platform/upstream/v8.git
14 years agoFix memory leaks in test-api.cc to make valgrind happy.
ricow@chromium.org [Tue, 13 Jul 2010 11:31:47 +0000 (11:31 +0000)]
Fix memory leaks in test-api.cc to make valgrind happy.

The memory leaks are all from the tests, not actually in V8.

I will put a nightly valgrind run on the fuzzer.

Review URL: http://codereview.chromium.org/2944015

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5052 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix JSON.parse typo which causes the input not to be string converted.
ager@chromium.org [Tue, 13 Jul 2010 10:57:45 +0000 (10:57 +0000)]
Fix JSON.parse typo which causes the input not to be string converted.

Review URL: http://codereview.chromium.org/2981004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5051 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix preparsing from a source string that is not external.
ager@chromium.org [Tue, 13 Jul 2010 10:29:31 +0000 (10:29 +0000)]
Fix preparsing from a source string that is not external.

This fixes issue 775.
Review URL: http://codereview.chromium.org/2959007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5050 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoSimplifying GenerateDictionaryNegativeLookup by making the 'extra' parameter mandatory.
serya@chromium.org [Tue, 13 Jul 2010 10:02:11 +0000 (10:02 +0000)]
Simplifying GenerateDictionaryNegativeLookup by making the 'extra' parameter mandatory.
Review URL: http://codereview.chromium.org/2982001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5049 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix compilation error on 64-bit MacOS V8 compilation. ASSERT_EQ macro needs unambigu...
whesse@chromium.org [Tue, 13 Jul 2010 08:55:06 +0000 (08:55 +0000)]
Fix compilation error on 64-bit MacOS V8 compilation.  ASSERT_EQ macro needs unambiguous similar integer types as arguments.
Review URL: http://codereview.chromium.org/2982005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5048 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoVirtually dispatched scavengers
vegorov@chromium.org [Tue, 13 Jul 2010 08:05:10 +0000 (08:05 +0000)]
Virtually dispatched scavengers

Use virtually dispatched specialized scavengers instead of single generic ScavengeObjectSlow implementation.

Rollback of r5041 with assertion checking callback alignment removed. Map space is iterated in a special fashion during scavenges so special callback alignment is not required.

Review URL: http://codereview.chromium.org/2950003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5047 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRevert revisions 5041 and 5042 introducing virtual scavenge
ager@chromium.org [Mon, 12 Jul 2010 16:57:07 +0000 (16:57 +0000)]
Revert revisions 5041 and 5042 introducing virtual scavenge
behavior. It breaks debug builds with snapshots on my machine.

TBR=vegorov@chromium.org
Review URL: http://codereview.chromium.org/2983001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5046 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRemove the special error message for overflows when using
ager@chromium.org [Mon, 12 Jul 2010 15:29:25 +0000 (15:29 +0000)]
Remove the special error message for overflows when using
Function.prototype.apply.  This avoids having more than one error
message for stack overflow situations which makes testing a pain.

Review URL: http://codereview.chromium.org/2967003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5045 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPort prototype-call-stubs for normal objects (http://codereview.chromium.org/2801018).
serya@chromium.org [Mon, 12 Jul 2010 13:23:42 +0000 (13:23 +0000)]
Port prototype-call-stubs for normal objects (codereview.chromium.org/2801018).

Review URL: http://codereview.chromium.org/2860049

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5044 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAllow to capture stack trace for uncaught exceptions
yurys@chromium.org [Mon, 12 Jul 2010 13:17:27 +0000 (13:17 +0000)]
Allow to capture stack trace for uncaught exceptions
Review URL: http://codereview.chromium.org/2961003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5043 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix Win64 build problem introduced by r5041.
vegorov@chromium.org [Mon, 12 Jul 2010 13:11:42 +0000 (13:11 +0000)]
Fix Win64 build problem introduced by r5041.

Review URL: http://codereview.chromium.org/2893009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5042 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoUse virtually dispatched specialized scavengers instead of single generic ScavengeObj...
vegorov@chromium.org [Mon, 12 Jul 2010 12:47:09 +0000 (12:47 +0000)]
Use virtually dispatched specialized scavengers instead of single generic ScavengeObjectSlow implementation.

Review URL: http://codereview.chromium.org/2895008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5041 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoCode cleanup: reorder functions in codegen-x64.cc to agree with the order in codegen...
whesse@chromium.org [Fri, 9 Jul 2010 13:20:12 +0000 (13:20 +0000)]
Code cleanup: reorder functions in codegen-x64.cc to agree with the order in codegen-ia32.cc.  If svn blame shows this change, run svn blame on the previous version of the file to find the actual author of the lines.
Review URL: http://codereview.chromium.org/2955004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5040 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoSecond half of number alloc change accidentally omitted from
erik.corry@gmail.com [Thu, 8 Jul 2010 13:50:13 +0000 (13:50 +0000)]
Second half of number alloc change accidentally omitted from
last change.
Review URL: http://codereview.chromium.org/2919001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5039 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd number allocation call to runtime.cc instead of misusing
erik.corry@gmail.com [Thu, 8 Jul 2010 13:16:51 +0000 (13:16 +0000)]
Add number allocation call to runtime.cc instead of misusing
UnaryMinus.
Review URL: http://codereview.chromium.org/2899004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5038 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoARM: Add support for the VFP mov literal instruction and mov
erik.corry@gmail.com [Thu, 8 Jul 2010 12:38:02 +0000 (12:38 +0000)]
ARM: Add support for the VFP mov literal instruction and mov
between single VFP registers.  Math.pow implementation has
been updated with the new instructions.  This is a commit
of http://codereview.chromium.org/2813046/show for Rodolph
Perfetta.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5037 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd ReleaseStore function on OpenBSD.
erik.corry@gmail.com [Thu, 8 Jul 2010 08:12:17 +0000 (08:12 +0000)]
Add ReleaseStore function on OpenBSD.
Review URL: http://codereview.chromium.org/2916001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5036 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix concurrent access to VMState::current_state_.
mikhail.naganov@gmail.com [Thu, 8 Jul 2010 08:00:38 +0000 (08:00 +0000)]
Fix concurrent access to VMState::current_state_.

The main fix is for current_state() and external_callback() accessors.
I also applied memory access ordering on current_state_ modification,
mainly to reflect the fact that it is being shared among VM and profiler
sampler threads.

BUG=361

Review URL: http://codereview.chromium.org/2852047

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5035 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPrepare push to trunk. We are now working on version 2.2.24.
erik.corry@gmail.com [Wed, 7 Jul 2010 12:09:22 +0000 (12:09 +0000)]
Prepare push to trunk. We are now working on version 2.2.24.
Review URL: http://codereview.chromium.org/2887011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5032 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoTest for correct exception message in invalid preparse data test in test-api.cc
whesse@chromium.org [Wed, 7 Jul 2010 11:31:33 +0000 (11:31 +0000)]
Test for correct exception message in invalid preparse data test in test-api.cc
Review URL: http://codereview.chromium.org/2829049

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5031 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPrevent invalid pre-parsing data passed in through the API from crashing V8.
whesse@chromium.org [Wed, 7 Jul 2010 10:28:22 +0000 (10:28 +0000)]
Prevent invalid pre-parsing data passed in through the API from crashing V8.
Review URL: http://codereview.chromium.org/2876046

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5030 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix crash introduced in r5019.
kaznacheev@chromium.org [Tue, 6 Jul 2010 13:48:51 +0000 (13:48 +0000)]
Fix crash introduced in r5019.

Note to self: never leave uninitialized objects in the code space.

Review URL: http://codereview.chromium.org/2800044

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5029 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoMake lint happy.
vitalyr@chromium.org [Tue, 6 Jul 2010 13:43:21 +0000 (13:43 +0000)]
Make lint happy.

Review URL: http://codereview.chromium.org/2832051

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5028 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRemove some extra calls in date.js.
vitalyr@chromium.org [Tue, 6 Jul 2010 12:57:46 +0000 (12:57 +0000)]
Remove some extra calls in date.js.

A few hot Date.prototype functions used to be implemented as
DateGetFoo calling GetFooFrom(this), and other callers of GetFooFrom
did repeated date value extraction.  In this change GetFooFrom
functions were inlined into the functions on the prototype and the
other callers switched to using date values directly.

Review URL: http://codereview.chromium.org/2841041

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5027 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoSimplify the transitions in the Binary Op ICs. Now a single call
erik.corry@gmail.com [Tue, 6 Jul 2010 12:56:55 +0000 (12:56 +0000)]
Simplify the transitions in the Binary Op ICs.  Now a single call
to the runtime will both patch in the more specialized binary op
stub and calculate the answer.  This eliminates the need to call
both the rest of the binary op and the patching runtime call.  The
runtime routines are altered to be more agressive in returning
Smis so we don't get spurious heap numbers as inputs to binary ops
while we are patching the binary op ICs.
Review URL: http://codereview.chromium.org/2843049

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5026 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix crash: handle all flat string types in regexp replace.
vitalyr@chromium.org [Tue, 6 Jul 2010 12:22:19 +0000 (12:22 +0000)]
Fix crash: handle all flat string types in regexp replace.

Review URL: http://codereview.chromium.org/2868046

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5025 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoDo not create handle in BreakLocationIterator::ClearDebugBreakAtIC
podivilov@chromium.org [Tue, 6 Jul 2010 12:10:49 +0000 (12:10 +0000)]
Do not create handle in BreakLocationIterator::ClearDebugBreakAtIC

Debug::HandleWeakDebugInfo weak callback is crashed when there is no handle scopes on the stack.

Review URL: http://codereview.chromium.org/2835035

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5024 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoConvert Unicode code points outside the basic multilingual plane to the replacement...
lrn@chromium.org [Tue, 6 Jul 2010 07:31:35 +0000 (07:31 +0000)]
Convert Unicode code points outside the basic multilingual plane to the replacement character.
Previous behavior was to silently truncate the value to 16 bits.

Review URL: http://codereview.chromium.org/2832050

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5023 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoJS2C tweaks:
vitalyr@chromium.org [Mon, 5 Jul 2010 14:28:37 +0000 (14:28 +0000)]
JS2C tweaks:

 o Only substitute constant values for full-words and not for
   accidental partial matches.

 o Keep the order of macros as in macros.py and allow depending on the
   previous macros.  We used to allow depending on macros that
   happened to sort after the current one in the dictionary.

Review URL: http://codereview.chromium.org/2800042

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5022 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRegExp replace with empty string optimization by Sandholm.
lrn@chromium.org [Mon, 5 Jul 2010 12:49:00 +0000 (12:49 +0000)]
RegExp replace with empty string optimization by Sandholm.

Review URL: http://codereview.chromium.org/2809048

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5021 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoExtracting relocation info from the code object.
kaznacheev@chromium.org [Mon, 5 Jul 2010 11:45:11 +0000 (11:45 +0000)]
Extracting relocation info from the code object.

It is now stored in a ByteArray referenced from the code object header.

Review URL: http://codereview.chromium.org/2812041

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5020 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoSpecialize GenericUnaryStub so that it knows whether it needs to
erik.corry@gmail.com [Mon, 5 Jul 2010 11:03:16 +0000 (11:03 +0000)]
Specialize GenericUnaryStub so that it knows whether it needs to
take negative zero into account.
Review URL: http://codereview.chromium.org/2850043

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5018 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPrepare push to trunk. Now working on version 2.2.23.
kmillikin@chromium.org [Mon, 5 Jul 2010 10:33:20 +0000 (10:33 +0000)]
Prepare push to trunk. Now working on version 2.2.23.

Review URL: http://codereview.chromium.org/2827046

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5016 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoDescribe LiveEdit changes and support preview mode
peter.rybin@gmail.com [Fri, 2 Jul 2010 20:46:04 +0000 (20:46 +0000)]
Describe LiveEdit changes and support preview mode

Review URL: http://codereview.chromium.org/2883020

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5015 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoMake ignore-negative-zeros analysis a little sharper.
erik.corry@gmail.com [Fri, 2 Jul 2010 20:24:40 +0000 (20:24 +0000)]
Make ignore-negative-zeros analysis a little sharper.
Review URL: http://codereview.chromium.org/2799048

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5014 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix bug in date code (issue 736) where -0 was not mapped to 0. This caused the
ager@chromium.org [Fri, 2 Jul 2010 20:09:33 +0000 (20:09 +0000)]
Fix bug in date code (issue 736) where -0 was not mapped to 0. This caused the
runtime system to throw an exception because it expected smi arguments.

Review URL: http://codereview.chromium.org/2848038

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5013 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoARM: Use VFP for non-Smi bitops results. This is basically Rodolph
erik.corry@gmail.com [Fri, 2 Jul 2010 20:03:07 +0000 (20:03 +0000)]
ARM: Use VFP for non-Smi bitops results.  This is basically Rodolph
Perfettas change from http://codereview.chromium.org/2829032 with
a tiny tweak to the unsigned shift right case.
Review URL: http://codereview.chromium.org/2861043

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5012 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd ES5 Object.isExtensible and Object.preventExtensions.
ricow@chromium.org [Fri, 2 Jul 2010 14:36:34 +0000 (14:36 +0000)]
Add ES5 Object.isExtensible and Object.preventExtensions.

Review URL: http://codereview.chromium.org/2819034

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5011 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoThis change allows generating call-stubs for objects with normal (non-fast) objects...
serya@chromium.org [Fri, 2 Jul 2010 14:15:04 +0000 (14:15 +0000)]
This change allows generating call-stubs for objects with normal (non-fast) objects in the prototype chain. StubCompiler::CheckPrototypes does ne

If the top level object is a normal object the stub is stored in its prototype map.

Lookup result of type NORMAL is not covered (since the normal stub currently doesn't check the prototype chain).
Review URL: http://codereview.chromium.org/2801018

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5010 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoX64: Added register holding Smi::FromInt(1).
lrn@chromium.org [Fri, 2 Jul 2010 14:09:35 +0000 (14:09 +0000)]
X64: Added register holding Smi::FromInt(1).
Don't use r15 for anything any more.

Review URL: http://codereview.chromium.org/2885018

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5009 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoUse the number of in-object properties when deciding how many fast
ager@chromium.org [Fri, 2 Jul 2010 11:27:57 +0000 (11:27 +0000)]
Use the number of in-object properties when deciding how many fast
properties to allow on an object. If there are many in-object
properties it is unlikely that the object is used as a dictionary and
we allow more map transitions to keep such objects in fast case.

Review URL: http://codereview.chromium.org/2818041

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5008 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoSimplify ToBoolean if we know we have a Smi.
erik.corry@gmail.com [Fri, 2 Jul 2010 08:42:22 +0000 (08:42 +0000)]
Simplify ToBoolean if we know we have a Smi.
Review URL: http://codereview.chromium.org/2808039

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5007 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoLanding for victorw@chromium.org.
ager@chromium.org [Fri, 2 Jul 2010 07:39:42 +0000 (07:39 +0000)]
Landing for victorw@chromium.org.

Fix issues so v8 could be built as a DLL.

-. get rid of all the compiler warning by moving dllexport/dllimport
to the individual members for classes which have inline members.

-. update v8 gyp to build v8.dll for chromium multi-dll version (win
and component==shared_library)

Note: most of the code are contributed by sjesse.

Code review URL: http://codereview.chromium.org/2882009/show

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5006 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix problem in r4998. Initialize target of break and continue in fast smi loop on...
whesse@chromium.org [Thu, 1 Jul 2010 16:00:51 +0000 (16:00 +0000)]
Fix problem in r4998.  Initialize target of break and continue in fast smi loop on X64.
Review URL: http://codereview.chromium.org/2830033

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5001 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoLet relational comparison operators quickly compare a string with itself.
whesse@chromium.org [Thu, 1 Jul 2010 15:18:07 +0000 (15:18 +0000)]
Let relational comparison operators quickly compare a string with itself.
Review URL: http://codereview.chromium.org/2802028

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@5000 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoARM: Don't emit a write barrier for an inlined keyed load
erik.corry@gmail.com [Thu, 1 Jul 2010 15:06:24 +0000 (15:06 +0000)]
ARM: Don't emit a write barrier for an inlined keyed load
if the right hand side is a literal like true, false, etc.
Also if the value is not a likely Smi we inline the newspace
check.
Review URL: http://codereview.chromium.org/2833048

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4999 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoX64: Made simpler version of fast-smi-loop code.
lrn@chromium.org [Thu, 1 Jul 2010 13:19:32 +0000 (13:19 +0000)]
X64: Made simpler version of fast-smi-loop code.
It avoids the overflow check on the increment.

Review URL: http://codereview.chromium.org/2835030

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4998 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoClear function breakpoints in Debug::HandleWeakDebugInfo callback.
podivilov@chromium.org [Thu, 1 Jul 2010 11:48:45 +0000 (11:48 +0000)]
Clear function breakpoints in Debug::HandleWeakDebugInfo callback.

When removing debug_info from SharedFunctionInfo, clear all breakpoints associated with that debug_info.
This is needed because function will live in heap until next gc, and therefore can be found
by Runtime::FindSharedFunctionInfoInScript.

Review URL: http://codereview.chromium.org/2848032

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4997 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoLiveEdit: Support debug break slots in frame dropping implementation
peter.rybin@gmail.com [Wed, 30 Jun 2010 17:29:00 +0000 (17:29 +0000)]
LiveEdit: Support debug break slots in frame dropping implementation

Review URL: http://codereview.chromium.org/2844030

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4995 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoARM: Correctness fix to Math.pow optimization
sgjesse@chromium.org [Wed, 30 Jun 2010 15:19:06 +0000 (15:19 +0000)]
ARM: Correctness fix to Math.pow optimization

The change in r4990 contained a bug in Math.pow when then exponent was a large negative smi. In that case calculating 1/Math.pow(x,-y) did not provide the correct result as Math.pow(x,-y) would overflow ti infinity. This was caught by Sputnik test S8.5_A13_T1.
Review URL: http://codereview.chromium.org/2815039

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4994 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoARM: Fix GC bug with previous change
sgjesse@chromium.org [Wed, 30 Jun 2010 13:39:03 +0000 (13:39 +0000)]
ARM: Fix GC bug with previous change

The change in r4990 contained a bug in Math.pow. One of the arguments for the call to runtime where clobbered if allocation of the heap number for the result failed.

Caught by a couple of Mozilla tests which does a lot of calls to Math.pow.

Fix presubmit error in x64.
Review URL: http://codereview.chromium.org/2847037

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4993 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoDecide that different detectable objects are not equal without calling runtime, on...
whesse@chromium.org [Wed, 30 Jun 2010 13:24:47 +0000 (13:24 +0000)]
Decide that different detectable objects are not equal without calling runtime, on all platforms.
Review URL: http://codereview.chromium.org/2805046

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4992 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoImplement IC for storing to dictionary case objects.
ager@chromium.org [Wed, 30 Jun 2010 12:27:49 +0000 (12:27 +0000)]
Implement IC for storing to dictionary case objects.

The IC stub is completely generic, so there will only be one such stub
in the system.

Added a new overloaded version of the macro assembler RecordWrite
method for cases where we have the address we store to computed up
front.

Review URL: http://codereview.chromium.org/2804029

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4991 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoARM: Special code for raising to the power of an integer
sgjesse@chromium.org [Wed, 30 Jun 2010 12:22:15 +0000 (12:22 +0000)]
ARM: Special code for raising to the power of an integer

When calculating Math.pow where the exponent is a smi use a simple loop to calculate the result.

Added support for the vmov instruction moving from one doubleword extension register to another.

Added some Math.pow tests which partially covers what is in the Sputnik tests.
Review URL: http://codereview.chromium.org/2804033

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4990 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPrepare push to trunk. We are now working on version 2.2.22.
lrn@chromium.org [Wed, 30 Jun 2010 08:57:23 +0000 (08:57 +0000)]
Prepare push to trunk. We are now working on version 2.2.22.

Review URL: http://codereview.chromium.org/2886005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4987 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoARM: Fix disassembly of double registers
sgjesse@chromium.org [Wed, 30 Jun 2010 08:46:19 +0000 (08:46 +0000)]
ARM: Fix disassembly of double registers

Added a few disassembler tests for vadd, vsub, vmul and vdiv.
Review URL: http://codereview.chromium.org/2823036

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4986 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix Chromium issue 47824.
lrn@chromium.org [Wed, 30 Jun 2010 07:40:40 +0000 (07:40 +0000)]
Fix Chromium issue 47824.
In rare cases a two-byte string was mistaken for an ascii-string.

Review URL: http://codereview.chromium.org/2858033

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4985 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoUpdate the mozilla tests to the newest version.
ricow@chromium.org [Wed, 30 Jun 2010 06:51:38 +0000 (06:51 +0000)]
Update the mozilla tests to the newest version.

This includes an update to the test expectations, and a new bug found
when investigating the new failures introduced by the updated tests
(http://code.google.com/p/v8/issues/detail?id=762)

Review URL: http://codereview.chromium.org/2838030

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4984 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoX64: Make the ToBoolean inline code do even less if the value is known to be a smi.
lrn@chromium.org [Tue, 29 Jun 2010 14:15:34 +0000 (14:15 +0000)]
X64: Make the ToBoolean inline code do even less if the value is known to be a smi.

Review URL: http://codereview.chromium.org/2834028

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4978 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoX64: A number of small tweaks.
lrn@chromium.org [Tue, 29 Jun 2010 14:03:11 +0000 (14:03 +0000)]
X64: A number of small tweaks.
Fix bug in disassembler output (using non-generic registers for memory operands in some cases).
Fix bug in movsxbq and make REX optional in movzx?q.

Review URL: http://codereview.chromium.org/2834027

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4977 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoType info: fix three bit representation of smi type.
vitalyr@chromium.org [Tue, 29 Jun 2010 13:48:20 +0000 (13:48 +0000)]
Type info: fix three bit representation of smi type.

Review URL: http://codereview.chromium.org/2802022

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4976 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agotools/stats-viewer.py: add counter filter.
vitalyr@chromium.org [Tue, 29 Jun 2010 11:30:34 +0000 (11:30 +0000)]
tools/stats-viewer.py: add counter filter.

Review URL: http://codereview.chromium.org/2838028

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4975 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoARM: Use the vsqrt instruction when available
sgjesse@chromium.org [Tue, 29 Jun 2010 09:40:36 +0000 (09:40 +0000)]
ARM: Use the vsqrt instruction when available

vsqrt is used to calculate Math.sqrt(x), Math.pow(x, 0.5) and Math.pow(x, -0.5). Code size doesn't matter, as %_MathSqrt and %_MathPow are only called in one place each.
Review URL: http://codereview.chromium.org/2885002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4974 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoEven more cleanup of es5 test expectations.
ricow@chromium.org [Tue, 29 Jun 2010 09:00:20 +0000 (09:00 +0000)]
Even more cleanup of es5 test expectations.

Removed expectations for tests that was removed in the latest
revision. Also changed FAIL_OK to FAIL for test that we should
implement but currently don't (thanks to lasse for noticing this).

Review URL: http://codereview.chromium.org/2827033

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4973 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoUpdate JSON.stringify to floor the space parameter (fixes issue 753).
ricow@chromium.org [Tue, 29 Jun 2010 07:22:40 +0000 (07:22 +0000)]
Update JSON.stringify to floor the space parameter (fixes issue 753).
Review URL: http://codereview.chromium.org/2877004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4972 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoEnsure that ToPrimitive is called on all objects involved in comparisons <, <=, ...
whesse@chromium.org [Tue, 29 Jun 2010 06:47:19 +0000 (06:47 +0000)]
Ensure that ToPrimitive is called on all objects involved in comparisons <, <=, >, >=.  Ensures that ToPrimitive is called when comparing an object to undefined. Fixes bugs on all platforms.
Review URL: http://codereview.chromium.org/2834022

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4971 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoUpdate the ES5 conformance tests to the latest version and fix test
ricow@chromium.org [Tue, 29 Jun 2010 05:56:15 +0000 (05:56 +0000)]
Update the ES5 conformance tests to the latest version and fix test
expectations.

We are currently using a rather old version of the ES5 conformance
tests and there has been a range of bug fixes in the test set. This
change brings us up to date.

In addition, our current test expectations discards the all chapters
but chapter 15. I have enabled the other chapters, cleaned up the
tests that we no longer fail on, and filed bugs for tests that we fail on.

It seems some of the bugs on the es5 conformance bug-tracker has been
fixed but not marked fixed. I will file bugs for the newly discovered
test bugs.

Review URL: http://codereview.chromium.org/2840023

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4970 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoUpdate the V8 benchmark suite with the following fixes:
kasperl@chromium.org [Mon, 28 Jun 2010 12:28:37 +0000 (12:28 +0000)]
Update the V8 benchmark suite with the following fixes:

*) Fix a couple of typos in DeltaBlue.
*) Make sure Splay doesn't always remove the node just added.
*) Run all benchmarks for at least 32 times (and warm up).
Review URL: http://codereview.chromium.org/2836031

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4966 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoProvide actual breakpoints locations in response to setBreakpoint and listBreakpoints...
podivilov@chromium.org [Mon, 28 Jun 2010 12:09:29 +0000 (12:09 +0000)]
Provide actual breakpoints locations in response to setBreakpoint and listBreakpoints requests.

Review URL: http://codereview.chromium.org/2799037

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4965 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoDo integer mod via sum-of-digits technique. This benefits the date
erik.corry@gmail.com [Mon, 28 Jun 2010 11:47:23 +0000 (11:47 +0000)]
Do integer mod via sum-of-digits technique.  This benefits the date
code.
Review URL: http://codereview.chromium.org/2876011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4964 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPrepare push to trunk. We are now working on version 2.2.21.
fschneider@chromium.org [Mon, 28 Jun 2010 11:17:55 +0000 (11:17 +0000)]
Prepare push to trunk. We are now working on version 2.2.21.

Review URL: http://codereview.chromium.org/2882002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4961 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRemove not-both-smis optimization on FloatingPointHelper::LoadNumbersAsIntegers on...
whesse@chromium.org [Mon, 28 Jun 2010 10:54:07 +0000 (10:54 +0000)]
Remove not-both-smis optimization on FloatingPointHelper::LoadNumbersAsIntegers on X64 platform.
Review URL: http://codereview.chromium.org/2861031

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4960 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix 15.2.3.4 es5conform test
yurys@chromium.org [Mon, 28 Jun 2010 09:25:09 +0000 (09:25 +0000)]
Fix 15.2.3.4 es5conform test

TBR=mnaganov,sgjesse
Review URL: http://codereview.chromium.org/2832029

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4959 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoObject.getOwnPropertyNames should be free of duplicates
yurys@chromium.org [Mon, 28 Jun 2010 08:31:31 +0000 (08:31 +0000)]
Object.getOwnPropertyNames should be free of duplicates

BUG=41243
Review URL: http://codereview.chromium.org/2825026

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4958 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAvoid a potential null dereference wrt the CPU profiler.
lukezarko [Fri, 25 Jun 2010 15:57:30 +0000 (15:57 +0000)]
Avoid a potential null dereference wrt the CPU profiler.

GetActiveProfilerModules()/PauseProfiler()/ResumeProfiler() can be reached
from the API when the --prof runtime flag is not set, leading to null
dereferences. Verify that Logger::profiler_ is non-NULL before using it.

Review URL: http://codereview.chromium.org/2843023

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4953 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix bug in CheckBothPositiveSmi macro.
lrn@chromium.org [Fri, 25 Jun 2010 12:31:49 +0000 (12:31 +0000)]
Fix bug in CheckBothPositiveSmi macro.
Fix lint.

Review URL: http://codereview.chromium.org/2826021

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4952 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd special case of taking a substring that is the entire string to stub.
lrn@chromium.org [Fri, 25 Jun 2010 11:24:41 +0000 (11:24 +0000)]
Add special case of taking a substring that is the entire string to stub.

Review URL: http://codereview.chromium.org/2800027

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4951 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFixes bug in Array.prototype.lastIndexOf when called with null or undefined as fromIn...
ricow@chromium.org [Fri, 25 Jun 2010 09:28:38 +0000 (09:28 +0000)]
Fixes bug in Array.prototype.lastIndexOf when called with null or undefined as fromIndex argument. (fixes issue 754).

Review URL: http://codereview.chromium.org/2840021

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4950 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoDon't make any decisions on old space limits after scavenges.
vegorov@chromium.org [Fri, 25 Jun 2010 08:59:52 +0000 (08:59 +0000)]
Don't make any decisions on old space limits after scavenges.

We just do not possess enough information at that point.

Review URL: http://codereview.chromium.org/2836026

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4949 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoLess intrusive patch for regexp bug.
lrn@chromium.org [Fri, 25 Jun 2010 07:51:46 +0000 (07:51 +0000)]
Less intrusive patch for regexp bug.

Review URL: http://codereview.chromium.org/2843024

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4947 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix bug in JSON.stringify where Boolean objects are incorrectly
ager@chromium.org [Fri, 25 Jun 2010 07:45:52 +0000 (07:45 +0000)]
Fix bug in JSON.stringify where Boolean objects are incorrectly
unwrapped.

This fixes issue 752.
Review URL: http://codereview.chromium.org/2845023

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4946 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix bug in regexp exec with global regexps.
lrn@chromium.org [Fri, 25 Jun 2010 07:00:29 +0000 (07:00 +0000)]
Fix bug in regexp exec with global regexps.

Review URL: http://codereview.chromium.org/2826020

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4945 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPort optimization of comparison with a trivial LHS from ia32 to x64.
whesse@chromium.org [Thu, 24 Jun 2010 14:22:25 +0000 (14:22 +0000)]
Port optimization of comparison with a trivial LHS from ia32 to x64.
Review URL: http://codereview.chromium.org/2868028

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4943 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix error in for-in on x64 platform using full compiler with keyed store IC.
whesse@chromium.org [Thu, 24 Jun 2010 14:02:36 +0000 (14:02 +0000)]
Fix error in for-in on x64 platform using full compiler with keyed store IC.

BUG=v8:748

http://code.google.com/p/v8/issues/detail?id=748

Review URL: http://codereview.chromium.org/2810027

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4942 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd "has fast elements" bit to maps and use it in inlined keyed loads.
vitalyr@chromium.org [Thu, 24 Jun 2010 13:56:35 +0000 (13:56 +0000)]
Add "has fast elements" bit to maps and use it in inlined keyed loads.

A potential issue with this change is creating lots of maps when
objects flip between fast/slow elements modes.  We could add special
transitions to avoid this.  Yet testing this on our benchmarks, gmail,
and wave seems to indicate that this is not a real problem.

Review URL: http://codereview.chromium.org/2870018

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4941 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFixing the regression introduced in r4716.
kaznacheev@chromium.org [Thu, 24 Jun 2010 12:31:49 +0000 (12:31 +0000)]
Fixing the regression introduced in r4716.

The regression made 2 tests fail on ia32 with --always-full-compiler.

Review URL: http://codereview.chromium.org/2862028

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4940 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix compilation failure on Win.
vegorov@chromium.org [Thu, 24 Jun 2010 12:11:24 +0000 (12:11 +0000)]
Fix compilation failure on Win.

Review URL: http://codereview.chromium.org/2821016

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4939 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoTake survival rates of young objects into account when choosing old generation limits.
vegorov@chromium.org [Thu, 24 Jun 2010 11:44:23 +0000 (11:44 +0000)]
Take survival rates of young objects into account when choosing old generation limits.

Stable high survival rates of young objects both during partial and
full collection indicate that mutator is either building or modifying
a structure with a long lifetime.

In this case we aggressively raise old generation memory limits to
postpone subsequent mark-sweep collection and trade memory space
for the mutation speed.

Review URL: http://codereview.chromium.org/2809032

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4938 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoChange kMinComplexMemCopy to 64 for IA32. It's just better.
sandholm@chromium.org [Thu, 24 Jun 2010 11:21:58 +0000 (11:21 +0000)]
Change kMinComplexMemCopy to 64 for IA32. It's just better.
Review URL: http://codereview.chromium.org/2818028

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4937 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoARM: Remove spilled scopes.
erik.corry@gmail.com [Thu, 24 Jun 2010 11:05:38 +0000 (11:05 +0000)]
ARM: Remove spilled scopes.
Review URL: http://codereview.chromium.org/2848023

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4936 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agofixed
mikhail.naganov@gmail.com [Thu, 24 Jun 2010 10:05:57 +0000 (10:05 +0000)]
fixed

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4935 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoX64: Remove more fpu code. Unroll more local initialization loops.
lrn@chromium.org [Thu, 24 Jun 2010 09:03:49 +0000 (09:03 +0000)]
X64: Remove more fpu code. Unroll more local initialization loops.

Review URL: http://codereview.chromium.org/2815028

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4934 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoMake SampleRateCalculator test resilient to float precision issues.
mikhail.naganov@gmail.com [Thu, 24 Jun 2010 08:57:24 +0000 (08:57 +0000)]
Make SampleRateCalculator test resilient to float precision issues.

Tested with kSamplingIntervalMs values 2 and 4.

Review URL: http://codereview.chromium.org/2811021

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4933 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoARM: Remove a bunch of spilled scopes. Still a lot to go.
erik.corry@gmail.com [Thu, 24 Jun 2010 07:54:48 +0000 (07:54 +0000)]
ARM: Remove a bunch of spilled scopes.  Still a lot to go.
Review URL: http://codereview.chromium.org/2828004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4932 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoX64: Change some fpu operations to use XMM registers.
lrn@chromium.org [Wed, 23 Jun 2010 14:05:18 +0000 (14:05 +0000)]
X64: Change some fpu operations to use XMM registers.

Review URL: http://codereview.chromium.org/2827022

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4930 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoARM: Fix bug introduced in 4783 (2.2.15) that caused the
erik.corry@gmail.com [Wed, 23 Jun 2010 13:44:11 +0000 (13:44 +0000)]
ARM: Fix bug introduced in 4783 (2.2.15) that caused the
result of 1 << x to be miscalculated for some inputs.
Review URL: http://codereview.chromium.org/2848021

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4929 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoX64: A bunch of small fixes.
lrn@chromium.org [Wed, 23 Jun 2010 11:48:30 +0000 (11:48 +0000)]
X64: A bunch of small fixes.
Make push/pop use emit_optional_rex32.
Fix bug in disassembler (swapped name of comisd/ucomisd).
Use fstp in FCmp macro.

Review URL: http://codereview.chromium.org/2818026

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4928 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoUse SSE2 registers when comparing identical heap numbers on X64.
whesse@chromium.org [Wed, 23 Jun 2010 09:21:32 +0000 (09:21 +0000)]
Use SSE2 registers when comparing identical heap numbers on X64.
Review URL: http://codereview.chromium.org/2850018

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4927 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRemove redundant checks in and around GenerateDictionaryLoad.
kaznacheev@chromium.org [Wed, 23 Jun 2010 09:10:21 +0000 (09:10 +0000)]
Remove redundant checks in and around GenerateDictionaryLoad.

Similar or duplicate checks are scattered around the code before doing the dictionary load.
Also the entire branch in GenerateCallNormal that handles global/builtin receiver is
guaranteed to bail out from GenerateDictionaryLoad, so there is no point in generating it at all.

The purpose of the patch is:
- making C++ code more compact and transparent,
- not generating dead code.

There is a tiny performance gain. The patch is ia32 only for now.

Please tell me if I am missing anything.

Review URL: http://codereview.chromium.org/2801007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4926 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPrepare push to trunk. Now working on version 2.2.20.
ricow@chromium.org [Wed, 23 Jun 2010 08:51:53 +0000 (08:51 +0000)]
Prepare push to trunk. Now working on version 2.2.20.
Review URL: http://codereview.chromium.org/2815025

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@4923 ce2b1a6d-e550-0410-aec6-3dcde31c8c00