were fixed in
d4fd5b1ec3bea01ad6985b7fe19a0f80b5fc970b.
Test: make test-art-host.
Change-Id: I45901836a92336ed1a337cdbf3fa9a0181c3d975
Narayan Kamath [Fri, 14 Oct 2016 17:49:12 +0000 (18:49 +0100)]
Fix handling of dex cache arrays for method types.
- Add support for relocating method type dex cache arrays.
- Add missing call to MethodTypeDexCachePair::Initialize in the
class_linker.
- Fix alignment calculation for method_types in
DexCacheArraysLayout.
- Removed unused MethodTypeOffset(uint32_t) method.
Fixes tests with --pic.
Test: make test-art-host
Bug:
30550796
Change-Id: I2bba7228762b9e9834e7659fe62090be78afdfa8
Brian Carlstrom [Mon, 17 Oct 2016 15:18:09 +0000 (15:18 +0000)]
Merge "Ensure OpenDexFilesFromImage closes file to prevent file descriptor leak"
Vladimir Marko [Mon, 17 Oct 2016 15:08:21 +0000 (15:08 +0000)]
Merge "Add DCHECK() for class loader in artResolveStringFromCode()."
Vladimir Marko [Mon, 17 Oct 2016 14:25:10 +0000 (15:25 +0100)]
Use FdFile::Copy() in dex2oat for better error checking.
Previously, read errors would have been silently ignored
and write errors would ungracefully crash.
Also improves performance on Linux, including Android,
by using sendfile(). (We still keep Mac builds around.)
Test: m test-art-host
Change-Id: I689e8b9fee4595c42b353bd10042b60a7d6c1bc7
Scott Wakeling [Mon, 17 Oct 2016 09:03:36 +0000 (10:03 +0100)]
ARM: VIXL32: Pass initial ART tests with new code generator.
- Implement enough codegen to pass ~70 art/tests.
- When ART_USE_VIXL_ARM_BACKEND is defined:
- Blacklist known-to-fail target tests
- interpret-only everything except the tests themselves
- Set a flag to use the VIXL based ARM backend
Test: export ART_USE_VIXL_ARM_BACKEND=true && mma test-art-target && mma test-art-host
Change-Id: Ic8bc095e8449f10f97fa0511284790f36c20e276
Vladimir Marko [Mon, 17 Oct 2016 12:27:55 +0000 (12:27 +0000)]
Merge "Do not write empty class table and intern table to image."
Treehugger Robot [Sat, 15 Oct 2016 04:06:52 +0000 (04:06 +0000)]
Merge "ART: Blacklist test 907 for target"
Andreas Gampe [Sat, 15 Oct 2016 02:05:06 +0000 (19:05 -0700)]
ART: Blacklist test 907 for target
Bug:
31684578
Test: m test-art-host
Change-Id: If2353bef8601ba16941a7b1df2befc215b493431
Treehugger Robot [Sat, 15 Oct 2016 02:02:37 +0000 (02:02 +0000)]
Merge changes If95ecaf4,Ifb33cae2
* changes:
ART: Temporarily use global references for GetLoadedClasses
ART: Add GetLoadedClasses support
Treehugger Robot [Sat, 15 Oct 2016 01:08:46 +0000 (01:08 +0000)]
Merge "ART: Make JNI global and weak global locks global"
Brian Carlstrom [Fri, 14 Oct 2016 08:04:26 +0000 (01:04 -0700)]
Ensure OpenDexFilesFromImage closes file to prevent file descriptor leak
(cherry picked from commit
645e477c2ce7a59ac98bc86463826cf082a13ad8)
Test: m -j32 test-art-host
Bug:
32095919
Change-Id: I44541edeb403a508e1b9794ee09fb1f55e3f42bf
Mathieu Chartier [Fri, 14 Oct 2016 23:40:46 +0000 (23:40 +0000)]
Merge "Move reference processor and mirror::Reference to ObjPtr"
Andreas Gampe [Fri, 14 Oct 2016 20:30:56 +0000 (13:30 -0700)]
ART: Temporarily use global references for GetLoadedClasses
Until we have enough local references, use global references.
Bug:
31684578
Bug:
32125344
Test: m test-art-host
Change-Id: If95ecaf424cfc7e8a84c7f99b071ac5d6bfb7797
Andreas Gampe [Wed, 12 Oct 2016 19:51:25 +0000 (12:51 -0700)]
ART: Add GetLoadedClasses support
Add support to retrieve all loaded classes. Use the standard
ClassLoader callback infrastructure.
Bug:
31684578
Test: m test-art-host-run-test-907-get-loaded-classes
Change-Id: Ifb33cae27dc32602bcd8fd930d3c7eda02598c26
Andreas Gampe [Fri, 14 Oct 2016 20:27:12 +0000 (13:27 -0700)]
ART: Make JNI global and weak global locks global
Add them to Locks. Give the JNI global table lock
a lock level, close to the weak global JNI references.
Bug:
31684578
Test: m test-art-host
Change-Id: I3857a3b0be69b16811d9999096b2c42e7a25d227
Hiroshi Yamauchi [Fri, 14 Oct 2016 21:46:31 +0000 (21:46 +0000)]
Merge "Add a mark bit check in the read barrier jni entrypoint."
Mathieu Chartier [Fri, 14 Oct 2016 21:28:47 +0000 (14:28 -0700)]
Move reference processor and mirror::Reference to ObjPtr
Bug:
31113334
Test: test-art-host
Change-Id: I2c7c3dfd88ebf12a0de271436f8a7781f997e061
Hiroshi Yamauchi [Fri, 14 Oct 2016 18:21:38 +0000 (11:21 -0700)]
Add a mark bit check in the read barrier jni entrypoint.
Bug:
30437917
Bug:
12687968
Test: test-art host/N9, N9 libartd boot, Ritz EAAC.
Change-Id: Ic5f8080959f243e80f00da73ba5f4b91a8d8b617
Mathieu Chartier [Fri, 14 Oct 2016 16:54:05 +0000 (16:54 +0000)]
Merge "Move mirror::ObjectArray to ObjPtr"
Vladimir Marko [Fri, 14 Oct 2016 15:28:52 +0000 (16:28 +0100)]
Add DCHECK() for class loader in artResolveStringFromCode().
Address late comment from
https://android-review.googlesource.com/281642
Test: m test-art-host
Bug:
20323084
Bug:
30627598
Change-Id: I3ba2617d2bc52ac8ba11307d206e6f5e6c017fb7
Narayan Kamath [Fri, 14 Oct 2016 14:29:03 +0000 (14:29 +0000)]
Merge "Revert "Revert "Temporarily blacklist 955-method-handles-smali."""
Narayan Kamath [Fri, 14 Oct 2016 14:27:33 +0000 (14:27 +0000)]
Revert "Revert "Temporarily blacklist 955-method-handles-smali.""
This reverts commit
f18d84699a7c18802f9774069143c173c6b927fb.
Tests are red on the bot. Blacklisting until this is investigated further.
Change-Id: Ibb9b4066abe28e8625c4eed5e0818e6a3fdd4f44
Treehugger Robot [Fri, 14 Oct 2016 13:23:03 +0000 (13:23 +0000)]
Merge "Revert "Temporarily blacklist 955-method-handles-smali.""
Roland Levillain [Fri, 14 Oct 2016 11:47:10 +0000 (11:47 +0000)]
Merge "Small clean up in art::x86_64::ReadBarrierMarkSlowPathX86_64."
Narayan Kamath [Thu, 13 Oct 2016 15:09:33 +0000 (16:09 +0100)]
Revert "Temporarily blacklist 955-method-handles-smali."
This reverts commit
adc2ef25244645a650111560d252e32b3e22f463.
The test is green, now that the required smali changes have been
merged into aosp/master.
Test: make test-art-host
Change-Id: I63eed37470d1def3bafccdd00116d7b49a4c9c78
Vladimir Marko [Thu, 13 Oct 2016 11:53:15 +0000 (12:53 +0100)]
Do not write empty class table and intern table to image.
And remove dead ClassLinker and InternTable code that would
be erroneously triggered by not writing the tables.
Test: m test-art-host
Change-Id: I2520daa8d1231e16112d44e364a72d782fa32527
Treehugger Robot [Fri, 14 Oct 2016 10:19:58 +0000 (10:19 +0000)]
Merge "interpreter: support for non-exact invokes."
Narayan Kamath [Wed, 3 Aug 2016 11:46:58 +0000 (12:46 +0100)]
interpreter: support for non-exact invokes.
Performs primitive argument conversions as well as boxing and unboxing
operations. Support for return value conversions will be added in a
follow up change.
Test: make test-art-host
Change-Id: I2e3348ff64a5826e477f87c12a7d5c390eb3a653
Nicolas Geoffray [Fri, 14 Oct 2016 08:37:36 +0000 (08:37 +0000)]
Merge "MIPS32: Improve integer comparison with constants."
Mathieu Chartier [Thu, 13 Oct 2016 20:48:23 +0000 (13:48 -0700)]
Move mirror::ObjectArray to ObjPtr
Also reduce calls to ObjPtr::Ptr.
Bug:
31113334
Test: test-art-host -j32
Change-Id: I73e1b356972bb079b66332f00f7f07451601044e
Treehugger Robot [Fri, 14 Oct 2016 00:37:37 +0000 (00:37 +0000)]
Merge "jni: Add read barrier fast path to jni compiler"
Mathieu Chartier [Thu, 13 Oct 2016 23:52:45 +0000 (23:52 +0000)]
Merge "Move common throws to ObjPtr"
Mathieu Chartier [Thu, 13 Oct 2016 20:59:58 +0000 (13:59 -0700)]
Move common throws to ObjPtr
Also a few other throws. Removed some calls to ObjPtr::Ptr.
Test: test-art-host
Bug:
31113334
Change-Id: I4082d64134a5a8961489f4309522c28a30968a18
Mathieu Chartier [Thu, 13 Oct 2016 20:08:36 +0000 (20:08 +0000)]
Merge "Move mirror::Class to use ObjPtr"