OSDN Git Service
Mathieu Chartier [Tue, 11 Oct 2016 00:59:03 +0000 (00:59 +0000)]
Merge "Use art::Atomic for CopyObject"
Andreas Gampe [Tue, 11 Oct 2016 00:43:31 +0000 (00:43 +0000)]
Merge "ART: Add allocation callback"
Treehugger Robot [Mon, 10 Oct 2016 23:39:05 +0000 (23:39 +0000)]
Merge "ART: Switch to name-based IMT hashing"
Andreas Gampe [Tue, 27 Sep 2016 19:19:45 +0000 (12:19 -0700)]
ART: Switch to name-based IMT hashing
Use a hash scheme based on the name. This keeps IMT slots stable
when dex tables change.
This incurs a severe performance penalty for computing the slot.
Measurements on host degraded from 30ns to an average of 85mus.
However, calls in compiled code will not incur this overhead.
Added a test comparing similar interfaces in similar dex files.
Bug:
31594153
Test: test-art-host
Change-Id: Ibb86679ee94bec561984ea25826e56b1a7964cd0
Hiroshi Yamauchi [Mon, 10 Oct 2016 20:42:18 +0000 (20:42 +0000)]
Merge "Avoid ThreadStress timeout with art-read-barrier-table-lookup."
Treehugger Robot [Mon, 10 Oct 2016 20:37:52 +0000 (20:37 +0000)]
Merge "ART: Add event callback support"
Andreas Gampe [Fri, 7 Oct 2016 22:05:24 +0000 (15:05 -0700)]
ART: Add allocation callback
Bug:
31684277
Test: m test-art-host
Change-Id: I959f44e23ca5fe55ed678315708895faf0aadb04
Bill Buzbee [Mon, 10 Oct 2016 20:27:34 +0000 (20:27 +0000)]
Merge "mterp: Placeholder support for invoke-polymorphic"
Mathieu Chartier [Fri, 7 Oct 2016 19:59:33 +0000 (12:59 -0700)]
Use art::Atomic for CopyObject
Just to be sure it doesn't get turned into memcpy.
Also avoid casting ObjectArray to IntArray. This is a strict aliasing
violation.
Bug:
32012820
Test: test-art-host
Change-Id: Icc5d4a758cb4f4e1686389bb0cb74ac08e554f04
Nicolas Geoffray [Mon, 10 Oct 2016 08:34:17 +0000 (08:34 +0000)]
Merge "MIPS64: Fix art_quick_aput_obj stubs"
Andreas Gampe [Fri, 7 Oct 2016 18:48:21 +0000 (11:48 -0700)]
ART: Add event callback support
Add basic event callback support infrastructure. Actual users will
follow.
Bug:
31684920
Test: m test-art-host
Change-Id: Ic496933ef3a94f9d27a2779b7f4fdc5b096eab22
Hiroshi Yamauchi [Sat, 8 Oct 2016 00:07:03 +0000 (17:07 -0700)]
Avoid ThreadStress timeout with art-read-barrier-table-lookup.
Disable MarkFromReadBarrierWithMeasurements for
art-read-barrier-table-lookup to avoid slowdown as it takes more read
barrier slow paths than art-read-barrier.
Increase the timeout.
Bug:
31679493
Bug:
12687968
Test: test-art-host with ART_READ_BARRIER_TYPE=TABLELOOKUP
Change-Id: I1a87864448e28099e522dc7fedce2208189248f2
buzbee [Fri, 7 Oct 2016 19:56:32 +0000 (12:56 -0700)]
mterp: Placeholder support for invoke-polymorphic
Force the mterp assembly interpreters to fall back to the
reference switch interpreter for invoke-polymorphic and
invoke-polymorphic-range while those new opcodes are under
development.
Test: test-art-host
Change-Id: I6167432164bd869c91d0804521d5edc9549f622e
Mathieu Chartier [Fri, 7 Oct 2016 19:45:02 +0000 (19:45 +0000)]
Merge "Use word copying for CopyObject"
Mathieu Chartier [Fri, 7 Oct 2016 17:51:53 +0000 (10:51 -0700)]
Use word copying for CopyObject
Prevent word tearing.
Bug:
32012820
Test: test-art-host
Change-Id: Ie3d5817f3bdf49e0d949caeccd7cd3512d61bdf7
Mathieu Chartier [Fri, 7 Oct 2016 16:57:40 +0000 (16:57 +0000)]
Merge "Move mirror::Object setters to ObjPtr"
Treehugger Robot [Fri, 7 Oct 2016 16:51:06 +0000 (16:51 +0000)]
Merge "Improved and simplified loop optimizations."
Goran Jakovljevic [Fri, 7 Oct 2016 15:03:28 +0000 (17:03 +0200)]
MIPS64: Fix art_quick_aput_obj stubs
Prevent situations where $gp is calculated and saved twice in a row
without restoring it. When restored in the end, its value was wrong.
This fixes test failure in stub_test.APutObj.
Test: mma test-art-target-gtest-stub_test64 on QEMU
Change-Id: I63b37ecee8d3576c57c0f6862d00bc03965b042d
Aart Bik [Thu, 6 Oct 2016 18:36:57 +0000 (11:36 -0700)]
Improved and simplified loop optimizations.
Rationale:
This CL merges some common cases into one, thereby simplifying
the code quite a bit. It also prepares for more general induction
cycles (rather than the simple phi-add currently used). Finally,
it generalizes the closed form elimination with empty loops.
As a result of the latter, elaborate but weird code like:
private static int waterFall() {
int i = 0;
for (; i < 10; i++);
for (; i < 20; i++);
for (; i < 30; i++);
for (; i < 40; i++);
for (; i < 50; i++);
return i;
}
now becomes just this (on x86)!
mov eax, 50
ret
Change-Id: I8d22ce63ce9696918f57bb90f64d9a9303a4791d
Test: m test-art-host
David Sehr [Fri, 7 Oct 2016 15:10:06 +0000 (15:10 +0000)]
Merge "Add "fixups" for ids referenced in code"
Nicolas Geoffray [Fri, 7 Oct 2016 08:41:37 +0000 (08:41 +0000)]
Merge "Blacklist test 903."
Nicolas Geoffray [Fri, 7 Oct 2016 08:29:20 +0000 (09:29 +0100)]
Blacklist test 903.
bug:
31681198
Change-Id: I2d5578242d594be791aedd99266263ec53182d77
Mathieu Chartier [Fri, 7 Oct 2016 00:31:27 +0000 (00:31 +0000)]
Merge "Move remaining jobject related functions to use ObjPtr"
David Sehr [Fri, 7 Oct 2016 00:09:32 +0000 (17:09 -0700)]
Add "fixups" for ids referenced in code
Allow better visualization by determining which type_id, string_id,
method_id, and field_id values are used by code_items.
Bug:
29921113
Change-Id: Ia6ff72064104cd5c0868e972ca65536dbeb37b09
Test: dexlayout -s {some favorite apks}
Mathieu Chartier [Thu, 6 Oct 2016 22:13:58 +0000 (15:13 -0700)]
Move mirror::Object setters to ObjPtr
Bug:
31113334
Test: test-art-host
Change-Id: I2c4c84645e194c3c435a4a6fd670176b0e98671f
Andreas Gampe [Thu, 6 Oct 2016 22:26:42 +0000 (22:26 +0000)]
Merge "ART: Register object tagging table"
Mathieu Chartier [Thu, 6 Oct 2016 01:32:08 +0000 (18:32 -0700)]
Move remaining jobject related functions to use ObjPtr
Also added ObjPtr::DownCast.
Bug:
31113334
Test: test-art-host
Change-Id: I59c253211dc435579ffdfd49f856861ab13d262c
Andreas Gampe [Thu, 6 Oct 2016 20:13:30 +0000 (13:13 -0700)]
ART: Register object tagging table
Actually register object tagging table with the runtime so it
gets swept properly.
Bug:
31385027
Test: m test-art-host
Test: m ART_USE_READ_BARRIER=true test-art-host
Change-Id: I4364458ece2893e6a85fe7518df39fb838199f20
Treehugger Robot [Thu, 6 Oct 2016 19:27:59 +0000 (19:27 +0000)]
Merge "test: Fix jni_compiler_test for const correctness"
Treehugger Robot [Thu, 6 Oct 2016 18:24:19 +0000 (18:24 +0000)]
Merge "Allow early abort logging"
Treehugger Robot [Thu, 6 Oct 2016 18:24:03 +0000 (18:24 +0000)]
Merge "ART: Support object tagging"
Treehugger Robot [Thu, 6 Oct 2016 18:22:32 +0000 (18:22 +0000)]
Merge "ObjPtr misc cleanup"
Igor Murashkin [Thu, 6 Oct 2016 17:51:11 +0000 (10:51 -0700)]
test: Fix jni_compiler_test for const correctness
Change-Id: Id0ea8f18f8fac3cf884a43a2ae6448bb0221209a
Hiroshi Yamauchi [Thu, 6 Oct 2016 17:19:17 +0000 (17:19 +0000)]
Merge "Temporarily add debug logging for b/
31357497"
Neil Fuller [Thu, 6 Oct 2016 08:55:03 +0000 (08:55 +0000)]
Merge "Revert "Blacklist flaky test.""
Neil Fuller [Thu, 6 Oct 2016 08:40:07 +0000 (08:40 +0000)]
Revert "Blacklist flaky test."
Reverts the blacklisting of NetworkSecurityPolicyTest
#testCleartextTrafficPolicyWithJarHttpURLConnection.
This reverts commit
71ceb721240d50d396ea1a61a2bf2832be345007.
Bug:
31944681
Change-Id: Ib0f13cfec021e258ceba480c43cced59c554b3f5
Treehugger Robot [Thu, 6 Oct 2016 02:57:57 +0000 (02:57 +0000)]
Merge "Fix a deadlock between thread flip and suspend request."
Mathieu Chartier [Wed, 5 Oct 2016 21:12:45 +0000 (14:12 -0700)]
Allow early abort logging
Print unattached threads with native stack traces for early abort.
Bug:
31855501
Test: test-art-host -j32
Test: thread_list_->Dump in Runtime::Start.
Change-Id: I5364625e798356c4fa1f88a8c646e374e59d93b0
Hiroshi Yamauchi [Thu, 6 Oct 2016 01:13:23 +0000 (18:13 -0700)]
Temporarily add debug logging for b/
31357497
Note the check can't be a DCHECK because 103-string-append runs with the
non-debug build (libart.so).
Bug:
31357497
Bug:
12687968
Test: run-test 103-string-append on N9.
Change-Id: I30ebe9111fa6c42796dcafa182ae066214805173
Treehugger Robot [Thu, 6 Oct 2016 01:12:23 +0000 (01:12 +0000)]
Merge "Revert "Revert "Add dex file writer to dexlayout tool."""
Hiroshi Yamauchi [Mon, 3 Oct 2016 22:32:01 +0000 (15:32 -0700)]
Fix a deadlock between thread flip and suspend request.
See
31683379#9 for the deadlock scenario.
Make ModifySuspendCount(+1) retry if the thread flip function is set.
Bug:
31683379
Bug:
12687968
Test: test-art, N9 libartd boot, Ritz EAAC with CC.
Test: 129-GetThreadId with gcstress and CC.
Change-Id: Id5cdfcd90a08a2ff497f9f0e2842fa4c613549bc
Treehugger Robot [Wed, 5 Oct 2016 23:41:27 +0000 (23:41 +0000)]
Merge "jni: Test that pushed handle scopes are popped after JNI transitions"
Treehugger Robot [Wed, 5 Oct 2016 23:32:05 +0000 (23:32 +0000)]
Merge "Change java_lang_reflect_Field.cc to use ObjPtr"
Mathieu Chartier [Wed, 5 Oct 2016 22:56:52 +0000 (15:56 -0700)]
ObjPtr misc cleanup
Check for subtypes in constructor.
Remove various calls to MakeObjPtr and ObjPtr::Ptr.
Bug:
31113334
Test: test-art-host
Change-Id: I8d680d514ea52cd64fc4f7c2e988926bc726174e
Mathieu Chartier [Wed, 5 Oct 2016 20:48:41 +0000 (13:48 -0700)]
Change java_lang_reflect_Field.cc to use ObjPtr
Also deleted some calls to ObjPtr::Ptr and fixed formatting.
Bug:
31113334
Test: test-art-host
Change-Id: I951b80886b723e1beff4783f28ee7c928cb10650
Treehugger Robot [Wed, 5 Oct 2016 21:55:20 +0000 (21:55 +0000)]
Merge "Change indirect reference table to use ObjPtr"
Igor Murashkin [Wed, 5 Oct 2016 21:33:30 +0000 (14:33 -0700)]
jni: Test that pushed handle scopes are popped after JNI transitions
There was previously a bug in generic JNI for @CriticalNative where new
handle scopes were pushed, but then not popped after the native call
was done. This caused a bug with stale pointers being found by GC.
Bug:
31933313
Change-Id: I423f9e7a7d391ba3a4771c830691931f0025b354
Treehugger Robot [Wed, 5 Oct 2016 20:29:58 +0000 (20:29 +0000)]
Merge "Refactoring of graph linearization and linear order."
Mathieu Chartier [Wed, 5 Oct 2016 02:06:30 +0000 (19:06 -0700)]
Change indirect reference table to use ObjPtr
Bug:
31113334
Test: test-art-host
Change-Id: I340fdf430897ebd790ea4e35f94bcee776e98445
Mathieu Chartier [Wed, 5 Oct 2016 19:59:08 +0000 (19:59 +0000)]
Merge "Initialize dex cache while holding dex_lock"
Andreas Gampe [Tue, 13 Sep 2016 02:58:13 +0000 (19:58 -0700)]
ART: Support object tagging
Add support for tagging in the JVMTI plugin.
Bug:
31385027
Test: m test-art-host
Change-Id: I4d8fb12cd23ca60dc0b0ce9051d1c77e5eb18aa9
Aart Bik [Wed, 5 Oct 2016 00:33:56 +0000 (17:33 -0700)]
Refactoring of graph linearization and linear order.
Rationale:
Ownership of graph's linear order and iterators was
a bit unclear now that other phases are using it.
New approach allows phases to compute their own
order, while ssa_liveness is sole owner for graph
(since it is not mutated afterwards).
Also shortens lifetime of loop's arena.
Test: test-art-host
Change-Id: Ib7137d1203a1e0a12db49868f4117d48a4277f30
Treehugger Robot [Wed, 5 Oct 2016 18:19:36 +0000 (18:19 +0000)]
Merge "Don't push handle scope for critical native generic JNI"
Vladimir Marko [Wed, 5 Oct 2016 17:00:41 +0000 (17:00 +0000)]
Merge "Remove #include "oat_file.h" from class_linker.h ."
Mathieu Chartier [Tue, 4 Oct 2016 22:41:42 +0000 (15:41 -0700)]
Don't push handle scope for critical native generic JNI
Leaving a stale handle scope causes problems for the GC the next
time roots are visited. At this point the stack will have other
contents and the GC will attempt to mark many invalid roots.
Bug:
31933313
Test: non preopt eng build booting.
Test: test-art-host
(cherry picked from commit
92879f9bbcfc034660ed1ff5ef741d7f2bcb116f)
Change-Id: I7d57964ccd2b59a05bb06f67062f646362ce1204
Mathieu Chartier [Thu, 15 Sep 2016 17:24:43 +0000 (10:24 -0700)]
Initialize dex cache while holding dex_lock
Fixes multiple threads calling RegisterDexFile occasionally getting
DCHECK failures due to the arrays not being null since the BSS ones
is per dex file.
Bug:
31369621
Test: test-art-host, no DCHECK failure during debug booting
Change-Id: I7b6e4cd03460dd1213eb4e044bdcf5f6103fd5f9
Treehugger Robot [Wed, 5 Oct 2016 16:03:30 +0000 (16:03 +0000)]
Merge "Make it possible to pass an arena allocator to HLoopOptimization."
Narayan Kamath [Wed, 5 Oct 2016 13:46:49 +0000 (13:46 +0000)]
Merge "Temporarily blacklist 955-method-handles-smali."
Narayan Kamath [Wed, 5 Oct 2016 13:15:01 +0000 (14:15 +0100)]
Temporarily blacklist 955-method-handles-smali.
Needs a companion smali change
7b348e4f323f44a7c44282.
Test: make test-art-host
Change-Id: Id245d9ead8467315c97cfe07103da2b406b08c49
Nicolas Geoffray [Wed, 5 Oct 2016 12:49:44 +0000 (13:49 +0100)]
Make it possible to pass an arena allocator to HLoopOptimization.
loop_optimization_test uses memory from HLoopOptimization's
allocator, which is scoped by the Run method.
Fix is to pass custom allocator.
test: m test-art-host-gtest
Change-Id: I359330e22202519f400a26da5403eeb00f0b2db4
Vladimir Marko [Tue, 4 Oct 2016 13:44:28 +0000 (14:44 +0100)]
Remove #include "oat_file.h" from class_linker.h .
Refactor the OatClass and OatMethod related functions from
ClassLinker to OatFile and ArtMethod, respectively. Refactor
the remaining ClassLinker dependencies on OatFile to break
the #include dependency and reduce incremental build times.
Test: m test-art-host
Change-Id: Iebc5b9f81b48fbcf79821cc827a5d7c4a0261bf6
Treehugger Robot [Wed, 5 Oct 2016 12:19:30 +0000 (12:19 +0000)]
Merge "Basic implementation of invoke / invoke-polymorphic."
Treehugger Robot [Wed, 5 Oct 2016 10:30:14 +0000 (10:30 +0000)]
Merge "Properly scope HLoopOptimization's allocator."
Narayan Kamath [Wed, 3 Aug 2016 11:46:58 +0000 (12:46 +0100)]
Basic implementation of invoke / invoke-polymorphic.
Basic switch interpreter support for invoke-polymorphic. This change
allows for virtual/interface and static invokes on method handles.
Support for direct invokes (including constructors) and field
getters and setters will be added in follow up changes.
Bug:
30550796
Test: make test-art-host
Change-Id: Ieb3a991d974060d930d56467908d5c7c11d0e38e
Nicolas Geoffray [Wed, 5 Oct 2016 08:55:42 +0000 (09:55 +0100)]
Properly scope HLoopOptimization's allocator.
HOptimization classes do not get their destructor called,
as they are arena objects. So the scope for the optimization
allocator needs to be the Run method.
Also anticipate bisection search breakage by adding
HLoopOptimization to the list of recognized optimizations.
Change-Id: I7770989c39d5700a3b6b0a20af5d4b874dfde111
Treehugger Robot [Wed, 5 Oct 2016 03:55:28 +0000 (03:55 +0000)]
Merge "MIPS32: Fix GenerateVirtualCall in the presence of intrinsics"
Colin Cross [Wed, 5 Oct 2016 01:38:09 +0000 (01:38 +0000)]
Merge "Revert "Compute PPID at ninja time""
Mathieu Chartier [Wed, 5 Oct 2016 00:54:38 +0000 (00:54 +0000)]
Merge "Fix build break from merge conflict"
Mathieu Chartier [Wed, 5 Oct 2016 00:48:45 +0000 (17:48 -0700)]
Fix build break from merge conflict
Bug:
31113334
Test: mm -j32
Change-Id: I5643a0ec22cf03a54671a5c7fa0ebda41cc9a8c3
Mathieu Chartier [Wed, 5 Oct 2016 00:39:07 +0000 (00:39 +0000)]
Merge "Rename ObjPtr::Decode to ObjPtr::Ptr"
Hiroshi Yamauchi [Wed, 5 Oct 2016 00:27:46 +0000 (00:27 +0000)]
Merge "Fix a CC crash with ThreadStress."
Colin Cross [Tue, 4 Oct 2016 23:06:16 +0000 (23:06 +0000)]
Revert "Compute PPID at ninja time"
This reverts commit
b1c17bb080ccf41b30909c0401df59e6f3a1ebb1.
It breaks make test-art-host when /bin/sh is not bash, as
/bin/sh -c /bin/bash -c echo $PPID
returns the pid of /bin/sh, not the pid of ninja.
Test: none
Change-Id: Ia4e5f76ea4c5f5b5c102ce54ae03c3fccb23a49f
Hiroshi Yamauchi [Tue, 4 Oct 2016 22:20:36 +0000 (15:20 -0700)]
Fix a CC crash with ThreadStress.
Add a read barrier on the locked object in Monitor::DescribeWait().
Bug:
31848261
Bug:
12687968
Test: test-art-host with CC and N9 libartd boot.
Change-Id: I569e1499c2aadf54d9074a8a790e5d512c2dad4f
Nicolas Geoffray [Tue, 4 Oct 2016 21:59:14 +0000 (21:59 +0000)]
Merge "Fix build breakage."
Nicolas Geoffray [Tue, 4 Oct 2016 21:56:35 +0000 (22:56 +0100)]
Fix build breakage.
Revert did not apply cleanly.
test: make test-art-host
Change-Id: Id94d377d19dc80237951ad40fd5e01f8fec67a47
Nicolas Geoffray [Tue, 4 Oct 2016 21:45:24 +0000 (21:45 +0000)]
Merge "Revert "Remove dead code from compiler driver.""
Nicolas Geoffray [Tue, 4 Oct 2016 21:44:45 +0000 (21:44 +0000)]
Revert "Remove dead code from compiler driver."
This reverts commit
44b3da25191052acc18528d8ade9cf3038505180.
Change-Id: I0cf049ed967bcaf8dfd89fc88288d992c63a4939
Jeff Hao [Tue, 4 Oct 2016 18:13:44 +0000 (18:13 +0000)]
Revert "Revert "Add dex file writer to dexlayout tool.""
This reverts commit
fd1a6c2a08ca3e2476b7424b9b0fa58e73b29e87.
Fixed output being clobbered during DexLayoutTest.DexFileOutput.
Option added to put dex output file in scratch directory.
Bug:
29921113
Test: mm test-art-host-gtest-dexlayout_test
Change-Id: I9e6b139cf06aaa39c83ad1e74329db266464a8e4
Aart Bik [Tue, 4 Oct 2016 20:40:23 +0000 (20:40 +0000)]
Merge "A first implementation of a loop optimization framework."
Andreas Gampe [Tue, 4 Oct 2016 19:08:14 +0000 (19:08 +0000)]
Merge changes I65bc7769,Ib24c7060
* changes:
ART: Reprint long messages on abort
ART: Clean up IRT-related abort messaging
Vladimir Marko [Tue, 4 Oct 2016 19:00:30 +0000 (19:00 +0000)]
Merge "Revert "Revert "Store resolved Strings for AOT code in .bss."""
David Sehr [Tue, 4 Oct 2016 18:06:48 +0000 (18:06 +0000)]
Merge "Option to gnuplot dex layout"
David Sehr [Mon, 26 Sep 2016 14:44:04 +0000 (07:44 -0700)]
Option to gnuplot dex layout
Display the portions of the dex file that are accessed by the classes in
the dex file. Optionally limit the display to only those classes that
are present in a runtime profile.
Bug:
29921113
Change-Id: I2a998fba448fec6c23941d3b4358531b1336e1b3
Test: dexlayout test
Mathieu Chartier [Tue, 4 Oct 2016 01:01:28 +0000 (18:01 -0700)]
Rename ObjPtr::Decode to ObjPtr::Ptr
Done to prevent ambiguity with ScopedObjectAccess::Decode.
Bug:
31113334
Test: test-art-host
Change-Id: I07a2497cc9cf66386311798933547471987fc316
Vladimir Marko [Mon, 3 Oct 2016 08:46:48 +0000 (08:46 +0000)]
Revert "Revert "Store resolved Strings for AOT code in .bss.""
Fixed oat_test to keep dex files alive. Fixed mips build.
Rewritten the .bss GC root visiting and added write barrier
to the artResolveStringFromCode().
Test: build aosp_mips-eng
Test: m ART_DEFAULT_GC_TYPE=SS test-art-target-host-gtest-oat_test
Test: Run ART test suite on host and Nexus 9.
Bug:
20323084
Bug:
30627598
This reverts commit
5f926055cb88089d8ca27243f35a9dfd89d981f0.
Change-Id: I07fa2278d82b8eb64964c9a4b66cb93726ccda6b
Colin Cross [Tue, 4 Oct 2016 16:45:44 +0000 (16:45 +0000)]
Merge "Compute PPID at ninja time"
Nicolas Geoffray [Tue, 4 Oct 2016 14:08:10 +0000 (14:08 +0000)]
Merge "Remove dead code from compiler driver."
Nicolas Geoffray [Mon, 3 Oct 2016 14:09:25 +0000 (15:09 +0100)]
Remove dead code from compiler driver.
test: m test-art-host
Change-Id: Ie4b6a4c371fff6c9c213d8d6604eb09a5e71d595
Nicolas Geoffray [Tue, 4 Oct 2016 11:01:53 +0000 (11:01 +0000)]
Merge "Revert "Add dex file writer to dexlayout tool.""
Nicolas Geoffray [Tue, 4 Oct 2016 11:01:17 +0000 (11:01 +0000)]
Revert "Add dex file writer to dexlayout tool."
Bug:
29921113
This reverts commit
69b58cf399c3507ca3debf9ec9b2db3c5a0fe7f2.
Change-Id: I8bf9b1222d99489059e45f47d220396a853170a4
Goran Jakovljevic [Tue, 4 Oct 2016 08:17:34 +0000 (10:17 +0200)]
MIPS32: Fix GenerateVirtualCall in the presence of intrinsics
Fix GenerateVirtualCall like it is done in the
I0ed88a48b313a8d28bc39fae40631123aadb13ef for all other architectures.
Test: mma -j2 test-art-target-run-test-optimizing on CI20
Change-Id: Ie55c62701b20c4f40badc95563b2aef9e83807de
Narayan Kamath [Tue, 4 Oct 2016 08:35:13 +0000 (08:35 +0000)]
Merge "interpreter: Remove unnecessary template arg from DoCallCommon."
Narayan Kamath [Mon, 3 Oct 2016 15:51:22 +0000 (16:51 +0100)]
interpreter: Remove unnecessary template arg from DoCallCommon.
The function was always called with a fixed size array so there's
no real advantage in templatizing it this way.
Test: make build-art-host
Change-Id: I782e113225f1c97877f2bcb5afe8df9ae8efd26e
Mathieu Chartier [Tue, 4 Oct 2016 08:02:59 +0000 (08:02 +0000)]
Merge "Add handle wrapper for interpreter field get/set"
Mathieu Chartier [Tue, 4 Oct 2016 05:49:46 +0000 (22:49 -0700)]
Add handle wrapper for interpreter field get/set
The listeners have thread suspension for jdwp tests.
Bug:
31113334
Test: tools/run-jdwp-tests.sh '--mode=host' '--variant=X32' --no-jit --debug
Change-Id: I08f17432cc0c79ee9a0163eb88d4d6355f6851ca
Andreas Gampe [Tue, 4 Oct 2016 02:47:08 +0000 (19:47 -0700)]
ART: Reprint long messages on abort
Add an abort message parameter to Runtime::Abort. In case the message
is multiline (and will thus not be completely preserved in the
Android abort reason), reprint the message after all threads have
been dumped.
Bug:
31893081
Test: m test-art-host
Change-Id: I65bc77691fec79f7c868a90d6132805fcc91e473
Andreas Gampe [Mon, 3 Oct 2016 18:42:31 +0000 (11:42 -0700)]
ART: Clean up IRT-related abort messaging
Change AbortIfNoCheckJNI to take the abort message as a parameter.
This way it can be passed down.
Bug:
31893081
Test: m test-art-host
Change-Id: Ib24c7060f1b32c9613ab84e6c1966082c866bc14
Mathieu Chartier [Tue, 4 Oct 2016 01:06:05 +0000 (01:06 +0000)]
Merge changes I50493bca,I8b266ad0
* changes:
Use ObjPtr for ResolvedFieldAccessTest and ResolvedMethodAccessTest
Move ArtField to ObjPtr
Colin Cross [Mon, 3 Oct 2016 23:40:13 +0000 (16:40 -0700)]
Compute PPID at ninja time
Using $(shell $$PPID) to compute PPID will return the pid of kati, which
will be the same for multiple sequential runs of make without changing
any makefiles. Replace it with a bash subshell command to return the pid
of ninja, which will be different for every call to make.
ART_HOST_TEST_DIR now contains $ that need to propagate to the final
bash command, and must be evaluated exactly once. Escape the $ in
${ART_HOST_TEST_DIR} in define-test-art-run-test so that
${ART_HOST_TEST_DIR} is not substituted in the function call to prevent
evaluating its contents in the $(eval) that wraps it, which would
substitute out the necessary $.
Test: m -j test-art-host
Change-Id: I3303805481cc1c77d1a321674e8c7c003043a7c4
Jeff Hao [Tue, 4 Oct 2016 00:46:26 +0000 (00:46 +0000)]
Merge "Add dex file writer to dexlayout tool."