platform/upstream/v8.git
10 years agoEnable concurrent sweeping. Added some extra debugging checks for concurrent sweeping.
hpayer@chromium.org [Tue, 21 Jan 2014 19:30:27 +0000 (19:30 +0000)]
Enable concurrent sweeping. Added some extra debugging checks for concurrent sweeping.

BUG=
R=titzer@chromium.org

Review URL: https://codereview.chromium.org/138903009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18722 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMIPS: Stay in fast enum case if the empty_slow_element_dictionary is used.
plind44@gmail.com [Tue, 21 Jan 2014 17:48:57 +0000 (17:48 +0000)]
MIPS: Stay in fast enum case if the empty_slow_element_dictionary is used.

Port r18705 (78b4ea3)

Original commit message:
This fixes slow for-in over frozen objects.

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/144023002

Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18721 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRevert "Unify calling to GenerateFastApiCallBody before stubbing it"
mvstanton@chromium.org [Tue, 21 Jan 2014 17:04:17 +0000 (17:04 +0000)]
Revert "Unify calling to GenerateFastApiCallBody before stubbing it"

This reverts commit r18714 for breaking webkit tests with an assert.

TBR=dcarney@chromium.org

Review URL: https://codereview.chromium.org/144143002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18720 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoZonify types in compiler frontend
rossberg@chromium.org [Tue, 21 Jan 2014 16:22:52 +0000 (16:22 +0000)]
Zonify types in compiler frontend

Clean up some zone/isolate handling in AST and its visitors on the way.

(Based on https://codereview.chromium.org/103743004/)

R=jkummerow@chromium.org, titzer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/102563004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18719 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoOnly refill free lists for spaces that are being swept
jochen@chromium.org [Tue, 21 Jan 2014 16:19:04 +0000 (16:19 +0000)]
Only refill free lists for spaces that are being swept

BUG=v8:3104
R=hpayer@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/143953005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18718 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoArray constructor shouldn't require a Cell, just an AllocationSite.
mvstanton@chromium.org [Tue, 21 Jan 2014 16:04:39 +0000 (16:04 +0000)]
Array constructor shouldn't require a Cell, just an AllocationSite.

The Array constructor has a needless dependency on an input argument
that is a Cell. It uses this to walk through to an AllocationSite.
The dependency hampers future work. Instead, pass the AllocationSite
as input to the Array constructor.

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/140963004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18716 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoUnify calling to GenerateFastApiCallBody before stubbing it
dcarney@chromium.org [Tue, 21 Jan 2014 15:06:46 +0000 (15:06 +0000)]
Unify calling to GenerateFastApiCallBody before stubbing it

R=verwaest@chromium.org

BUG=

Review URL: https://codereview.chromium.org/142973005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18714 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoSilence another VS warning
rossberg@chromium.org [Tue, 21 Jan 2014 14:41:48 +0000 (14:41 +0000)]
Silence another VS warning

R=titzer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/143973004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18712 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRetry landing "Implement zone-allocated types"
rossberg@chromium.org [Tue, 21 Jan 2014 14:14:12 +0000 (14:14 +0000)]
Retry landing "Implement zone-allocated types"

Works around apparent scoping bug in VS, the only change to before being a method rename in the test suite:

--- a/test/cctest/test-types.cc
+++ b/test/cctest/test-types.cc
@@ -153,7 +153,7 @@ struct ZoneRep {
     return reinterpret_cast<ZoneList<ZoneType*>*>(AsTagged(t));
   }

-  static Zone* Region(Zone* zone, Isolate* isolate) { return zone; }
+  static Zone* ToRegion(Zone* zone, Isolate* isolate) { return zone; }
 };

@@ -168,7 +168,7 @@ struct HeapRep {
   static Object* AsConstant(Handle<Type> t) { return Box::cast(*t)->value(); }
   static FixedArray* AsUnion(Handle<Type> t) { return FixedArray::cast(*t); }

-  static Isolate* Region(Zone* zone, Isolate* isolate) { return isolate; }
+  static Isolate* ToRegion(Zone* zone, Isolate* isolate) { return isolate; }
 };

@@ -183,7 +183,7 @@ struct Tests : Rep {
       isolate(CcTest::i_isolate()),
       scope(isolate),
       zone(isolate),
-      T(Rep::Region(&zone, isolate), isolate) {
+      T(Rep::ToRegion(&zone, isolate), isolate) {
   }

   static void CheckEqual(TypeHandle type1, TypeHandle type2) {

R=titzer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/143693003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18711 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRename MarkCompactCollector::StealMemoryFromSweeperThreads to RefillFreeLists
jochen@chromium.org [Tue, 21 Jan 2014 12:48:10 +0000 (12:48 +0000)]
Rename MarkCompactCollector::StealMemoryFromSweeperThreads to RefillFreeLists

The swept memory is now collected on the foreground thread, so it's not
stealing the memory from the threads anymore

BUG=v8:3104
R=hpayer@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/143463005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18710 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoAdd hydrogen support for ArrayPop, and remove the handwritten call stubs.
verwaest@chromium.org [Tue, 21 Jan 2014 12:42:24 +0000 (12:42 +0000)]
Add hydrogen support for ArrayPop, and remove the handwritten call stubs.
BUG=
R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/137783023

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18709 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoAdd infrastructure for skipping tests in GC stress mode.
machenbach@chromium.org [Tue, 21 Jan 2014 12:41:25 +0000 (12:41 +0000)]
Add infrastructure for skipping tests in GC stress mode.

Also move the GC stress configuration from the buildbot to the test runner.

BUG=
R=jkummerow@chromium.org, mvstanton@chromium.org

Review URL: https://codereview.chromium.org/141653008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18708 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMove the temporary private freelist closer to where it's used
jochen@chromium.org [Tue, 21 Jan 2014 11:48:51 +0000 (11:48 +0000)]
Move the temporary private freelist closer to where it's used

BUG=v8:3104
R=hpayer@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/143193002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18707 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMove the management of the already swept pages to MarkCompactCollector
jochen@chromium.org [Tue, 21 Jan 2014 11:41:18 +0000 (11:41 +0000)]
Move the management of the already swept pages to MarkCompactCollector

That way, MCC doesn't need to know about the state the threads are in to
steal free memory from them.

BUG=v8:3104
R=hpayer@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/143283002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18706 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoStay in fast enum case if the empty_slow_element_dictionary is used.
verwaest@chromium.org [Tue, 21 Jan 2014 11:20:11 +0000 (11:20 +0000)]
Stay in fast enum case if the empty_slow_element_dictionary is used.
This fixes slow for-in over frozen objects.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/135903014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18705 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFix invalid type assumption in GenerateNumberToString().
bmeurer@chromium.org [Tue, 21 Jan 2014 10:35:20 +0000 (10:35 +0000)]
Fix invalid type assumption in GenerateNumberToString().

We don't have any type feedback in %_NumberToString(), so it's
a bug to assume Type::Number here.

R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/143893002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18704 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoPrepare push to trunk. Now working on version 3.24.21.
machenbach@chromium.org [Tue, 21 Jan 2014 09:01:39 +0000 (09:01 +0000)]
Prepare push to trunk.  Now working on version 3.24.21.

R=jkummerow@chromium.org
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/134733005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18701 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years ago[Sheriff] Temporarily mark array-literal-feedback as flaky for GC stress.
machenbach@chromium.org [Tue, 21 Jan 2014 07:58:54 +0000 (07:58 +0000)]
[Sheriff] Temporarily mark array-literal-feedback as flaky for GC stress.

The test is blocking the v8 lkgr. It will be unmarked again after there is infrastructure to disable it on GC stress only.

BUG=
TBR=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/143463004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18700 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMIPS: Turn ArrayPush into a stub specialized on the elements kind and argc.
palfia@homejinni.com [Mon, 20 Jan 2014 23:08:52 +0000 (23:08 +0000)]
MIPS: Turn ArrayPush into a stub specialized on the elements kind and argc.

Port r18696 (6e4b51b4)

BUG=
R=gergely@homejinni.com

Review URL: https://codereview.chromium.org/143663002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18699 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years ago[Sheriff] Temporarily mark two mjsunit tests as flaky.
machenbach@chromium.org [Mon, 20 Jan 2014 19:05:31 +0000 (19:05 +0000)]
[Sheriff] Temporarily mark two mjsunit tests as flaky.

The tests are blocking the v8 lkgr. They will be unmarked again after there is infrastructure to disable them on GC stress only.

TBR=mvstanton@chromium.org
BUG=

Review URL: https://codereview.chromium.org/139343008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18698 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFix representation requirement in HReturn.
titzer@chromium.org [Mon, 20 Jan 2014 19:00:11 +0000 (19:00 +0000)]
Fix representation requirement in HReturn.

BUG=
R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/143523002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18697 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoTurn ArrayPush into a stub specialized on the elements kind and argc.
verwaest@chromium.org [Mon, 20 Jan 2014 17:09:24 +0000 (17:09 +0000)]
Turn ArrayPush into a stub specialized on the elements kind and argc.

BUG=
R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/143213003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18696 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoUpdate Isolate::RequestGarbageCollection after r18691
jochen@chromium.org [Mon, 20 Jan 2014 15:44:03 +0000 (15:44 +0000)]
Update Isolate::RequestGarbageCollection after r18691

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

Review URL: https://codereview.chromium.org/143403002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18695 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoEnable incremental marking when --expose-gc is turned on.
hpayer@chromium.org [Mon, 20 Jan 2014 14:14:07 +0000 (14:14 +0000)]
Enable incremental marking when --expose-gc is turned on.

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

Review URL: https://codereview.chromium.org/139343007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18691 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoExplicitly initialize MarkCompactCollector and ExternalStringTable
jochen@chromium.org [Mon, 20 Jan 2014 11:57:56 +0000 (11:57 +0000)]
Explicitly initialize MarkCompactCollector and ExternalStringTable

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

Review URL: https://codereview.chromium.org/143063003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18689 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoDoubleToIStub can't use ip on armv6, because the ubfx impl will clobber it
rmcilroy@chromium.org [Mon, 20 Jan 2014 11:30:48 +0000 (11:30 +0000)]
DoubleToIStub can't use ip on armv6, because the ubfx impl will clobber it

This previous change broke DoubleToIStub on armv6:
https://code.google.com/p/v8/source/detail?r=16322

The problem is that DoubleToIStub::Generate assumed that it could safely use the ip register, but on armv6 the ubfx implementation will clobber any previous value stored there.  So instead, pick another register.

Test case:
for (var i=0; i<2; i++) {
v = 4294967295;
v &= -2;
print(v)
}

This should print -2 twice, but on armv6 without this patch, it prints -2 followed by 2046.

This problem causes sunspider's bitops-nsieve-bit, crypto-md5 and crypto-sha1 tests to generate incorrect results (but the results are not checked for validity in sunspider-1.0 as available in chromium, but are checked and reported as incorrect in sunspider-1.0.2).

Thanks to Tomasz Kilarski for helping out with this.

R=bmeurer@chromium.org, rmcilroy@chromium.org

Review URL: https://codereview.chromium.org/131823004

Patch from Mostyn Bramley-Moore <mostynb@opera.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18688 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRevert "ES6: Implement Object.setPrototypeOf"
dslomov@chromium.org [Mon, 20 Jan 2014 10:59:07 +0000 (10:59 +0000)]
Revert "ES6: Implement Object.setPrototypeOf"

This reverts commit r18685 for breaking WebKit tests.

TBR=arv@chromium.org

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18687 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoES6: Implement Object.setPrototypeOf
dslomov@chromium.org [Mon, 20 Jan 2014 10:38:01 +0000 (10:38 +0000)]
ES6: Implement Object.setPrototypeOf

http://people.mozilla.org/~jorendorff/es6-draft.html#sec-object.setprototypeof

This just exposes the internal %SetPrototype and adds all the required
type checks as specified.

BUG=v8:2675
LOG=Y
R=dslomov@chromium.org, rossberg@chromium.org

Review URL: https://codereview.chromium.org/141913002

Patch from Erik Arvidsson <arv@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18685 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoString:WriteUtf8: Add REPLACE_INVALID_UTF8 option
dcarney@chromium.org [Mon, 20 Jan 2014 09:52:54 +0000 (09:52 +0000)]
String:WriteUtf8: Add REPLACE_INVALID_UTF8 option

This patch makes String::WriteUtf8 replace invalid code points (i.e. unmatched
surrogates) with the unicode replacement character when REPLACE_INVALID_UTF8 is
set.  This is done to avoid creating invalid UTF-8 output which can lead to
compatibility issues with software requiring valid UTF-8 inputs (e.g. the
WebSocket protocol requires valid UTF-8 and terminates connections when invalid
UTF-8 is encountered).

R=dcarney@chromium.org

BUG=

Review URL: https://codereview.chromium.org/121173009

Patch from Felix Geisendörfer <haimuiba@gmail.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18683 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoWe no longer need to recover type cells from the oracle.
mvstanton@chromium.org [Mon, 20 Jan 2014 09:48:05 +0000 (09:48 +0000)]
We no longer need to recover type cells from the oracle.

We only need the values within them. Function calls to Array from optimized code needed the cell in the past, but no longer.

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/141893002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18682 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoPrepare push to trunk. Now working on version 3.24.20.
machenbach@chromium.org [Mon, 20 Jan 2014 07:57:07 +0000 (07:57 +0000)]
Prepare push to trunk.  Now working on version 3.24.20.

R=jkummerow@chromium.org
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/143093002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18677 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoDe-virtualize GenerateNameCheck.
svenpanne@chromium.org [Mon, 20 Jan 2014 07:05:23 +0000 (07:05 +0000)]
De-virtualize GenerateNameCheck.

Currently there are two kinds of "vtables" used in the load/store stub
compilers: kind() and C++'s own vtables. This is a bit confusing, and
this is a step towards simplifying things.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/141763002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18676 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoReplace movq with movp for X64 when the operand size is kPointerSize
haitao.feng@intel.com [Mon, 20 Jan 2014 04:59:40 +0000 (04:59 +0000)]
Replace movq with movp for X64 when the operand size is kPointerSize

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/105503006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18675 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoUse kRegisterSize when passing arguments to the C++ runtime for X64
haitao.feng@intel.com [Mon, 20 Jan 2014 01:52:21 +0000 (01:52 +0000)]
Use kRegisterSize when passing arguments to the C++ runtime for X64

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/137443004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18674 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFix compilation with latest Xcode toolchain.
bmeurer@chromium.org [Sun, 19 Jan 2014 12:24:25 +0000 (12:24 +0000)]
Fix compilation with latest Xcode toolchain.

TBR=jarin@chromium.org

Review URL: https://codereview.chromium.org/142563002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18673 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRemove special String.fromCharCode call IC support.
verwaest@chromium.org [Fri, 17 Jan 2014 16:08:54 +0000 (16:08 +0000)]
Remove special String.fromCharCode call IC support.

BUG=
R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/141773002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18671 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years ago--trace-hydrogen fix
ishell@chromium.org [Fri, 17 Jan 2014 15:44:40 +0000 (15:44 +0000)]
--trace-hydrogen fix

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/141333006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18670 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRemove the special MathFloor / MathAbs call handlers.
verwaest@chromium.org [Fri, 17 Jan 2014 14:21:01 +0000 (14:21 +0000)]
Remove the special MathFloor / MathAbs call handlers.

BUG=
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/141733002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18669 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMoving logic to AstNode to determine how many type cells are required.
mvstanton@chromium.org [Fri, 17 Jan 2014 14:08:50 +0000 (14:08 +0000)]
Moving logic to AstNode to determine how many type cells are required.

With this change, we'll be able to discover how many type cells we
need at parse time, enabling future optimizations.

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/141533004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18668 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMinor bugfix in building inlined Array: bad argument to JSArrayBuilder.
mvstanton@chromium.org [Fri, 17 Jan 2014 12:18:57 +0000 (12:18 +0000)]
Minor bugfix in building inlined Array: bad argument to JSArrayBuilder.

An HConstant pointing to a Cell rather than an AllocationSite
was passed. The argument wasn't used because of fortuitous
flags. An assert was added to protect the argument.

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/141533003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18666 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFixed Windows build.
svenpanne@chromium.org [Fri, 17 Jan 2014 11:52:50 +0000 (11:52 +0000)]
Fixed Windows build.

TBR=dcarney@chromium.org

Review URL: https://codereview.chromium.org/140983005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18665 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoActivate calling push-to-trunk in auto-roll script.
machenbach@chromium.org [Fri, 17 Jan 2014 11:29:43 +0000 (11:29 +0000)]
Activate calling push-to-trunk in auto-roll script.

- Call push-to-trunk through python not through the shell
- Restore tree state on script errors
- Mock out python call in unit tests
- The actual call stays behind a flag

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/141633003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18664 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRemove special charAt and charCodeAt handling in the ICs.
verwaest@chromium.org [Fri, 17 Jan 2014 11:28:22 +0000 (11:28 +0000)]
Remove special charAt and charCodeAt handling in the ICs.

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/136443006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18663 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRemove special ArrayCode CallIC.
verwaest@chromium.org [Fri, 17 Jan 2014 11:24:36 +0000 (11:24 +0000)]
Remove special ArrayCode CallIC.
Once Call ICs are replaced by LoadIC + CallFunctionStub, we'll need a new way
of tracking this information.

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/141073006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18662 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFix logic error in assert in IsUndeclaredGlobal()
mvstanton@chromium.org [Fri, 17 Jan 2014 11:08:24 +0000 (11:08 +0000)]
Fix logic error in assert in IsUndeclaredGlobal()

Recent changes in IC logic meant that CallStubs no longer use the Contextual bit. IsUndeclaredGlobal() needed to adjust for that.

In fact, now the CL has morphed to remove the notion of storing contextual state in the IC at all, it just becomes some extra ic state of the load ic. This took some adjustment in harmony code to use the global receiver for certain stores.

Now it's clearer that only LoadICs actually record any information about contextual or not.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/140943002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18660 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoVarious extension-related cleanup and simplifications.
svenpanne@chromium.org [Fri, 17 Jan 2014 10:52:00 +0000 (10:52 +0000)]
Various extension-related cleanup and simplifications.

Removes the embarrassing "static"s, shuffles some code around, doing various cleanups on the way.

R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/130213009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18659 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFix stub-invoked setter callback handling.
dcarney@chromium.org [Fri, 17 Jan 2014 10:34:43 +0000 (10:34 +0000)]
Fix stub-invoked setter callback handling.

When invoking a setter callback for a property using
JSObject::SetPropertyWithCallback(),the callback arguments includes
a correct pair of receiver and holder objects.

Such a pair of _possibly different_ arguments (receiver, holder) must
also be supplied when invoking the same setter callback from JITed
code, when the setter is invoked through the StoreCallbackProperty
stub.

An example where this matters are the accessor properties kept on the
global scope of Worker (i.e., properties kept on the global object
itself, and not on its prototype.) Conflating the receiver with the
holder leads to general confusion when attempting to fetch out the
wrapper object.

LOG=N
R=dcarney@chromium.org, dcarney
BUG=239669

Review URL: https://codereview.chromium.org/139263008

Patch from Sigbjorn Finne <sigbjornf@opera.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18658 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRefactor string internalization.
ulan@chromium.org [Fri, 17 Jan 2014 10:27:57 +0000 (10:27 +0000)]
Refactor string internalization.

Remove duplicate code in StringTable and Heap. Instead of having many InternalizeXXXString methods, add a general InternalizeStringWithKey that takes a string table key.

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

Review URL: https://codereview.chromium.org/140953002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18657 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoIntroduce a setting to control the toolset for which d8 is compiled
jochen@chromium.org [Fri, 17 Jan 2014 09:11:22 +0000 (09:11 +0000)]
Introduce a setting to control the toolset for which d8 is compiled

BUG=v8:1775
R=machenbach@chromium.org
LOG=y

Review URL: https://codereview.chromium.org/138273016

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18656 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoPrepare push to trunk. Now working on version 3.24.19.
machenbach@chromium.org [Fri, 17 Jan 2014 08:10:15 +0000 (08:10 +0000)]
Prepare push to trunk.  Now working on version 3.24.19.

R=jkummerow@chromium.org
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/141603003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18653 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMIPS: Implement in-heap backing store for typed arrays.
palfia@homejinni.com [Thu, 16 Jan 2014 22:45:11 +0000 (22:45 +0000)]
MIPS: Implement in-heap backing store for typed arrays.

Port r18651 (49744859)

Original commit message:
This adds a fixed array sub-type that will represent a backing store for
typed arrays allocated with TypedArray(length) construtor.

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/141143003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18652 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoImplement in-heap backing store for typed arrays.
dslomov@chromium.org [Thu, 16 Jan 2014 17:08:45 +0000 (17:08 +0000)]
Implement in-heap backing store for typed arrays.

This adds a fixed array sub-type that will represent a backing store for
typed arrays allocated with TypedArray(length) construtor.

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

Review URL: https://codereview.chromium.org/101413006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18651 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRevert "Implement in-heap backing store for typed arrays."
dslomov@chromium.org [Thu, 16 Jan 2014 16:00:36 +0000 (16:00 +0000)]
Revert "Implement in-heap backing store for typed arrays."

This reverts commit r18649 for breaking Linux/nosnap and Win64 tests.

TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/140793003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18650 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoImplement in-heap backing store for typed arrays.
dslomov@chromium.org [Thu, 16 Jan 2014 15:01:27 +0000 (15:01 +0000)]
Implement in-heap backing store for typed arrays.

This adds a fixed array sub-type that will represent a backing store for
typed arrays allocated with TypedArray(length) construtor.

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

Committed: https://code.google.com/p/v8/source/detail?r=18646

Review URL: https://codereview.chromium.org/101413006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18649 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFix tree check in merge-to-branch script.
machenbach@chromium.org [Thu, 16 Jan 2014 14:40:56 +0000 (14:40 +0000)]
Fix tree check in merge-to-branch script.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/138963006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18648 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRevert "Implement in-heap backing store for typed arrays."
dslomov@chromium.org [Thu, 16 Jan 2014 14:26:15 +0000 (14:26 +0000)]
Revert "Implement in-heap backing store for typed arrays."

This reverts commit r18646 for breaking Win32 build.

TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/132233012

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18647 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoImplement in-heap backing store for typed arrays.
dslomov@chromium.org [Thu, 16 Jan 2014 14:18:37 +0000 (14:18 +0000)]
Implement in-heap backing store for typed arrays.

This adds a fixed array sub-type that will represent a backing store for
typed arrays allocated with TypedArray(length) construtor.

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

Review URL: https://codereview.chromium.org/101413006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18646 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRobustified extension installation a bit.
svenpanne@chromium.org [Thu, 16 Jan 2014 13:18:28 +0000 (13:18 +0000)]
Robustified extension installation a bit.

Removed ADDRESS_SANITIZER #ifdefs, the extension is so tiny that it's
not worth the clutter.

R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/138143014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18645 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoEnable allocation site pretenuring.
hpayer@chromium.org [Thu, 16 Jan 2014 11:59:34 +0000 (11:59 +0000)]
Enable allocation site pretenuring.

BUG=
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/133703015

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18643 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFix handling of last position for lazy deopt on X64, ARM and MIPS.
titzer@chromium.org [Thu, 16 Jan 2014 11:58:06 +0000 (11:58 +0000)]
Fix handling of last position for lazy deopt on X64, ARM and MIPS.

BUG=
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/140773002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18642 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoDeopt marked code at safe deoptimization point when pretenuring.
hpayer@chromium.org [Thu, 16 Jan 2014 11:54:12 +0000 (11:54 +0000)]
Deopt marked code at safe deoptimization point when pretenuring.

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

Review URL: https://codereview.chromium.org/138033012

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18641 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoUpdate load_rax and store_rax to support X32
haitao.feng@intel.com [Thu, 16 Jan 2014 11:35:30 +0000 (11:35 +0000)]
Update load_rax and store_rax to support X32

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/103843002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18640 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRefactor DeoptimizeDependentCodeGroup.
hpayer@chromium.org [Thu, 16 Jan 2014 10:00:03 +0000 (10:00 +0000)]
Refactor DeoptimizeDependentCodeGroup.

BUG=
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/137783016

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18639 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFix ChangeLog entry.
machenbach@chromium.org [Thu, 16 Jan 2014 08:28:42 +0000 (08:28 +0000)]
Fix ChangeLog entry.

Remove an issue from the ChangeLog that was reverted.

BUG=
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/139383007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18636 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoPrepare push to trunk. Now working on version 3.24.18.
machenbach@chromium.org [Thu, 16 Jan 2014 08:22:37 +0000 (08:22 +0000)]
Prepare push to trunk.  Now working on version 3.24.18.

R=jkummerow@chromium.org
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/140683002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18633 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRemoved apiutils.h and related cleanup.
svenpanne@chromium.org [Thu, 16 Jan 2014 08:17:40 +0000 (08:17 +0000)]
Removed apiutils.h and related cleanup.

ExtensionConfiguration is just a simple container for extension names
(in a perfect world we would use vector<string> and range-based for
loops), and HandleScopeData was in the totally wrong place. Some
additional cleanup on the way, e.g. using the null pattern behind our
external API.

R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/139393002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18632 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMIPS: Fix Win32 buildbreak (caused by overriden methods that have disappeared while...
palfia@homejinni.com [Thu, 16 Jan 2014 00:06:45 +0000 (00:06 +0000)]
MIPS: Fix Win32 buildbreak (caused by overriden methods that have disappeared while having the patch out for code review).

Port r18627 (c2ba7b25)

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/140203002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18631 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMIPS: This is a preview of a first step towards unification of the hydrogen call...
palfia@homejinni.com [Wed, 15 Jan 2014 23:37:49 +0000 (23:37 +0000)]
MIPS: This is a preview of a first step towards unification of the hydrogen call machinery.

Port r18626 (d3368a4c)

Original commit message:
The change replaces CallNamed, CallKeyed, CallConstantFunction and CallKnownGlobal hydrogen instructions with two new instructions with a more lower level semantics:

1. CallJSFunction for direct calls of JSFunction objects (no
   argument adaptation)

2. CallWithDescriptor for calls of a given Code object according to
   the supplied calling convention.

Details:

CallJSFunction should be straightforward, the main difference from the
existing InvokeFunction instruction is the absence of argument adaptor
handling. (As a next step, we will replace InvokeFunction with an
equivalent hydrogen code.)

For CallWithDescriptor, the calling conventions are represented by a
tweaked version of CallStubInterfaceDescriptor. In addition to the
parameter-register mapping, we also define parameter-representation
mapping there. The CallWithDescriptor instruction has variable number of
parameters now - this required some simple tweaks in Lithium, which
assumed fixed number of arguments in some places.

The calling conventions used in the calls are initialized in the
CallDescriptors class (code-stubs.h, <arch>/code-stubs-<arch>.cc), and
they live in a new table in the Isolate class. I should say I am not
quite sure about Representation::Integer32() representation for some of
the params of ArgumentAdaptorCall - it is not clear to me wether the
params could not end up on the stack and thus confuse the GC.

The change also includes an earlier small change to argument adaptor
(https://codereview.chromium.org/98463007) that avoids passing a naked
pointer to the code entry as a parameter. I am sorry for packaging that
with an already biggish change.

Performance implications:

Locally, I see a small regression (.2% or so). It is hard to say where
exactly it comes from, but I do see inefficient call sequences to the
adaptor trampoline. For example:

;;; <@78,#24> constant-t
bf85aa515a     mov edi,0x5a51aa85          ;; debug: position 29
;;; <@72,#53> load-named-field
8b7717         mov esi,[edi+0x17]          ;; debug: position 195
;;; <@80,#51> constant-s
b902000000     mov ecx,0x2                 ;; debug: position 195
;;; <@81,#51> gap
894df0         mov [ebp+0xf0],ecx
;;; <@82,#103> constant-i
bb01000000     mov ebx,0x1
;;; <@84,#102> constant-i
b902000000     mov ecx,0x2
;;; <@85,#102> gap
89d8           mov eax,ebx
89cb           mov ebx,ecx
8b4df0         mov ecx,[ebp+0xf0]
;;; <@86,#58> call-with-descriptor
e8ef57fcff     call ArgumentsAdaptorTrampoline  (0x2d80e6e0)    ;; code: BUILTIN

Note the silly handling of ecx; the hydrogen for this code is:

0 4 s27 Constant 1  range:1_1 <|@
0 3 t30 Constant 0x5bc1aa85 <JS Function xyz (SharedFunctionInfo 0x5bc1a919)> type:object <|@
0 1 t36 LoadNamedField t30.[in-object]@24 <|@
0 1 t38 Constant 0x2300e6a1 <Code> <|@
0 1 i102 Constant 2  range:2_2 <|@
0 1 i103 Constant 1  range:1_1 <|@
0 2 t41 CallWithDescriptor t38 t30 t36 s27 i103 i102 #2 changes[*] <|@

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/137663005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18630 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFix Win32 buildbreak (caused by overriden methods that have disappeared
jarin@chromium.org [Wed, 15 Jan 2014 17:51:09 +0000 (17:51 +0000)]
Fix Win32 buildbreak (caused by overriden methods that have disappeared
while having the patch out for code review).

R=danno@chromium.org
BUG=

Review URL: https://codereview.chromium.org/136303004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18627 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoThis is a preview of a first step towards unification of the hydrogen
jarin@chromium.org [Wed, 15 Jan 2014 17:00:35 +0000 (17:00 +0000)]
This is a preview of a first step towards unification of the hydrogen
call machinery.  The change replaces CallNamed, CallKeyed,
CallConstantFunction and CallKnownGlobal hydrogen instructions with two
new instructions with a more lower level semantics:

1. CallJSFunction for direct calls of JSFunction objects (no
   argument adaptation)

2. CallWithDescriptor for calls of a given Code object according to
   the supplied calling convention.

Details:

CallJSFunction should be straightforward, the main difference from the
existing InvokeFunction instruction is the absence of argument adaptor
handling. (As a next step, we will replace InvokeFunction with an
equivalent hydrogen code.)

For CallWithDescriptor, the calling conventions are represented by a
tweaked version of CallStubInterfaceDescriptor. In addition to the
parameter-register mapping, we also define parameter-representation
mapping there. The CallWithDescriptor instruction has variable number of
parameters now - this required some simple tweaks in Lithium, which
assumed fixed number of arguments in some places.

The calling conventions used in the calls are initialized in the
CallDescriptors class (code-stubs.h, <arch>/code-stubs-<arch>.cc), and
they live in a new table in the Isolate class. I should say I am not
quite sure about Representation::Integer32() representation for some of
the params of ArgumentAdaptorCall - it is not clear to me wether the
params could not end up on the stack and thus confuse the GC.

The change also includes an earlier small change to argument adaptor
(https://codereview.chromium.org/98463007) that avoids passing a naked
pointer to the code entry as a parameter. I am sorry for packaging that
with an already biggish change.

Performance implications:

Locally, I see a small regression (.2% or so). It is hard to say where
exactly it comes from, but I do see inefficient call sequences to the
adaptor trampoline. For example:

;;; <@78,#24> constant-t
bf85aa515a     mov edi,0x5a51aa85          ;; debug: position 29
;;; <@72,#53> load-named-field
8b7717         mov esi,[edi+0x17]          ;; debug: position 195
;;; <@80,#51> constant-s
b902000000     mov ecx,0x2                 ;; debug: position 195
;;; <@81,#51> gap
894df0         mov [ebp+0xf0],ecx
;;; <@82,#103> constant-i
bb01000000     mov ebx,0x1
;;; <@84,#102> constant-i
b902000000     mov ecx,0x2
;;; <@85,#102> gap
89d8           mov eax,ebx
89cb           mov ebx,ecx
8b4df0         mov ecx,[ebp+0xf0]
;;; <@86,#58> call-with-descriptor
e8ef57fcff     call ArgumentsAdaptorTrampoline  (0x2d80e6e0)    ;; code: BUILTIN

Note the silly handling of ecx; the hydrogen for this code is:

0 4 s27 Constant 1  range:1_1 <|@
0 3 t30 Constant 0x5bc1aa85 <JS Function xyz (SharedFunctionInfo 0x5bc1a919)> type:object <|@
0 1 t36 LoadNamedField t30.[in-object]@24 <|@
0 1 t38 Constant 0x2300e6a1 <Code> <|@
0 1 i102 Constant 2  range:2_2 <|@
0 1 i103 Constant 1  range:1_1 <|@
0 2 t41 CallWithDescriptor t38 t30 t36 s27 i103 i102 #2 changes[*] <|@

BUG=
R=verwaest@chromium.org, danno@chromium.org

Review URL: https://codereview.chromium.org/104663004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18626 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRevert of Make it possible to compile d8 for the host toolset as well (https://codere...
jochen@chromium.org [Wed, 15 Jan 2014 15:58:28 +0000 (15:58 +0000)]
Revert of Make it possible to compile d8 for the host toolset as well (https://codereview.chromium.org/139493002/)

Reason for revert:
still doesn't work on arm

Original issue's description:
> Make it possible to compile d8 for the host toolset as well
>
> 2nd attempt. Use a different output path for the host d8.
>
> BUG=v8:1775
> R=machenbach@chromium.org
> LOG=n
>
> Committed: https://code.google.com/p/v8/source/detail?r=18621

R=machenbach@chromium.org
TBR=machenbach@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=v8:1775

Review URL: https://codereview.chromium.org/139523003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18623 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoAdd tree control feature to auto-roll script.
machenbach@chromium.org [Wed, 15 Jan 2014 15:55:23 +0000 (15:55 +0000)]
Add tree control feature to auto-roll script.

This CL enables the auto-roll script to close and reopen the tree when pushing.

Modifies an auto-roll test so that the push-to-trunk part is executed in order to test the new tree control feature.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/130403006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18622 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMake it possible to compile d8 for the host toolset as well
jochen@chromium.org [Wed, 15 Jan 2014 15:35:10 +0000 (15:35 +0000)]
Make it possible to compile d8 for the host toolset as well

2nd attempt. Use a different output path for the host d8.

BUG=v8:1775
R=machenbach@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/139493002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18621 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRevert "Make it possible to compile d8 for the host toolset as well" and "For V8...
machenbach@chromium.org [Wed, 15 Jan 2014 14:24:32 +0000 (14:24 +0000)]
Revert "Make it possible to compile d8 for the host toolset as well" and "For V8, only build d8 on target"

This reverts commits r18618 and r18619 for breaking arm compilation.

BUG=
TBR=jochen@chromium.org

Review URL: https://codereview.chromium.org/139273004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18620 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFor V8, only build d8 on target
jochen@chromium.org [Wed, 15 Jan 2014 14:00:29 +0000 (14:00 +0000)]
For V8, only build d8 on target

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

Review URL: https://codereview.chromium.org/139403002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18619 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMake it possible to compile d8 for the host toolset as well
jochen@chromium.org [Wed, 15 Jan 2014 13:55:09 +0000 (13:55 +0000)]
Make it possible to compile d8 for the host toolset as well

BUG=v8:1775
R=jkummerow@chromium.org
LOG=y

Review URL: https://codereview.chromium.org/136763010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18618 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMake cells pointing to JSObjects weak in optimized code.
ulan@chromium.org [Wed, 15 Jan 2014 11:42:19 +0000 (11:42 +0000)]
Make cells pointing to JSObjects weak in optimized code.

This is done similar to weak embedded objects in optimized code (r17102). The
reference from optimized code to a cell is treated weakly in marking visitors
if the cell points to a JSObject. After marking we iterate over all cells
embedded in optimized code. If a cell is not marked but its value is marked,
then we revive the cell by marking it. Otherwise, the cell value is dead, so
we mark the code for deoptimization.

BUG=v8:2073
TEST=cctest/test-heap/CellsInOptimizedCodeAreWeak
LOG=Y
R=hpayer@chromium.org, mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/117483002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18616 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoPrepare push to trunk. Now working on version 3.24.17.
machenbach@chromium.org [Wed, 15 Jan 2014 10:29:29 +0000 (10:29 +0000)]
Prepare push to trunk.  Now working on version 3.24.17.

R=jkummerow@chromium.org
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/130143003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18613 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRevert "Eliminatable CheckMaps replaced with if(true) or if(false)."
jkummerow@chromium.org [Wed, 15 Jan 2014 09:53:54 +0000 (09:53 +0000)]
Revert "Eliminatable CheckMaps replaced with if(true) or if(false)."

This reverts r18592 for breaking the GC stress bots.

R=machenbach@chromium.org

Review URL: https://codereview.chromium.org/137783011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18611 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoReland "Remove flags argument from EmitCallWithStub" and "MIPS: Remove flags argument...
machenbach@chromium.org [Wed, 15 Jan 2014 09:05:21 +0000 (09:05 +0000)]
Reland "Remove flags argument from EmitCallWithStub" and "MIPS: Remove flags argument from EmitCallWithStub.""

This reverts commit r18606 since suspected problems still remained.

BUG=
R=jkummerow@chromium.org
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/139013005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18609 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRevert "Remove flags argument from EmitCallWithStub" and "MIPS: Remove flags argument...
machenbach@chromium.org [Wed, 15 Jan 2014 08:12:39 +0000 (08:12 +0000)]
Revert "Remove flags argument from EmitCallWithStub" and "MIPS: Remove flags argument from EmitCallWithStub."

This reverts commits r18598 and r18604 for breaking GC stress.

BUG=
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/133393003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18606 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years ago[Sheriff] Mark cpu profiler test as flaky on arm.
machenbach@chromium.org [Wed, 15 Jan 2014 08:06:02 +0000 (08:06 +0000)]
[Sheriff] Mark cpu profiler test as flaky on arm.

BUG=v8:2999
TBR=jkummerow@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/138183006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18605 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMIPS: Remove flags argument from EmitCallWithStub.
plind44@gmail.com [Tue, 14 Jan 2014 20:02:01 +0000 (20:02 +0000)]
MIPS: Remove flags argument from EmitCallWithStub.

Port r18598 (606541b7).

BUG=
R=gergely@homejinni.com

Review URL: https://codereview.chromium.org/133293003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18604 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMIPS: Remove HCallGlobal and merge uses with HCallNamed.
plind44@gmail.com [Tue, 14 Jan 2014 19:38:12 +0000 (19:38 +0000)]
MIPS: Remove HCallGlobal and merge uses with HCallNamed.

Port r18595 (957a383)

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/135053004

Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18603 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years ago[Sheriff] Mark cpu profiler test as flaky on windows.
machenbach@chromium.org [Tue, 14 Jan 2014 19:26:39 +0000 (19:26 +0000)]
[Sheriff] Mark cpu profiler test as flaky on windows.

This is fails constantly on windows since the XP->Win7 bot upgrade.

BUG=v8:3055
R=jkummerow@chromium.org
TBR=jkummerow@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/133153006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18602 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMIPS: Get rid of ContextualMode for call ICs.
plind44@gmail.com [Tue, 14 Jan 2014 18:33:36 +0000 (18:33 +0000)]
MIPS: Get rid of ContextualMode for call ICs.

Port r18594 (91e7cee)

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/133223003

Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18601 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMIPS: Fix loading the global receiver.
plind44@gmail.com [Tue, 14 Jan 2014 18:26:31 +0000 (18:26 +0000)]
MIPS: Fix loading the global receiver.

Port r18593 (2162d45)

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/137273006

Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18600 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMIPS: Remove CALL_AS_FUNCTION and CALL_AS_METHOD.
plind44@gmail.com [Tue, 14 Jan 2014 18:17:24 +0000 (18:17 +0000)]
MIPS: Remove CALL_AS_FUNCTION and CALL_AS_METHOD.

Port r18590 (566a7fd)

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/138553002

Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18599 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRemove flags argument from EmitCallWithStub
verwaest@chromium.org [Tue, 14 Jan 2014 17:31:31 +0000 (17:31 +0000)]
Remove flags argument from EmitCallWithStub

R=ishell@chromium.org

Review URL: https://codereview.chromium.org/135653003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18598 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRevert "Polymorphic named calls optimized for the case of repetitive call targets...
ishell@chromium.org [Tue, 14 Jan 2014 16:37:09 +0000 (16:37 +0000)]
Revert "Polymorphic named calls optimized for the case of repetitive call targets." for now because it prevents better results for checkmap elimination.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/138453002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18597 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoMIPS: Make the strict-mode calling convention for contextual calls the default one.
plind44@gmail.com [Tue, 14 Jan 2014 16:26:11 +0000 (16:26 +0000)]
MIPS: Make the strict-mode calling convention for contextual calls the default one.

Port r18581 (e5df8ae)

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/138383002

Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18596 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRemove HCallGlobal and merge uses with HCallNamed.
verwaest@chromium.org [Tue, 14 Jan 2014 16:15:52 +0000 (16:15 +0000)]
Remove HCallGlobal and merge uses with HCallNamed.

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/134333007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18595 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoGet rid of ContextualMode for call ICs.
verwaest@chromium.org [Tue, 14 Jan 2014 16:15:05 +0000 (16:15 +0000)]
Get rid of ContextualMode for call ICs.

BUG=
R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/137083002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18594 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFix loading the global receiver on ARM.
verwaest@chromium.org [Tue, 14 Jan 2014 16:13:18 +0000 (16:13 +0000)]
Fix loading the global receiver on ARM.

R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/138463002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18593 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoEliminatable CheckMaps replaced with if(true) or if(false).
ishell@chromium.org [Tue, 14 Jan 2014 16:06:40 +0000 (16:06 +0000)]
Eliminatable CheckMaps replaced with if(true) or if(false).

R=titzer@chromium.org

Review URL: https://codereview.chromium.org/130613003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18592 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoClosed generator returns a completed object instead of throwing a error
mstarzinger@chromium.org [Tue, 14 Jan 2014 15:19:34 +0000 (15:19 +0000)]
Closed generator returns a completed object instead of throwing a error

From ES6 rev20 draft, closed generator returns completed object (the
value is `undefined` and done is `true`).
Since a error thrown in generator is propagated to the caller without
setting status of a thrown generator to "completed", once a generator is
suspended by a error, status becomes "executing" forever. This is filed
as v8:3096

LOG=N
BUG=v8:3097
R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/136003003

Patch from Yusuke Suzuki <yusukesuzuki@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18591 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoRemove CALL_AS_FUNCTION and CALL_AS_METHOD.
verwaest@chromium.org [Tue, 14 Jan 2014 14:36:24 +0000 (14:36 +0000)]
Remove CALL_AS_FUNCTION and CALL_AS_METHOD.

BUG=
R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/136403005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18590 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoAnnotate mapped memory regions for LeakSanitizer.
mstarzinger@chromium.org [Tue, 14 Jan 2014 14:00:29 +0000 (14:00 +0000)]
Annotate mapped memory regions for LeakSanitizer.

Add in-code annotations for LeakSanitizer to treat any memory allocated through
the VirtualMemory class as a source of live pointers. This change eliminates
false positive leak reports when running Chromium under LSan.

BUG=chromium:328552
R=mstarzinger@chromium.org
LOG=Y

Review URL: https://codereview.chromium.org/137313002

Patch from Sergey Matveev <earthdok@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18589 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoFix test after r18586
jkummerow@chromium.org [Tue, 14 Jan 2014 14:00:10 +0000 (14:00 +0000)]
Fix test after r18586

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/138063003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18588 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

10 years agoDelete useless --loop-weight flag
jkummerow@chromium.org [Tue, 14 Jan 2014 13:45:43 +0000 (13:45 +0000)]
Delete useless --loop-weight flag

killing a static variable in the process.

BUG=chromium:333871
LOG=N
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/131223009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@18587 ce2b1a6d-e550-0410-aec6-3dcde31c8c00