platform/upstream/v8.git
11 years agoReintroduce runtime zone to Isolate.
bmeurer@chromium.org [Wed, 3 Jul 2013 11:40:30 +0000 (11:40 +0000)]
Reintroduce runtime zone to Isolate.

In case tcmalloc is not being used, the malloc()/free() overhead
can be significant for several runtime functions like StringReplace.
Therefore we reintroduce the runtime_zone into Isolate and reenable
the segment caching functionality of Zone.

There's now also a simpler version of ZoneScope w/o nesting capabilities.

BUG=v8:2759
R=danno@chromium.org, yangguo@chromium.org

Review URL: https://codereview.chromium.org/18635003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15465 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoImport the v8-i18n extension into v8
jkummerow@chromium.org [Wed, 3 Jul 2013 11:22:29 +0000 (11:22 +0000)]
Import the v8-i18n extension into v8

This adds the gyp flag v8_enable_i18n_support (off by default), and the
v8 flag FLAG_enable_i18n (on by default, but without effect if
v8_enable_i18n_support is off).

BUG=v8:2745
R=cira@chromium.org, danno@chromium.org, jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18487004

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

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15464 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFix small mistake in d8.js.
yangguo@chromium.org [Wed, 3 Jul 2013 10:38:20 +0000 (10:38 +0000)]
Fix small mistake in d8.js.

R=mvstanton@chromium.org
BUG=v8:2756

Review URL: https://codereview.chromium.org/18572009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15463 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoLog deopts with --log-timer-events.
yangguo@chromium.org [Wed, 3 Jul 2013 10:10:27 +0000 (10:10 +0000)]
Log deopts with --log-timer-events.

R=jkummerow@chromium.org
BUG=

Review URL: https://codereview.chromium.org/17599007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15462 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoDon't return frame by reference in ProcessTopOptimizedFrame.
mstarzinger@chromium.org [Wed, 3 Jul 2013 09:47:22 +0000 (09:47 +0000)]
Don't return frame by reference in ProcessTopOptimizedFrame.

R=ulan@chromium.org

Review URL: https://codereview.chromium.org/18592006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15461 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoMake MachO gdbjit support compile again
danno@chromium.org [Wed, 3 Jul 2013 08:29:46 +0000 (08:29 +0000)]
Make MachO gdbjit support compile again

R=mstarzinger@chromium.org

Review URL: https://chromiumcodereview.appspot.com/18621002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15460 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFix default type feedback returned from the oracle
jkummerow@chromium.org [Tue, 2 Jul 2013 16:31:39 +0000 (16:31 +0000)]
Fix default type feedback returned from the oracle

R=rossberg@chromium.org

Review URL: https://codereview.chromium.org/18465003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15459 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoHandlify JSObject::DefineAccessor method.
mstarzinger@chromium.org [Tue, 2 Jul 2013 16:24:23 +0000 (16:24 +0000)]
Handlify JSObject::DefineAccessor method.

R=rossberg@chromium.org

Review URL: https://codereview.chromium.org/18497003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15458 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoUse stored Isolate pointer instead of Isolate::Current()
yurys@chromium.org [Tue, 2 Jul 2013 15:44:30 +0000 (15:44 +0000)]
Use stored Isolate pointer instead of Isolate::Current()

Fixed a couple of places where stored pointer to the isolate can be used instead of reading from thread local storage.

BUG=None
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18418003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15457 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoInclude statistical profile into profviz.
yangguo@chromium.org [Tue, 2 Jul 2013 15:34:28 +0000 (15:34 +0000)]
Include statistical profile into profviz.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/17620008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15456 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoHandlify JSObject::SetPrototype method.
mstarzinger@chromium.org [Tue, 2 Jul 2013 15:32:46 +0000 (15:32 +0000)]
Handlify JSObject::SetPrototype method.

R=rossberg@chromium.org

Review URL: https://codereview.chromium.org/18089024

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15455 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRemove HFixedArrayBaseLength instruction and replace with regular HLoadNamedField.
titzer@chromium.org [Tue, 2 Jul 2013 15:31:17 +0000 (15:31 +0000)]
Remove HFixedArrayBaseLength instruction and replace with regular HLoadNamedField.

BUG=

Review URL: https://codereview.chromium.org/18508002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15454 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFix debuggersupport=off build.
yangguo@chromium.org [Tue, 2 Jul 2013 15:30:33 +0000 (15:30 +0000)]
Fix debuggersupport=off build.

R=ulan@chromium.org
BUG=v8:2754

Review URL: https://codereview.chromium.org/18500003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15453 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoEnable weak embedded maps in optimized code.
ulan@chromium.org [Tue, 2 Jul 2013 15:15:58 +0000 (15:15 +0000)]
Enable weak embedded maps in optimized code.

If the top optimized code in call stack is at the point that does not support
deoptimization, then treat the maps in the code as strong pointers.

Note that other optimized code in call stack must support deoptimization
because of the call instruction with side-effects.

BUG=217858,v8:2073
R=mstarzinger@chromium.org

Review URL: https://chromiumcodereview.appspot.com/16955008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15452 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoIntroduce -m64 flag for making x64 when the default gcc compiler is for X32
jkummerow@chromium.org [Tue, 2 Jul 2013 15:03:54 +0000 (15:03 +0000)]
Introduce -m64 flag for making x64 when the default gcc compiler is for X32

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18211004

Patch from Haitao Feng <haitao.feng@intel.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15451 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoNot consider the use representation in unintiliazed ArrayLiteral which always require...
jkummerow@chromium.org [Tue, 2 Jul 2013 15:00:13 +0000 (15:00 +0000)]
Not consider the use representation in unintiliazed ArrayLiteral which always requires SMI

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18263002

Patch from Weiliang Lin <weiliang.lin2@gmail.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15450 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoPrepare push to trunk. Now working on version 3.20.2.
jkummerow@chromium.org [Tue, 2 Jul 2013 14:41:58 +0000 (14:41 +0000)]
Prepare push to trunk.  Now working on version 3.20.2.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/18324010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15447 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoIncrease profiling interval for test-cpu-profiler/FunctionCallSample
yurys@chromium.org [Tue, 2 Jul 2013 14:06:39 +0000 (14:06 +0000)]
Increase profiling interval for test-cpu-profiler/FunctionCallSample

The test falkes on Windows bots as number of samples is not enough. This change increases sampling interval for the test on Windows.

BUG=2628
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18532003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15446 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFix invalid condition introduced with revision 15441.
bmeurer@chromium.org [Tue, 2 Jul 2013 12:16:30 +0000 (12:16 +0000)]
Fix invalid condition introduced with revision 15441.

R=danno@chromium.org

Review URL: https://codereview.chromium.org/18375004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15445 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRevert "Remove BinaryOp::result_type"
rossberg@chromium.org [Tue, 2 Jul 2013 11:57:43 +0000 (11:57 +0000)]
Revert "Remove BinaryOp::result_type"

This reverts https://code.google.com/p/v8/source/detail?r=15265, due to performance regression.

TBR=jkummerow@chromium.org
BUG=

Review URL: https://codereview.chromium.org/18525002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15444 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRevert "Derive synthetic type bounds for expressions"
rossberg@chromium.org [Tue, 2 Jul 2013 11:50:56 +0000 (11:50 +0000)]
Revert "Derive synthetic type bounds for expressions"

TBR=jkummerow@chromium.org
BUG=

Review URL: https://codereview.chromium.org/18524002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15443 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoMake sticky chromium revision available for perf tests.
machenbach@chromium.org [Tue, 2 Jul 2013 10:54:02 +0000 (10:54 +0000)]
Make sticky chromium revision available for perf tests.

This file will be regularly updated with the chromium revision to be used by performance test buildbots.

R=danno@chromium.org

Review URL: https://codereview.chromium.org/18244003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15442 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoTurn UInt32 analysis into a proper HPhase.
bmeurer@chromium.org [Tue, 2 Jul 2013 10:53:08 +0000 (10:53 +0000)]
Turn UInt32 analysis into a proper HPhase.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18487003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15441 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRelax test constraints from r15439 (Join threads after stopping).
yangguo@chromium.org [Tue, 2 Jul 2013 09:59:08 +0000 (09:59 +0000)]
Relax test constraints from r15439 (Join threads after stopping).

R=hpayer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/18509002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15440 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoJoin threads after stopping.
yangguo@chromium.org [Tue, 2 Jul 2013 09:04:45 +0000 (09:04 +0000)]
Join threads after stopping.

R=hpayer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/18287003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15439 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoTurn escape analysis into a proper HPhase.
bmeurer@chromium.org [Tue, 2 Jul 2013 08:29:33 +0000 (08:29 +0000)]
Turn escape analysis into a proper HPhase.

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

Review URL: https://codereview.chromium.org/18476004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15438 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoProperly cleanup the transcendental subcaches.
bmeurer@chromium.org [Tue, 2 Jul 2013 08:27:27 +0000 (08:27 +0000)]
Properly cleanup the transcendental subcaches.

R=danno@chromium.org

Review URL: https://codereview.chromium.org/18487002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15437 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoCorrectly report callstack when current function is FunctionCall builtin
yurys@chromium.org [Tue, 2 Jul 2013 07:51:09 +0000 (07:51 +0000)]
Correctly report callstack when current function is FunctionCall builtin

When current function is FunctionCall builtin we have no reliable way to determine its caller function (in many cases the top of the sampled stack contains address of the caller but sometimes it does not). Instead of dropping the sample or its two top frames we simply mark the caller frame as '(unresolved function)'. It seems like a better approach that dropping whole sample as knowing the top function and the rest of the stack the user should be able to figure out what the caller was.

This change adds builtin id to CodeEntry objects. It will be used later to add similar top frame analysis for FunctionApply and probably other builtins.

BUG=None
TBR=loislo@chromium.org

Review URL: https://codereview.chromium.org/18422003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15436 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoCleanup the default Isolate's entry stack.
bmeurer@chromium.org [Tue, 2 Jul 2013 07:21:07 +0000 (07:21 +0000)]
Cleanup the default Isolate's entry stack.

This adds an ASSERT that the entry_stack_ of the Isolate is
empty upon destruction. The default Isolate is an exception
here, as it may still contain an entry stack item upon
destruction.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18247004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15435 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRevert "Correctly report callstack when current function is FunctionCall builtin"
yurys@chromium.org [Tue, 2 Jul 2013 06:26:07 +0000 (06:26 +0000)]
Revert "Correctly report callstack when current function is FunctionCall builtin"

This reverts commit r15426. The new test fails in Debug mode.

BUG=None
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18478002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15434 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoCPUProfiler: propagate scriptId to the front-end
loislo@chromium.org [Tue, 2 Jul 2013 06:14:01 +0000 (06:14 +0000)]
CPUProfiler: propagate scriptId to the front-end

Each CpuProfileNode has resource_name string property.
It cost us N * strlen(resource_name) where N is number of functions in the collected profile.
We could transfer script_id instead of resource_name so it would reduce transfer
size and help us to solve the problem with evals and sourceURL.

BUG=none
TEST=test-cpu-profiler/CollectCpuProfile
R=jkummerow@chromium.org, yurys@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=15429

Review URL: https://codereview.chromium.org/17642009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15433 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoMIPS: Fix register usage in ProfileEntryHookStub
palfia@homejinni.com [Mon, 1 Jul 2013 21:05:03 +0000 (21:05 +0000)]
MIPS: Fix register usage in ProfileEntryHookStub

Different register must be used to pass the test on target hardware.

BUG=
TEST=cctest/test-api/SetFunctionEntryHook

Review URL: https://codereview.chromium.org/18368002
Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15432 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRevert "CPUProfiler: propagate scriptId to the front-end"
loislo@chromium.org [Mon, 1 Jul 2013 19:22:45 +0000 (19:22 +0000)]
Revert "CPUProfiler: propagate scriptId to the front-end"

This reverts commit d575f6bc8b262dac08f02913ae6e7c504c9dd900.

Check is failing on debug bots.

TBR= yurys@chromium.org

Review URL: https://chromiumcodereview.appspot.com/18332010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15431 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoCPUProfiler: propagate scriptId to the front-end
loislo@chromium.org [Mon, 1 Jul 2013 15:15:57 +0000 (15:15 +0000)]
CPUProfiler: propagate scriptId to the front-end

Each CpuProfileNode has resource_name string property.
It cost us N * strlen(resource_name) where N is number of functions in the collected profile.
We could transfer script_id instead of resource_name so it would reduce transfer
size and help us to solve the problem with evals and sourceURL.

BUG=none
TEST=test-cpu-profiler/CollectCpuProfile
R=jkummerow@chromium.org, yurys@chromium.org

Review URL: https://codereview.chromium.org/17642009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15429 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoAlso delete optimization jobs in the output queue when stopping thread.
yangguo@chromium.org [Mon, 1 Jul 2013 15:12:59 +0000 (15:12 +0000)]
Also delete optimization jobs in the output queue when stopping thread.

R=hpayer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/18364002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15428 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRevert r15419: "Generate StoreGlobal stubs with Hydrogen"
danno@chromium.org [Mon, 1 Jul 2013 15:12:21 +0000 (15:12 +0000)]
Revert r15419: "Generate StoreGlobal stubs with Hydrogen"

TBR=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/18357004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15427 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoCorrectly report callstack when current function is FunctionCall builtin
yurys@chromium.org [Mon, 1 Jul 2013 14:57:58 +0000 (14:57 +0000)]
Correctly report callstack when current function is FunctionCall builtin

When current function is FunctionCall builtin we have no reliable way to determine its caller function (in many cases the top of the sampled stack contains address of the caller but sometimes it does not). Instead of dropping the sample or its two top frames we simply mark the caller frame as '(unresolved function)'. It seems like a better approach that dropping whole sample as knowing the top function and the rest of the stack the user should be able to figure out what the caller was.

This change adds builtin id to CodeEntry objects. It will be used later to add similar top frame analysis for FunctionApply and probably other builtins.

BUG=None
R=jkummerow@chromium.org, loislo@chromium.org

Review URL: https://codereview.chromium.org/18316004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15426 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoDerive synthetic type bounds for expressions
rossberg@chromium.org [Mon, 1 Jul 2013 13:59:18 +0000 (13:59 +0000)]
Derive synthetic type bounds for expressions

Currently synthesizes Smi as the lower bound for numeric operations (except for +, which might result in a string). That would need to change in places where we want to track constants as type feedback.

Does not do anything about variables yet.

R=jkummerow@chromium.org
BUG=

Review URL: https://codereview.chromium.org/17842004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15422 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoImplement WeakMap.prototype.clear function.
mstarzinger@chromium.org [Mon, 1 Jul 2013 13:56:48 +0000 (13:56 +0000)]
Implement WeakMap.prototype.clear function.

R=rossberg@chromium.org
BUG=v8:2753
TEST=mjsunit/harmony/collections

Review URL: https://codereview.chromium.org/18352002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15421 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoDebug: support breakpoints set in the middle of statement (try #2 after rollback)
prybin@chromium.org [Mon, 1 Jul 2013 13:44:10 +0000 (13:44 +0000)]
Debug: support breakpoints set in the middle of statement (try #2 after rollback)

Review URL: https://codereview.chromium.org/18349004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15420 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoGenerate StoreGlobal stubs with Hydrogen
danno@chromium.org [Mon, 1 Jul 2013 13:22:13 +0000 (13:22 +0000)]
Generate StoreGlobal stubs with Hydrogen

- Constants globals are inlined into Hydrogen code using code dependencies that invalidate the Crankshafted code when global PropertyCells or the global object change.
- The more general case generates code that is just as good as the hand-written assembly stubs on all platforms.

R=ulan@chromium.org

Review URL: https://codereview.chromium.org/16925008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15419 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRevert "Debug: support breakpoints set in the middle of statement"
prybin@chromium.org [Mon, 1 Jul 2013 13:05:21 +0000 (13:05 +0000)]
Revert "Debug: support breakpoints set in the middle of statement"

Review URL: https://codereview.chromium.org/18326007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15418 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoEnsure CheckInitialized is present independent of define.
mstarzinger@chromium.org [Mon, 1 Jul 2013 12:57:15 +0000 (12:57 +0000)]
Ensure CheckInitialized is present independent of define.

This makes sure that the same symbols are present, independent of which
defines have been used while building V8. Otherwise only embedders with
compatible defines would be able to link against that binary.

R=danno@chromium.org
BUG=chromium:255779

Review URL: https://codereview.chromium.org/18305004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15417 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoDebug: support breakpoints set in the middle of statement
prybin@chromium.org [Mon, 1 Jul 2013 12:54:13 +0000 (12:54 +0000)]
Debug: support breakpoints set in the middle of statement

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/16093040

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15416 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoTest that profiler is stopped when isolate is being disposed
yurys@chromium.org [Mon, 1 Jul 2013 12:32:52 +0000 (12:32 +0000)]
Test that profiler is stopped when isolate is being disposed

The only way to get v8::CpuProfiler instance in the V8 public API is to call v8::Iolate::GetCpuProfiler(). The method will return NULL if the isolate has not been initialized yet or has been torn down already. It is the client's reponsibility to make sure that CPU profiling has been stopped before disposing of the isolate.

This CL adds a test for this and several ASSRTS enforcing that assumptions. This allowed to be sure that heap is always setup when CPU profiling is being started. Based on that the number of places where already compiled functions are reported to the profiler event processor boils down to the single place (CpuProfiler::StartProcessorIfNotStarted). I'm going to rely on this assumption in further changes.

BUG=None
R=loislo@chromium.org, yangguo@chromium.org

Review URL: https://codereview.chromium.org/18336002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15415 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoPlug leaked mutexes in OptimizingCompilerThread.
bmeurer@chromium.org [Mon, 1 Jul 2013 12:30:26 +0000 (12:30 +0000)]
Plug leaked mutexes in OptimizingCompilerThread.

BUG=
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18345004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15414 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoNo need to pass profiles to ProfilerEventsProcessor.
bmeurer@chromium.org [Mon, 1 Jul 2013 12:24:26 +0000 (12:24 +0000)]
No need to pass profiles to ProfilerEventsProcessor.

Following up on https://codereview.chromium.org/18353002, there's
no need to pass the profiles to ProfilerEventsProcessor's constructor.

BUG=
R=danno@chromium.org

Review URL: https://codereview.chromium.org/18332004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15413 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFix compiler error about unused profiles_ field in ProfilerEventsProcessor.
bmeurer@chromium.org [Mon, 1 Jul 2013 12:11:29 +0000 (12:11 +0000)]
Fix compiler error about unused profiles_ field in ProfilerEventsProcessor.

BUG=
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18353002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15412 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoPlug leaked string stream debug object cache in Isolate.
bmeurer@chromium.org [Mon, 1 Jul 2013 12:07:53 +0000 (12:07 +0000)]
Plug leaked string stream debug object cache in Isolate.

BUG=
R=danno@chromium.org

Review URL: https://codereview.chromium.org/18324007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15411 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoGrokdump: new shell command, "lm"
mvstanton@chromium.org [Mon, 1 Jul 2013 11:37:40 +0000 (11:37 +0000)]
Grokdump: new shell command, "lm"

lm - list matching modules and details such as product version number
found in the minidump. Also, enabled a mode to execute one command and
exit.

BUG=
R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/18310003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15410 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoAdd broader custom test expectations for blink tests to detect crashes. Remove also...
machenbach@chromium.org [Mon, 1 Jul 2013 11:08:01 +0000 (11:08 +0000)]
Add broader custom test expectations for blink tests to detect crashes. Remove also useless documentation.

Most of these tests had some problems (failure, crash or timeout), but which problem exactly cannot be retrieved anymore.

We let them rum for a while with a broader test expectation (pass fail slow) and analyse the results.

Tests that crash get a crash expectation. Timeouts will get a timeout expectation.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18319005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15409 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRestore message when rethrowing in TryCatch.
yangguo@chromium.org [Mon, 1 Jul 2013 10:54:39 +0000 (10:54 +0000)]
Restore message when rethrowing in TryCatch.

Based on a patch contributed by Andrew Paprocki <andrew@ishiboo.com>.

R=jkummerow@chromium.org
BUG=
TEST=cctest/test-api/TryCatchNestedSyntax

Review URL: https://codereview.chromium.org/17694002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15408 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoCpuProfiler: eliminate 2 layers of 4 for CodeCreateEvent calls.
loislo@chromium.org [Mon, 1 Jul 2013 10:12:03 +0000 (10:12 +0000)]
CpuProfiler: eliminate 2 layers of 4 for CodeCreateEvent calls.

The bodies of methods in ProfilerEventProcessor were moved into CpuProfiler.
Multiple NewCodeEntry methods in CpuProfilesCollection were replaced with one which
simply passes arguments to the CodeEntry constructor.
And CpuProfiler just calls this method when it needs a CodeEntry object.

This NewCodeEntry method is required because CpuProfilesCollection keeps ownership of CodeEntry objects.

BUG=255392
TEST=existing tests
R=yangguo@chromium.org, yurys@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=15405

Review URL: https://codereview.chromium.org/18053004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15407 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRevert "CpuProfiler: eliminate 2 layers of 4 for CodeCreateEvent calls."
loislo@chromium.org [Mon, 1 Jul 2013 09:51:44 +0000 (09:51 +0000)]
Revert "CpuProfiler: eliminate 2 layers of 4 for CodeCreateEvent calls."

This reverts commit 76adf84b83ec3c0b261cbc29369ce4ac83f9d002.

windows compilation failed

BUG=none
TBR=yurys@chromium.org

Review URL: https://codereview.chromium.org/18344002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15406 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoCpuProfiler: eliminate 2 layers of 4 for CodeCreateEvent calls.
loislo@chromium.org [Mon, 1 Jul 2013 09:39:15 +0000 (09:39 +0000)]
CpuProfiler: eliminate 2 layers of 4 for CodeCreateEvent calls.

The bodies of methods in ProfilerEventProcessor were moved into CpuProfiler.
Multiple NewCodeEntry methods in CpuProfilesCollection were replaced with one which
simply passes arguments to the CodeEntry constructor.
And CpuProfiler just calls this method when it needs a CodeEntry object.

This NewCodeEntry method is required because CpuProfilesCollection keeps ownership of CodeEntry objects.

BUG=255392
TEST=existing tests
R=yangguo@chromium.org, yurys@chromium.org

Review URL: https://codereview.chromium.org/18053004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15405 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFix infinite loop in regress-opt-after-deopt.
yangguo@chromium.org [Mon, 1 Jul 2013 09:14:15 +0000 (09:14 +0000)]
Fix infinite loop in regress-opt-after-deopt.

%CompleteOptimization attempts to install optimized functions
that the parallel thread has put on the output queue, as long as
the function is marked with a builtin.  However, activating the
debugger will set all functions to the lazy recompile builtin,
without the function being on the parallel recompilation pipeline.
So we wait for the function to finish parallel recompilation
while it's marked by a builtin that's unrelated to parallel
recompilation.

R=hpayer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/18129003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15404 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoTest fix - array-feedback.js has a test that only make sense when
mvstanton@chromium.org [Mon, 1 Jul 2013 09:00:14 +0000 (09:00 +0000)]
Test fix - array-feedback.js has a test that only make sense when
running crankshaft. Allow the test to tolerate --nocrankshaft.

BUG=
R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/18328002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15403 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoAdd dependency to ICU 4.6
jkummerow@chromium.org [Mon, 1 Jul 2013 08:32:41 +0000 (08:32 +0000)]
Add dependency to ICU 4.6

We use the patched version from the chromium project, as it already has
a gyp file and allows for using the system icu instead of the bundled
version if desired.

This is in preparation of bringing in v8-i18n.

R=jkummerow@chromium.org
BUG=v8:2745

Review URL: https://codereview.chromium.org/17850002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15402 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRemove irrelevant parser test that has SVN issues.
machenbach@chromium.org [Mon, 1 Jul 2013 08:11:30 +0000 (08:11 +0000)]
Remove irrelevant parser test that has SVN issues.

The file tests different CR and LF combinations that cannot be stored correctly in SVN.

Since we do not need coverage here, we rather remove the test completely.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18333002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15401 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRemove test from blacklist that should pass.
machenbach@chromium.org [Mon, 1 Jul 2013 07:54:56 +0000 (07:54 +0000)]
Remove test from blacklist that should pass.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18324003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15398 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoMIPS: Fix wrong checkin.
palfia@homejinni.com [Fri, 28 Jun 2013 19:52:44 +0000 (19:52 +0000)]
MIPS: Fix wrong checkin.

This commit fixes r15396.

BUG=

Review URL: https://codereview.chromium.org/18054020

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15397 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoMIPS: Improved function entry hook coverage.
palfia@homejinni.com [Fri, 28 Jun 2013 18:34:08 +0000 (18:34 +0000)]
MIPS: Improved function entry hook coverage.

Port r15384 (d553efd)

Original commit message:
Adds more coverage for function entry hook, sufficient to capture profiles
that are contiguous from C++, through JS and back out to C++.

BUG=

Review URL: https://codereview.chromium.org/18184011
Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15396 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoMIPS: Hydrogen array constructor cleanup and improvements.
palfia@homejinni.com [Fri, 28 Jun 2013 17:16:17 +0000 (17:16 +0000)]
MIPS: Hydrogen array constructor cleanup and improvements.

Port r15383 (3e90aaf)

Original commit message:
* Cleanup of LCallNewArray::PrintDataTo() method
* Created HCallNewArray::PrintDataTo()
* Created many more tests in array-constructor-feedback.js
* Removed redundant instructions in
GenerateRecordCallTarget
* Bugfix in CreateArrayDispatchOneArgument: on a call to
new Array(0), we'd like to set the type feedback cell to
a packed elements kind, but we shouldn't do it if the
cell contains the megamorphic sentinel.
* When used from crankshaft, ArrayConstructorStubs can
avoid verifying that the function being called is the
array function from the current native context, relying
instead on the fact that crankshaft issues an
HCheckFunction to protect the constructor call. (this
new minor key is used in LCodeGen::DoCallNewArray(), and
influences code generation in
CodeStubGraphBuilderBase::BuildArrayConstructor()).
* Optimization: the array constructor specialized for
FAST_SMI_ELEMENTS can save some instructions by looking
up the correct map on the passed in constructor, rather
than indexing into the array of cached maps per element
kind.

BUG=

Review URL: https://codereview.chromium.org/18191007
Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15395 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoUse HInstructionIterator more broadly for hydrogen.
mstarzinger@chromium.org [Fri, 28 Jun 2013 16:31:11 +0000 (16:31 +0000)]
Use HInstructionIterator more broadly for hydrogen.

R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/18068002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15394 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFirst simplistic implementation of escape analysis.
mstarzinger@chromium.org [Fri, 28 Jun 2013 16:09:54 +0000 (16:09 +0000)]
First simplistic implementation of escape analysis.

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

Review URL: https://codereview.chromium.org/17914002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15393 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoAdd %_DebugBreakInOptimizedCode() pseudo function call to insert int3/stop instructio...
jkummerow@chromium.org [Fri, 28 Jun 2013 15:48:38 +0000 (15:48 +0000)]
Add %_DebugBreakInOptimizedCode() pseudo function call to insert int3/stop instructions into optimized code

R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/17870002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15392 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoClean up the usage of V8_TARGET_ARCH_${arch} and V8_HOST_ARCH_${arch}
jkummerow@chromium.org [Fri, 28 Jun 2013 15:34:48 +0000 (15:34 +0000)]
Clean up the usage of V8_TARGET_ARCH_${arch} and V8_HOST_ARCH_${arch}

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18037002

Patch from Haitao Feng <haitao.feng@intel.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15391 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoUse macros instead of using number directly for x64
jkummerow@chromium.org [Fri, 28 Jun 2013 15:26:01 +0000 (15:26 +0000)]
Use macros instead of using number directly for x64

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18076002

Patch from Weiliang Lin <weiliang.lin2@gmail.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15390 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoIntroduce a notion of "v8_code" in gyp config files.
jkummerow@chromium.org [Fri, 28 Jun 2013 15:22:46 +0000 (15:22 +0000)]
Introduce a notion of "v8_code" in gyp config files.

This allows for compiling third-party code (such as ICU) with less
strict flags.

BUG=v8:2745
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18181006

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

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15389 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRoll gyp to r1656
jkummerow@chromium.org [Fri, 28 Jun 2013 15:20:07 +0000 (15:20 +0000)]
Roll gyp to r1656

r1656 contains a bug fix for the make generator to correctly regenerate
Makefiles if a gyp was touched. Before, it would assume that it is
running in the top-level directory. However, the v8 standalone build
puts the generated Makefile into the out subdirectory, and so gyp
failed to locate the gyp binary.

BUG=none
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18053015

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

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15388 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoPrepare push to trunk. Now working on version 3.20.1.
mstarzinger@chromium.org [Fri, 28 Jun 2013 13:58:15 +0000 (13:58 +0000)]
Prepare push to trunk.  Now working on version 3.20.1.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18052019

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15385 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoImproved function entry hook coverage
danno@chromium.org [Fri, 28 Jun 2013 13:40:41 +0000 (13:40 +0000)]
Improved function entry hook coverage

Adds more coverage for function entry hook, sufficient to capture profiles that are contiguous from C++, through JS and back out to C++.

R=danno@chromium.org

Committed: http://code.google.com/p/v8/source/detail?r=15361

Review URL: https://codereview.chromium.org/16578008

Patch from Sigurður Ásgeirsson <siggi@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15384 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoHydrogen array constructor cleanup and improvements
mvstanton@chromium.org [Fri, 28 Jun 2013 13:16:14 +0000 (13:16 +0000)]
Hydrogen array constructor cleanup and improvements

* Cleanup of LCallNewArray::PrintDataTo() method
* Created HCallNewArray::PrintDataTo()
* Created many more tests in array-constructor-feedback.js
* Removed redundant instructions in
  GenerateRecordCallTarget
* Bugfix in CreateArrayDispatchOneArgument: on a call to
  new Array(0), we'd like to set the type feedback cell to
  a packed elements kind, but we shouldn't do it if the
  cell contains the megamorphic sentinel.
* When used from crankshaft, ArrayConstructorStubs can
  avoid verifying that the function being called is the
  array function from the current native context, relying
  instead on the fact that crankshaft issues an
  HCheckFunction to protect the constructor call. (this
  new minor key is used in LCodeGen::DoCallNewArray(), and
  influences code generation in
  CodeStubGraphBuilderBase::BuildArrayConstructor()).
* Optimization: the array constructor specialized for
  FAST_SMI_ELEMENTS can save some instructions by looking
  up the correct map on the passed in constructor, rather
  than indexing into the array of cached maps per element
  kind.

BUG=
R=danno@chromium.org

Review URL: https://codereview.chromium.org/17091002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15383 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRepair test file with newline problems.
machenbach@chromium.org [Fri, 28 Jun 2013 13:10:58 +0000 (13:10 +0000)]
Repair test file with newline problems.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18051011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15382 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoHeap snapshot: Update user roots definition.
alph@chromium.org [Fri, 28 Jun 2013 12:53:52 +0000 (12:53 +0000)]
Heap snapshot: Update user roots definition.

Do not define an object as a user root if its context is not
present in Global handles.

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

Review URL: https://codereview.chromium.org/17881005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15381 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFix GCC 4.2 errors introduced with revision 15379.
bmeurer@chromium.org [Fri, 28 Jun 2013 12:28:39 +0000 (12:28 +0000)]
Fix GCC 4.2 errors introduced with revision 15379.

BUG=
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18184005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15380 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoMake use of templatized convienience functions for adding Hydrogen instructions.
bmeurer@chromium.org [Fri, 28 Jun 2013 11:44:16 +0000 (11:44 +0000)]
Make use of templatized convienience functions for adding Hydrogen instructions.

Building on the previous changes from https://codereview.chromium.org/18050004/
this patch makes even more use of the templatized functions for adding
Hydrogen instructions.

R=danno@chromium.org
BUG=

Review URL: https://codereview.chromium.org/18051010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15379 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoAbort optimization when debugger is turned on.
yangguo@chromium.org [Fri, 28 Jun 2013 11:34:51 +0000 (11:34 +0000)]
Abort optimization when debugger is turned on.

BUG=v8:2751
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18198003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15378 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoDelete test file with newline problems.
machenbach@chromium.org [Fri, 28 Jun 2013 11:25:26 +0000 (11:25 +0000)]
Delete test file with newline problems.

The file contains the combinations CRLF LFCR CR and LF, and it looks like git automatically changed them to LF in the last dcommit.

See also http://www.dont-panic.cc/capi/2009/02/16/how-to-force-git-to-consider-a-file-as-binary/

The file is going to be added again in a different issue with an additional .gitattributes file to treat it as binary.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18080013

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15377 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoAvoid data race in debug mode on the parallel thread.
yangguo@chromium.org [Fri, 28 Jun 2013 11:24:27 +0000 (11:24 +0000)]
Avoid data race in debug mode on the parallel thread.

R=jkummerow@chromium.org
BUG=

Review URL: https://codereview.chromium.org/18194004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15376 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFast-forward version number on bleeding_edge to 3.20.0
jkummerow@chromium.org [Fri, 28 Jun 2013 10:55:25 +0000 (10:55 +0000)]
Fast-forward version number on bleeding_edge to 3.20.0

R=danno@chromium.org

Review URL: https://codereview.chromium.org/18194005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15375 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoSkip webkit tests in debug that are very slow.
machenbach@chromium.org [Fri, 28 Jun 2013 10:32:43 +0000 (10:32 +0000)]
Skip webkit tests in debug that are very slow.

Duration for each is between 14 s and 160 s.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18034017

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15374 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRefactor HInferRepresentation into an HPhase and use the phase zone.
bmeurer@chromium.org [Fri, 28 Jun 2013 07:40:35 +0000 (07:40 +0000)]
Refactor HInferRepresentation into an HPhase and use the phase zone.

Rename HInferRepresentation to HInferRepresentationPhase,
following naming scheme suggested by danno@chromium.org in
https://codereview.chromium.org/17458002

The HInferRepresentationPhase now uses the phase zone for all its
allocations.

Depends on https://codereview.chromium.org/17587008

R=danno@chromium.org
BUG=

Review URL: https://codereview.chromium.org/18054002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15373 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFix zone usage within LAllocator.
bmeurer@chromium.org [Fri, 28 Jun 2013 07:36:43 +0000 (07:36 +0000)]
Fix zone usage within LAllocator.

Previously all allocations in the register allocator were done on the
chunk (and graph) zone, even those that did not need to survive past
the register allocation. Now only allocations that need to survive
use the chunk zone, all other allocations use a separate zone local
to the LAllocator.

Depends on https://codereview.chromium.org/18022002

R=danno@chromium.org
BUG=

Review URL: https://codereview.chromium.org/17553004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15372 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFix Android V8 path for testing on devices.
ulan@chromium.org [Fri, 28 Jun 2013 07:31:23 +0000 (07:31 +0000)]
Fix Android V8 path for testing on devices.

This allows to run tests in production Android builds.

R=yangguo@chromium.org

Review URL: https://chromiumcodereview.appspot.com/18087002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15371 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoMigrated several tests from blink to V8 repository.
machenbach@chromium.org [Thu, 27 Jun 2013 20:02:04 +0000 (20:02 +0000)]
Migrated several tests from blink to V8 repository.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18068003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15370 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoMake webkit test output comparison compatible to stress testing.
machenbach@chromium.org [Thu, 27 Jun 2013 19:45:45 +0000 (19:45 +0000)]
Make webkit test output comparison compatible to stress testing.

In stress testing, the output is repeated several times. In this case, it is now compared several times to the actual output.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/18062002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15369 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoMIPS: The check for internalized strings relied on the fact that we had less than...
palfia@homejinni.com [Thu, 27 Jun 2013 19:19:48 +0000 (19:19 +0000)]
MIPS: The check for internalized strings relied on the fact that we had less than 64 distinct InstanceTypes.

Port r15358 (3af77be)

Original commit message:
The check for internalized strings relied on the fact that we had less
than 64 distinct InstanceTypes. We are hitting that boundary, so this
check needs to be more comprehensive. In fact, two bits need to be
tested: verify that kNotStringTag isn't set, and that kInternalizedTag
is set.

BUG=

Review URL: https://codereview.chromium.org/18083004
Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15368 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRemove superfluous HInnerAllocatedObject in BuildEmitDeepCopy
danno@chromium.org [Thu, 27 Jun 2013 16:07:55 +0000 (16:07 +0000)]
Remove superfluous HInnerAllocatedObject in BuildEmitDeepCopy

R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/17881004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15367 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoType handling for special cases and optimization
danno@chromium.org [Thu, 27 Jun 2013 15:31:49 +0000 (15:31 +0000)]
Type handling for special cases and optimization

R=rossberg@chromium.org

Review URL: https://codereview.chromium.org/18078002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15366 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRevert r15361 "Improved function entry hook coverage" because of ARM build error.
ulan@chromium.org [Thu, 27 Jun 2013 15:31:06 +0000 (15:31 +0000)]
Revert r15361 "Improved function entry hook coverage" because of ARM build error.

R=siggi@chromium.org
BUG=

Review URL: https://chromiumcodereview.appspot.com/18062006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15365 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoTurn off stress compaction for break point tests.
hpayer@chromium.org [Thu, 27 Jun 2013 15:19:04 +0000 (15:19 +0000)]
Turn off stress compaction for break point tests.

BUG=
R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/17880005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15364 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFix check style breakage.
hpayer@chromium.org [Thu, 27 Jun 2013 15:17:55 +0000 (15:17 +0000)]
Fix check style breakage.

BUG=
R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/18056005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15363 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoFix constructor call of relocation lock.
yangguo@chromium.org [Thu, 27 Jun 2013 14:44:36 +0000 (14:44 +0000)]
Fix constructor call of relocation lock.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/18020002

Patch from Haitao Feng <haitao.feng@intel.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15362 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoImproved function entry hook coverage
danno@chromium.org [Thu, 27 Jun 2013 14:39:37 +0000 (14:39 +0000)]
Improved function entry hook coverage

Adds more coverage for function entry hook, sufficient to capture profiles that are contiguous from C++, through JS and back out to C++.

R=danno@chromium.org

Review URL: https://codereview.chromium.org/16578008

Patch from Sigurður Ásgeirsson <siggi@chromium.org>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15361 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoAdd templatized convienience functions for adding hydrogen instructions
danno@chromium.org [Thu, 27 Jun 2013 14:36:14 +0000 (14:36 +0000)]
Add templatized convienience functions for adding hydrogen instructions

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/18050004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15360 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoMake sure HCallNewArray uses right ElementsKind
danno@chromium.org [Thu, 27 Jun 2013 14:22:07 +0000 (14:22 +0000)]
Make sure HCallNewArray uses right ElementsKind

Close a small gc window that exists between when a HNewArrayCall instruction is
created and the ElementsKind type feedback for the construction function is
accesses.

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/18059005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15359 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoThe check for internalized strings relied on the fact that we had less
mvstanton@chromium.org [Thu, 27 Jun 2013 13:39:44 +0000 (13:39 +0000)]
The check for internalized strings relied on the fact that we had less
than 64 distinct InstanceTypes. We are hitting that boundary, so this
check needs to be more comprehensive. In fact, two bits need to be
tested: verify that kNotStringTag isn't set, and that kInternalizedTag
is set.

BUG=
R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/17895002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15358 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoGeneralize utilities to allow code templatization
danno@chromium.org [Thu, 27 Jun 2013 13:36:15 +0000 (13:36 +0000)]
Generalize utilities to allow code templatization

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/17853004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15357 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

11 years agoRefactor Hydrogen environment liveness analysis into an HPhase.
bmeurer@chromium.org [Thu, 27 Jun 2013 13:15:10 +0000 (13:15 +0000)]
Refactor Hydrogen environment liveness analysis into an HPhase.

Rename EnvironmentSlotLivenessAnalyzer to HEnvironmentLivenessAnalysisPhase,
following naming scheme suggested by danno@chromium.org in
https://codereview.chromium.org/17458002

The environment slot liveness analysis now uses the phase zone for
all its allocations.

Depends on https://codereview.chromium.org/18034003

R=danno@chromium.org
BUG=

Review URL: https://codereview.chromium.org/17587008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@15356 ce2b1a6d-e550-0410-aec6-3dcde31c8c00