X87: When serializer is enalbed, we disable the X87 stack check to reduce the code size.

     The reason is that the big code size will leads to test-serialize test cases fail
     when debug mode and snapshot=off.

BUG=
R=weiliang.lin@intel.com

Review URL: https://codereview.chromium.org/582373002

Patch from Chunyang Dai <chunyang.dai@intel.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24070 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

The --optimize-for-size flag should imply a small semi-space

R=hpayer@chromium.org, mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/585523002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24069 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix leak in cctest/test-heap/ReleaseStackTraceData

TBR=svenpanne@chromium.org
BUG=none
LOG=n

Review URL: https://codereview.chromium.org/585843002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24068 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Reland 24052 - Require V8 to be explicitly initialized before an Isolate is created

> We also initialize the Isolate on creation.
>
> This should allow for getting rid of the last remaining default isolate
> traces. Also, it'll speed up several isolate related operations that no
> longer require locks.
>
> Embedders that relied on v8::Isolate to return an uninitialized Isolate
> (so they can set ResourceConstraints for example, or set flags that
> modify the way the isolate is created) should either do the setup before
> creating the isolate, or use the recently added CreateParams to pass e.g.
> ResourceConstraints.
>
> BUG=none
> LOG=y
> R=svenpanne@chromium.org
>
> Review URL: https://codereview.chromium.org/469783002

BUG=none
LOG=y
TBR=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/583153002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24067 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Perform scavenges only for small idle times.

BUG=
R=jochen@chromium.org

Review URL: https://codereview.chromium.org/586643002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24066 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Reland sticky regexps https://codereview.chromium.org/567313003/

R=svenpanne@chromium.org, yangguo@chromium.org
BUG=

Review URL: https://codereview.chromium.org/580383003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24065 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Removed a few dead template functions.

R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/583833006

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24064 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

X87: Convert KeyedLoad indexed interceptor case to a Handler.

port r24042.

original commit message:

  Convert KeyedLoad indexed interceptor case to a Handler.

  Currently, KeyedLoads on objects with indexed interceptors are handled with a
  special stub. Instead, key on the map and handler mechanism for more uniform
  treatment.

BUG=
R=weiliang.lin@intel.com

Review URL: https://codereview.chromium.org/581273005

Patch from Jing Bao <jing.bao@intel.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24063 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

X87: Move state sentinels into TypeFeedbackVector.

port r24037.

original commit message:

  Move state sentinels into TypeFeedbackVector.

  These sentinels were in the wrong place, living in only tangentially related class TypeFeedbackInfo, but they codify state in the TypeFeedbackVector.

BUG=
R=weiliang.lin@intel.com

Review URL: https://codereview.chromium.org/585743002

Patch from Jing Bao <jing.bao@intel.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24062 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Emit comment with instruction+reason before deopt calls.

Note that we still need to migrate from sometimes emitting those
comments by hand to passing a reason explicitly, but this can be done
incrementally in separate CLs.

R=jarin@chromium.org

Review URL: https://codereview.chromium.org/582743002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24061 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

[Auto-roll] Bump up version to 3.29.79.0

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24059 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "Require V8 to be explicitly initialized before an Isolate is created"

LOG=n
TBR=svenpanne@chromium.org
BUG=none

Review URL: https://codereview.chromium.org/582953002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24055 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Skip math-floor-of-div-nosudiv on arm

TBR=svenpanne@chromium.org
BUG=none
LOG=n

Review URL: https://codereview.chromium.org/587453002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24054 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

[Auto-roll] Bump up version to 3.29.77.0

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24053 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Require V8 to be explicitly initialized before an Isolate is created

We also initialize the Isolate on creation.

This should allow for getting rid of the last remaining default isolate
traces. Also, it'll speed up several isolate related operations that no
longer require locks.

Embedders that relied on v8::Isolate to return an uninitialized Isolate
(so they can set ResourceConstraints for example, or set flags that
modify the way the isolate is created) should either do the setup before
creating the isolate, or use the recently added CreateParams to pass e.g.
ResourceConstraints.

BUG=none
LOG=y
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/469783002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24052 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Cleanup class parsing a bit

BUG=v8:3330
LOG=Y
R=marja@chromium.org

Review URL: https://codereview.chromium.org/575083002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24051 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Update cctest OWNERS file for mips committers.

BUG=
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/576613003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24050 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "filter cross context eval"

(Reverts https://codereview.chromium.org/294073002/)

BUG=chromium:415051
LOG=y
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/582093002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24049 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

ES6: Implement generator method shorthand

https://people.mozilla.org/~jorendorff/es6-draft.html#sec-method-definitions

BUG=v8:3516
LOG=Y
R=dslomov@chromium.org

Review URL: https://codereview.chromium.org/577973002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24048 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Convert KeyedLoad indexed interceptor case to a Handler.

Port r24042 (eb9b9ec)

Original commit message:
Currently, KeyedLoads on objects with indexed interceptors are handled with a
special stub. Instead, key on the map and handler mechanism for more uniform
treatment.

BUG=
R=dusan.milosavljevic@imgtec.com

Review URL: https://codereview.chromium.org/579273002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24047 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Move state sentinels into TypeFeedbackVector.

Port r24037 (d821bf9)

Original commit message:
These sentinels were in the wrong place, living in only tangentially related class TypeFeedbackInfo, but they codify state in the TypeFeedbackVector.

BUG=
R=dusan.milosavljevic@imgtec.com

Review URL: https://codereview.chromium.org/583573003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24046 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "RegExp: Add support for the ES6-proposed sticky flag"

Causes a flaky failure on buildbots. Here is the (deterministic) repro step (thanks to Michael Stanton):

first go to flag-definitions.h and set this to false.

DEFINE_BOOL(enable_sse4_1, false,
            "enable use of SSE4.1 instructions if available")

Run the following and it should fail:

tools/run-tests.py --arch=ia32 --mode=release cctest/test-api/Regress2107

R=yangguo@chromium.org
BUG=

Review URL: https://codereview.chromium.org/580123002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24045 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Avoid idle times smaller than 1ms.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/580083002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24044 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Implement generator mirror

R=yangguo@chromium.org, aandrey@chromium.org

BUG=v8:3292
LOG=N

Review URL: https://codereview.chromium.org/580823002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24043 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Convert KeyedLoad indexed interceptor case to a Handler.

Currently, KeyedLoads on objects with indexed interceptors are handled with a
special stub. Instead, key on the map and handler mechanism for more uniform
treatment.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/575373004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24042 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix gcmole warning.

TBR=ishell@chromium.org

Review URL: https://codereview.chromium.org/582033002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24041 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Capacity returns allocatable memory and TotalCapacity returns allocatable plus non-allocatable memory for the new space.
BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/577223002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24040 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Move state sentinels into TypeFeedbackVector.

These sentinels were in the wrong place, living in only tangentially related class TypeFeedbackInfo, but they codify state in the TypeFeedbackVector.

R=ishell@chromium.org

Review URL: https://codereview.chromium.org/579153003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24037 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Implement ToLength from ES6 section 7.1.15

BUG=
R=rossberg@chromium.org

Review URL: https://codereview.chromium.org/552273002

Patch from Caitlin Potter <caitpotter88@gmail.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24036 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Turn keyed loads with string-based (non-convertible to array-index) key into named loads

BUG=
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/585433002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24032 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

RegExp: Add support for the ES6-proposed sticky flag

R=yangguo@chromium.org, rossberg@chromium.org
BUG=

Review URL: https://codereview.chromium.org/567313003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24031 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove dead CompilationInfo::DisableOptimization predicate.

R=titzer@chromium.org

Review URL: https://codereview.chromium.org/585443002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24030 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Don't inline polymorphic cases if not all cases can be handled inline.

BUG=
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/578233002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24029 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

[Auto-roll] Bump up version to 3.29.76.0

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24028 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Introduce TypeFeedbackVector, as FixedArray grew constrictive.

The TypeFeedbackVector is poised to host significant functionality. While it
remains a FixedArray under the covers, we need a place to hold logic and
definitions unique to its function.

BUG=
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/581993002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24027 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Thread the Lithium instruction down to DeoptimizeIf and friends.

This is a purely mechanical refactoring and a first step towards being
able to report more helpful deoptimization reasons. With this
refactoring, we know at least the mnemonic of the instruction causing
the deopt, although this is not used yet. Future steps will be using
the mnemonic, passing additional explicit deopt reasons and removing
the fragile machinery of searching for code comments.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/559143003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24026 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove Compiler::RecordFunctionCompilation from interface.

R=sigurds@chromium.org

Review URL: https://codereview.chromium.org/582593003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24025 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Unify use-sites of EnsureDeoptimizationSupport.

R=sigurds@chromium.org

Review URL: https://codereview.chromium.org/563123004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24024 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Reland "Add handling for argument adaptor frames to inlining."

Original: https://codereview.chromium.org/573703002/

Reland Fixes:
- Add deopt framestate to CollectStackTrace runtime call

R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/544953006

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24023 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Don't update the compilation size statistics when doing an optimized compile.

R=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/575393002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24022 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Use allocation throughput to estimate next scavenge event in idle notification.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/573943004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24021 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "Don't use OwnPrototypeChainLength in GetOwnPropertyNames"

This reverts commit r23997 for causing check failures in
layout tests:
http://build.chromium.org/p/client.v8/builders/V8-Blink%20Linux%2064%20%28dbg%29/builds/498

TBR=verwaest@chromium.org

Review URL: https://codereview.chromium.org/581013003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24018 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

[Auto-roll] Bump up version to 3.29.73.0

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24014 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS64: Fix move operations from coprocessor in simulator.

This resolves calculation errors for trigonometric functions.

TEST=test262/S15.8.2.7_A6.js
BUG=
R=jkummerow@chromium.org, paul.lind@imgtec.com

Review URL: https://codereview.chromium.org/558163006

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24013 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Rename Runtime_CompileUnoptimized to Runtime_CompileLazy, because that is what it does. Split Compiler::GetUnoptimizedCode into two variants, one for lazy compilation (which can return optimized code!) and the other that actually returns unoptimized code.

R=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/547293004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24012 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "Add handling for argument adaptor frames to inlining."

This reverts commit r24008.

TBR=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/581673002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24010 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Introduce simplified BooleanToNumber operator.

R=titzer@chromium.org
TEST=cctest/test-simplified-lowering/LowerBooleanToNumber

Review URL: https://codereview.chromium.org/579723004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24009 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Add handling for argument adaptor frames to inlining.
- Lazy deopt from inlined calls
- Lazy deopt from inlined calls with parameter mismatch

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

Review URL: https://codereview.chromium.org/573703002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24008 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix the IsTheHole bug.

For Double constant of non nan, object_ may be uninitialized.

(This patch comes from weiliang.lin@intel.com )

BUG=
R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/565183004

Patch from Chunyang Dai <chunyang.dai@intel.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24007 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

This is a reland of issue 563813002:

https://codereview.chromium.org/575473002/

This time with additional include headers and MSVC specific guards. Previously, only V8 Win32 - nosnap - shared failed:

http://build.chromium.org/p/client.v8/builders/V8%20Win32%20-%20nosnap%20-%20shared/builds/2916

R=marja@chromium.org

Review URL: https://codereview.chromium.org/576733002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24006 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Array.prototype.sort: Unchecked calls to hasOwnProperty and push and sort

BUG=v8:3537
LOG=
R=arv@chromium.org, wingo@igalia.com, yangguo@chromium.org

Review URL: https://codereview.chromium.org/555173002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24005 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

[Auto-roll] Bump up version to 3.29.72.0

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24004 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix compile error introduced in 24002.

TBR=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/547293003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24003 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Serialize code stubs using stub key.

Also add some tracing to the code serializer.

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/556243005

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24002 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Minor compiler pipeline refactoring. Inline UpdateSharedFunctionInfo and make Parser::Parse responsible for setting the strict mode of the CompilationInfo.

R=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/555553003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24001 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Support stepping into generator function.

R=aandrey@chromium.org, wingo@igalia.com
BUG=v8:3572
LOG=Y

Review URL: https://codereview.chromium.org/544953005

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@24000 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

X87: turn on some tests of test-serialize

They were disabled by r21661. We turn on it because x87 snapshot can work now.

BUG=
R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/568833003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23999 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Don't use OwnPrototypeChainLength in GetOwnPropertyNames

BUG=
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/574753002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23997 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fixed deopt reasons in TaggedToI.

Every DeoptimizeIf should be preceded by a RecordComment explaining
the reason. In the long run, the reason should be an argument of
DeoptimizeIf, but we're not there yet.

On x87, things are a bit ugly due to some pushing/popping, so if
somebody feels inclined to improve this: Feel free. Probably the right
approach would be reloading instead of the push/pop horror.

Another thing is our inconsistent handling of the "done" continuation
of deferred code on the various platforms, I made tiny changes on the
way, but this should better be unified somehow, with all those
micro-optimizations removed.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/578583002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23996 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Unbreak build

TBR=hpayer@chromium.org

Review URL: https://codereview.chromium.org/568433003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23995 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Measure new space allocation throughput.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/578453003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23994 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

[Auto-roll] Bump up version to 3.29.71.0

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23993 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

X87: Removing ic.h from code-stubs.h

port r23977.

original commit message:

  Removing ic.h from code-stubs.h

  CodeStubs use state types defined in ic.h, but this has the unfortunate effect of spreading ic.h all over the place. Instead, define these shared state types in ic-public.h and allow ic.h to concern itself with internal state change of the ICs.

  More work could/should be done here, but this is a first step.

BUG=
R=weiliang.lin@intel.com

Review URL: https://codereview.chromium.org/573123003

Patch from Jing Bao <jing.bao@intel.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23992 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Class syntax parsing

This implements parsing for ClassExpression and ClassDeclaration.
The runtime is not yet implemented and the value is currently
hard coded to undefined.

BUG=v8:3330
LOG=Y
R=dslomov@chromium.org, marja@chromium.org, rossberg@chromium.org

Review URL: https://codereview.chromium.org/561913002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23988 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

[Auto-roll] Bump up version to 3.29.69.0

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23987 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS64: Fixed int vs. uintptr_t confusion (plus some cleanup on the way).

Port r23966 (a498d5e3)

BUG=
R=jkummerow@chromium.org, paul.lind@imgtec.com

Review URL: https://codereview.chromium.org/570563006

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23986 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS64: Removing ic.h from code-stubs.h

Port r23977 (d0cba839)

Original commit message:
CodeStubs use state types defined in ic.h, but this has the unfortunate effect of spreading ic.h all over the place. Instead, define these shared state types in ic-public.h and allow ic.h to concern itself with internal state change of the ICs.

More work could/should be done here, but this is a first step.

BUG=
R=paul.lind@imgtec.com

Review URL: https://codereview.chromium.org/574493004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23984 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

[Auto-roll] Bump up version to 3.29.68.0

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23983 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Avoid usage of temporary MachineOperatorBuilder.

R=titzer@chromium.org

Review URL: https://codereview.chromium.org/562203004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23982 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Track how much we miss the idle notification limit

We can't report negative values using histograms, so we split the data
up into two histograms

BUG=chromium:397026
LOG=n
R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/572293002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23981 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove unused histogram definitions

R=marja@chromium.org
LOG=n
BUG=none

Review URL: https://codereview.chromium.org/572923002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23980 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Allow IdleNotification to trigger Scavenges.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/563493002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23979 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Re-reland "Use unsigned type bitsets to limit undefined behaviour"

Change: make bitset uint32_t, to avoid weird sign extension bugs on Win64.

BUG=
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/577563002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23978 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Removing ic.h from code-stubs.h

CodeStubs use state types defined in ic.h, but this has the unfortunate effect of spreading ic.h all over the place. Instead, define these shared state types in ic-public.h and allow ic.h to concern itself with internal state change of the ICs.

More work could/should be done here, but this is a first step.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/565873002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23977 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Use the regular start incremental marking strategy in the idle notification.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/568023002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23976 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Simplify the LookupIterator

BUG=
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/570293002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23975 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Enable ES6 generators

R=rossberg@chromium.org

BUG=v8:2355
LOG=Y

Review URL: https://codereview.chromium.org/573963003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23974 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Adapt some typed lowering unit tests after r23972.

R=rossberg@chromium.org
TEST=cctest/test-js-typed-lowering

Review URL: https://codereview.chromium.org/574713002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23973 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix typed lowering to number comparison.

R=titzer@chromium.org
TEST=mjsunit/regress/regress-3564
BUG=v8:3564
LOG=N

Review URL: https://codereview.chromium.org/574653002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23972 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Allow trace-codegen to dereference handles.

This fixes --print-all-code for some tests.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/574673002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23971 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

ARM64: Remove the unused 'SmiAbs'.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/572903003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23970 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

The value register in storefield(smi) is not used as a temp
BUG=
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/551803005

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23969 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

[Auto-roll] Bump up version to 3.29.67.0

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23967 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fixed int vs. uintptr_t confusion (plus some cleanup on the way).

BUG=v8:3556
LOG=y
R=jochen@chromium.org

Review URL: https://codereview.chromium.org/567573004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23966 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Introduce DONE state in idle notification handler.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/577573002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23965 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Move configuration of ResourceConstraints to Isolate construction

We can only set resource constraints before the isolate is initialized.
Since in the future, we want to initialize isolates at construction
time, we need to set constraints there.

It's possible to later modify the stack limit, so introduce an
Isolate::SetStackLimit method for that.

The SetResourceConstraints method will be deprecated. Users should pass
ResourceConstraints to Isolate::New, and use Isolate::SetStackLimit to
later modify the stack limit.

BUG=none
R=svenpanne@chromium.org
LOG=y

Review URL: https://codereview.chromium.org/559993005

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23964 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert r23962 "Configure d8 & cctest to suppress Windows error dialogs."

Breaks compilation on Windows

TBR=sigurds@chromium.org
BUG=none
LOG=n

Review URL: https://codereview.chromium.org/572873004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23963 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Configure d8 & cctest to suppress Windows error dialogs.

R=machenbach@chromium.org

Review URL: https://codereview.chromium.org/575473002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23962 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Disable lowering to StringAdd due to various issues.

R=titzer@chromium.org

Review URL: https://codereview.chromium.org/566303003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23961 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Clean up regexp code gen a little

R=yangguo@chromium.org
BUG=

Review URL: https://codereview.chromium.org/565043003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23958 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Skip long running tests with TurboFan and GC Stress.

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/559343004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23957 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Only include compile-only win64 trybot in default set.

BUG=chromium:353487
LOG=n
R=jkummerow@chromium.org
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/570113002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23956 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Re-revert "Use unsigned type bitsets to limit undefined behaviour"

...and the various debug CLs.

R=machenbach@chromium.org
BUG=

Review URL: https://codereview.chromium.org/564413004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23955 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Use an upper bound for marking time estimation in idle notification.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/571913003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23954 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Exact black-list of AST nodes for TurboFan.

R=rossberg@chromium.org

Review URL: https://codereview.chromium.org/564203003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23953 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Compile V8 with extra optimization in GN Release mode.

This matches the GYP build which does 'optimize': 'max' on the corresponding targets.

R=jochen@chromium.org

Review URL: https://codereview.chromium.org/567073003

Patch from Brett Wilson <brettw@chromium.org>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23952 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Skip octane/pdfjs for now in debug mode because slow.

TBR=rossberg@chromium.org
TEST=benchmarks/octane/pdfjs

Review URL: https://codereview.chromium.org/569243002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23950 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix schedule for interleaved floating control by wiring one floating control structure per block.

R=mstarzinger@chromium.org, bmeurer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/564403003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23949 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Reset context disposed counter after executing the idle garbage collection operation.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/552723004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23948 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Explicitly mask enum values to work around probable VS/Win64 compiler bug

TBR=svenpanne@chromium.org
BUG=

Review URL: https://codereview.chromium.org/564413003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@23947 ce2b1a6d-e550-0410-aec6-3dcde31c8c00