OSDN Git Service

android-x86/art.git
8 years agoMerge "Optimizing: Improve shift simplification, x >>> 64."
Vladimir Marko [Tue, 15 Mar 2016 16:06:03 +0000 (16:06 +0000)]
Merge "Optimizing: Improve shift simplification, x >>> 64."

8 years agoMerge "runtime: delete native/java_lang_Runtime.cc"
Narayan Kamath [Tue, 15 Mar 2016 15:39:10 +0000 (15:39 +0000)]
Merge "runtime: delete native/java_lang_Runtime.cc"

8 years agoOptimizing: Improve shift simplification, x >>> 64.
Vladimir Marko [Tue, 15 Mar 2016 14:57:32 +0000 (14:57 +0000)]
Optimizing: Improve shift simplification, x >>> 64.

Simplify shifts by a multiple of bit size, not just 0.
ARM codegen does not expect to see such shifts and it
is guarding against them with a DCHECK().

Bug: 27638111
Change-Id: I3ae8383d7edefa0facd375ce511e7a226d5468a1

8 years agoruntime: delete native/java_lang_Runtime.cc
Narayan Kamath [Tue, 15 Mar 2016 14:47:29 +0000 (14:47 +0000)]
runtime: delete native/java_lang_Runtime.cc

We're now using Runtime.c from ojluni, which does the same set
of things via the JVM interface. This removes unnecessary duplication
of code between Runtime.cc and OpenJdkJvm.cc.

Also includes a few changes to JVM_NativeLoad to bring it in sync
with the version in Runtime.cc.

bug: 27387202
Change-Id: I92aac81d131661aca588e352aee7a167593e17d4

8 years agoMerge "Optimizing: Prevent potential valgrind error."
Vladimir Marko [Tue, 15 Mar 2016 13:23:38 +0000 (13:23 +0000)]
Merge "Optimizing: Prevent potential valgrind error."

8 years agoOptimizing: Prevent potential valgrind error.
Vladimir Marko [Tue, 15 Mar 2016 10:40:33 +0000 (10:40 +0000)]
Optimizing: Prevent potential valgrind error.

This CL preemptively extends the workaround from
    https://android-review.googlesource.com/208230
to an almost identical bit of code.

Bug: 27651442
Change-Id: I7683d42b46b16f2293916defc6ef1d871dc9af6c

8 years agoMerge "Optimizing: Fix valgrind error in image_test64."
Vladimir Marko [Tue, 15 Mar 2016 09:51:10 +0000 (09:51 +0000)]
Merge "Optimizing: Fix valgrind error in image_test64."

8 years agoMerge "Fix zygote handling of when secondary image is out of date."
Jeff Hao [Tue, 15 Mar 2016 02:41:14 +0000 (02:41 +0000)]
Merge "Fix zygote handling of when secondary image is out of date."

8 years agoMerge "Fixed bug in BCE, with regression test."
Aart Bik [Tue, 15 Mar 2016 02:40:15 +0000 (02:40 +0000)]
Merge "Fixed bug in BCE, with regression test."

8 years agoFix zygote handling of when secondary image is out of date.
Jeff Hao [Tue, 15 Mar 2016 01:50:49 +0000 (18:50 -0700)]
Fix zygote handling of when secondary image is out of date.

Simple fix for case when primary image is current, but a secondary image
is out of date. Will want to do a cleaner fix in the future.

Bug: 27340451

(cherry-pick of commit b8510d2311964489342d8875ef7f5b681ba874b4)

Change-Id: Ia6db863e3d3c72120630bcbae8d399a78dda2710

8 years agoMerge "Add oatdump support for app images"
Mathieu Chartier [Tue, 15 Mar 2016 00:11:27 +0000 (00:11 +0000)]
Merge "Add oatdump support for app images"

8 years agoAdd oatdump support for app images
Mathieu Chartier [Wed, 9 Mar 2016 00:49:58 +0000 (16:49 -0800)]
Add oatdump support for app images

Example usage on host:
oatdumpd --app-oat=art/plus32.odex --app-image=art/plus32.art
--image=art/oats/system@framework@boot.art --instruction-set=arm

TODO: Add to oatdump test.

Bug: 27408512
Bug: 22858531

(cherry picked from commit bcb6a72569a1401b36a3ad3b6aa4d13e29966cf0)

Change-Id: I9d1aa7eaa16795e5fbabc6974d245849e16b1d03

8 years agoMerge "ART: Update dalvik-cache for zygote after OTA"
Andreas Gampe [Mon, 14 Mar 2016 23:37:15 +0000 (23:37 +0000)]
Merge "ART: Update dalvik-cache for zygote after OTA"

8 years agoART: Update dalvik-cache for zygote after OTA
Andreas Gampe [Sat, 21 Nov 2015 03:26:51 +0000 (19:26 -0800)]
ART: Update dalvik-cache for zygote after OTA

Detect OTA artifacts and move them over to the regular dalvik-cache
on boot.

Bug: 25612095

(cherry picked from commit 754479f7aa056602de8282f4f03718b42c20b89b)

Change-Id: I9a1a22641cc7833435098627a5cc6d954521c4ea

8 years agoFixed bug in BCE, with regression test.
Aart Bik [Mon, 14 Mar 2016 21:11:26 +0000 (14:11 -0700)]
Fixed bug in BCE, with regression test.

The fix is twofold:
(1) Ensure that bound checks are never eliminated more than once
    to guard against any conceivable situation in which the same
    bounds check appears multiple times in array length use list.
(2) Specially reject BoundsCheck(x,x) since that always goes OOB.

BUG=27628526

Change-Id: I399ec4254323e0cfcd0a68898f403cfab7b35135

8 years agoOptimizing: Fix valgrind error in image_test64.
Vladimir Marko [Mon, 14 Mar 2016 21:21:29 +0000 (21:21 +0000)]
Optimizing: Fix valgrind error in image_test64.

Bug: 27651442
Change-Id: Id9b80c6015dbc3b82966766ca4ad010be770f116

8 years agoMerge "Do not include image header in decompressed size"
Mathieu Chartier [Mon, 14 Mar 2016 20:50:00 +0000 (20:50 +0000)]
Merge "Do not include image header in decompressed size"

8 years agoDo not include image header in decompressed size
Mathieu Chartier [Mon, 14 Mar 2016 18:02:59 +0000 (11:02 -0700)]
Do not include image header in decompressed size

Could cause a buffer overflow since we told LZ4 to decompress more
bytes than necessary.

Bug: 27561308

(cherry picked from commit 324eb2c6b049f1677133f0b708730e904c5e73ab)

Change-Id: I524c03b8f0e8a84814dbd8678285184e9d4da9f1

8 years agoMerge "Fix cases where we miss instrumentation changes"
Mathieu Chartier [Mon, 14 Mar 2016 20:14:30 +0000 (20:14 +0000)]
Merge "Fix cases where we miss instrumentation changes"

8 years agoFix cases where we miss instrumentation changes
Mathieu Chartier [Sat, 12 Mar 2016 01:22:23 +0000 (17:22 -0800)]
Fix cases where we miss instrumentation changes

Moved allocation stack push to after we record the allocation since
it can cause thread suspension.

Added handling in entrypoint utils for thread suspension cases.

Keep the AllocRecordObjectMap around since we do not want to delete
it if there are any threads waiting on new_record_condition_. The
condition guards adding stack traces while the GC is running. If we
delete the map and there are still waiters that did not resume, it
caused a CHECK failure. This could happen in cases where one thread
disables allocation tracking while other threads are about to
resume from the condition.

Bug: 27506909

(cherry picked from commit 0f394021d6abfdf9ebea6c7f405ec936a812ea62)

Change-Id: I1dc51e9f000684b4032b57beab59d317ece26f06

8 years agoMerge "Optimizing: Fix TypeConversion(And(x, const)) simplification."
Vladimir Marko [Mon, 14 Mar 2016 19:13:14 +0000 (19:13 +0000)]
Merge "Optimizing: Fix TypeConversion(And(x, const)) simplification."

8 years agoOptimizing: Fix TypeConversion(And(x, const)) simplification.
Vladimir Marko [Mon, 14 Mar 2016 18:00:05 +0000 (18:00 +0000)]
Optimizing: Fix TypeConversion(And(x, const)) simplification.

Avoid introducing implicit conversions when simplifying the
expression TypeConversion(And(x, const)). Previously, when
we dropped the And, we could end up with a TypeConversion to
the same type which should be eliminated on subsequent pass
of the block's instructions; however, a subsequent dependent
TypeConversion in the same block would be processed earlier
and we would unexpectedly see its input as the conversion to
the same type, failing a DCHECK().

Bug: 27626509
Change-Id: I5874a9ceafbf635cf3391beea807ede8468ab5c3

8 years agoMerge "Clean up the valgrind test targets."
Hiroshi Yamauchi [Mon, 14 Mar 2016 18:47:50 +0000 (18:47 +0000)]
Merge "Clean up the valgrind test targets."

8 years agoClean up the valgrind test targets.
Hiroshi Yamauchi [Mon, 14 Mar 2016 18:18:36 +0000 (11:18 -0700)]
Clean up the valgrind test targets.

Allow the 32-bit or the 64-bit valgrind tests to run separately or both.

Bug: 27156726
Change-Id: Ib439eb2f57d3302be152cb5af1a8960a41725f31

8 years agoMerge "Optimizing: Do not re-record standby checks for dynamic BCE."
Vladimir Marko [Mon, 14 Mar 2016 17:42:19 +0000 (17:42 +0000)]
Merge "Optimizing: Do not re-record standby checks for dynamic BCE."

8 years agoMerge "Revert "Enable compilation of secondary dex files""
Nicolas Geoffray [Mon, 14 Mar 2016 15:54:54 +0000 (15:54 +0000)]
Merge "Revert "Enable compilation of secondary dex files""

8 years agoRevert "Enable compilation of secondary dex files"
Nicolas Geoffray [Mon, 14 Mar 2016 15:54:39 +0000 (15:54 +0000)]
Revert "Enable compilation of secondary dex files"

Reverting until we fix the buildbot failures on jdwp and libcore.

Bug: 26719109

This reverts commit 770b39eb564ec16c672ace83515535b055edb2b1.

Change-Id: Ifba366d2ddc9311737b6ea29c007475a40cc803f

8 years agoMerge "ART: Check alignment of section offsets"
Andreas Gampe [Mon, 14 Mar 2016 15:39:01 +0000 (15:39 +0000)]
Merge "ART: Check alignment of section offsets"

8 years agoMerge "Add symbolizer option to generate debug information only."
David Srbecky [Mon, 14 Mar 2016 14:19:40 +0000 (14:19 +0000)]
Merge "Add symbolizer option to generate debug information only."

8 years agoMerge "Ignore empty maps when emitting DWARF variable locations."
David Srbecky [Mon, 14 Mar 2016 14:10:24 +0000 (14:10 +0000)]
Merge "Ignore empty maps when emitting DWARF variable locations."

8 years agoOptimizing: Do not re-record standby checks for dynamic BCE.
Vladimir Marko [Mon, 14 Mar 2016 13:56:02 +0000 (13:56 +0000)]
Optimizing: Do not re-record standby checks for dynamic BCE.

Adding the checks to the standby vector invalidates the
vector storage used by range-based loop in the
BCEVisitor::Finish() as exposed by valgrind in image_test.

Bug: 27597089
Change-Id: Ib0f0e8cdfdb7211a64a26836e085cb99fb2ce8b8

8 years agoMerge "Enable compilation of secondary dex files"
Calin Juravle [Mon, 14 Mar 2016 11:38:24 +0000 (11:38 +0000)]
Merge "Enable compilation of secondary dex files"

8 years agoMerge "Revert "Re-enable test that was causing TimeoutExceptions on ARM64.""
Roland Levillain [Mon, 14 Mar 2016 10:18:29 +0000 (10:18 +0000)]
Merge "Revert "Re-enable test that was causing TimeoutExceptions on ARM64.""

8 years agoIgnore empty maps when emitting DWARF variable locations.
David Srbecky [Sat, 12 Mar 2016 21:10:04 +0000 (21:10 +0000)]
Ignore empty maps when emitting DWARF variable locations.

This is rewrite of https://android-review.googlesource.com/#/c/202115

The aim in both cases is to avoid gaps in generated locations.
(which helps to keep the size of the generated DWARF down).

However, the previous CL was a bit too eager with extending of
variable scope and reporting locations. We might have reported
variable in scope when in fact, it was not.

This CL implements simpler solution by just filtering stackmaps
without dex register maps at first opportunity. This should
ensure that locations for breakpoint locations are completely
accurate as originally intended.

Change-Id: I98378716c0ef5ef46b12181502904621eb6ecf2f

8 years agoRevert "Re-enable test that was causing TimeoutExceptions on ARM64."
Roland Levillain [Sat, 12 Mar 2016 13:42:33 +0000 (13:42 +0000)]
Revert "Re-enable test that was causing TimeoutExceptions on ARM64."

The test is failing again on the ARMv8 concurrent collector
configuration
(https://build.chromium.org/p/client.art/builders/armv8-concurrent-collector/builds/1833/steps/test%20libcore/logs/stdio).

This reverts commit 9e9411cc6edf5fa8be0cabd11203fac09e4bac13.

Bug: 26155567
Change-Id: I81b543a48cbbc1eca99efcc430965222ca6edfec

8 years agoMerge "ART: mterp arm/arm64 cleanup"
Bill Buzbee [Fri, 11 Mar 2016 23:05:59 +0000 (23:05 +0000)]
Merge "ART: mterp arm/arm64 cleanup"

8 years agoART: mterp arm/arm64 cleanup
buzbee [Fri, 11 Mar 2016 17:51:11 +0000 (09:51 -0800)]
ART: mterp arm/arm64 cleanup

Assembly code cleanup in response to comments from already-submitted
CL: https://android-review.googlesource.com/#/c/188977/

Change-Id: I0ea85c5759a08cb50ef3e97dc5cf79b3ba041640

8 years agoMerge "Enable clang for ARM builds."
Stephen Hines [Fri, 11 Mar 2016 21:19:48 +0000 (21:19 +0000)]
Merge "Enable clang for ARM builds."

8 years agoMerge "Assembly region TLAB allocation fast path for arm."
Hiroshi Yamauchi [Fri, 11 Mar 2016 21:18:21 +0000 (21:18 +0000)]
Merge "Assembly region TLAB allocation fast path for arm."

8 years agoAssembly region TLAB allocation fast path for arm.
Hiroshi Yamauchi [Wed, 24 Feb 2016 20:51:18 +0000 (12:51 -0800)]
Assembly region TLAB allocation fast path for arm.

This is for the CC collector.

Share the common fast path code with the tlab fast path code.

Speedup (on N5):
        BinaryTrees:  2291 ->  902 ms (-60%)
        MemAllocTest: 2137 -> 1845 ms (-14%)

Bug: 9986565
Bug: 12687968

Change-Id: Ica63094ec2f85eaa4fd04d202a20090399275d85

8 years agoMerge "ART: Allow unwinding unattached threads"
Christopher Ferris [Fri, 11 Mar 2016 20:14:18 +0000 (20:14 +0000)]
Merge "ART: Allow unwinding unattached threads"

8 years agoMerge "Integrate BitwiseNegated into shared framework."
Vladimir Marko [Fri, 11 Mar 2016 19:48:08 +0000 (19:48 +0000)]
Merge "Integrate BitwiseNegated into shared framework."

8 years agoMerge "Typo in comment in art::mirror::Object::IdentityHashCode."
Roland Levillain [Fri, 11 Mar 2016 19:28:44 +0000 (19:28 +0000)]
Merge "Typo in comment in art::mirror::Object::IdentityHashCode."

8 years agoMerge "Fix DCHECK in OatQuickMethodHeader::GetFrameSizeInBytes()"
David Srbecky [Fri, 11 Mar 2016 19:25:45 +0000 (19:25 +0000)]
Merge "Fix DCHECK in OatQuickMethodHeader::GetFrameSizeInBytes()"

8 years agoFix DCHECK in OatQuickMethodHeader::GetFrameSizeInBytes()
David Srbecky [Fri, 11 Mar 2016 18:48:55 +0000 (18:48 +0000)]
Fix DCHECK in OatQuickMethodHeader::GetFrameSizeInBytes()

We should check whether the frame is divisible by kStackAlignment.

Change-Id: I5936a6e90b38c09775b4c3a8212641e639c4181c

8 years agoMerge "Tests for round() method."
Aart Bik [Fri, 11 Mar 2016 18:43:28 +0000 (18:43 +0000)]
Merge "Tests for round() method."

8 years agoMerge "Suppress a valgrind libc issue."
Hiroshi Yamauchi [Fri, 11 Mar 2016 18:24:22 +0000 (18:24 +0000)]
Merge "Suppress a valgrind libc issue."

8 years agoSuppress a valgrind libc issue.
Hiroshi Yamauchi [Thu, 10 Mar 2016 21:49:14 +0000 (13:49 -0800)]
Suppress a valgrind libc issue.

To prevent our valgrind tests from failing.

Bug: 27596582
Bug: 27156726
Change-Id: Ic55256bd09fdffcf2a0c8fa4fa7951bc648b86a8

8 years agoMerge "Revert "Revert "Allow duplicated methods in different DWARF line tables."""
David Srbecky [Fri, 11 Mar 2016 18:19:11 +0000 (18:19 +0000)]
Merge "Revert "Revert "Allow duplicated methods in different DWARF line tables."""

8 years agoRevert "Revert "Allow duplicated methods in different DWARF line tables.""
David Srbecky [Fri, 11 Mar 2016 14:35:45 +0000 (14:35 +0000)]
Revert "Revert "Allow duplicated methods in different DWARF line tables.""

This reverts commit 8862fac4a0b97d827d2808146d2d79b8d799b998.

Change-Id: I4d5629df4580b7ac08a5cb04924c56eecad3ad25

8 years agoMerge "DWARF: Compensate for compiler's off-by-one instruction."
David Srbecky [Fri, 11 Mar 2016 18:16:35 +0000 (18:16 +0000)]
Merge "DWARF: Compensate for compiler's off-by-one instruction."

8 years agoMerge "Symbolize native debug information for AOTed methods in boot.oat."
David Srbecky [Fri, 11 Mar 2016 17:41:52 +0000 (17:41 +0000)]
Merge "Symbolize native debug information for AOTed methods in boot.oat."

8 years agoSymbolize native debug information for AOTed methods in boot.oat.
David Srbecky [Mon, 7 Mar 2016 20:47:29 +0000 (20:47 +0000)]
Symbolize native debug information for AOTed methods in boot.oat.

The oatdump symbolizer will produce all DWARF information now,
not just the symtab. This allows us to recover almost all native
debug information after the fact, even for non-debuggable oat files.

The main drawback is that non-debuggable oat files do not have
enough stackmaps to provide reasonable line stepping experience.
We might also be missing inlined methods from backtraces.

Change-Id: I73ff6f43a3a0e5a0907af353f74f04e9b0129590

8 years agoMerge "Add trampoline symbols to native debug info."
David Srbecky [Fri, 11 Mar 2016 17:35:48 +0000 (17:35 +0000)]
Merge "Add trampoline symbols to native debug info."

8 years agoAdd trampoline symbols to native debug info.
David Srbecky [Fri, 11 Mar 2016 17:11:44 +0000 (17:11 +0000)]
Add trampoline symbols to native debug info.

The debugger needs them to unwind through the trampolines and to
understand what is happening in the call stack.

Change-Id: Ia554058c3796788adcd7336d620a7734eb366905

8 years agoMerge "Minor optimizations of debug::WriteCFISection."
David Srbecky [Fri, 11 Mar 2016 17:33:22 +0000 (17:33 +0000)]
Merge "Minor optimizations of debug::WriteCFISection."

8 years agoMinor optimizations of debug::WriteCFISection.
David Srbecky [Fri, 11 Mar 2016 16:54:22 +0000 (16:54 +0000)]
Minor optimizations of debug::WriteCFISection.

The method might be passed method infos without CFI in some cases.
Use the sorting phase as a chance to filter them out.
This makes sure we do not allocate memory and sort methods if there
is in fact no work to do.

Also change the sort to stable - for the sake of determinism.

Change-Id: I97d57d77e8b709d0d49d6971f66b955efcbb57b0

8 years agoTypo in comment in art::mirror::Object::IdentityHashCode.
Roland Levillain [Fri, 11 Mar 2016 16:27:27 +0000 (16:27 +0000)]
Typo in comment in art::mirror::Object::IdentityHashCode.

Change-Id: I5380d1c34deda32c668e48ed5f80e96b29f5b0b5

8 years agoMerge "MIPS: Don't use $t8 for calling entrypoints"
Roland Levillain [Fri, 11 Mar 2016 16:47:02 +0000 (16:47 +0000)]
Merge "MIPS: Don't use $t8 for calling entrypoints"

8 years agoMerge "Revert "Revert "Revert "Revert "Change condition to opposite if lhs is constan...
Nicolas Geoffray [Fri, 11 Mar 2016 16:37:23 +0000 (16:37 +0000)]
Merge "Revert "Revert "Revert "Revert "Change condition to opposite if lhs is constant"""""

8 years agoDWARF: Compensate for compiler's off-by-one instruction.
David Srbecky [Fri, 11 Mar 2016 14:25:00 +0000 (14:25 +0000)]
DWARF: Compensate for compiler's off-by-one instruction.

The compiler generates stackmaps with PC value which is *after*
the instruction rather PC of the instruction itself. This causes
trouble when generating native line number mapping since the
branch instruction itself will not be mapped, and consequently
we get incorrect line number in backtraces. Add code to compensate
for this. See the in-code comment for more details.

Change-Id: I72c992e6d08a767f314290d562421b251ae60732

8 years agoMIPS: Don't use $t8 for calling entrypoints
Goran Jakovljevic [Fri, 11 Mar 2016 14:22:18 +0000 (15:22 +0100)]
MIPS: Don't use $t8 for calling entrypoints

When jumping to entrypoints, $t9 register must be used. Value of $gp
is calculated based on value from $t9 and it should contain the
address of the first instruction of the entrypoint.

This CL enables MIPS32 and MIPS64 to boot.

Change-Id: Idf0f7f479cceafabb88b1b087da190dbda86e1c3

8 years agoMerge "Fix invariant in reference type propagation."
Nicolas Geoffray [Fri, 11 Mar 2016 16:15:32 +0000 (16:15 +0000)]
Merge "Fix invariant in reference type propagation."

8 years agoAdd symbolizer option to generate debug information only.
David Srbecky [Thu, 10 Mar 2016 15:32:37 +0000 (15:32 +0000)]
Add symbolizer option to generate debug information only.

There is no need to copy the .rodata and .text sections since the
consumers can get their data form the original ELF file. We mark
those section as SHT_NOBITS instead, which is the standard way to
create ELF file with separate debug information only.

The new option is called --only-keep-debug and it essentially has
the same effect as --only-keep-debug option for "strip" command.

Change-Id: I4f7488c04bbad86288beeda11c3bf295890e740d

8 years agoMerge "Revert "Allow duplicated methods in different DWARF line tables.""
David Srbecky [Fri, 11 Mar 2016 14:35:33 +0000 (14:35 +0000)]
Merge "Revert "Allow duplicated methods in different DWARF line tables.""

8 years agoRevert "Allow duplicated methods in different DWARF line tables."
David Srbecky [Fri, 11 Mar 2016 14:34:47 +0000 (14:34 +0000)]
Revert "Allow duplicated methods in different DWARF line tables."

This reverts commit 3c5462364daee17e651299f1c370f965f34baed8.

Change-Id: Icc63c47cd53abeff2a470ae84715012a1ef988ac

8 years agoMerge "Allow duplicated methods in different DWARF line tables."
David Srbecky [Fri, 11 Mar 2016 14:28:00 +0000 (14:28 +0000)]
Merge "Allow duplicated methods in different DWARF line tables."

8 years agoMerge "Allow generation of native debug info for multiple methods."
David Srbecky [Fri, 11 Mar 2016 14:25:56 +0000 (14:25 +0000)]
Merge "Allow generation of native debug info for multiple methods."

8 years agoMerge "Refactor MethodDebugInfo (input of DWARF writer)."
David Srbecky [Fri, 11 Mar 2016 14:25:40 +0000 (14:25 +0000)]
Merge "Refactor MethodDebugInfo (input of DWARF writer)."

8 years agoAllow duplicated methods in different DWARF line tables.
David Srbecky [Fri, 11 Mar 2016 11:49:20 +0000 (11:49 +0000)]
Allow duplicated methods in different DWARF line tables.

This makes the compilation units more self-contained.
If method is mentioned in the compilation unit, we can
also find its line table there. Otherwise, we would have
to search through all of them.

Change-Id: I0cdfb9006e796e41e123fc1f4fecd15312570068

8 years agoFix invariant in reference type propagation.
Nicolas Geoffray [Fri, 11 Mar 2016 13:35:51 +0000 (13:35 +0000)]
Fix invariant in reference type propagation.

Also some cleanups.

Change-Id: I7f0ec7d06b4bab10dbfa230c757447d311658f93

8 years agoIntegrate BitwiseNegated into shared framework.
Artem Serov [Tue, 9 Feb 2016 17:15:29 +0000 (17:15 +0000)]
Integrate BitwiseNegated into shared framework.

Share implementation between arm and arm64.

Change-Id: I0dd12e772cb23b4c181fd0b1e2a447470b1d8702

8 years agoRevert "Revert "Revert "Revert "Change condition to opposite if lhs is constant""""
Anton Shamin [Mon, 15 Feb 2016 06:48:36 +0000 (12:48 +0600)]
Revert "Revert "Revert "Revert "Change condition to opposite if lhs is constant""""

This reverts commit d4aee949b3dd976295201b5310f13aa2df40afa1.

Change-Id: I505b8c9863c310a3a708f580b00d425b750c9541

8 years agoMerge "Fix thread race when fetching the ProfilingInfo object."
Nicolas Geoffray [Fri, 11 Mar 2016 11:33:13 +0000 (11:33 +0000)]
Merge "Fix thread race when fetching the ProfilingInfo object."

8 years agoFix thread race when fetching the ProfilingInfo object.
Nicolas Geoffray [Fri, 11 Mar 2016 09:57:57 +0000 (09:57 +0000)]
Fix thread race when fetching the ProfilingInfo object.

Problem is:
1) Compiler fetches the ProfilingInfo of A, it's null.
2) Mutator creates the ProfilingInfo.
3) Compiler notifies it's not using A anymore, calls
   ProfilingInfo::DecrementInlineUse -> Crash as we expected
   ProfilingInfo::IncrementUse to be called before.

Also update some namings to better reflect what is going on.

Change-Id: I55ea4c5d81988131467095e18a0d13a8be9d0ef7

8 years agoART: Allow unwinding unattached threads
Andreas Gampe [Tue, 1 Mar 2016 22:38:37 +0000 (14:38 -0800)]
ART: Allow unwinding unattached threads

Partial revert of commit ed8b723c5f3989d2593ec21c65c96d6d8bf25579.

Make it (constexpr) configurable whether we allow unwinding native
stacks of unattached threads.

Bug: 27449879

Change-Id: I307a80c9a0166f33fa8a41e492d7eb0d77d0e44f

8 years agoMerge "Log when why an Unwind failed."
Christopher Ferris [Thu, 10 Mar 2016 22:12:34 +0000 (22:12 +0000)]
Merge "Log when why an Unwind failed."

8 years agoMerge "Allocate interrupted exception before re-acquiring lock"
Mathieu Chartier [Thu, 10 Mar 2016 20:07:55 +0000 (20:07 +0000)]
Merge "Allocate interrupted exception before re-acquiring lock"

8 years agoMerge "ART: Release all resource on MonitorPool destruction"
Andreas Gampe [Thu, 10 Mar 2016 20:06:27 +0000 (20:06 +0000)]
Merge "ART: Release all resource on MonitorPool destruction"

8 years agoART: Release all resource on MonitorPool destruction
Andreas Gampe [Thu, 10 Mar 2016 16:33:45 +0000 (08:33 -0800)]
ART: Release all resource on MonitorPool destruction

To be valgrind-clean, we need to release the current metadata and
all chunks on destruction.

Bug: 27156726
Change-Id: Ia51cea139a6e9669975b6ac045f5223cd68f1f6a

8 years agoAllocate interrupted exception before re-acquiring lock
Mathieu Chartier [Thu, 10 Mar 2016 18:49:35 +0000 (10:49 -0800)]
Allocate interrupted exception before re-acquiring lock

Allocating the monitor exception after acquiring the lock can cause
a deadlock in the following scenario:

Reference queue daemon interrupted from wait() on
ReferenceQueue.class. Tries to allocate exception and starts a GC
that blocks on WaitForGcToComplete.

Some other thread is already doing a GC, and tries to enqueue the
cleared refereneces. This deadlocks trying to lock Reference.class
in ReferenceQueue.add.

Bug: 27508829
Change-Id: Icbc18b6b8d1e906c3f7413810d6cefdda06eb921

8 years agoLog when why an Unwind failed.
Christopher Ferris [Thu, 10 Mar 2016 00:03:00 +0000 (16:03 -0800)]
Log when why an Unwind failed.

Bug: 27449879
Change-Id: If7ae5f0991da6fd64dd45e22d69eecef1388a28e

8 years agoTests for round() method.
Aart Bik [Thu, 10 Mar 2016 18:59:10 +0000 (10:59 -0800)]
Tests for round() method.

Rationale:
While working on some ideas on implementing round(),
I wrote these tests. We might as well use them to
ensure any future work does not break (test also
provides an easy place to add future "difficult" values).

bug=26327751

Change-Id: I911a3e1e5fe45147bc1b7214225c9aec4e979965

8 years agoMerge "ART: Do not use vixld - workaround to fix dex2oatds."
Vladimir Marko [Thu, 10 Mar 2016 18:48:37 +0000 (18:48 +0000)]
Merge "ART: Do not use vixld - workaround to fix dex2oatds."

8 years agoMerge "Avoid generating dead code on frame enter/exit. This includes stack operations...
Aart Bik [Thu, 10 Mar 2016 18:39:18 +0000 (18:39 +0000)]
Merge "Avoid generating dead code on frame enter/exit. This includes stack operations and, on x86, call/pop to read PC."

8 years agoART: Do not use vixld - workaround to fix dex2oatds.
Vladimir Marko [Thu, 10 Mar 2016 17:25:50 +0000 (17:25 +0000)]
ART: Do not use vixld - workaround to fix dex2oatds.

This is a quick workaround for ODR violations caused
by linking libvixl.a compiled without VIXL_DEBUG with
the libartd-compiler.a compiled with VIXL_DEBUG.

Bug: 27588884
Change-Id: Ib1af165f177f125f03cdd99777dff4c2912f6405

8 years agoAvoid generating dead code on frame enter/exit.
Aart Bik [Thu, 3 Mar 2016 00:06:13 +0000 (16:06 -0800)]
Avoid generating dead code on frame enter/exit.
This includes stack operations and, on x86, call/pop to read PC.

bug=26997690

Rationale:
(1) If method is fully intrinsified, and makes no calls in slow
    path or uses special input, no need to require current method.
(2) Invoke instructions with HasPcRelativeDexCache() generate code
    that reads the PC (call/pop) on x86. However, if the invoke is
    an intrinsic that is later replaced with actual code, this PC
    reading code may be dead.

Example X86 (before/after):

0x0000108c: 83EC0C      sub esp, 12
0x0000108f: 890424      mov [esp], eax       <-- not needed
0x00001092: E800000000  call +0 (0x00001097)
0x00001097: 58          pop eax              <-- dead code to read PC
0x00001098: F30FB8C1    popcnt eax, ecx
0x0000109c: F30FB8DA    popcnt ebx, edx
0x000010a0: 03D8        add ebx, eax
0x000010a2: 89D8        mov eax, ebx
0x000010a4: 83C40C      add esp, 12          <-- not needed
0x000010a7: C3          ret

0x0000103c: F30FB8C1    popcnt eax, ecx
0x00001040: F30FB8DA    popcnt ebx, edx
0x00001044: 03D8        add ebx, eax
0x00001046: 89D8        mov eax, ebx
0x00001048: C3          ret

Example ARM64 (before/after):

0x0000103c: f81e0fe0      str x0, [sp, #-32]!
0x00001040: f9000ffe      str lr, [sp, #24]
0x00001044: dac01020      clz x0, x1
0x00001048: f9400ffe      ldr lr, [sp, #24]
0x0000104c: 910083ff      add sp, sp, #0x20 (32)
0x00001050: d65f03c0      ret

0x0000103c: dac01020      clz x0, x1
0x00001040: d65f03c0      ret

Change-Id: I8377db80c9a901a08fff4624927cf4a6e585da0c

8 years agoMerge "Add Mips abiflags section to the oatfile."
Vladimir Marko [Thu, 10 Mar 2016 17:11:44 +0000 (17:11 +0000)]
Merge "Add Mips abiflags section to the oatfile."

8 years agoMerge "MIPS32: java.lang.Thread, and java.lang.String intrinsics:"
Roland Levillain [Thu, 10 Mar 2016 17:09:37 +0000 (17:09 +0000)]
Merge "MIPS32: java.lang.Thread, and java.lang.String intrinsics:"

8 years agoMerge "Disable some image_test cases with concurrent collector."
Roland Levillain [Thu, 10 Mar 2016 15:28:34 +0000 (15:28 +0000)]
Merge "Disable some image_test cases with concurrent collector."

8 years agoMerge "Tweak JIT hotness threshold."
Nicolas Geoffray [Thu, 10 Mar 2016 15:25:02 +0000 (15:25 +0000)]
Merge "Tweak JIT hotness threshold."

8 years agoDisable some image_test cases with concurrent collector.
Roland Levillain [Thu, 10 Mar 2016 15:16:31 +0000 (15:16 +0000)]
Disable some image_test cases with concurrent collector.

The following image_test cases sometimes fail on ART Builbot's
x86 concurrent collector configuration:
- ImageTest.WriteReadUncompressed
- ImageTest.WriteReadLZ4
- ImageTest.WriteReadLZ4HC
Disable them to make the build turn green again, while we
investigate the failures.

Bug: 27578460
Change-Id: I46126e4690e6300e3bfa771c1b6a560272ecb1da

8 years agoTweak JIT hotness threshold.
Nicolas Geoffray [Tue, 8 Mar 2016 16:50:21 +0000 (16:50 +0000)]
Tweak JIT hotness threshold.

bug:23128949
bug:27398183

Change-Id: Iffde6ba064e54546827cb8fc2a670baedf2e2409

8 years agoMerge "MIPS32: Fix test 082-inline-execute"
Nicolas Geoffray [Thu, 10 Mar 2016 14:23:35 +0000 (14:23 +0000)]
Merge "MIPS32: Fix test 082-inline-execute"

8 years agoMerge "Fix and enable java.lang.StringFactory intrinsics."
Roland Levillain [Thu, 10 Mar 2016 14:21:40 +0000 (14:21 +0000)]
Merge "Fix and enable java.lang.StringFactory intrinsics."

8 years agoMIPS32: Fix test 082-inline-execute
Goran Jakovljevic [Tue, 8 Mar 2016 14:26:51 +0000 (15:26 +0100)]
MIPS32: Fix test 082-inline-execute

This is fixing the test when run with --debuggable flag.

Change-Id: Ifd11c28a77f1c6f6fe5951c5c27309187b29d3ee

8 years agoMerge "Clear inline caches at each full GC."
Nicolas Geoffray [Thu, 10 Mar 2016 10:47:57 +0000 (10:47 +0000)]
Merge "Clear inline caches at each full GC."

8 years agoAllow generation of native debug info for multiple methods.
David Srbecky [Wed, 9 Mar 2016 11:44:44 +0000 (11:44 +0000)]
Allow generation of native debug info for multiple methods.

Slight tweak in the API which I will need in the near future.

Change-Id: I45954ae16710bc941a9a06a3a17c70798315ca53

8 years agoRefactor MethodDebugInfo (input of DWARF writer).
David Srbecky [Mon, 7 Mar 2016 17:33:57 +0000 (17:33 +0000)]
Refactor MethodDebugInfo (input of DWARF writer).

Do not pass CompiledMethod pointer through since it is only available
during AOT compile but not during JIT compile or at runtime. Creating
mock CompiledMethod just pass data is proving increasingly tricky, so
copy the fields that we need to MethodDebugInfo instead.

Change-Id: I820297b41e769fcac488c0ff2d2ea0492bb13ed8

8 years agoMerge "ART: Fix assumption in class profile collection"
Andreas Gampe [Thu, 10 Mar 2016 01:44:24 +0000 (01:44 +0000)]
Merge "ART: Fix assumption in class profile collection"