HasOnlyAsciiChars can return incorrect results. Fixup usages and rename.

R=ulan@chromium.org
BUG=

Review URL: https://codereview.chromium.org/14509012

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14453 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

forgot parends around macro arg

R=svenpanne@chromium.org
BUG=

Review URL: https://codereview.chromium.org/14334009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14452 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Bootstrapper code for constructed arrays needs to be handlefied

BUG=

Review URL: https://codereview.chromium.org/14008004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14451 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Implement support for Math.imul in Crankshaft.

R=jkummerow@chromium.org
TEST=mjsunit/math-imul

Review URL: https://codereview.chromium.org/14471041

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14450 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix test/build failures (win64 compile warning, isolate test parallel compilation failure, gcmole issue, isolate test intermittant failure)

BUG=

Review URL: https://codereview.chromium.org/14265024

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14449 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Better handling of Phi nodes with constant inputs

Review URL: https://codereview.chromium.org/14471034

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14448 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Prevent false positive from GCMole in DeferredFormatStackTrace.

R=yangguo@chromium.org
TEST=gcmole

Review URL: https://codereview.chromium.org/14416010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14447 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Avoid unnecessary indirection when creating CodeEntries

BUG=None

Review URL: https://codereview.chromium.org/14471035

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14446 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Do not dereference handles during relocation.

R=hpayer@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/13982023

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14445 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Defer slow branch of GetAlignedPointerFromInternalField

BUG=

Review URL: https://codereview.chromium.org/14137026

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14444 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Constructed arrays can be created with Hydrogen code stubs. The feature is still off by default (--optimize-constructed-arrays).

Port r14441 (0c30d023)

BUG=

Review URL: https://codereview.chromium.org/13905009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14443 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Capture receiver in generator object

Port r14434 (04f254d1)

Original commit message:
Previously there has been no reason to context-allocate the receiver, so
access to the receiver always goes through the stack.  This was failing
with generators, which assumed that forcing context allocation would
relieve the need of storing anything but the context and the function on
the stack.

This CL adds a slot in generator objects to capture the receiver, and
restores it when resuming a generator.

BUG=v8:2355
TEST=mjsunit/harmony/generators-iteration

Review URL: https://codereview.chromium.org/14195033

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14442 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Constructed arrays can be created with Hydrogen code stubs. The feature is still off by default (--optimize-constructed-arrays).

BUG=

Review URL: https://codereview.chromium.org/12385014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14441 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Win64 build

R=rossberg@chromium.org,mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/14392013

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14440 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Tell the GC prologues to construct RetainedObjectInfos only when needed.

The GC prologue is called by the GC, but also by the heap snapshotter. The
RetainedObjectInfos are only needed by the heap snapshotter, so it's wasteful to
construct them always. (And it will be even more wasteful when Blink migrates to
the new GC APIs, since after that point it no longer knows about object groups.)

BUG=

Review URL: https://codereview.chromium.org/14471028
Patch from Marja Hölttä <marja@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14439 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

First cut at API for ES6 ArrayBuffers

R=rossberg@chromium.org
BUG=

Review URL: https://codereview.chromium.org/13958007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14438 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Relax --debug-code checks that runtime returns are not the hole

ia32, unlike the other architectures, includes a --debug-code check that
asserts that runtime functions do not return the hole.  However the new
SuspendJSGeneratorObject runtime does return the hole at times.

This CL adds a wee hack that only signals an error if the callee was not
SuspendJSGeneratorObject.

R=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/13856011

Patch from Andy Wingo <wingo@igalia.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14437 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Capture receiver in generator object

Previously there has been no reason to context-allocate the receiver, so
access to the receiver always goes through the stack.  This was failing
with generators, which assumed that forcing context allocation would
relieve the need of storing anything but the context and the function on
the stack.

This CL adds a slot in generator objects to capture the receiver, and
restores it when resuming a generator.

BUG=v8:2355
TEST=mjsunit/harmony/generators-iteration

Review URL: https://codereview.chromium.org/14158006

Patch from Andy Wingo <wingo@igalia.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14434 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Get isolate from thread local instead of (potentially missing) block in HConstant.

R=ulan@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/14284011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14433 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix evaluation order problems in Object.observe hooks.

R=rossberg@chromium.org
TEST=gcmole

Review URL: https://codereview.chromium.org/14493012

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14432 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Grow the old generation faster in high promotion mode.

BUG=

Review URL: https://codereview.chromium.org/14493011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14431 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix overflow check in mul-i which was missing since r14322

Review URL: https://codereview.chromium.org/14471012

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14430 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Clean up VMState a little bit.

Port r14421 (31cd7dda)

BUG=

Review URL: https://codereview.chromium.org/14483002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14429 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Generators can resume

Port r14415 (d358defa) and r14424 (7b549ce7)

Original commit message:
The generator object methods "next", "send", and "throw" now
include some inline assembly to set up a resumed stack frame.  In some
common cases, we can just jump back into the frame to resume it.
Otherwise the resume code calls out to a runtime to fill in the operand
stack, rewind the handlers, and possibly to throw an exception.

BUG=v8:2355
TESTS=mjsunit/harmony/generators-iteration

Review URL: https://codereview.chromium.org/13864010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14428 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Add monomorphic CompareNilICs and Crankshaft support

Port r14407 (cae6596b)

BUG=

Review URL: https://codereview.chromium.org/13840006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14427 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Fix NaN handling for start index in ArraySplice.

Casting NaN to int is unpredictable, on different architectures it produces different int value.

TEST=test262/S15.4.4.10_A2.1_T2, S15.4.4.10_A2.2_T2, S15.4.4.12_A2.1_T2

BUG=

Review URL: https://codereview.chromium.org/14257006
Patch from Dusan Milosavljevic <Dusan.Milosavljevic@rt-rk.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14426 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove bogus forward declaration of ObjectGroup.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/14333016

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14425 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix generator suspend and resume on ARM

This CL fixes a number of rookie mistakes in
FullCodeGenerator::VisitYield and
FullCodeGenerator::EmitGeneratorResume, and re-enables the
generators-iteration test on ARM.

R=mstarzinger@chromium.org
BUG=v8:2355
TEST=mjsunit/harmony/generators-iteration

Review URL: https://codereview.chromium.org/13843022

Patch from Andy Wingo <wingo@igalia.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14424 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

New GC APIs, try 2.

With these APIs, the embedder doesn't need to copy Persistent handles around.

BUG=

Review URL: https://codereview.chromium.org/14007008

Patch from Marja Hölttä <marja@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14423 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fixed HUnaryMathOperation regarding its possible operations.

Review URL: https://codereview.chromium.org/14333013

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14422 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Clean up VMState a little bit.

R=svenpanne@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/14139033

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14421 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Change cctest/test-lockers to not copy persistent handles around.

Instead, create Local handles to pass them around. This also means that the
code needs to be shifted around a bit such that a handle scope exists when
creating threads.

Review URL: https://codereview.chromium.org/14150017

Patch from Jochen Eisinger <jochen@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14420 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix windows build.

R=svenpanne@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/14476003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14417 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove Isolate::Current() from histograms.

R=svenpanne@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/14471007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14416 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Generators can resume

The generator object methods "next", "send", and "throw" now
include some inline assembly to set up a resumed stack frame.  In some
common cases, we can just jump back into the frame to resume it.
Otherwise the resume code calls out to a runtime to fill in the operand
stack, rewind the handlers, and possibly to throw an exception.

BUG=v8:2355
TESTS=mjsunit/harmony/generators-iteration

Review URL: https://codereview.chromium.org/14066016

Patch from Andy Wingo <wingo@igalia.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14415 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MathSqrt and MathPowHalf do not cause any promotion.

Review URL: https://codereview.chromium.org/13829009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14414 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Properly consider Double inputs for representation inference when they are not marked as kFlexibleRepresentation

Review URL: https://codereview.chromium.org/14408006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14413 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix segmentation fault in CodeMap::Print

CodeEntry* is always NULL for SharedFunctionInfo entries in the CodeMap. Take this into account when printing the map.

Drive-by: removed CodeEntry::shared_id() which is never called.
BUG=None

Review URL: https://codereview.chromium.org/14387004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14410 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Simplify some code related to x64 calling convention.

R=mvstanton@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/13940014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14409 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix clang build for x64.

R=ulan@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/14471004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14408 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Add monomorphic CompareNilICs and Crankshaft support

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/14367018

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14407 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Replace qsort with std::sort.

std::sort is a template, so it can be inlined more aggressively. Also, it's
O(n log n), while libc's qsort is O(n^2)

BUG=2639

Review URL: https://codereview.chromium.org/14315005

Patch from Jochen Eisinger <jochen@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14406 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Prepare push to trunk.  Now working on version 3.18.4.

R=yangguo@chromium.org

Review URL: https://chromiumcodereview.appspot.com/14139031

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14403 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Removed unbalanced brackets when printing an LEnvironment.

Review URL: https://codereview.chromium.org/14286005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14402 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

If the input to an HChange is of type smi, then no need to mark gvn flag changes new space promotion.

BUG=

Review URL: https://codereview.chromium.org/14253010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14401 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Refactor ExternalReference::isolate_address() to not rely on Isolate::Current().

R=svenpanne@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/14429003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14400 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Disallow dereferencing deferred handles when generating optimized code.

Port r14388 (6843e0c9), r14389 (331ea32d) and r14390 (c8528d1c)

BUG=

Review URL: https://codereview.chromium.org/14305011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14399 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Adds EXTERNAL_DOUBLE_ARRAY to a list of instance types

BUG=v8:2646

Patch by Andrei Kashcha <anvaka@gmail.com>

Review URL: https://codereview.chromium.org/14042008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14398 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Expose the GC under a name that is less collision prone than window.gc.

This also keeps backwards compatibility (window.gc() still works).

BUG=2641

Review URL: https://codereview.chromium.org/14075012

Patch from Marja Hölttä <marja@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14397 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Do not emit Simulates in HandlePolymorphicElementAccess.

BUG=v8:2653
R=jkummerow@chromium.org
TEST=mjsunit/regress/regress-2653.js

Review URL: https://chromiumcodereview.appspot.com/14081025

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14396 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Add a flag to deoptimize all functions every n garbage collections.

R=mstarzinger@chromium.org

Review URL: https://chromiumcodereview.appspot.com/14091013

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14395 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Do not emit double values at their use sites.

Revert part of r14179. From the regression test's comment:

Currently, the gap resolver doesn't handle moves from a ConstantOperand to a
DoubleRegister, but these kind of moves appeared when HConstant::EmitAtUses
was changed to allow special double values (-0, NaN, hole). So we should
either enhance the gap resolver or make sure that such moves don't happen.

BUG=chrome:234101

Review URL: https://codereview.chromium.org/14429002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14394 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Make it clear that GCMole depends on Clang 2.9 currently.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/13950017

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14393 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix ugly typo in r14391.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/14425006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14392 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Switch GYP script to not use utils.py anymore.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/14108010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14391 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix typo.

R=mvstanton@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/14425005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14390 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix clang build.

R=mvstanton@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/14091011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14389 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Disallow dereferencing deferred handles when generating optimized code.

R=mvstanton@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/14403015

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14388 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Change deprecated comments that still talk about SCons.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/14404012

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14387 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Print ProfileNode id when dumping CPU profile to console

Node ids are monotonically increasing and looking at them we can
say e.g. when the node was added before another one. We need this
to diagnose sporadic failures of cctest/test-cpu-profiler/CollectCpuProfile

BUG=v8:2628

Review URL: https://codereview.chromium.org/14402014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14386 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove the executable bit from scanner.cc file.

Patch from Thiago Farina <tfarina@chromium.org>.

TBR=yangguo@chromium.org

Review URL: https://codereview.chromium.org/14081021

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14385 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: ARM: Fix erroneous detection of Double result type in Integer32 BinaryOpStubs

Port r14371 (b5c4a990)

BUG=

Review URL: https://codereview.chromium.org/13973013

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14384 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Improvements in lithium code generation. Recognizing if some operands are constants, we can often save on registers and instructions.

Port r14364 (2819e5ee)

BUG=

Review URL: https://codereview.chromium.org/14159015

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14383 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

ARM: VFP cleanup now that VFP2 is the baseline.

BUG=none
TEST=none

Review URL: https://chromiumcodereview.appspot.com/14119008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14382 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Ignore observed Double output in binary operations when all uses are truncating to Integer32

BUG=v8:2424

Review URL: https://codereview.chromium.org/14320021

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14381 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fixed issue in StoreNamedField codegen where integer32 constants were not converted to a smi.

BUG=

Review URL: https://codereview.chromium.org/14075014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14380 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

create uniform string api

R=yangguo@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/12426015

Patch from Dan Carney <dcarney@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14379 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Add methods to allow resuming execution after calling TerminateExecution().

R=mvstanton@chromium.org
BUG=v8:2361

Review URL: https://chromiumcodereview.appspot.com/14401008

Patch from Andrew Paprocki <andrew@ishiboo.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14378 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Ensure callbacks transitions are actually followed.

Review URL: https://chromiumcodereview.appspot.com/14371007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14377 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Disable ParserSync test in GC stress mode for now.

R=jkummerow@chromium.org
TEST=cctest/test-parsing/ParserSync

Review URL: https://codereview.chromium.org/14075017

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14376 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix outdated build/README.txt

BUG=v8:2649

Review URL: https://codereview.chromium.org/13911011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14375 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Make Clang plugins directory configurable for GCMole.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/13956027

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14374 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Add d8 functionality for switching between realms (a.k.a. contexts), 2nd try

The only difference to the previous implementation (https://codereview.chromium.org/14295011/) is in patch set 2.

R=yangguo@chromium.org
BUG=

Review URL: https://codereview.chromium.org/14404005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14373 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Replace CheckBuilder with IfBuilder everywhere.

This deprecates and removes the CheckBuilder which is less powerful
than the generic IfBuilder which can deopt as well by now.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/14075013

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14372 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

ARM: Fix erroneous detection of Double result type in Integer32 BinaryOpStubs

BUG=v8:2648

Review URL: https://codereview.chromium.org/14178013

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14371 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Various improvements regarding the way we print code code comments.

* All Lithium instructions have an associated Hydrogen instruction now,
  simplifying things.

* Consistently print <Lithium instruction number,Hydrogen value id> prefixes.

* Do not print uninteresting Lithium instructions like empty gaps, jumps to the
  next instruction, etc.

* Removed special handling of HChange-like instructions, it is totally unclear
  why they had this special treatment. If we really want to print more
  information about Lithium instructions, we should do it in a totally way,
  anyway (e.g. by unifying things with the generation of hydrogen*.cfg files).

* Made deferred code and the jump table stand out a little bit more.

* Print info about special blocks like loop headers and OSR entries.

Review URL: https://codereview.chromium.org/14371005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14370 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Prepare push to trunk.  Now working on version 3.18.3.

R=jkummerow@chromium.org

Review URL: https://chromiumcodereview.appspot.com/14403005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14367 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Skip one generators test because of no-snap failures.

R=rossberg@chromium.org
BUG=v8:2355
TEST=mjsunit/harmony/generators-objects

Review URL: https://codereview.chromium.org/13956023

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14366 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Infer smi-range for smi-typed values.

Review URL: https://chromiumcodereview.appspot.com/14192034

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14365 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Improvements in lithium code generation. Recognizing if some operands are constants, we can often save on registers and instructions.

BUG=

Review URL: https://codereview.chromium.org/14022011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14364 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Generator objects can suspend

Port r14353 (f4bb81d1)

Original commit message:
* src/ast.h:
* src/parser.cc: Differentiate between the different kinds of yields, in
  anticipation of boxing return values.  Parse `return' into `yield' in
  a generator.

* src/runtime.h:
* src/runtime.cc (Runtime_SuspendJSGeneratorObject): New horrible
  runtime function: saves continuation, context, and operands into the
  generator object.

* src/arm/full-codegen-arm.cc (VisitYield):
* src/ia32/full-codegen-ia32.cc (VisitYield):
* src/x64/full-codegen-x64.cc (VisitYield): Arrange to call
  SuspendJSGeneratorObject.  If the call returns the hole, we suspend.
  Otherwise we resume.

BUG=v8:2355
TEST=These codepaths are tested when the generator is first invoked, and so
are covered by mjsunit/harmony/generators-objects.js.

Review URL: https://codereview.chromium.org/14091006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14363 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Refactor HCheckMaps to have a private constructor, removing duplicated code and simplifying calls in clients.
Ignore bsuite directory.

Review URL: https://codereview.chromium.org/14367023

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14362 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

OS::MemMove/OS::MemCopy: Don't call through to generated code when size == 0 to avoid prefetching invalid memory

BUG=chromium:233500

Review URL: https://codereview.chromium.org/14365011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14361 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Switch elements access to use IfBuilder instead of CheckBuilder.

R=danno@chromium.org

Review URL: https://codereview.chromium.org/14365014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14360 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

RecordWriteStubs on ARM need to save FP registers

A previous change erroneously stopped saving fp registers for write stubs in
snapshotted code. This CL restores correct behavior, and makes sure code that
saves fp registers checks at runtime for 16 or 32 fp registers. Also fix a bug
in the arm simulator debugger.

BUG=

Review URL: https://codereview.chromium.org/14246032

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14359 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Propagate OOM exception instead of crashing in CALL_HEAP_FUNCTION_PASS_EXCEPTION macro.

R=jkummerow@chromium.org
BUG=webkit/fast/js/concat-large-strings-crash.html

Review URL: https://chromiumcodereview.appspot.com/14365017

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14358 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix build failures on Win64 since r14354.

R=ulan@chromium.org

Review URL: https://codereview.chromium.org/14246034

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14357 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove heap snapshot size limit.

As long as frontend now stores snapshots in external arrays
it doesn't make sense to check the collected snapshot size on the
backend side. In other words if backend has managed to fit the
collected snapshot into its memory, it should also fit into the
frontend's memory.

R=yurys@chromium.org
BUG=chromium:232305

Review URL: https://codereview.chromium.org/14246029

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14356 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Update arm and mips simulator to also use cmath

Review URL: https://codereview.chromium.org/14241029

Patch from Jochen Eisinger <jochen@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14355 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Update tests to use std::isnan instead of isnan

Should fix compilation breakage after r14352

Review URL: https://codereview.chromium.org/14328044

Patch from Jochen Eisinger <jochen@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14354 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Generator objects can suspend

* src/ast.h:
* src/parser.cc: Differentiate between the different kinds of yields, in
  anticipation of boxing return values.  Parse `return' into `yield' in
  a generator.

* src/runtime.h:
* src/runtime.cc (Runtime_SuspendJSGeneratorObject): New horrible
  runtime function: saves continuation, context, and operands into the
  generator object.

* src/arm/full-codegen-arm.cc (VisitYield):
* src/ia32/full-codegen-ia32.cc (VisitYield):
* src/x64/full-codegen-x64.cc (VisitYield): Arrange to call
  SuspendJSGeneratorObject.  If the call returns the hole, we suspend.
  Otherwise we resume.

BUG=v8:2355
TEST=These codepaths are tested when the generator is first invoked, and so
are covered by mjsunit/harmony/generators-objects.js.

Review URL: https://codereview.chromium.org/13704010

Patch from Andy Wingo <wingo@igalia.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14353 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Replace math.h with cmath

This will make it easier to use other STL headers in the future

Review URL: https://codereview.chromium.org/14362023

Patch from Jochen Eisinger <jochen@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14352 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove unused member variables in IfBuilder

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/14328042

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14351 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Switch GCMole to use GYP build file instead of SCons.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/14320024

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14350 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Simplified LCodeGen::GetNextEmittedBlock and LCodeGen::EmitGoto a bit.

GetNextEmittedBlock is always called with the same argument (an instance
variable), so let's remove it. In EmitGoto, avoid assignment to an argument.

This CL is split off another CL for easier reviewing.

Review URL: https://codereview.chromium.org/14246031

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14349 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert r14252 as it broke --prof for some cases

R=jkummerow
BUG=v8:2642

Review URL: https://codereview.chromium.org/14367020

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14348 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Handle retry-after-gc failures within LoadIC::Load and StoreIC::Store.

Follow-up for r14321, makes the remaining unsafe calls to runtime functions during ic computation safe.

R=verwaest@chromium.org
BUG=222301

Review URL: https://chromiumcodereview.appspot.com/13976015

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14347 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Removed redundant instruction in VerifyX87StackDepth()

BUG=

Review URL: https://codereview.chromium.org/14353004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14346 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Add infrastructure to automatucally generate IC miss stubs

Port r14342 (24df6264)

BUG=

Review URL: https://codereview.chromium.org/13896016

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14345 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: HArgument instructions currently require a frame. In Lithium we can ensure a frame is created for these instructions via a compile info flag.

Port r14339 (7d544be1)

BUG=

Review URL: https://codereview.chromium.org/14367003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@14344 ce2b1a6d-e550-0410-aec6-3dcde31c8c00