OSDN Git Service

android-x86/art.git
8 years agoWait for GC to finish in ThreadList::~ThreadList
Mathieu Chartier [Wed, 12 Aug 2015 23:40:32 +0000 (16:40 -0700)]
Wait for GC to finish in ThreadList::~ThreadList

Added a DisableGC to the heap instead of using the runtime shutdown
boolean. The runtime shutting down boolean is set to true before the
non daemon threads have exited. These threads may still be doing
allocations and we don't want to throw OOME if unnecessary.

Bug: 18577101
Change-Id: Iceb7048e6bd799aa2716099459c54f8dc0fb8feb

8 years agoMerge "Visit class roots from ClassLoader::VisitReferences"
Mathieu Chartier [Thu, 13 Aug 2015 01:01:38 +0000 (01:01 +0000)]
Merge "Visit class roots from ClassLoader::VisitReferences"

8 years agoMerge "ART: Add correct -inl file"
Andreas Gampe [Wed, 12 Aug 2015 23:53:44 +0000 (23:53 +0000)]
Merge "ART: Add correct -inl file"

8 years agoART: Add correct -inl file
Andreas Gampe [Wed, 12 Aug 2015 23:42:06 +0000 (16:42 -0700)]
ART: Add correct -inl file

Follow-up to f10b6e109bfb595b6752d1b59db680694ac1684d.

Bug: 20843113
Change-Id: Idfb9479c2cd242c2fbf748177177ae1cba0508d6

8 years agoMerge "ART: Change UninitializedThis tracking in the verifier"
Andreas Gampe [Wed, 12 Aug 2015 23:28:54 +0000 (23:28 +0000)]
Merge "ART: Change UninitializedThis tracking in the verifier"

8 years agoART: Change UninitializedThis tracking in the verifier
Andreas Gampe [Wed, 12 Aug 2015 17:48:12 +0000 (10:48 -0700)]
ART: Change UninitializedThis tracking in the verifier

Only relying on register types is error-prone. For example, we may
inadvertently reject correct code when the constructor terminates
abnormally.

Bug: 20843113
Change-Id: I8826cd167780df25a6166740f183d216483fa550

8 years agoVisit class roots from ClassLoader::VisitReferences
Mathieu Chartier [Thu, 6 Aug 2015 22:34:15 +0000 (15:34 -0700)]
Visit class roots from ClassLoader::VisitReferences

This causes the classes of a class loader to get marked when that
class loader gets marked instead of during class root visiting.

Bug: 22720414

Change-Id: If53f042aff1d9f7bf94ecbe6886601edda029b7d

8 years agoMerge "Don't allow GC to start if the runtime is shutting down"
Mathieu Chartier [Wed, 12 Aug 2015 22:02:02 +0000 (22:02 +0000)]
Merge "Don't allow GC to start if the runtime is shutting down"

8 years agoDon't allow GC to start if the runtime is shutting down
Mathieu Chartier [Wed, 12 Aug 2015 20:53:06 +0000 (13:53 -0700)]
Don't allow GC to start if the runtime is shutting down

Daemon threads may be trying to do allocations as the runtime is
shutting down, don't allow GC to start since it touches the heap
and may cause crashes if it is running while we delete the heap
or thread list.

Bug: 18577101
Change-Id: I1404b41f2aee13d3c26acffdf588c985628beefc

8 years agoMerge "Delete AllocationTimer"
Mathieu Chartier [Wed, 12 Aug 2015 20:22:34 +0000 (20:22 +0000)]
Merge "Delete AllocationTimer"

8 years agoDelete AllocationTimer
Mathieu Chartier [Wed, 12 Aug 2015 17:20:27 +0000 (10:20 -0700)]
Delete AllocationTimer

Not very useful, running benchmarks is better.

Bug: 19534862
Change-Id: I725830274e306621bffdabffdee602a03daf016f

8 years agoMerge "Add a GVN dependency 'GC' for garbage collection."
Roland Levillain [Wed, 12 Aug 2015 17:46:35 +0000 (17:46 +0000)]
Merge "Add a GVN dependency 'GC' for garbage collection."

8 years agoAdd a GVN dependency 'GC' for garbage collection.
Alexandre Rames [Wed, 12 Aug 2015 12:43:29 +0000 (13:43 +0100)]
Add a GVN dependency 'GC' for garbage collection.

This will be used by incoming architecture specific optimizations. The
dependencies must be conservative. When an HInstruction is created we
may not be sure whether it can trigger GC. In that case the
'ChangesGC' dependency must be set. We control at code-generation time
that HInstructions that can call have the 'ChangesGC' dependency
set.

Change-Id: Iea6a7f430009f37a9599b0a0039207049906e45d

8 years agoMerge "[optimizing] More x86_64 code improvements"
Roland Levillain [Wed, 12 Aug 2015 11:43:20 +0000 (11:43 +0000)]
Merge "[optimizing] More x86_64 code improvements"

8 years agoMerge "Some heap cleanup"
Mathieu Chartier [Wed, 12 Aug 2015 03:22:25 +0000 (03:22 +0000)]
Merge "Some heap cleanup"

8 years agoSome heap cleanup
Mathieu Chartier [Wed, 12 Aug 2015 00:35:25 +0000 (17:35 -0700)]
Some heap cleanup

Bug: 19534862
Change-Id: Ia63f489d26ec8813a263ce877bdbbc8c4e8fe5f4

8 years agoMerge "Add daemon threads to 004-ThreadStress"
Mathieu Chartier [Wed, 12 Aug 2015 00:12:18 +0000 (00:12 +0000)]
Merge "Add daemon threads to 004-ThreadStress"

8 years agoMerge "[MIPS] Avoid using odd numbered fp registers"
Roland Levillain [Tue, 11 Aug 2015 22:34:21 +0000 (22:34 +0000)]
Merge "[MIPS] Avoid using odd numbered fp registers"

8 years agoMerge "Address some comments"
Mathieu Chartier [Tue, 11 Aug 2015 21:13:37 +0000 (21:13 +0000)]
Merge "Address some comments"

8 years agoAddress some comments
Mathieu Chartier [Tue, 11 Aug 2015 20:01:23 +0000 (13:01 -0700)]
Address some comments

Change-Id: I0262304cc720a0e93015955d0a7fb05dfebe213e

8 years agoMerge "Revert "Revert "Pass the verified method to the inlinee DexCompilationUnit."""
Nicolas Geoffray [Tue, 11 Aug 2015 17:42:21 +0000 (17:42 +0000)]
Merge "Revert "Revert "Pass the verified method to the inlinee DexCompilationUnit."""

8 years ago[optimizing] More x86_64 code improvements
Mark Mendell [Mon, 25 May 2015 20:02:44 +0000 (16:02 -0400)]
[optimizing] More x86_64 code improvements

Use the constant area some more, use 32-bit immediates in movq
instructions when possible, and other small tweaks.

Remove the commented out code for Math.Abs(float/double) as it would
fail for baseline compiler due to the output being the same as the
input.

Change-Id: Ifa39f1865b94cec2e1c0a99af3066a645e9d3618
Signed-off-by: Mark Mendell <mark.p.mendell@intel.com>
8 years agoMerge "ART: Fix Quick's DCE+GVN"
Vladimir Marko [Tue, 11 Aug 2015 14:28:04 +0000 (14:28 +0000)]
Merge "ART: Fix Quick's DCE+GVN"

8 years agoMerge "ART-Optimizing: Fix the type of HDivZeroCheck"
Calin Juravle [Tue, 11 Aug 2015 14:25:45 +0000 (14:25 +0000)]
Merge "ART-Optimizing: Fix the type of HDivZeroCheck"

8 years agoART: Fix Quick's DCE+GVN
Serguei Katkov [Mon, 10 Aug 2015 06:59:02 +0000 (12:59 +0600)]
ART: Fix Quick's DCE+GVN

DCE_GVN does not take into account the following case:
  mov a, b
  ...
  mov c, b
when optimization tries to replace a with c it must ensure that
for all uses of a there is no new definition of c before use.
Otherwise that use will incorrectly substituted with new c instead
of original b.

Bug: 23102860
Change-Id: I48ee0bd4386419b0f9c814c21b3537a392518cd1
Signed-off-by: Serguei Katkov <serguei.i.katkov@intel.com>
8 years agoART-Optimizing: Fix the type of HDivZeroCheck
Serguei Katkov [Mon, 3 Aug 2015 07:55:33 +0000 (13:55 +0600)]
ART-Optimizing: Fix the type of HDivZeroCheck

HDivZeroCheck is created during the building CFG and at this moment
its type is not known completely. So it sets the type to int or long.
However, later SSA builder can insert the type conversion and type
of input of HDivZeroCheck can become byte or short while the type
of HDivZeroCheck remains the same.

In reality the type of HDivZeroCheck should be always equal to
its input parameter.
To fix this inconsistency we return the type of HDivZeroCheck as its
input type. Code generators are updated accordingly.

Change-Id: I6a5aedc8d479cfc6328704e7ddf252bca830076b
Signed-off-by: Serguei Katkov <serguei.i.katkov@intel.com>
8 years agoRevert "Revert "Pass the verified method to the inlinee DexCompilationUnit.""
Nicolas Geoffray [Tue, 11 Aug 2015 02:14:38 +0000 (02:14 +0000)]
Revert "Revert "Pass the verified method to the inlinee DexCompilationUnit.""

bug:22866717

This reverts commit f25b661b3708c865cdaa0a018372a42881353bea.

Change-Id: I95e8310e6c3d8e6edb736fdefcabf602cce3a4e0

8 years agoMerge "ART: Change UnresolvedMergedType merge"
Andreas Gampe [Tue, 11 Aug 2015 01:18:15 +0000 (01:18 +0000)]
Merge "ART: Change UnresolvedMergedType merge"

8 years agoART: Change UnresolvedMergedType merge
Andreas Gampe [Tue, 11 Aug 2015 00:15:32 +0000 (17:15 -0700)]
ART: Change UnresolvedMergedType merge

Change to use IsUnresolvedTypes.

Bug: 22881413

(cherry picked from commit 168387df8dcbae1b51f124eed84daf8e8de6a974)

Change-Id: Ic27b97b475b28a5bdf994ea5178767e0efb0e752

8 years agoAdd daemon threads to 004-ThreadStress
Man Cao [Mon, 10 Aug 2015 22:51:27 +0000 (15:51 -0700)]
Add daemon threads to 004-ThreadStress

Bug: 19537976
Change-Id: I129e311aa6f7ca4f49d64ec50c0e07857b12b6d8

8 years agoMerge "ART: Fix reg_type_test"
Andreas Gampe [Mon, 10 Aug 2015 18:46:55 +0000 (18:46 +0000)]
Merge "ART: Fix reg_type_test"

8 years agoART: Fix reg_type_test
Andreas Gampe [Mon, 10 Aug 2015 18:45:24 +0000 (11:45 -0700)]
ART: Fix reg_type_test

Follow-up to 750f7c2827318f6d07620f2ef0321218ea4d8670.

Bug: 22881413
Change-Id: Iaea4cc29ed680710d01532124bc472547ee18bdb

8 years agoMerge "ART: Add complex UnresolvedMergedType test case"
Andreas Gampe [Mon, 10 Aug 2015 18:26:59 +0000 (18:26 +0000)]
Merge "ART: Add complex UnresolvedMergedType test case"

8 years agoMerge "ART: Change UnresolvedMergedType internal representation"
Andreas Gampe [Mon, 10 Aug 2015 18:26:46 +0000 (18:26 +0000)]
Merge "ART: Change UnresolvedMergedType internal representation"

8 years agoMerge "Do read barriers on native roots in CopyClassVisitor"
Mathieu Chartier [Mon, 10 Aug 2015 18:16:03 +0000 (18:16 +0000)]
Merge "Do read barriers on native roots in CopyClassVisitor"

8 years agoDo read barriers on native roots in CopyClassVisitor
Mathieu Chartier [Mon, 10 Aug 2015 17:13:02 +0000 (10:13 -0700)]
Do read barriers on native roots in CopyClassVisitor

Fixes a race condition caused by making a copy of a class then
clearing the native root arrays of the original class. If the
original class was gray and the new class was black then the GC
would miss the native roots.

Bug: 22957957

Change-Id: I706110018220af12a2ad0d72eb803f1cfe3580b9

8 years agoART: Change UnresolvedMergedType internal representation
Andreas Gampe [Thu, 6 Aug 2015 18:31:56 +0000 (11:31 -0700)]
ART: Change UnresolvedMergedType internal representation

Use a BitVector instead of the tree representation. This
avoids flattening the components and other instances.

Bug: 22881413
Change-Id: Ibf7cfb54443affeb1753bf114c0f306125391c62

8 years agoART: Add complex UnresolvedMergedType test case
Andreas Gampe [Fri, 7 Aug 2015 15:07:16 +0000 (08:07 -0700)]
ART: Add complex UnresolvedMergedType test case

Add a test case that stresses the verifier. Add dex2oat timeout
to the host prebuild configuration.

Bug: 22881413
Change-Id: Ic4a0d16a4dc69f3d43bac40a560b6e6884df263f

8 years agoMerge "Fix a bug in the register allocator around pair allocation."
Nicolas Geoffray [Mon, 10 Aug 2015 15:07:43 +0000 (15:07 +0000)]
Merge "Fix a bug in the register allocator around pair allocation."

8 years ago[MIPS] Avoid using odd numbered fp registers
Goran Jakovljevic [Tue, 28 Jul 2015 12:26:12 +0000 (14:26 +0200)]
[MIPS] Avoid using odd numbered fp registers

Don't use odd numbered fp registers for single precision data on
MIPS32r6 (64-bit FPU).

Bug: 23050326
Change-Id: I35cc19df091149773411e2336b01c170929376bc

8 years agoFix a bug in the register allocator around pair allocation.
Nicolas Geoffray [Sat, 8 Aug 2015 01:13:03 +0000 (18:13 -0700)]
Fix a bug in the register allocator around pair allocation.

We may get hints that do not work with the current implementation
of register pairs, which forces the allocation of (low + 1)
for the high register. For example, if the hint is EBX, we will
allocate ESP for the high register.

bug:23043730

Change-Id: I371ebb0c61568f09d12eb9ab815c0bf0ea02d49b

8 years agoMerge "ART: Clean up unnecessary ArtMethod**"
Andreas Gampe [Sat, 8 Aug 2015 00:23:51 +0000 (00:23 +0000)]
Merge "ART: Clean up unnecessary ArtMethod**"

8 years agoART: Clean up unnecessary ArtMethod**
Andreas Gampe [Sat, 8 Aug 2015 00:20:11 +0000 (17:20 -0700)]
ART: Clean up unnecessary ArtMethod**

ArtMethods are no longer Java objects, so the additional indirection
is no longer necessary here.

Change-Id: If76756d875b418b3f6e83f51b3225a158e9ce29b

8 years agoMerge "ART: Remove TODO in BitVector"
Andreas Gampe [Sat, 8 Aug 2015 00:14:42 +0000 (00:14 +0000)]
Merge "ART: Remove TODO in BitVector"

8 years agoART: Remove TODO in BitVector
Andreas Gampe [Fri, 7 Aug 2015 15:29:13 +0000 (08:29 -0700)]
ART: Remove TODO in BitVector

Refactor the BitVector constructor: split it up to remove the
possibility to provide contradicting parameters, and add a custom
copying constructor.

Change-Id: Ie943f279baa007db578aea0f2f33fa93311612ee

8 years agoMerge "Add class table field to class loader"
Mathieu Chartier [Fri, 7 Aug 2015 23:33:43 +0000 (23:33 +0000)]
Merge "Add class table field to class loader"

8 years agoMerge "Fix C++14 bitrot."
Dan Albert [Fri, 7 Aug 2015 23:21:25 +0000 (23:21 +0000)]
Merge "Fix C++14 bitrot."

8 years agoFix C++14 bitrot.
Dan Albert [Fri, 7 Aug 2015 23:19:01 +0000 (16:19 -0700)]
Fix C++14 bitrot.

Change-Id: I22f7925a87bca45646c563971342a9a1b2090ba6

8 years agoAdd class table field to class loader
Mathieu Chartier [Wed, 5 Aug 2015 22:08:12 +0000 (15:08 -0700)]
Add class table field to class loader

Fixes bug with the class table where the comparator would cause read
barriers and break the map strict ordering properties.

Bug: 22957957

Change-Id: I8dbc042db6e22e2172ab4ec58ddf1db0345dcaaa

8 years agoMerge "Ensure coherency of call kinds for LocationSummary."
Nicolas Geoffray [Fri, 7 Aug 2015 16:04:33 +0000 (16:04 +0000)]
Merge "Ensure coherency of call kinds for LocationSummary."

8 years agoMerge "Revert "Pass the verified method to the inlinee DexCompilationUnit.""
Nicolas Geoffray [Fri, 7 Aug 2015 15:31:20 +0000 (15:31 +0000)]
Merge "Revert "Pass the verified method to the inlinee DexCompilationUnit.""

8 years agoRevert "Pass the verified method to the inlinee DexCompilationUnit."
Nicolas Geoffray [Fri, 7 Aug 2015 15:31:10 +0000 (15:31 +0000)]
Revert "Pass the verified method to the inlinee DexCompilationUnit."

Crashes testing libcore on host/x86.

Bug:22866717

This reverts commit 08624c50f488ecd845e48683e84726ff8c53e6a7.

Change-Id: Ia8f130d9ea32e171e554bd3e10c44f8453111b52

8 years agoMerge "Pass the verified method to the inlinee DexCompilationUnit."
Nicolas Geoffray [Fri, 7 Aug 2015 14:47:18 +0000 (14:47 +0000)]
Merge "Pass the verified method to the inlinee DexCompilationUnit."

8 years agoMerge "Tighten default inlining settings when using the space filter."
Roland Levillain [Fri, 7 Aug 2015 10:50:06 +0000 (10:50 +0000)]
Merge "Tighten default inlining settings when using the space filter."

8 years agoTighten default inlining settings when using the space filter.
Roland Levillain [Fri, 7 Aug 2015 10:38:32 +0000 (11:38 +0100)]
Tighten default inlining settings when using the space filter.

Bug: 21868508
Change-Id: Ic83813a966cef18e59447083926bf033aa587154

8 years agoEnsure coherency of call kinds for LocationSummary.
Alexandre Rames [Fri, 7 Aug 2015 09:26:17 +0000 (10:26 +0100)]
Ensure coherency of call kinds for LocationSummary.

The coherency is enforced with checks added in the `InvokeRuntime`
helper, that we now also use on x86 and x86_64.

Change-Id: I8cb92b042f25dc3c5fd390e9c61a45b477d081f4

8 years agoMerge "Introduce arch-specific checker tests."
David Brazdil [Fri, 7 Aug 2015 07:23:23 +0000 (07:23 +0000)]
Merge "Introduce arch-specific checker tests."

8 years agoPass the verified method to the inlinee DexCompilationUnit.
Nicolas Geoffray [Fri, 7 Aug 2015 04:10:30 +0000 (21:10 -0700)]
Pass the verified method to the inlinee DexCompilationUnit.

Sharpening and other optimizations in compiler driver rely
on having it there.

Also fixes a warning for the intrinsics code.

bug:22866717
Change-Id: I6ac4f55e9e77452a4ea9e0d0dd0ad70124a95a55

8 years agoMerge "ART: Fix the simplifier for add/sub"
Calin Juravle [Thu, 6 Aug 2015 17:49:55 +0000 (17:49 +0000)]
Merge "ART: Fix the simplifier for add/sub"

8 years agoART: Fix the simplifier for add/sub
Serguei Katkov [Wed, 5 Aug 2015 11:03:30 +0000 (17:03 +0600)]
ART: Fix the simplifier for add/sub

Instruction simplifier for add/sub should not proceed with floats
because that might cause the incorrect behavior with signed zero.

Change-Id: If0c9bf3931bcbf96b0814f8605a86997aea37145
Signed-off-by: Serguei Katkov <serguei.i.katkov@intel.com>
8 years agoMerge "Fix image loading in interpeter-only mode"
Sebastien Hertz [Thu, 6 Aug 2015 14:46:41 +0000 (14:46 +0000)]
Merge "Fix image loading in interpeter-only mode"

8 years agoMerge "Fix DDM reply buffer deallocation"
Sebastien Hertz [Thu, 6 Aug 2015 14:43:58 +0000 (14:43 +0000)]
Merge "Fix DDM reply buffer deallocation"

8 years agoIntroduce arch-specific checker tests.
Alexandre Rames [Thu, 6 Aug 2015 13:49:28 +0000 (14:49 +0100)]
Introduce arch-specific checker tests.

- The '.cfg' output is now created on target.
- Arch-specific checker tests can be created by inserting a
  suffix. For example:
      /// CHECK-START-ARM64: int Main.foo(int) register (after)
      /// CHECK-DAG:   <<Arg:i\d+>>     ParameterValue

Change-Id: I55cdb37f8e806c7ffdde6b676c8f44ac30b59051

8 years agoFix image loading in interpeter-only mode
Sebastien Hertz [Thu, 6 Aug 2015 10:52:43 +0000 (12:52 +0200)]
Fix image loading in interpeter-only mode

Follow-up https://android-review.googlesource.com/162935. We now need
an ArtMethodVisitor to update entrypoints of methods in the image.

Bug: 22832610
Change-Id: I30ca9c369a73c3372694b446e73afa2e37890a65

8 years agoFix DDM reply buffer deallocation
Sebastien Hertz [Thu, 6 Aug 2015 09:37:02 +0000 (11:37 +0200)]
Fix DDM reply buffer deallocation

Reply buffer is an uint8_t* allocated with new uint8_t[...] so it
must be deallocated with delete[], not free.

Change-Id: Ib5d6387aa63f1e0393113f82602f661bbada4aa2

8 years agoMerge "Move ArtFields and ArtMethods to be a length prefixed array"
Mathieu Chartier [Thu, 6 Aug 2015 03:12:39 +0000 (03:12 +0000)]
Merge "Move ArtFields and ArtMethods to be a length prefixed array"

8 years agoMove ArtFields and ArtMethods to be a length prefixed array
Mathieu Chartier [Thu, 30 Jul 2015 23:20:06 +0000 (16:20 -0700)]
Move ArtFields and ArtMethods to be a length prefixed array

Fixes race conditions between changing method and fields arrays
being seen in the wrong order by the GC.

Bug: 22832610
Change-Id: Ia21d6698f73ba207a6392c3d6b9be2658933073f

8 years agoMerge "ART: Extend ISA features option for (host) run tests"
Andreas Gampe [Wed, 5 Aug 2015 23:06:02 +0000 (23:06 +0000)]
Merge "ART: Extend ISA features option for (host) run tests"

8 years agoART: Extend ISA features option for (host) run tests
Andreas Gampe [Wed, 5 Aug 2015 16:24:45 +0000 (09:24 -0700)]
ART: Extend ISA features option for (host) run tests

To allow to test advanced intrinsics only available on (fairly)
modern processors, and because our build requirements are still
too low to enable them in the build system, pipe through the
existing make variables to the run-tests to allow to override
the dex2oat autodetection scheme.

Usage (standalone run-test): use the new --instruction-set-features
command line parameter to set the value for dex2oat (in the prebuild
configuration) or as an -Xcompiler-option (for no-prebuild). Note:
this is general and also works for target tests.

Usage (through make): set DEX2OAT_HOST_INSTRUCTION_SET_FEATURES and
2ND_DEX2OAT_HOST_INSTRUCTION_SET_FEATURES to set instruction set
features, as was possible before for the compilation of core files.
This option is now forwarded to host run-tests.

Change-Id: I9d89c82b6a5c8853748e5d1c153dde5d4050757a

8 years agoMerge "ART: Wire up DexToDexCompiler without extern"
Andreas Gampe [Wed, 5 Aug 2015 18:59:25 +0000 (18:59 +0000)]
Merge "ART: Wire up DexToDexCompiler without extern"

8 years agoART: Wire up DexToDexCompiler without extern
Andreas Gampe [Thu, 23 Jul 2015 08:19:26 +0000 (01:19 -0700)]
ART: Wire up DexToDexCompiler without extern

The extern declaration actually had the wrong signature, a parameter
was missing, and we never noticed.

Remove the function field and extern. Add a header for the compiler.
Use the header file.

Change-Id: Ia4d4dd86211da6045709a45c7bf8430471d1b62b

8 years agoMerge "Revert "cleanup: Replace pointers with out-parameters and fix-up formatting""
Igor Murashkin [Wed, 5 Aug 2015 18:18:59 +0000 (18:18 +0000)]
Merge "Revert "cleanup: Replace pointers with out-parameters and fix-up formatting""

8 years agoMerge "Address some comments and clean up"
Mathieu Chartier [Wed, 5 Aug 2015 18:02:53 +0000 (18:02 +0000)]
Merge "Address some comments and clean up"

8 years agoMerge "Fix mod-union logic for native roots"
Mathieu Chartier [Wed, 5 Aug 2015 17:42:56 +0000 (17:42 +0000)]
Merge "Fix mod-union logic for native roots"

8 years agoFix mod-union logic for native roots
Mathieu Chartier [Wed, 5 Aug 2015 17:10:07 +0000 (10:10 -0700)]
Fix mod-union logic for native roots

Bug: 22949217
Change-Id: I9e4631870160d5a9412df1eec2a7b79c65668542

8 years agoMerge "Have constant folding be more flexible."
Roland Levillain [Wed, 5 Aug 2015 12:52:12 +0000 (12:52 +0000)]
Merge "Have constant folding be more flexible."

8 years agoMerge "Add constant folding for long unary operations in opt. compiler."
Roland Levillain [Wed, 5 Aug 2015 10:03:46 +0000 (10:03 +0000)]
Merge "Add constant folding for long unary operations in opt. compiler."

8 years agoHave constant folding be more flexible.
Roland Levillain [Wed, 5 Aug 2015 09:21:34 +0000 (10:21 +0100)]
Have constant folding be more flexible.

- Have Evaluate methods take as argument(s) and return value
  instances of HConstant (instead of built-in 32- or 64-bit
  integer values), to let the evaluated instruction choose
  the type of the statically evaluated node; for instance,
  art::HEqual::Evaluate shall return a HIntConstant
  node (as implementation of a Boolean constant) whatever
  the type of its inputs (a pair of HIntConstant or a pair
  of HLongConstant).
- Split the evaluation job from the operation logic: the
  former is addressed by Evaluate methods, while the latter
  is done by a generic Compute method.
- Adress valid BinOp(int, long) and BinOp(long, int) cases.
- Add a constructor to art::HIntConstant to build an integer
  constant from a `bool` value.

Change-Id: If84b6fe8406bb94ddb1aa8b02e36628dff526db3

8 years agoAdd constant folding for long unary operations in opt. compiler.
Roland Levillain [Thu, 11 Dec 2014 12:14:33 +0000 (12:14 +0000)]
Add constant folding for long unary operations in opt. compiler.

Add tests to exercise the constant folding of these
instructions.

Also, prevent Java methods from run-tests exercising the
code generation of these instruction from being inlined,
so that they continue to check the generated code (and
not the code produced by the constant folding pass).

Change-Id: I28efca7cdb5142ac2b6d158ba296fb9136d62481

8 years agoMerge "ART: Move exception clearing into own instruction"
David Brazdil [Wed, 5 Aug 2015 08:34:24 +0000 (08:34 +0000)]
Merge "ART: Move exception clearing into own instruction"

8 years agoMerge "Revert "Revert "ART: Use bionic TLS slot for thread-self"""
Andreas Gampe [Wed, 5 Aug 2015 02:55:46 +0000 (02:55 +0000)]
Merge "Revert "Revert "ART: Use bionic TLS slot for thread-self"""

8 years agoRevert "Revert "ART: Use bionic TLS slot for thread-self""
Andreas Gampe [Wed, 5 Aug 2015 01:08:53 +0000 (01:08 +0000)]
Revert "Revert "ART: Use bionic TLS slot for thread-self""

This reverts commit 50832b6804d15ea5ddfe99a2753a09111b5bfe37.

Fix clearing the TLS on detach.

Change-Id: I75ed5ccccb4f31800e5f893055e96067a9695258

8 years agoMerge "ART: x86_64 RoundDouble/Float intrinsics should initialize out value."
Nicolas Geoffray [Wed, 5 Aug 2015 02:11:01 +0000 (02:11 +0000)]
Merge "ART: x86_64 RoundDouble/Float intrinsics should initialize out value."

8 years agoAddress some comments and clean up
Mathieu Chartier [Tue, 4 Aug 2015 22:19:41 +0000 (15:19 -0700)]
Address some comments and clean up

Change-Id: I538cf204f1c89d5fc81f8fc5e5800fcf1cf87359

8 years agoMerge "Revert "ART: Use bionic TLS slot for thread-self""
Andreas Gampe [Wed, 5 Aug 2015 00:36:05 +0000 (00:36 +0000)]
Merge "Revert "ART: Use bionic TLS slot for thread-self""

8 years agoRevert "ART: Use bionic TLS slot for thread-self"
Andreas Gampe [Wed, 5 Aug 2015 00:31:02 +0000 (00:31 +0000)]
Revert "ART: Use bionic TLS slot for thread-self"

This reverts commit b81535875ff46d1630c76091fe969a9eb7fc4d16.

Breaking tests, need to investigate.

Change-Id: Iab8c4f6136e6cb9bd6cf3c3495aa33962ef890ee

8 years agoMerge "Parallel Move Resolver: Perform Stack/Stack first"
Nicolas Geoffray [Wed, 5 Aug 2015 00:10:01 +0000 (00:10 +0000)]
Merge "Parallel Move Resolver: Perform Stack/Stack first"

8 years agoMerge "ART: Use bionic TLS slot for thread-self"
Andreas Gampe [Tue, 4 Aug 2015 22:35:34 +0000 (22:35 +0000)]
Merge "ART: Use bionic TLS slot for thread-self"

8 years agoMerge "Update expectation."
Nicolas Geoffray [Tue, 4 Aug 2015 19:06:16 +0000 (19:06 +0000)]
Merge "Update expectation."

8 years agoRevert "cleanup: Replace pointers with out-parameters and fix-up formatting"
Igor Murashkin [Tue, 4 Aug 2015 18:18:43 +0000 (11:18 -0700)]
Revert "cleanup: Replace pointers with out-parameters and fix-up formatting"

This reverts commit a315f5c546b796f55f4872bb6efc15eb858d9639.

--

Revert "runtime: cleanup class_linker out-parameters and formatting"

This reverts commit bc1d78daa463572c5a770cdca858a3b51d8e1b7b.

--

Revert "base: replace raw pointers for out-parameters with safer out<T>"

This reverts commit fb326cffc679cab8eb873b9e44795706f023cb3c.

8 years agoUpdate expectation.
Nicolas Geoffray [Tue, 4 Aug 2015 17:16:43 +0000 (18:16 +0100)]
Update expectation.

Vogar wants # between class and method.

Change-Id: I3e4d54e2cc22bac6b73fe611fcbfcbe39095159c

8 years agoMerge "ART: Use Clang default integrated assembler."
Andreas Gampe [Tue, 4 Aug 2015 16:47:39 +0000 (16:47 +0000)]
Merge "ART: Use Clang default integrated assembler."

8 years agoART: Use Clang default integrated assembler.
Chih-Hung Hsieh [Mon, 27 Jul 2015 17:10:44 +0000 (10:10 -0700)]
ART: Use Clang default integrated assembler.

Issues with x86, arm32 and arm64 have been fixed.
Allow the use of the integrated assembler.

Bug: 22874917
Change-Id: Iacfffda29d57bfc931da568a9436c2e34cb38df6

8 years agoART: Move exception clearing into own instruction
David Brazdil [Tue, 4 Aug 2015 15:22:25 +0000 (16:22 +0100)]
ART: Move exception clearing into own instruction

Runtime delivers exceptions only to catch blocks which begin with a
MOVE_EXCEPTION instruction (in DEX). In that case, the catch block is
expected to clear the thread-local exception storage after having
read the exception reference.

This patch changes Optimizing to represent MOVE_EXCEPTION with two
instructions - HLoadException and HClearException - instead of one.
If the exception reference is not used, HLoadException can be safely
removed, saving a memory load without breaking the runtime behaviour.

Change-Id: Idad8a714467bf9d9d5fccefbc43c0bd8ae13ddba

8 years agoMerge "ART: Fix arm32 assembly for Clang"
Andreas Gampe [Tue, 4 Aug 2015 15:16:33 +0000 (15:16 +0000)]
Merge "ART: Fix arm32 assembly for Clang"

8 years agoART: Fix arm32 assembly for Clang
Chih-Hung Hsieh [Mon, 27 Jul 2015 17:10:44 +0000 (10:10 -0700)]
ART: Fix arm32 assembly for Clang

Clang ignores .thumb_func and requires an explicit
.thumb directive, as given by the ARM Unified Assembler
Language. Conditionally add .thumb to thumb2 functions.

Bug: 22762844
Change-Id: I08abdb5f4e0ba5135d4ed850ca3e284203f95a03

8 years agoART: x86_64 RoundDouble/Float intrinsics should initialize out value.
Pavel Vyssotski [Fri, 31 Jul 2015 07:03:17 +0000 (13:03 +0600)]
ART: x86_64 RoundDouble/Float intrinsics should initialize out value.

x86_64 RoundDouble intrinsic should initialize output register for the case of
"inPlusPointFive >= maxLong" as expected. The same for the RoundFloat intrinsic.
Fixed also the out register type in CreateSSE41FPToIntLocations provoking
a DCHECK failure.

Change-Id: I0a910682e2917214861683c678ffba8e0f4bfed8
Signed-off-by: Pavel Vyssotski <pavel.n.vyssotski@intel.com>
8 years agoMerge "Added repe_cmpsq instruction to x86_64 assembler"
Vladimir Marko [Tue, 4 Aug 2015 11:20:31 +0000 (11:20 +0000)]
Merge "Added repe_cmpsq instruction to x86_64 assembler"

8 years agoMerge "ARM/ARM64: Implement numberOfLeadingZeros intrinsic."
Vladimir Marko [Tue, 4 Aug 2015 09:05:51 +0000 (09:05 +0000)]
Merge "ARM/ARM64: Implement numberOfLeadingZeros intrinsic."

8 years agoARM/ARM64: Implement numberOfLeadingZeros intrinsic.
Scott Wakeling [Fri, 10 Jul 2015 10:42:06 +0000 (11:42 +0100)]
ARM/ARM64: Implement numberOfLeadingZeros intrinsic.

Change-Id: I4042fb7a0b75140475dcfca23e8f79d310f5333b