platform/upstream/v8.git
9 years agoRemove superflous SharedFunctionInfo::uses_super call.
mstarzinger [Thu, 20 Nov 2014 16:21:56 +0000 (08:21 -0800)]
Remove superflous SharedFunctionInfo::uses_super call.

R=danno@chromium.org

Review URL: https://codereview.chromium.org/736383002

Cr-Commit-Position: refs/heads/master@{#25445}

9 years agoWhen optimizing deserialized code, make sure IC state is preserved.
yangguo [Thu, 20 Nov 2014 16:20:48 +0000 (08:20 -0800)]
When optimizing deserialized code, make sure IC state is preserved.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/737373003

Cr-Commit-Position: refs/heads/master@{#25444}

9 years ago[turbofan] use double temp registers for stack-stack swap on arm64
Dan Carney [Thu, 20 Nov 2014 16:11:22 +0000 (17:11 +0100)]
[turbofan] use double temp registers for stack-stack swap on arm64

TBR=vogelheim@chromium.org

BUG=

Review URL: https://codereview.chromium.org/744743002

Cr-Commit-Position: refs/heads/master@{#25443}

9 years agoFix Regress3540 after r25441.
Ulan Degenbaev [Thu, 20 Nov 2014 15:42:32 +0000 (16:42 +0100)]
Fix Regress3540 after r25441.

BUG=
TBR=hpayer@chromium.org

Review URL: https://codereview.chromium.org/746443002

Cr-Commit-Position: refs/heads/master@{#25442}

9 years agoReserve code range block for evacuation.
ulan [Thu, 20 Nov 2014 14:52:23 +0000 (06:52 -0800)]
Reserve code range block for evacuation.

If we run out of code range, then GC wouldn't be able to compact code space,
because it wouldn't be able to allocate a new page. This can cause code space
fragmentation and OOM crashes.

BUG=chromium:430118
LOG=Y

Review URL: https://codereview.chromium.org/742733002

Cr-Commit-Position: refs/heads/master@{#25441}

9 years ago[turbofan] More aggressive reuse of spill slots in the register allocator.
dcarney [Thu, 20 Nov 2014 14:40:56 +0000 (06:40 -0800)]
[turbofan] More aggressive reuse of spill slots in the register allocator.

BUG=

Review URL: https://codereview.chromium.org/725083004

Cr-Commit-Position: refs/heads/master@{#25440}

9 years ago[turbofan]: More optimizations to add and subtract operations on x64
Daniel Clifford [Thu, 20 Nov 2014 13:48:34 +0000 (14:48 +0100)]
[turbofan]: More optimizations to add and subtract operations on x64

- Use "leal" for subtraction of integer constant when non-constant input to
  subtract is used more than once.
- Use "incl", "decl", and "addl" when they are smaller/cheaper than their
  leal/addl/subl equivalant.

R=titzer@chromium.org

Review URL: https://codereview.chromium.org/738073002

Cr-Commit-Position: refs/heads/master@{#25439}

9 years ago[turbofan] remove redundant '& 0x1F' for shifts.
Rodolph Perfetta [Thu, 20 Nov 2014 13:09:04 +0000 (13:09 +0000)]
[turbofan] remove redundant '& 0x1F' for shifts.

JavaScript shifts perform an implicit '& 0x1F' on their right operand, this
patch removes it when the underlying architecture already does it.

BUG=
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/732103002

Cr-Commit-Position: refs/heads/master@{#25438}

9 years agoEnable TurboFan for Win64 targets.
mstarzinger [Thu, 20 Nov 2014 11:51:04 +0000 (03:51 -0800)]
Enable TurboFan for Win64 targets.

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/743963002

Cr-Commit-Position: refs/heads/master@{#25437}

9 years agoFix lower bound violation
Andreas Rossberg [Thu, 20 Nov 2014 11:22:43 +0000 (12:22 +0100)]
Fix lower bound violation

R=jarin@chromium.org
BUG=433332
LOG=N

Review URL: https://codereview.chromium.org/739563002

Cr-Commit-Position: refs/heads/master@{#25436}

9 years agoDisable classes in sloppy mode unless --harmony-sloppy is set
Andreas Rossberg [Thu, 20 Nov 2014 10:51:49 +0000 (11:51 +0100)]
Disable classes in sloppy mode unless --harmony-sloppy is set

Also clean up flag names a little.

Baseline: https://codereview.chromium.org/713413003/

R=arv@chromium.org, dslomov@chromium.org
BUG=

Review URL: https://codereview.chromium.org/722203006

Cr-Commit-Position: refs/heads/master@{#25435}

9 years agoConditional flags for tests - set JS stack size for simulators.
balazs.kilvady [Thu, 20 Nov 2014 09:46:55 +0000 (01:46 -0800)]
Conditional flags for tests - set JS stack size for simulators.

There are separated JS and C stacks on simulators so for some stack
extensive tests (like mozilla/js1_5/extensions/regress-355497) might
cause a C stack overflow and that overflow is not caught by V8. It is
not an issue on real HW. Increasing the C stack also solves the problem
but we have already FLAG_sim_stack_size flag to control the JS stack
size.

This patch makes it possible to add flags to tests conditionally in
.status files.

TEST=mozilla/js1_5/extensions/regress-355497
BUG=v8:3152
LOG=N

Review URL: https://codereview.chromium.org/735723006

Cr-Commit-Position: refs/heads/master@{#25434}

9 years agoImplement log10 via fdlibm port.
yangguo [Thu, 20 Nov 2014 09:37:18 +0000 (01:37 -0800)]
Implement log10 via fdlibm port.

R=rtoy@chromium.org
BUG=v8:3579
LOG=N

Review URL: https://codereview.chromium.org/739913003

Cr-Commit-Position: refs/heads/master@{#25433}

9 years agoRevert of Rename String.prototype.contains to 'includes'. (patchset #1 id:1 of https...
machenbach [Thu, 20 Nov 2014 09:03:44 +0000 (01:03 -0800)]
Revert of Rename String.prototype.contains to 'includes'. (patchset #1 id:1 of https://codereview.chromium.org/742963002/)

Reason for revert:
Breaks test262-es6:
http://build.chromium.org/p/client.v8/builders/V8%20Linux/builds/1289

Original issue's description:
> Rename String.prototype.contains to 'includes'.
>
> Per TC39 Nov 2014 decison.
>
> R=arv@chromium.org,yangguo@chromium.org
> LOG=Y
>
> Committed: https://chromium.googlesource.com/v8/v8/+/b5379216e23ff0ec734e70361dd07057c2421c96

TBR=arv@chromium.org,yangguo@chromium.org,dslomov@chromium.org
NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/745543002

Cr-Commit-Position: refs/heads/master@{#25432}

9 years agoRemove --clever-optimizations
jochen [Thu, 20 Nov 2014 08:54:27 +0000 (00:54 -0800)]
Remove --clever-optimizations

We should always be clever

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

Review URL: https://codereview.chromium.org/723023003

Cr-Commit-Position: refs/heads/master@{#25431}

9 years agoRename String.prototype.contains to 'includes'.
dslomov [Thu, 20 Nov 2014 00:31:59 +0000 (16:31 -0800)]
Rename String.prototype.contains to 'includes'.

Per TC39 Nov 2014 decison.

R=arv@chromium.org,yangguo@chromium.org
LOG=Y

Review URL: https://codereview.chromium.org/742963002

Cr-Commit-Position: refs/heads/master@{#25430}

9 years agoRemove Weak{Map,Set}.prototype.clear.
dslomov [Wed, 19 Nov 2014 21:52:49 +0000 (13:52 -0800)]
Remove Weak{Map,Set}.prototype.clear.

Per Nov 2014 TC39 decision.

R=adamk@chromium.org
LOG=Y

Review URL: https://codereview.chromium.org/739303002

Cr-Commit-Position: refs/heads/master@{#25429}

9 years agoReduce ErrorsFutureStrictReservedWords test to make it faster
arv [Wed, 19 Nov 2014 20:31:01 +0000 (12:31 -0800)]
Reduce ErrorsFutureStrictReservedWords test to make it faster

Since checking all possible combinations is taking so long this
reduces the test to test the odd cases (let, yield and static) as
well as a single ordinary case.

BUG=v8:3707
LOG=n

Review URL: https://codereview.chromium.org/724713004

Cr-Commit-Position: refs/heads/master@{#25428}

9 years agoRemoved unnecessary generalization of all fields in Map::GeneralizeRepresentation...
ishell [Wed, 19 Nov 2014 18:10:12 +0000 (10:10 -0800)]
Removed unnecessary generalization of all fields in Map::GeneralizeRepresentation() (introduced in r25082).

This generalization caused unnecessary map deprecation when the transition tree of the split map is full.

BUG=chromium:431807
LOG=N

Review URL: https://codereview.chromium.org/736953003

Cr-Commit-Position: refs/heads/master@{#25427}

9 years ago[turbofan]: delay ssa deconstruction in register allocator
dcarney [Wed, 19 Nov 2014 16:23:33 +0000 (08:23 -0800)]
[turbofan]: delay ssa deconstruction in register allocator

BUG=

Review URL: https://codereview.chromium.org/738853002

Cr-Commit-Position: refs/heads/master@{#25426}

9 years agoMIPS64: Prepare additonal code for turbofan landing.
Dusan Milosavljevic [Wed, 19 Nov 2014 16:08:23 +0000 (17:08 +0100)]
MIPS64: Prepare additonal code for turbofan landing.

TEST=
BUG=
R=paul.lind@imgtec.com

Review URL: https://codereview.chromium.org/735033002

Cr-Commit-Position: refs/heads/master@{#25425}

9 years agoMIPS64: Add turbofan support for mips64.
Dusan Milosavljevic [Wed, 19 Nov 2014 15:44:39 +0000 (16:44 +0100)]
MIPS64: Add turbofan support for mips64.

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

Review URL: https://codereview.chromium.org/732403002

Cr-Commit-Position: refs/heads/master@{#25424}

9 years agoAllow stepping into Object.observe handlers.
aandrey [Wed, 19 Nov 2014 15:16:30 +0000 (07:16 -0800)]
Allow stepping into Object.observe handlers.

BUG=chromium:432468
R=yangguo@chromium.org, adamk@chromium.org
LOG=N

Review URL: https://codereview.chromium.org/739523002

Cr-Commit-Position: refs/heads/master@{#25423}

9 years agoOne instead of two runtime calls when initializing regexp.
yangguo [Wed, 19 Nov 2014 14:13:44 +0000 (06:13 -0800)]
One instead of two runtime calls when initializing regexp.

R=ulan@chromium.org

Review URL: https://codereview.chromium.org/719403005

Cr-Commit-Position: refs/heads/master@{#25422}

9 years agoRevert "Re-land r25392 Use a stub in crankshaft for grow store arrays."
Michael Stanton [Wed, 19 Nov 2014 14:08:30 +0000 (15:08 +0100)]
Revert "Re-land r25392 Use a stub in crankshaft for grow store arrays."

Due to performance issue.
TBR=danno@chromium.org

Review URL: https://codereview.chromium.org/735323002

Cr-Commit-Position: refs/heads/master@{#25421}

9 years agoEnable "strict mode"; for debugger scripts
yurys [Wed, 19 Nov 2014 13:29:45 +0000 (05:29 -0800)]
Enable "strict mode"; for debugger scripts

BUG=v8:3708

Review URL: https://codereview.chromium.org/736583007

Cr-Commit-Position: refs/heads/master@{#25420}

9 years agoUpdate default trybots.
machenbach [Wed, 19 Nov 2014 13:00:06 +0000 (05:00 -0800)]
Update default trybots.

Add new gn trybot and remove nosnap dbg for being too slow.
It can be used on demand.

BUG=chromium:353487
LOG=n

Review URL: https://codereview.chromium.org/743643002

Cr-Commit-Position: refs/heads/master@{#25419}

9 years agoMIPS: Skip some tests on big-endian boards.
paul.lind [Wed, 19 Nov 2014 12:52:05 +0000 (04:52 -0800)]
MIPS: Skip some tests on big-endian boards.

Skip one hard fail, and two frequent flakes while we investigate.

BUG=
NOTRY=true

Review URL: https://codereview.chromium.org/728653004

Cr-Commit-Position: refs/heads/master@{#25418}

9 years agoPropertyDetails cleanup: NORMAL property type merged with FIELD.
ishell [Wed, 19 Nov 2014 11:45:17 +0000 (03:45 -0800)]
PropertyDetails cleanup: NORMAL property type merged with FIELD.

First step towards replacing PropertyType with two enums: {DATA,ACCESSOR} x {CONST,WRITABLE}.

Review URL: https://codereview.chromium.org/733253004

Cr-Commit-Position: refs/heads/master@{#25417}

9 years agoDo fewer encoding checks in FlatStringReader used in the JSON stringifier.
yangguo [Wed, 19 Nov 2014 11:16:47 +0000 (03:16 -0800)]
Do fewer encoding checks in FlatStringReader used in the JSON stringifier.

R=ishell@chromium.org

Review URL: https://codereview.chromium.org/740673002

Cr-Commit-Position: refs/heads/master@{#25416}

9 years agoFix Unhandled ReferenceError in debug-debugger.js
Yury Semikhatsky [Wed, 19 Nov 2014 10:25:21 +0000 (13:25 +0300)]
Fix Unhandled ReferenceError in debug-debugger.js

This fixes following exception in Sky on attempt to set a breakpoint
"Unhandled: Uncaught ReferenceError: break_point is not defined"
I think this happens in Sky but not in Chrome because Sky scripts are executed in strict mode.

BUG=None
LOG=N
R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/741683002

Cr-Commit-Position: refs/heads/master@{#25415}

9 years agoX87: Re-land r25392 Use a stub in crankshaft for grow store arrays.
chunyang.dai [Wed, 19 Nov 2014 09:28:10 +0000 (01:28 -0800)]
X87: Re-land r25392 Use a stub in crankshaft for grow store arrays.

port 47f55baeaff0d59366772db95b6c0328a216f611 (r25397)

original commit message:

  Re-land r25392 Use a stub in crankshaft for grow store arrays.

  Code was vulnerable to different evaluation order in Clang.

BUG=

Review URL: https://codereview.chromium.org/739823002

Cr-Commit-Position: refs/heads/master@{#25414}

9 years agofix typo in pipeline.cc
Dan Carney [Wed, 19 Nov 2014 08:55:13 +0000 (09:55 +0100)]
fix typo in pipeline.cc

TBR=bmeurer@chromium.org

BUG=

Review URL: https://codereview.chromium.org/743603003

Cr-Commit-Position: refs/heads/master@{#25413}

9 years ago[turbofan] Smartify typing of NumberToInt32 and NumberToUint32.
Benedikt Meurer [Wed, 19 Nov 2014 07:40:22 +0000 (08:40 +0100)]
[turbofan] Smartify typing of NumberToInt32 and NumberToUint32.

According to ES5 9.5 and 9.6, NaN, -inf, +inf, -0 and 0 all truncate to
zero for both ToInt32 and ToUint32, so we can be a lot smarter in the
typer, loosing less information upon truncation (i.e. x | 0 and x >>> 0).

R=jarin@chromium.org

Review URL: https://codereview.chromium.org/739743003

Cr-Commit-Position: refs/heads/master@{#25412}

9 years agoRevert "Forward declaration for Isolate / Platform in libplatform.h." and "Fixes...
Benedikt Meurer [Wed, 19 Nov 2014 07:27:48 +0000 (08:27 +0100)]
Revert "Forward declaration for Isolate / Platform in libplatform.h." and "Fixes d8 on windows following.".

This reverts commit 0f57ce8f1a3d6b4f3a4b65739d719b8467ff5d9c.
and 3245b8cb832f4a948aff9fb1c271cc08c4690d3b for breaking Chromium
compile.

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/743533003

Cr-Commit-Position: refs/heads/master@{#25411}

9 years agoImplement ES6 String.raw behind --harmony-templates
caitpotter88 [Wed, 19 Nov 2014 04:44:35 +0000 (20:44 -0800)]
Implement ES6 String.raw behind --harmony-templates

BUG=

Review URL: https://codereview.chromium.org/731573004

Cr-Commit-Position: refs/heads/master@{#25410}

9 years agoFixes d8 on windows following.
bulach [Tue, 18 Nov 2014 21:11:19 +0000 (13:11 -0800)]
Fixes d8 on windows following.

Broke by https://codereview.chromium.org/740493002/

BUG=
NOTRY=true
NOTREECHECKS=true

Review URL: https://codereview.chromium.org/734853004

Cr-Commit-Position: refs/heads/master@{#25409}

9 years agoForward declaration for Isolate / Platform in libplatform.h.
bulach [Tue, 18 Nov 2014 20:29:23 +0000 (12:29 -0800)]
Forward declaration for Isolate / Platform in libplatform.h.

Reduces dependencies on #include files, making it easier for other
build systems to include this library.

BUG=

Review URL: https://codereview.chromium.org/740493002

Cr-Commit-Position: refs/heads/master@{#25408}

9 years agoClasses: Expand test to cover strict runtime behavior
arv [Tue, 18 Nov 2014 19:00:08 +0000 (11:00 -0800)]
Classes: Expand test to cover strict runtime behavior

This tests that the extends expression is treated as strict at
runtime and not just at parse time.

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

Review URL: https://codereview.chromium.org/729323003

Cr-Commit-Position: refs/heads/master@{#25407}

9 years agoharmony-scoping: better error messages for let declarations in sloppy mode.
dslomov [Tue, 18 Nov 2014 18:51:20 +0000 (10:51 -0800)]
harmony-scoping: better error messages for let declarations in sloppy mode.

R=rossberg@chromium.org
BUG=v8:2198
LOG=N

Review URL: https://codereview.chromium.org/713413003

Cr-Commit-Position: refs/heads/master@{#25406}

9 years agoCopy unittests binary in android-sync.sh
cullinan [Tue, 18 Nov 2014 18:36:47 +0000 (10:36 -0800)]
Copy unittests binary in android-sync.sh

Without this change, `make android_arm.release.check` fails since the
unittests binary does not exist on the device.

BUG=v8:3695
LOG=

Review URL: https://codereview.chromium.org/722373003

Cr-Commit-Position: refs/heads/master@{#25405}

9 years agoMIPS: Re-land r25392 Use a stub in crankshaft for grow store arrays.
Balazs Kilvady [Tue, 18 Nov 2014 18:22:11 +0000 (10:22 -0800)]
MIPS: Re-land r25392 Use a stub in crankshaft for grow store arrays.

Port 47f55baeaff0d59366772db95b6c0328a216f611

Original commit message:
Code was vulnerable to different evaluation order in Clang.

BUG=
R=paul.lind@imgtec.com

Review URL: https://codereview.chromium.org/735783002

Patch from Balazs Kilvady <balazs.kilvady@imgtec.com>.

Cr-Commit-Position: refs/heads/master@{#25404}

9 years agoUse ZoneVector instead of ZoneList in the graph builder.
mstarzinger [Tue, 18 Nov 2014 17:48:44 +0000 (09:48 -0800)]
Use ZoneVector instead of ZoneList in the graph builder.

R=titzer@chromium.org

Review URL: https://codereview.chromium.org/742433002

Cr-Commit-Position: refs/heads/master@{#25403}

9 years agov8::String::Concat must not throw.
yangguo [Tue, 18 Nov 2014 15:51:05 +0000 (07:51 -0800)]
v8::String::Concat must not throw.

R=dcarney@chromium.org
BUG=chromium:420240
LOG=Y
API=v8::String::Concat may return empty handle on overflow.

Review URL: https://codereview.chromium.org/735763002

Cr-Commit-Position: refs/heads/master@{#25402}

9 years agoRemove ambiguous getter for operator value counts.
mstarzinger [Tue, 18 Nov 2014 15:45:22 +0000 (07:45 -0800)]
Remove ambiguous getter for operator value counts.

R=titzer@chromium.org

Review URL: https://codereview.chromium.org/735583003

Cr-Commit-Position: refs/heads/master@{#25401}

9 years agoFix disabling all break points from within the debug event callback.
yangguo [Tue, 18 Nov 2014 14:57:39 +0000 (06:57 -0800)]
Fix disabling all break points from within the debug event callback.

BUG=chromium:432493
LOG=Y

Review URL: https://codereview.chromium.org/728103008

Cr-Commit-Position: refs/heads/master@{#25400}

9 years agoFix loop information computation for floating loops.
mstarzinger [Tue, 18 Nov 2014 14:55:43 +0000 (06:55 -0800)]
Fix loop information computation for floating loops.

R=jarin@chromium.org
TEST=cctest/test-scheduler/LoopedFloatingDiamond3

Review URL: https://codereview.chromium.org/726953002

Cr-Commit-Position: refs/heads/master@{#25399}

9 years agoPrototype for a client-side performance trybot script.
Michael Achenbach [Tue, 18 Nov 2014 14:39:34 +0000 (15:39 +0100)]
Prototype for a client-side performance trybot script.

BUG=chromium:374740
LOG=n
TEST=tools/try_perf.py sunspider octane
R=phajdan.jr@chromium.org

Review URL: https://codereview.chromium.org/734403002

Cr-Commit-Position: refs/heads/master@{#25398}

9 years agoRe-land r25392 Use a stub in crankshaft for grow store arrays.
Michael Stanton [Tue, 18 Nov 2014 14:30:45 +0000 (15:30 +0100)]
Re-land r25392 Use a stub in crankshaft for grow store arrays.

Code was vulnerable to different evaluation order in Clang.

R=danno@chromium.org

Review URL: https://codereview.chromium.org/726693004

Cr-Commit-Position: refs/heads/master@{#25397}

9 years agoUse 16 instead of 4 for determining member offsets.
rickyz [Tue, 18 Nov 2014 14:09:23 +0000 (06:09 -0800)]
Use 16 instead of 4 for determining member offsets.

This avoids UBSan warnings about unaligned memory access.

BUG=

Review URL: https://codereview.chromium.org/734873002

Cr-Commit-Position: refs/heads/master@{#25396}

9 years agoFix include mistakes
alexanderk [Tue, 18 Nov 2014 13:46:57 +0000 (05:46 -0800)]
Fix include mistakes

1) Missing <stdio.h> while using fopen, fclose, fseek etc
2) platform-linux.cc includes <sys/types.h> wtice

BUG=v8:3691

Review URL: https://codereview.chromium.org/716123003

Cr-Commit-Position: refs/heads/master@{#25395}

9 years agoFix IC handler cache clearing for JSBuiltinsObject
Jakob Kummerow [Tue, 18 Nov 2014 13:35:40 +0000 (14:35 +0100)]
Fix IC handler cache clearing for JSBuiltinsObject

The clearing mechanism relies on comparing the cached handler with the installed handler. If we try to preserve monomorphism by pretending that the IC was in uninitialized state, then it will go premonomorphic first, which means on the next miss there's no installed handler available to compare against. Instead, pretend it was premonomorphic, so the comparison will happen right away, will fail as it should, and the cached handler will be cleared.

Thanks to Weiliang for starting the investigation that led to this.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/730383002

Cr-Commit-Position: refs/heads/master@{#25394}

9 years agoRevert "Use a stub in crankshaft for grow store arrays."
Michael Stanton [Tue, 18 Nov 2014 13:07:57 +0000 (14:07 +0100)]
Revert "Use a stub in crankshaft for grow store arrays."

This reverts commit d40204f84c0f6fad171ba8ace587f902bd5b14bc.

TBR=danno@chromium.org

Review URL: https://codereview.chromium.org/735653003

Cr-Commit-Position: refs/heads/master@{#25393}

9 years agoUse a stub in crankshaft for grow store arrays.
Michael Stanton [Tue, 18 Nov 2014 12:32:05 +0000 (13:32 +0100)]
Use a stub in crankshaft for grow store arrays.

We were deopting without learning anything.

BUG=v8:3417
LOG=N
R=danno@chromium.org

Review URL: https://codereview.chromium.org/368263003

Cr-Commit-Position: refs/heads/master@{#25392}

9 years agoUse FlatStringReader in JSON stringifier.
yangguo [Tue, 18 Nov 2014 12:03:25 +0000 (04:03 -0800)]
Use FlatStringReader in JSON stringifier.

R=ishell@chromium.org

Review URL: https://codereview.chromium.org/736543003

Cr-Commit-Position: refs/heads/master@{#25391}

9 years ago[turbofan] Push JSToNumber conversions into Phis.
Benedikt Meurer [Tue, 18 Nov 2014 11:34:09 +0000 (12:34 +0100)]
[turbofan] Push JSToNumber conversions into Phis.

This essentially performs the following transformation

  JSToNumber(phi(x1,...,xn,control):primitive)
    => phi(JSToNumber(x1),...,JSToNumber(xn),control):number

which is similar to what we already do for JSToBoolean.

TEST=mjsunit/asm
R=svenpanne@chromium.org
BUG=

Review URL: https://codereview.chromium.org/732463003

Cr-Commit-Position: refs/heads/master@{#25390}

9 years agoAllow stepping into Promise handlers.
aandrey [Tue, 18 Nov 2014 09:50:12 +0000 (01:50 -0800)]
Allow stepping into Promise handlers.

BUG=chromium:432468
R=yangguo@chromium.org, adamk@chromium.org, arv@chromium.org
LOG=N

Review URL: https://codereview.chromium.org/734163002

Cr-Commit-Position: refs/heads/master@{#25389}

9 years ago[turbofan] Avoid useless sign extension after sign extended load.
Benedikt Meurer [Tue, 18 Nov 2014 09:27:47 +0000 (10:27 +0100)]
[turbofan] Avoid useless sign extension after sign extended load.

TEST=unittests
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/735623002

Cr-Commit-Position: refs/heads/master@{#25388}

9 years agoHydrogen should recognize literal smi arrays as fast literals.
Michael Stanton [Tue, 18 Nov 2014 09:27:34 +0000 (10:27 +0100)]
Hydrogen should recognize literal smi arrays as fast literals.

R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/726423002

Cr-Commit-Position: refs/heads/master@{#25387}

9 years ago[x64] Recognize MOVSXBL.
Benedikt Meurer [Tue, 18 Nov 2014 09:11:29 +0000 (10:11 +0100)]
[x64] Recognize MOVSXBL.

TEST=mjsunit/asm
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/728103007

Cr-Commit-Position: refs/heads/master@{#25386}

9 years ago[x64] Recognize MOVSXWL.
Benedikt Meurer [Tue, 18 Nov 2014 07:51:41 +0000 (08:51 +0100)]
[x64] Recognize MOVSXWL.

Also add some debug code to verify correct zero extension of 32-bit
moves.

TEST=mjsunit/asm
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/736623002

Cr-Commit-Position: refs/heads/master@{#25385}

9 years agoAdd a cctest for using a C++ FunctionCallback as an Object.observe observer
Adam Klein [Mon, 17 Nov 2014 20:29:00 +0000 (12:29 -0800)]
Add a cctest for using a C++ FunctionCallback as an Object.observe observer

R=rossberg@chromium.org
BUG=v8:3076
LOG=n

Review URL: https://codereview.chromium.org/733483003

Cr-Commit-Position: refs/heads/master@{#25384}

9 years agoharmony-scoping: Implement debugger support for script scope.
dslomov [Mon, 17 Nov 2014 17:57:56 +0000 (09:57 -0800)]
harmony-scoping: Implement debugger support for script scope.

We add a new ScopeType, ScopeType.Script. The scope with
ScopeType.Script is always present in the scope chain (ScopeIterator
fakes it if neededi - i.e. if ScriptContext for a script has not been
allocated since that script has no lexical declarations).
ScriptScope reflects ScriptContextTable.

R=yurys@chromium.org,yangguo@chromium.org
BUG=v8:3690
LOG=N

Review URL: https://codereview.chromium.org/726643002

Cr-Commit-Position: refs/heads/master@{#25383}

9 years ago[turbofan] add ForTesting to pipeline entry points that are for testing only.
Dan Carney [Mon, 17 Nov 2014 14:46:41 +0000 (15:46 +0100)]
[turbofan] add ForTesting to pipeline entry points that are for testing only.

R=bmeurer@chromium.org

BUG=

Review URL: https://codereview.chromium.org/727373002

Cr-Commit-Position: refs/heads/master@{#25382}

9 years ago[turbofan] More useful typing for And, Or and Shr.
jarin [Mon, 17 Nov 2014 13:55:27 +0000 (05:55 -0800)]
[turbofan] More useful typing for And, Or and Shr.

BUG=

Review URL: https://codereview.chromium.org/714413004

Cr-Commit-Position: refs/heads/master@{#25381}

9 years agoExpose internal properties of map/set iterators via mirrors.
Andrey Adaikin [Mon, 17 Nov 2014 12:48:11 +0000 (15:48 +0300)]
Expose internal properties of map/set iterators via mirrors.

R=yangguo@chromium.org, vsevik
LOG=Y

Committed: https://code.google.com/p/v8/source/detail?r=d5f5d38f73f43eba9658d91ffbe511af8c340d78

Review URL: https://codereview.chromium.org/710273002

Cr-Commit-Position: refs/heads/master@{#25380}

9 years ago[turbofan] move register allocation phases to pipeline
Dan Carney [Mon, 17 Nov 2014 12:36:58 +0000 (13:36 +0100)]
[turbofan] move register allocation phases to pipeline

BUG=
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/727323002

Cr-Commit-Position: refs/heads/master@{#25379}

9 years agoReland "Soft fail for invalid cache data."
yangguo [Mon, 17 Nov 2014 12:16:27 +0000 (04:16 -0800)]
Reland "Soft fail for invalid cache data."

Review URL: https://codereview.chromium.org/733023003

Cr-Commit-Position: refs/heads/master@{#25378}

9 years ago[turbofan] Smartify the GraphReducer.
Benedikt Meurer [Mon, 17 Nov 2014 12:12:24 +0000 (13:12 +0100)]
[turbofan] Smartify the GraphReducer.

Don't use the generic algorithm, but instead start going into the
direction of ControlReducer, using a stack plus a revisit queue to
not miss any more possibilities for reductions anymore.

TEST=cctest,unittests
R=dcarney@chromium.org

Committed: https://chromium.googlesource.com/v8/v8/+/f047507370634155113d78685372630a230613cf

Committed: https://chromium.googlesource.com/v8/v8/+/6e148989a4227a5290a7f8ca72c71f5740870afe

Review URL: https://codereview.chromium.org/726513002

Cr-Commit-Position: refs/heads/master@{#25377}

9 years ago[turbofan] Fix pushing of JSToBooleans into Phis.
Benedikt Meurer [Mon, 17 Nov 2014 12:10:28 +0000 (13:10 +0100)]
[turbofan] Fix pushing of JSToBooleans into Phis.

Now we actually implement it the way it is meant to be, that is:

  JSToBoolean(Phi(x1,...,xn):primitive)
    => Phi(JSToBoolean(x1),...,JSToBoolean(xn)):boolean

This also fixes the endless recursion within JSTypedLowering when
the GraphReducer does all possible reductions instead of using the
generic algorithm.

R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/730043002

Cr-Commit-Position: refs/heads/master@{#25376}

9 years ago[turbofan] Factor out common code between ReduceSelect and ReduceBranch in ControlRed...
Ben L. Titzer [Mon, 17 Nov 2014 11:34:53 +0000 (12:34 +0100)]
[turbofan] Factor out common code between ReduceSelect and ReduceBranch in ControlReducer.

R=bmeurer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/734593003

Cr-Commit-Position: refs/heads/master@{#25375}

9 years agoFix Windows build after 25373
Daniel Clifford [Mon, 17 Nov 2014 11:26:27 +0000 (12:26 +0100)]
Fix Windows build after 25373

R=bmeurer@chromium.org
TBR=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/733783003

Cr-Commit-Position: refs/heads/master@{#25374}

9 years agoGeneralize ScaledWithOffsetMatcher to support 64 bits
Daniel Clifford [Mon, 17 Nov 2014 10:50:56 +0000 (11:50 +0100)]
Generalize ScaledWithOffsetMatcher to support 64 bits

Preparation for supporting more addressing modes in instruction selection.

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/729853005

Cr-Commit-Position: refs/heads/master@{#25373}

9 years agoRemove left-over generated runtime test from mjsunit.
Michael Starzinger [Mon, 17 Nov 2014 10:50:44 +0000 (11:50 +0100)]
Remove left-over generated runtime test from mjsunit.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/734633002

Cr-Commit-Position: refs/heads/master@{#25372}

9 years agoFix one more missing c0_ < 0 check in scanner
Jakob Kummerow [Mon, 17 Nov 2014 09:43:17 +0000 (10:43 +0100)]
Fix one more missing c0_ < 0 check in scanner

BUG=chromium:433766
LOG=n
R=jarin@chromium.org

Review URL: https://codereview.chromium.org/731953003

Cr-Commit-Position: refs/heads/master@{#25371}

9 years agoFix concurrent sweeping in predictable mode and bring --concurrent-sweeping flag...
Hannes Payer [Mon, 17 Nov 2014 09:39:42 +0000 (10:39 +0100)]
Fix concurrent sweeping in predictable mode and bring --concurrent-sweeping flag back.

BUG=
R=jochen@chromium.org

Review URL: https://codereview.chromium.org/726363002

Cr-Commit-Position: refs/heads/master@{#25370}

9 years agoAdded --trace_idle_notification_verbose which prints out the idle notificatino heap...
Hannes Payer [Mon, 17 Nov 2014 09:16:14 +0000 (10:16 +0100)]
Added --trace_idle_notification_verbose which prints out the idle notificatino heap state.

BUG=
R=jochen@chromium.org

Review URL: https://codereview.chromium.org/732473002

Cr-Commit-Position: refs/heads/master@{#25369}

9 years ago[turbofan] Remove int32 narrowing during typed lowering.
Jaroslav Sevcik [Mon, 17 Nov 2014 09:04:30 +0000 (10:04 +0100)]
[turbofan] Remove int32 narrowing during typed lowering.

With Int32Add we lose the int/uint distinction, so later, in simplified lowering we can make a wrong decision. E.g., see the attached test case, where we lower NumberAdd -> Int32Add because inputs are Uint32, but during simplified lowering we change the inputs to Int32, so we get a wrong result.

Simplified lowering will lower the NumberAdd operations anyway, so we should lose performance.

BUG=
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/721723004

Cr-Commit-Position: refs/heads/master@{#25368}

9 years agoReland "Optimize function across closures." (again).
yangguo [Mon, 17 Nov 2014 08:42:45 +0000 (00:42 -0800)]
Reland "Optimize function across closures." (again).

Review URL: https://codereview.chromium.org/707463002

Cr-Commit-Position: refs/heads/master@{#25367}

9 years agoClasses: Add support for stepping through default constructors
arv [Sat, 15 Nov 2014 19:48:32 +0000 (11:48 -0800)]
Classes: Add support for stepping through default constructors

If a class extends another class and it doesn't provide a constructor,
one is created for them. We therefore need to ensure that stepping into
the constructor steps into the super class constructor.

BUG=v8:3674
LOG=Y
R=dslomov@chromium.org, aandrey , yurys

Review URL: https://codereview.chromium.org/725983002

Cr-Commit-Position: refs/heads/master@{#25366}

9 years agoLet git ignore the third_party dir to avoid unnecessary clean-ups.
machenbach [Sat, 15 Nov 2014 19:33:04 +0000 (11:33 -0800)]
Let git ignore the third_party dir to avoid unnecessary clean-ups.

BUG=chromium:433273
LOG=n
NOTRY=true

Review URL: https://codereview.chromium.org/728983004

Cr-Commit-Position: refs/heads/master@{#25365}

9 years agoThrow as per spec when modifying an Array with builtin methods
adamk [Fri, 14 Nov 2014 19:42:03 +0000 (11:42 -0800)]
Throw as per spec when modifying an Array with builtin methods

BUG=v8:3684
LOG=n

Review URL: https://codereview.chromium.org/726773002

Cr-Commit-Position: refs/heads/master@{#25364}

9 years agoFix desugaring of let bindings in for loops to handle continue properly
adamk [Fri, 14 Nov 2014 19:32:53 +0000 (11:32 -0800)]
Fix desugaring of let bindings in for loops to handle continue properly

This requires putting the original loop's body inside an inner for loop (with
the same labels as the original loop) and re-binding the temp variables in its
"next" expression. A second flag is added to the desugared code to ensure the
loop body executes at most once per loop.

BUG=v8:3683
LOG=y

Review URL: https://codereview.chromium.org/720863002

Cr-Commit-Position: refs/heads/master@{#25363}

9 years agoImplement ES6 Template Literals
caitpotter88 [Fri, 14 Nov 2014 18:53:41 +0000 (10:53 -0800)]
Implement ES6 Template Literals

BUG=v8:3230

Review URL: https://codereview.chromium.org/663683006

Cr-Commit-Position: refs/heads/master@{#25362}

9 years ago[turbofan] refactor pipeline to use hydrogen like Run calls
dcarney [Fri, 14 Nov 2014 16:44:38 +0000 (08:44 -0800)]
[turbofan] refactor pipeline to use hydrogen like Run calls

BUG=

Review URL: https://codereview.chromium.org/727733002

Cr-Commit-Position: refs/heads/master@{#25361}

9 years agoClasses: Implement correct name binding
arv [Fri, 14 Nov 2014 15:05:05 +0000 (07:05 -0800)]
Classes: Implement correct name binding

Named class declarations and class expression have a const binding for
the name that is in TDZ for the extends expression.

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

Review URL: https://codereview.chromium.org/722793005

Cr-Commit-Position: refs/heads/master@{#25360}

9 years agoExtend typed lowering to cover JSStrictEqual on differing types.
Michael Starzinger [Fri, 14 Nov 2014 14:25:08 +0000 (15:25 +0100)]
Extend typed lowering to cover JSStrictEqual on differing types.

R=rossberg@chromium.org, titzer@chromium.org
TEST=unittests/JSTypedLoweringTest.JSStrictEqualWithTheHole

Review URL: https://codereview.chromium.org/722223003

Cr-Commit-Position: refs/heads/master@{#25359}

9 years agoImplement 'setVariableValue' for debugger block scopes.
Dmitry Lomov [Fri, 14 Nov 2014 13:39:06 +0000 (14:39 +0100)]
Implement 'setVariableValue' for debugger block scopes.

R=aandrey@chromium.org, rossberg@chromium.org, yurys@chromium.org
BUG=v8:3690
LOG=N

Review URL: https://codereview.chromium.org/732543002

Cr-Commit-Position: refs/heads/master@{#25358}

9 years agoRemove AstConstructionVisitor/AstNullVisitor
Andy Wingo [Fri, 14 Nov 2014 13:13:09 +0000 (14:13 +0100)]
Remove AstConstructionVisitor/AstNullVisitor

R=svenpanne@chromium.org
BUG=

Review URL: https://codereview.chromium.org/679943004

Cr-Commit-Position: refs/heads/master@{#25357}

9 years agoRevert "[turbofan] Smartify the GraphReducer."
Benedikt Meurer [Fri, 14 Nov 2014 11:48:37 +0000 (12:48 +0100)]
Revert "[turbofan] Smartify the GraphReducer."

This reverts commit 6e148989a4227a5290a7f8ca72c71f5740870afe for
breaking Massive/Embenchen.

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/727743002

Cr-Commit-Position: refs/heads/master@{#25356}

9 years agoDo not do field type tracking with GC stress.
Ben L. Titzer [Fri, 14 Nov 2014 11:36:26 +0000 (12:36 +0100)]
Do not do field type tracking with GC stress.

R=verwaest@chromium.org
BUG=

Review URL: https://codereview.chromium.org/718393004

Cr-Commit-Position: refs/heads/master@{#25355}

9 years agoMake TurboFan inlining work without deoptimization.
Michael Starzinger [Fri, 14 Nov 2014 10:34:44 +0000 (11:34 +0100)]
Make TurboFan inlining work without deoptimization.

R=titzer@chromium.org

Review URL: https://codereview.chromium.org/726823002

Cr-Commit-Position: refs/heads/master@{#25354}

9 years agoRemove arguments arity check in mjsunit because it vastly slows down the tests.
Ben L. Titzer [Fri, 14 Nov 2014 10:30:12 +0000 (11:30 +0100)]
Remove arguments arity check in mjsunit because it vastly slows down the tests.

R=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/724133004

Cr-Commit-Position: refs/heads/master@{#25353}

9 years agoRevert of Whitespace change to test CQ. (patchset #1 id:1 of https://codereview.chrom...
machenbach [Fri, 14 Nov 2014 09:36:07 +0000 (01:36 -0800)]
Revert of Whitespace change to test CQ. (patchset #1 id:1 of https://codereview.chromium.org/725873002/)

Reason for revert:
Test revert.

Original issue's description:
> Whitespace change to test CQ.
>
> TBR=machenbach@chromium.org
>
> Committed: https://chromium.googlesource.com/v8/v8/+/42340bdfbd8b4f002257e55464f2e0b81c6ea424

TBR=
NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/725893002

Cr-Commit-Position: refs/heads/master@{#25352}

9 years agoWhitespace change to test CQ.
machenbach [Fri, 14 Nov 2014 09:32:03 +0000 (01:32 -0800)]
Whitespace change to test CQ.

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/725873002

Cr-Commit-Position: refs/heads/master@{#25351}

9 years ago[turbofan] small cleanups to aid register allocator debugging
Dan Carney [Fri, 14 Nov 2014 08:56:56 +0000 (09:56 +0100)]
[turbofan] small cleanups to aid register allocator debugging

BUG=
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/727693002

Cr-Commit-Position: refs/heads/master@{#25350}

9 years agoFix releases script after git migration.
Michael Achenbach [Fri, 14 Nov 2014 08:38:57 +0000 (09:38 +0100)]
Fix releases script after git migration.

BUG=chromium:410721
LOG=n
TBR=tandrii@chromium.org
TEST=script_test.py

Review URL: https://codereview.chromium.org/725073002

Cr-Commit-Position: refs/heads/master@{#25349}

9 years agoMove feedback slot allocation to post-pass
Andy Wingo [Fri, 14 Nov 2014 08:21:13 +0000 (09:21 +0100)]
Move feedback slot allocation to post-pass

R=mvstanton@chromium.org, svenpanne@chromium.org

Review URL: https://codereview.chromium.org/670953003

Cr-Commit-Position: refs/heads/master@{#25348}

9 years ago[turbofan] Optimize remainder of integer division by unknown power of two.
Benedikt Meurer [Fri, 14 Nov 2014 08:21:04 +0000 (09:21 +0100)]
[turbofan] Optimize remainder of integer division by unknown power of two.

Drive-by-Fix: minint % 0 was broken on ARM, but we didn't notice because
there was no test covering that case...

TEST=msjunit
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/727673002

Cr-Commit-Position: refs/heads/master@{#25347}

9 years agoAllow idle notification when incremental marking is turned off.
Hannes Payer [Fri, 14 Nov 2014 08:13:03 +0000 (09:13 +0100)]
Allow idle notification when incremental marking is turned off.

BUG=
R=jochen@chromium.org

Review URL: https://codereview.chromium.org/725583005

Cr-Commit-Position: refs/heads/master@{#25346}