OSDN Git Service

android-x86/art.git
9 years agoFix checks for kLiteral in local optimizations.
Dave Allison [Wed, 6 Aug 2014 04:32:46 +0000 (21:32 -0700)]
Fix checks for kLiteral in local optimizations.

The check for kLiteral (literal load) just checked the kLiteral
bit in the def mask.  The kEncodeAll mask has the kLiteral bit
set so this check was triggering.  The fix is to check for
only the kLiteral bit being set and no other special bits.

The semantics of the special bits in the use/def mask is that
only one of them can be set at the same time.

Bug: 16824330

Change-Id: I0f1c1157e017870414ffef11767e5433d1fd4401

9 years agoMerge "Add card table test."
Mathieu Chartier [Wed, 6 Aug 2014 18:35:57 +0000 (18:35 +0000)]
Merge "Add card table test."

9 years agoAdd card table test.
Mathieu Chartier [Wed, 6 Aug 2014 01:15:56 +0000 (18:15 -0700)]
Add card table test.

Tests some of the functionality supported by the card table.

Removed some logcat spam from monitor_pool.

Change-Id: I1423816a72572f78aca44552effa2b4c6aac46c8

9 years agoMerge "Revert "Rules depend on DEX2OATD, not DEX2OATD_DEPENDENCY.""
Brian Carlstrom [Wed, 6 Aug 2014 16:42:26 +0000 (16:42 +0000)]
Merge "Revert "Rules depend on DEX2OATD, not DEX2OATD_DEPENDENCY.""

9 years agoRevert "Rules depend on DEX2OATD, not DEX2OATD_DEPENDENCY."
Brian Carlstrom [Mon, 28 Jul 2014 23:12:58 +0000 (23:12 +0000)]
Revert "Rules depend on DEX2OATD, not DEX2OATD_DEPENDENCY."

This reverts commit b1efe9746724f28773d185c7137c16ca8207ed47.

Change-Id: I5af2fae2c7b72e6ded72f00d3f7d4f306f6703ee

9 years agoMerge "When performing a long jump, update SP last."
Jeff Hao [Wed, 6 Aug 2014 16:01:53 +0000 (16:01 +0000)]
Merge "When performing a long jump, update SP last."

9 years agoWhen performing a long jump, update SP last.
Jeff Hao [Wed, 6 Aug 2014 01:59:16 +0000 (18:59 -0700)]
When performing a long jump, update SP last.

Bug: 16571371

(cherry picked from commit 76a0723959c2185ace249e705774af9a679a0e0d)

Change-Id: Ib3a7bcda8aecdf7e92ea40a76de32aed83a42ee0

9 years agoMerge "Fix SIB for base + index addressing in x86_64 assembler."
Nicolas Geoffray [Wed, 6 Aug 2014 15:31:19 +0000 (15:31 +0000)]
Merge "Fix SIB for base + index addressing in x86_64 assembler."

9 years agoFix SIB for base + index addressing in x86_64 assembler.
Nicolas Geoffray [Tue, 5 Aug 2014 23:24:09 +0000 (00:24 +0100)]
Fix SIB for base + index addressing in x86_64 assembler.

Change-Id: Ib630bc28e6d694ffbe4a4a71cc988e36d00f6633

9 years agoMerge "Fix movw in x86_64 assembler."
Nicolas Geoffray [Tue, 5 Aug 2014 22:19:45 +0000 (22:19 +0000)]
Merge "Fix movw in x86_64 assembler."

9 years agoFix movw in x86_64 assembler.
Nicolas Geoffray [Tue, 5 Aug 2014 21:52:45 +0000 (22:52 +0100)]
Fix movw in x86_64 assembler.

Change-Id: Ibceb03fd57adea09643aa77a9399be196fa14709

9 years agoMerge "x86: Fix assembler for Pextr"
Ian Rogers [Tue, 5 Aug 2014 21:57:45 +0000 (21:57 +0000)]
Merge "x86: Fix assembler for Pextr"

9 years agoMerge "X86: Assembler: Correct r8_form for some cases"
Ian Rogers [Tue, 5 Aug 2014 21:32:29 +0000 (21:32 +0000)]
Merge "X86: Assembler: Correct r8_form for some cases"

9 years agoMerge "ART: Fix lock max count definition"
Ian Rogers [Tue, 5 Aug 2014 21:44:41 +0000 (21:44 +0000)]
Merge "ART: Fix lock max count definition"

9 years agoMerge "Add x86_64 optimized __memcmp16 implementation; fix tabs in 32-bit implementat...
Andreas Gampe [Wed, 6 Aug 2014 15:10:58 +0000 (15:10 +0000)]
Merge "Add x86_64 optimized __memcmp16 implementation; fix tabs in 32-bit implementation."

9 years agoAdd x86_64 optimized __memcmp16 implementation;
Varvara Rainchik [Wed, 30 Jul 2014 13:01:24 +0000 (17:01 +0400)]
Add x86_64 optimized __memcmp16 implementation;
fix tabs in 32-bit implementation.

Change-Id: I7bbfb344074aed66511c1a845998dc38798116ea
Signed-off-by: Varvara Rainchik <varvara.rainchik@intel.com>
9 years agoMerge "Fix byte_cas to use uintptr_t Atomic."
Mathieu Chartier [Wed, 6 Aug 2014 01:12:05 +0000 (01:12 +0000)]
Merge "Fix byte_cas to use uintptr_t Atomic."

9 years agoFix byte_cas to use uintptr_t Atomic.
Mathieu Chartier [Tue, 5 Aug 2014 22:59:43 +0000 (15:59 -0700)]
Fix byte_cas to use uintptr_t Atomic.

Previously we used int32_t Atomic with a uintptr_t bit shift.
This was a mismatch on 64 bit and resulted in occasionally
having the cas return without having succeeded.

Bug: 16819816

Change-Id: I10fa9c2f09ae8fe0997a4c9293fa123c6bde5920

9 years agoMerge "x86_64: Load double constant using core register"
Ian Rogers [Tue, 5 Aug 2014 20:38:02 +0000 (20:38 +0000)]
Merge "x86_64: Load double constant using core register"

9 years agox86_64: Load double constant using core register
Alexei Zavjalov [Thu, 10 Jul 2014 11:34:23 +0000 (18:34 +0700)]
x86_64: Load double constant using core register

This changes the double const loading scheme for x86_64 in order to
reduce number of memory accesses.

Change-Id: If6ee34b890287ed616ca4ea5d1c13f06a06293a9
Signed-off-by: Alexei Zavjalov <alexei.zavjalov@intel.com>
9 years agoMerge "Create allocator adapter for using Arena in std containers."
Vladimir Marko [Tue, 5 Aug 2014 09:13:39 +0000 (09:13 +0000)]
Merge "Create allocator adapter for using Arena in std containers."

9 years agoCreate allocator adapter for using Arena in std containers.
Vladimir Marko [Thu, 31 Jul 2014 14:33:43 +0000 (15:33 +0100)]
Create allocator adapter for using Arena in std containers.

Create ArenaAllocatorAdapter, similar to the existing
ScopedArenaAllocatorAdapter, for allocating memory for
standard containers via the ArenaAllocator. Add the ability
to specify allocation kind rather than just kArenaAllocSTL
to both adapters. Move the scoped arena allocator to the
scoped_arena_containers.h header file.

Define template aliases for containers using the new adapter
and change a few MIRGraph and Mir2Lir members to use them.

Change-Id: I9bbc50248e0fed81729497b848cb29bf68444268

9 years agoMerge changes Ibe92d8b5,I9bada598
Brian Carlstrom [Tue, 5 Aug 2014 17:49:28 +0000 (17:49 +0000)]
Merge changes Ibe92d8b5,I9bada598

* changes:
  Make system use patchoat to relocate during runtime.
  Temporary fix for bad LIR generation in GenInlinedGet.

9 years agoMake system use patchoat to relocate during runtime.
Alex Light [Fri, 18 Jul 2014 21:57:04 +0000 (14:57 -0700)]
Make system use patchoat to relocate during runtime.

Change dalvik_system_DexFile.cc so that isDexOptNeededInternal will be
able to indicate that a patchoat is required. Change default of relocate
option to be on.

Bug: 15358152

Change-Id: Ibe92d8b55a24bbf718b0416a21b76e5df7a2de26

9 years agoTemporary fix for bad LIR generation in GenInlinedGet.
Alex Light [Tue, 29 Jul 2014 22:22:40 +0000 (15:22 -0700)]
Temporary fix for bad LIR generation in GenInlinedGet.

In GenInlinedGet we will sometimes try to load a class directly. If
we are doing that and find one, but cannot get the direct pointer
(due to image compilation or including patch information) we might
end up generating incorrect patching code due to the changing of the
cu_->dex_file attribute.

Bug: 16656190
Change-Id: I9bada598bf4fd972d6164163eedd4a496da306d6

9 years agoMerge "ART: Enable x86_64 bit support for intrinsic for System.arraycopy(char[],...
Ian Rogers [Tue, 5 Aug 2014 17:30:49 +0000 (17:30 +0000)]
Merge "ART: Enable x86_64 bit support for intrinsic for System.arraycopy(char[], ..)"

9 years agoART: Enable x86_64 bit support for intrinsic for System.arraycopy(char[], ..)
DaniilSokolov [Fri, 18 Jul 2014 01:58:15 +0000 (18:58 -0700)]
ART: Enable x86_64 bit support for intrinsic for System.arraycopy(char[], ..)

Implements x86_64 support for intrinsic for java.lang.System.arraycopy(char[], int, char[], int, int).

With this fix the intrinsic works on x86 and x86_64 architectures.

Change-Id: Icc2889ccd0cf7d821522abb7437893e3149e7c99
Signed-off-by: Daniil Sokolov <daniil.y.sokolov@intel.com>
9 years agoMerge "ART: Fix memory unmapped twice issue in ElfFile::Load(bool)"
Brian Carlstrom [Tue, 5 Aug 2014 05:45:06 +0000 (05:45 +0000)]
Merge "ART: Fix memory unmapped twice issue in ElfFile::Load(bool)"

9 years agoART: Fix memory unmapped twice issue in ElfFile::Load(bool)
Jim_Guo [Mon, 28 Apr 2014 03:11:57 +0000 (11:11 +0800)]
ART: Fix memory unmapped twice issue in ElfFile::Load(bool)

Root Cause:
  The overlapped memory region will be unmapped by
  (1) ~MemMap() of reservation MemMap (reserve) and
  (2) ~MemMap() of "reuse" MemMap (segment).
  Someone takes the memory region after (1) and it will be unmapped in (2).
  So, SIGSEGV occurs when using the unmapped memory region.

Solution:
  Fixes this issue by skip unmap "reuse" MemMap in destructor.
  And always create reservation MemMap before "reuse" MemMap. (It also solved
  the fixupELF case which does not reserve the whole needed memory region).

Bug: 16486685
Change-Id: I8f2538861d5c3fa7b9a04d2c3f516319cc060291

9 years agoMerge "Use the uncached form of BacktraceMap::Create."
Christopher Ferris [Tue, 5 Aug 2014 02:56:06 +0000 (02:56 +0000)]
Merge "Use the uncached form of BacktraceMap::Create."

9 years agoUse the uncached form of BacktraceMap::Create.
Christopher Ferris [Tue, 5 Aug 2014 00:15:21 +0000 (17:15 -0700)]
Use the uncached form of BacktraceMap::Create.

The code assumes that calling BacktraceMap::Create creates a map based
on the current state of the maps. This is not true by default, so
call the Create function indicating we want an uncached BacktraceMap object.

(cherry picked from 27a10f618357cf85cc0677a04f0a5a3a8a437aed)

Change-Id: Ic7a2dedaea9950174b24ede4dcda30ffe036fe93

9 years agoMerge "ART: Refactor long ops in quick compiler"
Andreas Gampe [Tue, 5 Aug 2014 01:44:22 +0000 (01:44 +0000)]
Merge "ART: Refactor long ops in quick compiler"

9 years agoART: Refactor long ops in quick compiler
Andreas Gampe [Mon, 4 Aug 2014 23:30:03 +0000 (16:30 -0700)]
ART: Refactor long ops in quick compiler

Make GenArithOpLong virtual. Let the implementation in gen_common be
very basic, without instruction-set checks, and meant as a fall-back.
Backends should implement and dispatch to code for better implementations.
This allows to remove the GenXXXLong virtual methods from Mir2Lir, and
clean up the backends (especially removing some LOG(FATAL) implementations).

Change-Id: I6366443c0c325c1999582d281608b4fa229343cf

9 years agoMerge "Set REG0_USED on X86 Set8R instruction"
Ian Rogers [Tue, 5 Aug 2014 16:56:44 +0000 (16:56 +0000)]
Merge "Set REG0_USED on X86 Set8R instruction"

9 years agoSet REG0_USED on X86 Set8R instruction
Mark Mendell [Thu, 31 Jul 2014 18:36:54 +0000 (14:36 -0400)]
Set REG0_USED on X86 Set8R instruction

Since this instruction only affects the low byte of the register, it is
preceded by an XOR to zero the upper 3 bytes.  Set8R isn't marked as
using operand 0 as an input.  In practice, this works for now, as the
Xor sets the CC and Set8R uses the CC (although not that of the Xor, but
of a Cmp generally).

This just marks REG0 as using the previous contents of the register, as
it is modifying only 1 byte of 4.

Change-Id: I7a69cbdb06979da5d5d2ae17fabd7c22c5a17701
Signed-off-by: Mark Mendell <mark.p.mendell@intel.com>
9 years agoMerge "ART: Correct disassembling of 64bit immediates on x86_64"
Ian Rogers [Tue, 5 Aug 2014 14:57:33 +0000 (14:57 +0000)]
Merge "ART: Correct disassembling of 64bit immediates on x86_64"

9 years agoMerge "x86_64: Fix Test32RM"
Ian Rogers [Tue, 5 Aug 2014 14:53:08 +0000 (14:53 +0000)]
Merge "x86_64: Fix Test32RM"

9 years agox86_64: Fix Test32RM
Chao-ying Fu [Thu, 24 Jul 2014 19:08:28 +0000 (12:08 -0700)]
x86_64: Fix Test32RM

This patch fixes Test32RM use flags and the format.

Change-Id: I486cb7f27e65caeefccbd3bbcc38257ddca033c8
Signed-off-by: Chao-ying Fu <chao-ying.fu@intel.com>
9 years agoMerge "Use canonical paths when searching for dex files"
Calin Juravle [Tue, 5 Aug 2014 10:47:26 +0000 (10:47 +0000)]
Merge "Use canonical paths when searching for dex files"

9 years agoMerge "AArch64: Implement InexpensiveConstant methods."
Andreas Gampe [Mon, 4 Aug 2014 23:22:12 +0000 (23:22 +0000)]
Merge "AArch64: Implement InexpensiveConstant methods."

9 years agoAArch64: Implement InexpensiveConstant methods.
Matteo Franchin [Fri, 4 Jul 2014 11:53:27 +0000 (12:53 +0100)]
AArch64: Implement InexpensiveConstant methods.

Implement IsInexpensiveConstant and friends for A64.
Also extending the methods to take the opcode with respect to which
the constant is inexpensive. Additionally, logical operations (i.e.
and, or, xor) can now handle the immediates 0 and ~0 (which are not
logical immediates).

Change-Id: I46ce1287703765c5ab54983d13c1b3a1f5838622

9 years agoUse canonical paths when searching for dex files
Calin Juravle [Tue, 15 Jul 2014 22:56:47 +0000 (23:56 +0100)]
Use canonical paths when searching for dex files

Apps which use the DexPathClassLoader directly may
pass symlinks when trying to load dex files. This
will not work as we use string comparision to find
the dex in an oat file. The CL fixes this issue by
using using dex conical paths for comparisons.

Bug: 15313272

Change-Id: Ic314374b17612c3afbcadec93a88b2515a0aca5e

9 years agoART: Fix lock max count definition
Dmitry Petrochenko [Mon, 28 Jul 2014 10:40:16 +0000 (17:40 +0700)]
ART: Fix lock max count definition

The lock max count should utilize 14 bits since 2 highest bits are reserved for lock state.

Change-Id: I9d562f7bca9c0853231800a706a8523204e8aa9d
Signed-off-by: Dmitry Petrochenko <dmitry.petrochenko@intel.com>
Signed-off-by: Serguei Katkov <serguei.i.katkov@intel.com>
9 years agoMerge "Add trampoline invoke and ART callback tests for native-bridge"
Andreas Gampe [Mon, 4 Aug 2014 23:32:31 +0000 (23:32 +0000)]
Merge "Add trampoline invoke and ART callback tests for native-bridge"

9 years agoAdd trampoline invoke and ART callback tests for native-bridge
Yong WU [Sun, 3 Aug 2014 08:06:52 +0000 (16:06 +0800)]
Add trampoline invoke and ART callback tests for native-bridge

Create trampolines for JNI functions in native-bridge. Test trampoline
invoking from ART side.

Add tests for ART callbacks which called from native-bridge. In this test
they are invoked from the trampoline of JNI_OnLoad.

Change-Id: Ibaa5103fdf49a6b29a1c12f45410fe5f82d46b1e

9 years agoMerge "AArch64: Add inlining support for ceil(), floor(), rint(), round()"
Andreas Gampe [Mon, 4 Aug 2014 22:56:32 +0000 (22:56 +0000)]
Merge "AArch64: Add inlining support for ceil(), floor(), rint(), round()"

9 years agoAArch64: Add inlining support for ceil(), floor(), rint(), round()
Serban Constantinescu [Thu, 31 Jul 2014 18:07:17 +0000 (19:07 +0100)]
AArch64: Add inlining support for ceil(), floor(), rint(), round()

This patch adds inlining support for the following Math, StrictMath
methods in the ARM64 backend:
  * double ceil(double)
  * double floor(double)
  * double rint(double)
  * long round(double)
  * int round(float)

Also some cleanup.

Change-Id: I9f5a2f4065b1313649f4b0c4380b8176703c3fe1
Signed-off-by: Serban Constantinescu <serban.constantinescu@arm.com>
9 years agoMerge "Make verifier identify unresolved methods as constructors by <init>."
Jeff Hao [Mon, 4 Aug 2014 22:17:40 +0000 (22:17 +0000)]
Merge "Make verifier identify unresolved methods as constructors by <init>."

9 years agoMake verifier identify unresolved methods as constructors by <init>.
Jeff Hao [Mon, 4 Aug 2014 21:47:17 +0000 (14:47 -0700)]
Make verifier identify unresolved methods as constructors by <init>.

Check was previously missing brackets in the method name.

Bug: 16560516

(cherry picked from commit 04ce113458b6b01f96d58dbb4c41682ee22d8b5d)

Change-Id: Ie98c5b84e73ffc43ea08e1b08ae9c6e41bcd5891

9 years agoMerge "ART: Implement the easy long division/remainder by a constant"
Andreas Gampe [Mon, 4 Aug 2014 19:36:53 +0000 (19:36 +0000)]
Merge "ART: Implement the easy long division/remainder by a constant"

9 years agoART: Implement the easy long division/remainder by a constant
Alexei Zavjalov [Mon, 14 Jul 2014 19:19:41 +0000 (02:19 +0700)]
ART: Implement the easy long division/remainder by a constant

Also optimizes long/int divisions by power-of-two values.

Also do some clean-up.

Change-Id: Ie414e64aac251c81361ae107d157c14439e6dab5
Signed-off-by: Alexei Zavjalov <alexei.zavjalov@intel.com>
9 years agoMerge "Move image classes options from art to AndroidRuntime [art]"
Brian Carlstrom [Mon, 4 Aug 2014 18:27:31 +0000 (18:27 +0000)]
Merge "Move image classes options from art to AndroidRuntime [art]"

9 years agoMerge "Move from dalvik.vm.image-dex2oat-flags to dalvik.vm.dex2oat-filter [art]"
Brian Carlstrom [Mon, 4 Aug 2014 18:27:30 +0000 (18:27 +0000)]
Merge "Move from dalvik.vm.image-dex2oat-flags to dalvik.vm.dex2oat-filter [art]"

9 years agoMerge "AArch64: Remove unnecessary work around for sp."
Andreas Gampe [Mon, 4 Aug 2014 17:46:05 +0000 (17:46 +0000)]
Merge "AArch64: Remove unnecessary work around for sp."

9 years agoAArch64: Remove unnecessary work around for sp.
Zheng Xu [Tue, 1 Jul 2014 01:53:22 +0000 (09:53 +0800)]
AArch64: Remove unnecessary work around for sp.

Use RegRegRegExtend to encode instruction for "sub/add sp, sp, Xm".

Change-Id: I13d3d2d386a7bd827e1396f291a7dcb9bffd5a29

9 years agoMerge "Inform about System.exit() call in logcat"
Narayan Kamath [Mon, 4 Aug 2014 17:31:49 +0000 (17:31 +0000)]
Merge "Inform about System.exit() call in logcat"

9 years agoInform about System.exit() call in logcat
Dmitry Petrochenko [Thu, 3 Jul 2014 10:07:45 +0000 (17:07 +0700)]
Inform about System.exit() call in logcat

This diagnostic message helps to filter out crashes during System.exit()
when thread's behaviour is undefined.

Change-Id: I772932a35b503b1efca695e7da7bf2e562c03a7c
Signed-off-by: Dmitry Petrochenko <dmitry.petrochenko@intel.com>
9 years agoMerge "ART: Rework ARM64 entry sequence"
Andreas Gampe [Mon, 4 Aug 2014 16:32:25 +0000 (16:32 +0000)]
Merge "ART: Rework ARM64 entry sequence"

9 years agoART: Rework ARM64 entry sequence
Andreas Gampe [Tue, 29 Jul 2014 07:35:00 +0000 (00:35 -0700)]
ART: Rework ARM64 entry sequence

Try to fold one sub of SP in the ARM64 entry sequence. When the
framesize is small, generate a sub over the full frame-size, and
adjust the spill offsets accordingly. If the framesize is too
large, use a pre-indexed store and fill upwards from there.

Change-Id: I1c15ac6276fb62b8164372de02fd92437f605938

9 years agoMerge "Fix broken JNI IsAssignableFrom."
Narayan Kamath [Mon, 4 Aug 2014 11:24:20 +0000 (11:24 +0000)]
Merge "Fix broken JNI IsAssignableFrom."

9 years agoFix broken JNI IsAssignableFrom.
Narayan Kamath [Fri, 11 Jul 2014 18:15:11 +0000 (19:15 +0100)]
Fix broken JNI IsAssignableFrom.

The expected argument order was the opposite of what the spec
dictated.

Reported-By: Mikhail Naganov <mnaganov@google.com>
bug: 16531674

Change-Id: I1ebc6d5814ebb4feeb96f3ecdf1a4602da4ba7ce

9 years agoMerge "Revert "Revert "Enable Load Store Elimination for ARM and ARM64"""
Bill Buzbee [Thu, 31 Jul 2014 21:02:56 +0000 (21:02 +0000)]
Merge "Revert "Revert "Enable Load Store Elimination for ARM and ARM64"""

9 years agoX86: Assembler: Correct r8_form for some cases
Mark Mendell [Thu, 31 Jul 2014 17:20:21 +0000 (13:20 -0400)]
X86: Assembler: Correct r8_form for some cases

Set r8_form to false for instruction formats that don't reference
registers.

Change-Id: Ib01edef4ef7f22de25a31dc4207889bff97d163d
Signed-off-by: Mark Mendell <mark.p.mendell@intel.com>
9 years agoMove image classes options from art to AndroidRuntime [art]
Brian Carlstrom [Wed, 30 Jul 2014 22:13:25 +0000 (15:13 -0700)]
Move image classes options from art to AndroidRuntime [art]

Bug: 15165413
Change-Id: I3dfa702ac29279cf76d4cc451138c03a5158ac3a

9 years agoMove from dalvik.vm.image-dex2oat-flags to dalvik.vm.dex2oat-filter [art]
Brian Carlstrom [Tue, 29 Jul 2014 02:13:27 +0000 (19:13 -0700)]
Move from dalvik.vm.image-dex2oat-flags to dalvik.vm.dex2oat-filter [art]

This will allow us to conditionally change the compiler-filter based on other properties.

Bug: 15165413
Change-Id: I6877a6eab221ab142909b595fd28191b5d3a095b

9 years agoART: Correct disassembling of 64bit immediates on x86_64
Vladimir Kostyukov [Wed, 23 Jul 2014 05:10:07 +0000 (12:10 +0700)]
ART: Correct disassembling of 64bit immediates on x86_64

The patch fixes an issue with disassembling 'movsxd' and 'movabsq'
instructions altered with 64bit immediates: not only a REX.W prefix
may be prepended to these instructions.

Change-Id: Ida7c7b368327a6b5cae1ff12ec00ceb0769c0a3d
Signed-off-by: Vladimir Kostyukov <vladimir.kostyukov@intel.com>
9 years agoMerge "Fix image not regenerating on changes to bootclasspath"
Narayan Kamath [Thu, 31 Jul 2014 16:56:26 +0000 (16:56 +0000)]
Merge "Fix image not regenerating on changes to bootclasspath"

9 years agoFix image not regenerating on changes to bootclasspath
Alex Light [Thu, 31 Jul 2014 16:32:13 +0000 (09:32 -0700)]
Fix image not regenerating on changes to bootclasspath

Bug: 16367410
Change-Id: I10e81d068ffc2ab712a287f88ef67f6a1be4df4c

9 years agoMerge "ART: LoadConstWide should clobber temp reg"
Bill Buzbee [Thu, 31 Jul 2014 14:08:33 +0000 (14:08 +0000)]
Merge "ART: LoadConstWide should clobber temp reg"

9 years agoART: LoadConstWide should clobber temp reg
Alexei Zavjalov [Wed, 30 Jul 2014 12:31:04 +0000 (19:31 +0700)]
ART: LoadConstWide should clobber temp reg

If we have 2+ LoadConstWide(FP) calls in one method it is possible
that LoadConstWide will load the method poiner only once. In some
cases, for example, if we have branches, initialization might not be
done and it may lead to a segmentation fault.

Change-Id: If45fc2d1109d7ce9bd272f5c56446b2a6884daac
Signed-off-by: Alexei Zavjalov <alexei.zavjalov@intel.com>
9 years agoMerge "Add read barriers for the roots in the verifier."
Hiroshi Yamauchi [Thu, 31 Jul 2014 18:32:43 +0000 (18:32 +0000)]
Merge "Add read barriers for the roots in the verifier."

9 years agoAdd read barriers for the roots in the verifier.
Hiroshi Yamauchi [Wed, 30 Jul 2014 21:26:22 +0000 (14:26 -0700)]
Add read barriers for the roots in the verifier.

Note: Because the roots (the class references in RegType objects) can
be updated by the read barriers, a lot of uses of type "const RegType"
were replaced with "RegType".

Bug: 12687968
Change-Id: I6cf37a87f352938d43fb51560a8d927ada104f50

9 years agoMerge "Make ImageSpace skip validation if we are using a relocated image"
Narayan Kamath [Thu, 31 Jul 2014 18:06:57 +0000 (18:06 +0000)]
Merge "Make ImageSpace skip validation if we are using a relocated image"

9 years agoMake ImageSpace skip validation if we are using a relocated image
Alex Light [Thu, 31 Jul 2014 17:48:46 +0000 (10:48 -0700)]
Make ImageSpace skip validation if we are using a relocated image

Change-Id: I0682d1870f5b3e27adb9c9de803fbb7d364ad00d

9 years agoMerge "Make 32bit run-tests work correctly on 64bit default devices."
Brian Carlstrom [Wed, 30 Jul 2014 17:41:27 +0000 (17:41 +0000)]
Merge "Make 32bit run-tests work correctly on 64bit default devices."

9 years agoMake 32bit run-tests work correctly on 64bit default devices.
Alex Light [Wed, 30 Jul 2014 01:31:18 +0000 (18:31 -0700)]
Make 32bit run-tests work correctly on 64bit default devices.

Change-Id: I7a5ca303ab91dec0df95358435e4a56c18a7fce3

9 years agoMerge "Reduce time and memory usage of GVN."
Vladimir Marko [Thu, 31 Jul 2014 09:31:40 +0000 (09:31 +0000)]
Merge "Reduce time and memory usage of GVN."

9 years agoReduce time and memory usage of GVN.
Vladimir Marko [Tue, 29 Jul 2014 11:04:10 +0000 (12:04 +0100)]
Reduce time and memory usage of GVN.

Filter out dead sregs in GVN. Reclaim memory after each LVN
in the GVN modification phase.

Bug: 16398693
Change-Id: I8c88c3009663754e1b66c0ef3f62c3b93276e385

9 years agoMerge "ART: Build fix"
Andreas Gampe [Thu, 31 Jul 2014 06:24:06 +0000 (06:24 +0000)]
Merge "ART: Build fix"

9 years agoART: Build fix
Andreas Gampe [Thu, 31 Jul 2014 06:21:36 +0000 (23:21 -0700)]
ART: Build fix

Make lint happy, as comments should be separated from code.

Change-Id: I4bfd88357302be9a6a104f1152e3b1fda386371e

9 years agoMerge "1. Fix CFI for quick compiled code in x86 & x86_64; 2. Emit CFI in .eh_frame...
Ian Rogers [Thu, 31 Jul 2014 04:21:44 +0000 (04:21 +0000)]
Merge "1. Fix CFI for quick compiled code in x86 & x86_64; 2. Emit CFI in .eh_frame instead of .debug_frame."

9 years ago1. Fix CFI for quick compiled code in x86 & x86_64;
Tong Shen [Wed, 30 Jul 2014 16:31:22 +0000 (09:31 -0700)]
1. Fix CFI for quick compiled code in x86 & x86_64;
2. Emit CFI in .eh_frame instead of .debug_frame.

With CFI, we can correctly unwind past quick generated code.
Now gdb should unwind to main() for both x86 & x86_64 host-side ART.

Note that it does not work with relocation yet.

Testing:
1. art/test/run-test --host --gdb [--64] --no-relocate 005
2. In gdb, run 'b art_quick_invoke_stub', then 'r', then 'c' a few times
3. In gdb, run 'bt'. You should see stack frames down to main()

Change-Id: I5350d4097dc3d360a60cb17c94f1d02b99bc58bb

9 years agoMerge "Add --always-clean flag to run-test."
Ian Rogers [Thu, 31 Jul 2014 03:56:05 +0000 (03:56 +0000)]
Merge "Add --always-clean flag to run-test."

9 years agoMerge "ART: Fix verifier mishandling erroneous array component types"
Andreas Gampe [Thu, 31 Jul 2014 03:39:34 +0000 (03:39 +0000)]
Merge "ART: Fix verifier mishandling erroneous array component types"

9 years agoART: Fix verifier mishandling erroneous array component types
Andreas Gampe [Thu, 31 Jul 2014 01:59:05 +0000 (18:59 -0700)]
ART: Fix verifier mishandling erroneous array component types

The verifier must not assume that component types are not erroneous.

Bug: 16661259
Change-Id: I23b2f517259ca9c0b8a1aa38f6348fcd61e0b22e

9 years agoMerge "Add support for bump pointer spaces to DdmSendHeapSegments."
Mathieu Chartier [Thu, 31 Jul 2014 00:15:37 +0000 (00:15 +0000)]
Merge "Add support for bump pointer spaces to DdmSendHeapSegments."

9 years agoAdd support for bump pointer spaces to DdmSendHeapSegments.
Mathieu Chartier [Wed, 30 Jul 2014 21:59:56 +0000 (14:59 -0700)]
Add support for bump pointer spaces to DdmSendHeapSegments.

Added support for bump pointer spaces as well as
differentiating between DlMallocSpaces and RosAllocSpaces.
Added logic to reset the start of next chunk to prevent
inbetween space regions counted as free.

Fixed a debug build bug where we were doing an mprotect after
creating a rosalloc space. In debug builds, this writes a magic
value in the page. This was faulting since it was set to PROT_NONE.
The fix moves the mprotect before the RosAlloc space creation.

Bug: 16408851
Change-Id: Ibe18198d05ff353fbb57d10b2f7719fdcbf1c5f0

9 years agoAdd --always-clean flag to run-test.
Alex Light [Wed, 30 Jul 2014 16:41:21 +0000 (09:41 -0700)]
Add --always-clean flag to run-test.

This flag makes run-test remove the test-artifacts even if it fails. Also
enable this option by default when doing run-tests with make. Add
a ART_TEST_RUN_TEST_ALWAYS_CLEAN environment variable to control this
option.

Change-Id: I7867b400d570d8d679d9822d1fa65f49eb3522ae

9 years agoRevert "Revert "Enable Load Store Elimination for ARM and ARM64""
Serban Constantinescu [Tue, 15 Jul 2014 16:44:21 +0000 (17:44 +0100)]
Revert "Revert "Enable Load Store Elimination for ARM and ARM64""

This patch refactors the implementation of the LoadStoreElimination
optimisation pass. Please note that this pass was disabled and not
functional for any of the backends.

The current implementation tracks aliases and handles DalvikRegs as well
as Heap memory regions. It has been tested and it is known to optimise
out the following:
  * Load - Load
  * Store - Load
  * Store - Store
  * Load Literals

Change-Id: I3aadb12a787164146a95bc314e85fa73ad91e12b

9 years agoMerge "ART: Check for duplicate class definitions in dex file verifier"
Andreas Gampe [Tue, 29 Jul 2014 23:07:05 +0000 (23:07 +0000)]
Merge "ART: Check for duplicate class definitions in dex file verifier"

9 years agoART: Check for duplicate class definitions in dex file verifier
Andreas Gampe [Tue, 29 Jul 2014 08:22:07 +0000 (01:22 -0700)]
ART: Check for duplicate class definitions in dex file verifier

Reject dex files that try to define two classes with the same type id.

Bug: 16016492
Change-Id: I82cc99c00b3e3cb2b3d537ec14beed9049fe900f

9 years agoMerge "Stop thread from reattaching during runtime shutdown while tracing."
Jeff Hao [Tue, 29 Jul 2014 22:06:32 +0000 (22:06 +0000)]
Merge "Stop thread from reattaching during runtime shutdown while tracing."

9 years agoStop thread from reattaching during runtime shutdown while tracing.
Jeff Hao [Tue, 29 Jul 2014 01:35:38 +0000 (18:35 -0700)]
Stop thread from reattaching during runtime shutdown while tracing.

Bug: 16024763
Change-Id: Iad5ba180241ff74b15baf5c3a15ed2d2ed60fcf0

9 years agoMerge "Add GcRoot to clean up and enforce read barriers."
Hiroshi Yamauchi [Tue, 29 Jul 2014 22:00:56 +0000 (22:00 +0000)]
Merge "Add GcRoot to clean up and enforce read barriers."

9 years agoAdd GcRoot to clean up and enforce read barriers.
Hiroshi Yamauchi [Wed, 23 Jul 2014 01:08:23 +0000 (18:08 -0700)]
Add GcRoot to clean up and enforce read barriers.

Introduce a value-type wrapper around Object* for GC roots so that 1)
we won't have to directly add the read barrier code in many places and
2) we can avoid accidentally bypassing/missing read barriers on GC
roots (the GcRoot interface ensures that the read barrier is executed
on a read).

The jdwp test passed.

Bug: 12687968
Change-Id: Ib167c7c325b3c7e3900133578815f04d219972a1

9 years agoMerge "ART: Remove bogus dcheck"
Andreas Gampe [Wed, 30 Jul 2014 15:25:03 +0000 (15:25 +0000)]
Merge "ART: Remove bogus dcheck"

9 years agoART: Remove bogus dcheck
Andreas Gampe [Wed, 30 Jul 2014 06:03:55 +0000 (23:03 -0700)]
ART: Remove bogus dcheck

Mea culpa.

Change-Id: I6fd3b678e30354c5ab5ec6da5b3619c26d486563

9 years agoMerge "ART: Rework quick entrypoint code in Mir2Lir, cleanup"
Andreas Gampe [Mon, 28 Jul 2014 16:41:14 +0000 (16:41 +0000)]
Merge "ART: Rework quick entrypoint code in Mir2Lir, cleanup"

9 years agoART: Rework quick entrypoint code in Mir2Lir, cleanup
Andreas Gampe [Mon, 28 Jul 2014 02:44:50 +0000 (19:44 -0700)]
ART: Rework quick entrypoint code in Mir2Lir, cleanup

To reduce the complexity of calling trampolines in generic code,
introduce an enumeration for entrypoints. Introduce a header that lists
the entrypoint enum and exposes a templatized method that translates an
enum value to the corresponding thread offset value.

Call helpers are rewritten to have an enum parameter instead of the
thread offset. Also rewrite LoadHelper and GenConversionCall this way.
It is now LoadHelper's duty to select the right thread offset size.

Introduce InvokeTrampoline virtual method to Mir2Lir. This allows to
further simplify the call helpers, as well as make OpThreadMem specific
to X86 only (removed from Mir2Lir).

Make GenInlinedCharAt virtual, move a copy to X86 backend, and simplify
both copies. Remove LoadBaseIndexedDisp and OpRegMem from Mir2Lir, as they
are now specific to X86 only.

Remove StoreBaseIndexedDisp from Mir2Lir, as it was only ever used in the
X86 backend.

Remove OpTlsCmp from Mir2Lir, as it was only ever used in the X86 backend.

Remove OpLea from Mir2Lir, as it was only ever defined in the X86 backend.

Remove GenImmedCheck from Mir2Lir as it was neither used nor implemented.

Change-Id: If0a6182288c5d57653e3979bf547840a4c47626e

9 years agoMerge "ART: Fix target test paths"
Andreas Gampe [Sun, 27 Jul 2014 21:56:42 +0000 (21:56 +0000)]
Merge "ART: Fix target test paths"