OSDN Git Service

android-x86/art.git
7 years agoMerge "Update known buildbot failures for ProcessBuilderTest"
Treehugger Robot [Thu, 4 Aug 2016 13:44:17 +0000 (13:44 +0000)]
Merge "Update known buildbot failures for ProcessBuilderTest"

7 years agoUpdate known buildbot failures for ProcessBuilderTest
Tobias Thierer [Wed, 3 Aug 2016 18:12:37 +0000 (19:12 +0100)]
Update known buildbot failures for ProcessBuilderTest

Also updated the comment string based on our latest understanding.
This is a native crash. Both of the affected tests as well as the
other tests in ProcessBuilderTest run 'sh' as a child process.
Only one of the tests that are affected to far accesses /proc/,
which had previously been suspected as related to the cause.

The statement that this is a native crash is based on there not
being a relevant stacktrace in logcat.

A 'sh' related native crash failure occurs in the logs but that
is not specific to these tests so it's not clear if it is related.
If not, it's not clear why the additional native crash doesn't
show up in the logs.

Test: Nothing

Bug: 30657148

Change-Id: I8347edab3cacd060baf1087a1939926f70fdd349

7 years agoMerge "Note known failures for TCP_USER_TIMEOUT tests on host"
Tobias Thierer [Thu, 4 Aug 2016 10:51:13 +0000 (10:51 +0000)]
Merge "Note known failures for TCP_USER_TIMEOUT tests on host"

7 years agoMerge "libcore_failures: Remove obsolete entries."
Narayan Kamath [Thu, 4 Aug 2016 09:18:08 +0000 (09:18 +0000)]
Merge "libcore_failures: Remove obsolete entries."

7 years agolibcore_failures: Remove obsolete entries.
Narayan Kamath [Wed, 3 Aug 2016 09:13:22 +0000 (10:13 +0100)]
libcore_failures: Remove obsolete entries.

These tests were failing with the OpenJDK switch, but have since
been fixed.

bug: 26326992
test: art/tools/run-libcore-tests.sh --mode=host --variant=X32
Change-Id: I9210d3fd428e0e462d95c2a620fe014e26bb942c

7 years agoMerge "Fix CC table lookup tests"
Treehugger Robot [Thu, 4 Aug 2016 03:35:08 +0000 (03:35 +0000)]
Merge "Fix CC table lookup tests"

7 years agoFix CC table lookup tests
Mathieu Chartier [Thu, 4 Aug 2016 01:51:34 +0000 (18:51 -0700)]
Fix CC table lookup tests

Only set the mark bits for baker barrier, not sure why this
doesn't work for table lookup barrier yet.

Bug: 30162165

Test: test-art-host with CC table lookup

Change-Id: I4309ac22bfc0f8a5c9bf081365e3e38da43eee30

7 years agoMerge "Change one read barrier bit to mark bit"
Treehugger Robot [Wed, 3 Aug 2016 23:40:55 +0000 (23:40 +0000)]
Merge "Change one read barrier bit to mark bit"

7 years agoChange one read barrier bit to mark bit
Mathieu Chartier [Fri, 29 Jul 2016 01:08:51 +0000 (18:08 -0700)]
Change one read barrier bit to mark bit

Optimization to help slow path performance. When the GC marks an
object through the read barrier slow path. The GC sets the mark bit
in the lock word of that reference. This bit is checked from the
assembly entrypoint the common case is that it is set. If the bit is
set, the read barrier knows the object is already marked and there is
no work to do.

To prevent dirty pages in zygote and image, the bit is set by the
image writer and zygote space creation.

EAAC score (lower is better):
N9: 777 -> 700 (average 31 of runs)
N6P (960000 mhz): 1737.48 -> 1442.31 (average of 25 runs)

Bug: 30162165
Bug: 12687968

Test: N9, N6P booting, test-art-host, test-art-target all with CC

Change-Id: Iae0cacfae221e33151d3c0ab65338d1c822ab63d

7 years agoMerge "Use ArrayInstance.getLength method."
Treehugger Robot [Wed, 3 Aug 2016 22:21:08 +0000 (22:21 +0000)]
Merge "Use ArrayInstance.getLength method."

7 years agoMerge "ARM64: Use the zero register in the parallel-move resolver."
Roland Levillain [Wed, 3 Aug 2016 17:56:02 +0000 (17:56 +0000)]
Merge "ARM64: Use the zero register in the parallel-move resolver."

7 years agoNote known failures for TCP_USER_TIMEOUT tests on host
Tobias Thierer [Wed, 3 Aug 2016 17:33:33 +0000 (18:33 +0100)]
Note known failures for TCP_USER_TIMEOUT tests on host

TCP_USER_TIMEOUT is not currently defined on the prebuilt tcp.h (at least
not on x86-64), so its value at the non-native level remains 0, breaking
tests that rely on it.

Devices kernels and bionic support it, therefore these tests are only
expected to break on host.

Test: Nothing. Buildbot will let us know soon enough if the expected
      failures do not match reality.

Bug: 30402085
Change-Id: Iacaf63ab341c6e4a41c7d21fb3064d58ada94800

7 years agoMerge "Fix constant type"
Treehugger Robot [Wed, 3 Aug 2016 17:08:51 +0000 (17:08 +0000)]
Merge "Fix constant type"

7 years agoMerge "Fix an assertion in the non-Baker read barrier ARM64 slow path."
Roland Levillain [Wed, 3 Aug 2016 17:03:25 +0000 (17:03 +0000)]
Merge "Fix an assertion in the non-Baker read barrier ARM64 slow path."

7 years agoUse ArrayInstance.getLength method.
Richard Uhler [Wed, 3 Aug 2016 16:26:35 +0000 (09:26 -0700)]
Use ArrayInstance.getLength method.

Now that perflib provides it as a better way to get the length of an
array instance.

Change-Id: I6bf8bb5541229e16f36fa8ede0a07cd0eb7b931d
Test: m ahat-test
Bug: 30481345

7 years agoFix constant type
David Sehr [Wed, 3 Aug 2016 16:05:20 +0000 (09:05 -0700)]
Fix constant type

Remove the unnecessary cast to long for DCHECK, a nit that was omitted
from the last CL, https://android-review.googlesource.com/#/c/248243/.

Bug: b/28856653
Test: test-art-host-gtest-dex_file_test
Change-Id: I8ca18d19cbac61e40ddfbe32729aef5c20347174

7 years agoMerge "ART: Extend run-test timeout logging time"
Treehugger Robot [Wed, 3 Aug 2016 16:09:45 +0000 (16:09 +0000)]
Merge "ART: Extend run-test timeout logging time"

7 years agoMerge "Scope the compilation and Create the ProfilingInfo in the loop."
Nicolas Geoffray [Wed, 3 Aug 2016 15:21:40 +0000 (15:21 +0000)]
Merge "Scope the compilation and Create the ProfilingInfo in the loop."

7 years agoFix an assertion in the non-Baker read barrier ARM64 slow path.
Roland Levillain [Wed, 3 Aug 2016 14:02:10 +0000 (15:02 +0100)]
Fix an assertion in the non-Baker read barrier ARM64 slow path.

Test: m ART_USE_READ_BARRIER=true ART_READ_BARRIER_TYPE=TABLELOOKUP test-art-target (ARM64).
Bug: 30563701
Change-Id: Ib7969bd78e9c14cc2be9d5527d80385486b4f409

7 years agoARM64: Use the zero register in the parallel-move resolver.
Alexandre Rames [Tue, 2 Aug 2016 12:45:28 +0000 (13:45 +0100)]
ARM64: Use the zero register in the parallel-move resolver.

When moving zero to a stack slot, this will change

    mov temp, #0
    str temp, [...]

into

    str zr, [...]

Change-Id: I2211b00d70f3fa0a02e781c90198757290b2bf89

7 years agoMerge "Return error directly when uncompressed length is 0"
Treehugger Robot [Tue, 2 Aug 2016 22:41:39 +0000 (22:41 +0000)]
Merge "Return error directly when uncompressed length is 0"

7 years agoReturn error directly when uncompressed length is 0
ganxiaolin [Mon, 18 Jul 2016 03:21:44 +0000 (11:21 +0800)]
Return error directly when uncompressed length is 0

If uncompressed length is 0, in function MemMap::MapAnonymous, it will generate a MemMap instance whose base_begin is nullptr.
Then, when OpenMemory, it will create a DexFile instance.
At DexFile::DexFile constructor, header_ is asigned to 0, then header_->string_ids_off_ will cause crash.

Bug: b/28856653
Test: test-art-host-gtest-dex_file_test
Signed-off-by: ganxiaolin <ganxiaolin@xiaomi.com>
Change-Id: Id37f7629f4646cbc385ef054cb83b15be4c59b00

7 years agoScope the compilation and Create the ProfilingInfo in the loop.
Nicolas Geoffray [Tue, 2 Aug 2016 17:49:25 +0000 (18:49 +0100)]
Scope the compilation and Create the ProfilingInfo in the loop.

We need to release the mutator lock to avoid being in a deadlock.

Also allocating the profiling just before compiling in the unlickely
event that it got collected after releasing the mutator lock.

bug:30351473
Change-Id: I98dba646536d857152949b92f6c69ff9a03cf783

7 years agoMerge "Improve the graph visualizer's output for constant locations."
Nicolas Geoffray [Tue, 2 Aug 2016 20:18:27 +0000 (20:18 +0000)]
Merge "Improve the graph visualizer's output for constant locations."

7 years agoMerge "ARM64: Make the VIXL macro assembler part of ART ARM64's assembler."
Nicolas Geoffray [Tue, 2 Aug 2016 20:06:39 +0000 (20:06 +0000)]
Merge "ARM64: Make the VIXL macro assembler part of ART ARM64's assembler."

7 years agoART: Extend run-test timeout logging time
Andreas Gampe [Tue, 2 Aug 2016 04:53:57 +0000 (21:53 -0700)]
ART: Extend run-test timeout logging time

Double the time a test is allowed to attempt logging after it
timed out to two minutes.

Also change from using nested timeouts to using "-k" in general,
not just for host dex2oat.

Bug: 30351473
Test: m test-art-host
Change-Id: I134c7b220f69eaab7ea430a5c7ce45d77b091996

7 years agoMerge "Split art-run-tests between building and packaging"
Dan Willemsen [Tue, 2 Aug 2016 17:41:59 +0000 (17:41 +0000)]
Merge "Split art-run-tests between building and packaging"

7 years agoMerge "Align method code rather than method header in oat files."
Vladimir Marko [Tue, 2 Aug 2016 16:59:35 +0000 (16:59 +0000)]
Merge "Align method code rather than method header in oat files."

7 years agoMerge "ART: Fix Mips Thread offset checks"
Treehugger Robot [Tue, 2 Aug 2016 16:52:14 +0000 (16:52 +0000)]
Merge "ART: Fix Mips Thread offset checks"

7 years agoARM64: Make the VIXL macro assembler part of ART ARM64's assembler.
Alexandre Rames [Tue, 2 Aug 2016 12:45:28 +0000 (13:45 +0100)]
ARM64: Make the VIXL macro assembler part of ART ARM64's assembler.

This avoids a dynamic allocation of the VIXL macro assembler.

Change-Id: I4cd62678d0978f1ad6f32ea0ce7279e09152be38

7 years agoART: Fix Mips Thread offset checks
Andreas Gampe [Tue, 2 Aug 2016 15:44:20 +0000 (08:44 -0700)]
ART: Fix Mips Thread offset checks

Follow-up to commit 542451cc546779f5c67840e105c51205a1b0a8fd.

Bug: 30373134
Test: m test-art-host
Change-Id: I1d0202896e683cb20db30e8aac1c5ebcacc96467

7 years agoImprove the graph visualizer's output for constant locations.
Alexandre Rames [Tue, 2 Aug 2016 12:45:28 +0000 (13:45 +0100)]
Improve the graph visualizer's output for constant locations.

Change-Id: I423fb378ee61fb53c3b328fc74f4e95cdef0992a

7 years agoMerge "DexFile: Add basic support for v38 dex files."
Narayan Kamath [Tue, 2 Aug 2016 12:36:14 +0000 (12:36 +0000)]
Merge "DexFile: Add basic support for v38 dex files."

7 years agoMerge "ARM: Embed 0.0 in VCMP."
Vladimir Marko [Tue, 2 Aug 2016 12:31:10 +0000 (12:31 +0000)]
Merge "ARM: Embed 0.0 in VCMP."

7 years agoARM: Embed 0.0 in VCMP.
Vladimir Marko [Mon, 1 Aug 2016 16:41:45 +0000 (17:41 +0100)]
ARM: Embed 0.0 in VCMP.

Test: Run ART test suite on Nexus 5.
Change-Id: I5cbbd98c4d64a4d9213e27adcae929ead5099a39

7 years agoAlign method code rather than method header in oat files.
Vladimir Marko [Mon, 1 Aug 2016 15:33:16 +0000 (16:33 +0100)]
Align method code rather than method header in oat files.

This has always been the intent. For example, the function
name CompiledMethod::AlignCode() shows this intent.

Test: Run ART test suite on host and Nexus 9.
Change-Id: I11ae8963fc537fee853fe82e3aca22e77907eae7

7 years agoDexFile: Add basic support for v38 dex files.
Narayan Kamath [Mon, 1 Aug 2016 13:20:31 +0000 (14:20 +0100)]
DexFile: Add basic support for v38 dex files.

Makes sure they aren't rejected, essentially. Also adds positive
and negative test cases in dex_file_test.

bug: 30550796
Change-Id: I8c95055a22f8a1435868c8bf3497f9641b7bb900
test: make test-art-host

7 years agoMerge "dex_instruction: remove unused parameter from instruction definitions."
Narayan Kamath [Tue, 2 Aug 2016 08:09:35 +0000 (08:09 +0000)]
Merge "dex_instruction: remove unused parameter from instruction definitions."

7 years agodex_instruction: remove unused parameter from instruction definitions.
Narayan Kamath [Mon, 1 Aug 2016 16:32:37 +0000 (17:32 +0100)]
dex_instruction: remove unused parameter from instruction definitions.

bug: 30550796
test: make test-art-host
Change-Id: Ib80a2d5c548578cefb0552b597e1d50b813ce8ef

7 years agoSplit art-run-tests between building and packaging
Dan Willemsen [Tue, 2 Aug 2016 05:58:31 +0000 (22:58 -0700)]
Split art-run-tests between building and packaging

We remove the 'obj/PACKAGING' directory during an installclean. Instead
of forcing a rebuild of all of these tests after an installclean, build
them in a different obj directory, then copy them into the PACKAGING
directory to be gathered together by LOCAL_PICKUP_FILES.

The built directory isn't used directly for LOCAL_PICKUP_FILES, since we
don't clean up tests that are no longer defined, so an installclean can
fix that. This matches the behavior of modules installed into the system
image, etc.

This brings my build time for "m -j55 tests" from 3m30s to 35s after
"m installclean".

The total size of these binaries are small (~10MB), so the extra copy is
insignificant.

Change-Id: I20f1c09df25b70df8c7f86167480b3a39e99642a

7 years agoMerge "ART: Convert pointer size to enum"
Treehugger Robot [Tue, 2 Aug 2016 03:47:35 +0000 (03:47 +0000)]
Merge "ART: Convert pointer size to enum"

7 years agoMerge "Revert "cpp-define-generator: "make art-update-cpp-defines" to regen asm_support""
Nicolas Geoffray [Tue, 2 Aug 2016 02:07:19 +0000 (02:07 +0000)]
Merge "Revert "cpp-define-generator: "make art-update-cpp-defines" to regen asm_support""

7 years agoRevert "cpp-define-generator: "make art-update-cpp-defines" to regen asm_support"
Nicolas Geoffray [Tue, 2 Aug 2016 02:07:07 +0000 (02:07 +0000)]
Revert "cpp-define-generator: "make art-update-cpp-defines" to regen asm_support"

Reverting to unbreak golem builds.

This reverts commit 3c92b129047b12f0887a2068b3800d7a7dcb7d78.

Change-Id: I78ef481f57fbcd3b3a9f4802dd6c21dabb186856

7 years agoART: Convert pointer size to enum
Andreas Gampe [Tue, 26 Jul 2016 16:02:02 +0000 (09:02 -0700)]
ART: Convert pointer size to enum

Move away from size_t to dedicated enum (class).

Bug: 30373134
Bug: 30419309
Test: m test-art-host
Change-Id: Id453c330f1065012e7d4f9fc24ac477cc9bb9269

7 years agoMerge "Fixed bug in disassembly of roundss/roundsd"
Treehugger Robot [Tue, 2 Aug 2016 00:32:16 +0000 (00:32 +0000)]
Merge "Fixed bug in disassembly of roundss/roundsd"

7 years agoFixed bug in disassembly of roundss/roundsd
Aart Bik [Mon, 1 Aug 2016 22:55:36 +0000 (15:55 -0700)]
Fixed bug in disassembly of roundss/roundsd

Rationale:
These instructions should be marked as load, so that, using
Intel syntax, destination (xmm0) appears at left hand side, as in
   roundss xmm0, xmm1
and not the other way around. First I suspected a bug in the
encoding (hence the test) and even the register allocator, but
since the code behaved correctly, only disassembly was really wrong.

Test: disassembler_x86_test (but nothing for actual disassembly)

BUG=26327751

Change-Id: I060ef57f4d5a64cdc04b97ae8a799d1c0d22da05

7 years agoMerge "Added direct memory operand support for comiss/sd on x86."
Treehugger Robot [Mon, 1 Aug 2016 22:45:01 +0000 (22:45 +0000)]
Merge "Added direct memory operand support for comiss/sd on x86."

7 years agoMerge "libsigchain: intercept bsd_signal on lp32"
Treehugger Robot [Mon, 1 Aug 2016 22:39:15 +0000 (22:39 +0000)]
Merge "libsigchain: intercept bsd_signal on lp32"

7 years agolibsigchain: intercept bsd_signal on lp32
Dimitry Ivanov [Mon, 1 Aug 2016 18:19:03 +0000 (11:19 -0700)]
libsigchain: intercept bsd_signal on lp32

bsd_signal is a deprecated synonym for the signal. Nevertheless
there are apps and libraries calling this function.

libsigchain needs to intercept these calls and handle them the same
way it handles signal() call.

Bug: http://b/30562229
Test: readelf --dyn-sym app_process32 and check that bsd_signal is exported
      readelf --dyn-sym app_process64 and check that bsd_signal is not exported
Change-Id: I3aa41632bb015568d8524a82f1c97e4443ec0d6c

7 years agoAdded direct memory operand support for comiss/sd on x86.
Aart Bik [Mon, 1 Aug 2016 21:11:20 +0000 (14:11 -0700)]
Added direct memory operand support for comiss/sd on x86.

Rationale:
Memory operands allows for shorter instruction sequences.
Something that may be used e.g. the upcoming implementation
of the round intrinsic.

Bug=26327751

Change-Id: Ifd976ed86917de1879fa036cbbbff5edf4fb741d
Test: assembler_x86_test

7 years agoMerge "Put the deletion of profiling info under a GC critical section."
Nicolas Geoffray [Mon, 1 Aug 2016 19:45:05 +0000 (19:45 +0000)]
Merge "Put the deletion of profiling info under a GC critical section."

7 years agoMerge "Revert "Revert "Scan immune spaces using mod union tables."""
Hiroshi Yamauchi [Mon, 1 Aug 2016 19:05:43 +0000 (19:05 +0000)]
Merge "Revert "Revert "Scan immune spaces using mod union tables."""

7 years agoMerge "ART: remove gcc cruft from the makefiles"
Colin Cross [Mon, 1 Aug 2016 18:14:07 +0000 (18:14 +0000)]
Merge "ART: remove gcc cruft from the makefiles"

7 years agoMerge "cpp-define-generator: "make art-update-cpp-defines" to regen asm_support"
Treehugger Robot [Mon, 1 Aug 2016 17:46:49 +0000 (17:46 +0000)]
Merge "cpp-define-generator: "make art-update-cpp-defines" to regen asm_support"

7 years agoMerge "Fix art::ArchTest::FinalizeSetup not being called."
Przemyslaw Szczepaniak [Mon, 1 Aug 2016 10:54:37 +0000 (10:54 +0000)]
Merge "Fix art::ArchTest::FinalizeSetup not being called."

7 years agoFix art::ArchTest::FinalizeSetup not being called.
Przemyslaw Szczepaniak [Mon, 25 Jul 2016 16:31:06 +0000 (17:31 +0100)]
Fix art::ArchTest::FinalizeSetup not being called.

CommonRuntimeTestImpl::SetUp() is calling FinalizeSetup. Since
CommonRuntimeTestImpl::FinalizeSetup is not virtual (it's virtual
one step higher in class hierarchy in CommonRuntimeTestBase), overridden
version in ArchTest was never used. This is causing ArchTest to
fail with openjdk8u60 version of java.lang.reflect (it's calling
Class#getDeclaredField in clinit, which fails on 32 bit
version because of different pointer sizes [arch_test forces
64-bit ISA])

Bug: 28666126
Test:  make -j 32 test-art-host-gtest
Change-Id: Ief370f5b18ef787575ac0f88ecbe17ebbf037542

7 years agoMerge "ART: Fix RegTypeCache for instance field declaring class"
David Brazdil [Mon, 1 Aug 2016 08:39:04 +0000 (08:39 +0000)]
Merge "ART: Fix RegTypeCache for instance field declaring class"

7 years agoPut the deletion of profiling info under a GC critical section.
Nicolas Geoffray [Sat, 30 Jul 2016 21:49:11 +0000 (22:49 +0100)]
Put the deletion of profiling info under a GC critical section.

Otherwise the GC could see dangling pointers.

bug:30033802
Change-Id: I839c4f6e537aba98769e99e4d75ff869c9ff13eb

7 years agoMerge "Fix for interpreter crash on new instance of class"
Aart Bik [Sat, 30 Jul 2016 02:53:25 +0000 (02:53 +0000)]
Merge "Fix for interpreter crash on new instance of class"

7 years agoMerge "Revert "Revert "Use try lock to fix class resolution race"""
Treehugger Robot [Sat, 30 Jul 2016 02:10:00 +0000 (02:10 +0000)]
Merge "Revert "Revert "Use try lock to fix class resolution race"""

7 years agoRevert "Revert "Use try lock to fix class resolution race""
Mathieu Chartier [Fri, 29 Jul 2016 23:26:01 +0000 (16:26 -0700)]
Revert "Revert "Use try lock to fix class resolution race""

Fix possible deadlock in EnsureResolved caused by interaction with
GC. Since we were sleeping while holding the mutator lock, it could
block thread suspension. This would deadlock if the thread that
had locked h_class is already suspended since we would spin forever
and not make progress.

Bug: 27417671
Bug: 30500547

Test: test-art-host ART_TEST_GC_STRESS=true

This reverts commit 69bf969c055c31a75d17ea92aeee756042678114.

Change-Id: Id8750df065dc3b9ef7dc874f2eb2cc2c58e5d1eb

7 years agoMerge "Create OatFileAssistant::OatFileInfo inner class."
Richard Uhler [Sat, 30 Jul 2016 00:10:18 +0000 (00:10 +0000)]
Merge "Create OatFileAssistant::OatFileInfo inner class."

7 years agoFix for interpreter crash on new instance of class
Aart Bik [Tue, 26 Jul 2016 00:52:22 +0000 (17:52 -0700)]
Fix for interpreter crash on new instance of class

Rationale:
Fuzz testing found divergences between the compiler and interpreter
which turned out to be caused by calling instance on java.lang.Class
(this worked for compiler but crashed interpeter). Since RI does not
allow this construct, solution is to force interpreter in this
unlikely case and throw run time exception like RI. This fixes
two cases found with fuzz testing. CL also includes
fail-before/pass-after test.

Test: 600-verifier-fails

BUG=29758098

Change-Id: Ie80f7758def44e6655d28fec4c10c34ffa0fd60b

7 years agoRevert "Revert "Scan immune spaces using mod union tables.""
Hiroshi Yamauchi [Fri, 29 Jul 2016 22:07:05 +0000 (15:07 -0700)]
Revert "Revert "Scan immune spaces using mod union tables.""

This reverts commit ccf7c8a68972f3b7bfa38af0c6b792f712f83765.

Change-Id: Iee618d9ae78e1dcdea4a7ecbe9253821d4d4d015
Test: N9 boot, art tests, Ritz EAAC.
Bug: 29516968
Bug: 29517059
Bug: 12687968

7 years agoMerge "Revert "Use try lock to fix class resolution race""
Treehugger Robot [Fri, 29 Jul 2016 21:23:49 +0000 (21:23 +0000)]
Merge "Revert "Use try lock to fix class resolution race""

7 years agoRevert "Use try lock to fix class resolution race"
Mathieu Chartier [Fri, 29 Jul 2016 20:06:45 +0000 (20:06 +0000)]
Revert "Use try lock to fix class resolution race"

This reverts commit a704eda0078989a73cac111ed309aca50d2e289b.

Bug: 27417671
Bug: 30500547
Change-Id: Ieea05236b9e61c722660cd9497c9d55d13ccd010

7 years agoMerge "Add arena stack and priority queue type aliases"
Treehugger Robot [Fri, 29 Jul 2016 18:57:26 +0000 (18:57 +0000)]
Merge "Add arena stack and priority queue type aliases"

7 years agoMerge "Revert "Scan immune spaces using mod union tables.""
Treehugger Robot [Fri, 29 Jul 2016 18:52:16 +0000 (18:52 +0000)]
Merge "Revert "Scan immune spaces using mod union tables.""

7 years agoMerge "ART: remove ART_JIT makefile variable"
Colin Cross [Fri, 29 Jul 2016 17:57:57 +0000 (17:57 +0000)]
Merge "ART: remove ART_JIT makefile variable"

7 years agoRevert "Scan immune spaces using mod union tables."
Hiroshi Yamauchi [Fri, 29 Jul 2016 17:41:01 +0000 (17:41 +0000)]
Revert "Scan immune spaces using mod union tables."

This reverts commit 8256da35feac0ba7589fcd4b692a4b9a936679a4.

Test failing.

Change-Id: I391dcbdba19c6f6313e06075b0a69ea406c7c50e

7 years agoAdd arena stack and priority queue type aliases
Matthew Gharrity [Fri, 29 Jul 2016 16:13:33 +0000 (09:13 -0700)]
Add arena stack and priority queue type aliases

These will be used in the graph coloring register allocator.

Test: m test-art-host

Change-Id: I4d3d063ae5a02e697f07fc8212f391b5347270e5

7 years agocpp-define-generator: "make art-update-cpp-defines" to regen asm_support
Igor Murashkin [Fri, 29 Jul 2016 16:19:44 +0000 (09:19 -0700)]
cpp-define-generator: "make art-update-cpp-defines" to regen asm_support

Change-Id: I1a550eda6c77f38e63bb1c8fb943541cbd3f9cdf

7 years agoMerge "Pass the right class loader when inlining."
Nicolas Geoffray [Fri, 29 Jul 2016 14:28:44 +0000 (14:28 +0000)]
Merge "Pass the right class loader when inlining."

7 years agoART: remove gcc cruft from the makefiles
Colin Cross [Thu, 28 Jul 2016 23:01:42 +0000 (16:01 -0700)]
ART: remove gcc cruft from the makefiles

ART only builds with clang now, delete all logic to determine gcc
vs. clang, and move all LOCAL_CLANG_CFLAGS into LOCAL_CFLAGS.

Test: mma, make test-art-host, make test-art-target
Change-Id: I578615233ad3e71af18afb47bb11f3be8417216c

7 years agoART: remove ART_JIT makefile variable
Colin Cross [Thu, 28 Jul 2016 23:03:13 +0000 (16:03 -0700)]
ART: remove ART_JIT makefile variable

The ART_JIT variable isn't used (and was never used?).

Test: mma, make test-art-host, make test-art-target
Change-Id: Ie6463c3c13fab43504f3c60fd5a15f696e2cf481

7 years agoCreate OatFileAssistant::OatFileInfo inner class.
Richard Uhler [Tue, 19 Apr 2016 22:39:37 +0000 (15:39 -0700)]
Create OatFileAssistant::OatFileInfo inner class.

To share code reused for getting info about both the odex and oat files.

Test: oat file assistant tests.
Change-Id: Ifcc48e0ac481899285e39e3b02f3ade180e38728

7 years agoMerge "Avoid read barrier for IntArray::GetArrayClass"
Mathieu Chartier [Thu, 28 Jul 2016 22:27:34 +0000 (22:27 +0000)]
Merge "Avoid read barrier for IntArray::GetArrayClass"

7 years agoAvoid read barrier for IntArray::GetArrayClass
Mathieu Chartier [Thu, 28 Jul 2016 18:02:38 +0000 (11:02 -0700)]
Avoid read barrier for IntArray::GetArrayClass

Changed the code use Mark instead of read barrier, this showed an
existing lock violation and possible deadlock which was fixed.
Prevent DCHECK failure from the assert.

Bug: 30469265
Test: test-art-host with CC

Change-Id: I275f953f06f6d13262043fc62eb88dca0356465a

7 years agoMerge "Use try lock to fix class resolution race"
Mathieu Chartier [Thu, 28 Jul 2016 18:12:56 +0000 (18:12 +0000)]
Merge "Use try lock to fix class resolution race"

7 years agoMerge "Scan immune spaces using mod union tables."
Hiroshi Yamauchi [Thu, 28 Jul 2016 17:24:43 +0000 (17:24 +0000)]
Merge "Scan immune spaces using mod union tables."

7 years agoMerge "Disable CheckSwapUsage test with read barriers on x86."
Roland Levillain [Thu, 28 Jul 2016 16:54:49 +0000 (16:54 +0000)]
Merge "Disable CheckSwapUsage test with read barriers on x86."

7 years agoMerge "Remove two ReadBarrierMarkRegX entrypoints."
Roland Levillain [Thu, 28 Jul 2016 16:20:19 +0000 (16:20 +0000)]
Merge "Remove two ReadBarrierMarkRegX entrypoints."

7 years agoPass the right class loader when inlining.
Nicolas Geoffray [Thu, 28 Jul 2016 02:49:14 +0000 (03:49 +0100)]
Pass the right class loader when inlining.

Otherwise, method and type resolution can resolve to the wrong
things and as a side effect update the dex cache with wrong data.

bug:30403437
test:./art/test/run-test --host --jit --dev --no-prebuild 613

(cherry picked from commit 0a210d9b108c87c0e7c1d430a92ce6fc89790c95)

Change-Id: I8fdca96fccae079c8e99b8e86e7b6935acfce89d

7 years agoART: Fix RegTypeCache for instance field declaring class
David Brazdil [Wed, 27 Jul 2016 14:29:25 +0000 (15:29 +0100)]
ART: Fix RegTypeCache for instance field declaring class

For bytecode instruction:

  iget(-*) ClassB->field:type

where the resolved field is actually declared in ClassA (ClassB
extends ClassA), MethodVerifier would create a RegType representing
ClassA but cache it under the descriptor "ClassB".

This is a bug but does not have any implications on correctness
because earlier resolution of ClassB (part of IGET handling) creates
another cache entry with the "ClassB" descriptor. Because the latter,
wrong entry is always imprecise (ClassA cannot be final because ClassB
extends it), either the earlier entry will be discovered or neither
of them will be a match.

Descriptor was replaced with the correct one and a DCHECK added when
creating a RegType to make sure the descriptor matches.

Bug: 30458218
Change-Id: I19e1bdd5dd79e5eac558122a06b9128d0a5c021f
Test: m test-art-host-run-test
Test: art/test/run-test --host 800-smali

7 years agoDisable CheckSwapUsage test with read barriers on x86.
Roland Levillain [Thu, 28 Jul 2016 15:37:28 +0000 (16:37 +0100)]
Disable CheckSwapUsage test with read barriers on x86.

This test does not produce the expected results on
some x86 systems when read barriers are enabled.

Also remove unused TEST_DISABLED_FOR_READ_BARRIER macro
definition.

Test: m test-art-host-gtest
Bug: 29259363
Bug: 12687968
Change-Id: If7c124226b32c10425725e62bb684150aba108ab

7 years agoMerge "ARM64: Add comment to mterp int-to-long."
Vladimir Marko [Thu, 28 Jul 2016 14:10:54 +0000 (14:10 +0000)]
Merge "ARM64: Add comment to mterp int-to-long."

7 years agoMerge "Add ProcessBuilderTest#testRedirectInherit() to known failures"
Tobias Thierer [Thu, 28 Jul 2016 13:49:20 +0000 (13:49 +0000)]
Merge "Add ProcessBuilderTest#testRedirectInherit() to known failures"

7 years agoARM64: Add comment to mterp int-to-long.
Vladimir Marko [Thu, 28 Jul 2016 13:04:18 +0000 (14:04 +0100)]
ARM64: Add comment to mterp int-to-long.

Test: Build for Nexus 9.
Change-Id: I649e71fc1cdc2a4a103159bf72ceeb005231e66f

7 years agoMerge "MIPS32: Improve string and class loads"
Vladimir Marko [Thu, 28 Jul 2016 12:43:29 +0000 (12:43 +0000)]
Merge "MIPS32: Improve string and class loads"

7 years agoMerge "ARM64: More mterp improvements."
Vladimir Marko [Thu, 28 Jul 2016 09:00:59 +0000 (09:00 +0000)]
Merge "ARM64: More mterp improvements."

7 years agoMerge "ART: Use old operator<< for MemBarrierKind"
Treehugger Robot [Thu, 28 Jul 2016 05:35:28 +0000 (05:35 +0000)]
Merge "ART: Use old operator<< for MemBarrierKind"

7 years agoART: Use old operator<< for MemBarrierKind
Andreas Gampe [Thu, 28 Jul 2016 04:25:41 +0000 (21:25 -0700)]
ART: Use old operator<< for MemBarrierKind

To fix assumptions in the tests.

Test: m test-art-host
Change-Id: Ie2e738524ee8f75a3c41730f6a1449dab490e535

7 years agoMIPS32: Improve string and class loads
Alexey Frunze [Tue, 19 Jul 2016 22:00:40 +0000 (15:00 -0700)]
MIPS32: Improve string and class loads

Tested:
- MIPS32 Android boots in QEMU
- test-art-host-gtest
- test-art-target-run-test-optimizing in QEMU, on CI20
- test-art-target-gtest on CI20

Change-Id: I70fd5d5267f8594c3b29d5a4ccf66b8ca8b09df3

7 years agoMerge "ART: Delete old compiler_enums.h"
Treehugger Robot [Thu, 28 Jul 2016 01:54:06 +0000 (01:54 +0000)]
Merge "ART: Delete old compiler_enums.h"

7 years agoMerge "Reduce unnecessary read barriers in GC"
Mathieu Chartier [Thu, 28 Jul 2016 01:05:07 +0000 (01:05 +0000)]
Merge "Reduce unnecessary read barriers in GC"

7 years agoART: Delete old compiler_enums.h
Andreas Gampe [Thu, 28 Jul 2016 00:53:11 +0000 (17:53 -0700)]
ART: Delete old compiler_enums.h

Holdover from the Quick days. Move the two enums that are still
used closer to the actual users (and prune no longer used cases).

Test: m test-art-host
Change-Id: I88aa49961a54635788cafac570ddc3125aa38262

7 years agoScan immune spaces using mod union tables.
Hiroshi Yamauchi [Wed, 27 Jul 2016 21:08:19 +0000 (14:08 -0700)]
Scan immune spaces using mod union tables.

It's faster to scan the immune spaces using the mod union tables (gray
objects on dirty pages) than the live bitmaps (gray objects).

The immune space scan time goes from ~13ms down to ~10ms in the
testWidgetsContainerFling test on angler at 1 GHz.

Change-Id: I0914d7bdb50cbf8a49c859070a3501b48eb79f3e
Test: N9 boot, art tests, Ritz EAAC.
Bug: 29516968
Bug: 29517059
Bug: 12687968

7 years agoUse try lock to fix class resolution race
Mathieu Chartier [Wed, 13 Jul 2016 16:53:35 +0000 (09:53 -0700)]
Use try lock to fix class resolution race

There was some possible deadlocks related to EnsureResolved caused by
acquiring an object lock.

Scenario:
Thread 1 acquires lock on obj1
Thread 1 begins to resolve / initialize class1
Thread 1 blocks since it sees that class1 is already being resolved and
gets preempted before it can acquire the object lock on class1
Thread 2 finishes resolving and initializing class1 and locks class1
Thread 2 blocks attempting to lock obj1
Thread 1 blocks attempting to lock class1
Deadlock

Fixed the deadlock by changing EnsureResolved to use a try lock for the
unresolved case.

Added a test.

Test: Device boot, test-art-host, monitor_test

Bug: 27417671
Change-Id: Ic6e1c3ca6f45490cf8a7bf8e137dee71ac83ff64

7 years agoMerge "MIPS32: Block callee save fp registers in debuggable"
Treehugger Robot [Wed, 27 Jul 2016 21:06:44 +0000 (21:06 +0000)]
Merge "MIPS32: Block callee save fp registers in debuggable"

7 years agoReduce unnecessary read barriers in GC
Mathieu Chartier [Wed, 27 Jul 2016 17:45:47 +0000 (10:45 -0700)]
Reduce unnecessary read barriers in GC

Removed read barrier from IsUnprocessed, DequeuePendingReference,
EnqueueReference, and a few other places.

Hard to tell if GC time goes down.

EAAC:
Before GC slow path count: 254857
After GC slow path count: 1005

Bug: 30162165
Bug: 12687968

Test: test-art-host, volantis boot with CC

Change-Id: Ic2add3a9b1e1d7561b0b167f2218b10f8dbff76c