platform/upstream/v8.git
14 years agoFix DebugEvaluateWithoutStack test on Windows
peter.rybin@gmail.com [Tue, 19 Jan 2010 15:12:43 +0000 (15:12 +0000)]
Fix DebugEvaluateWithoutStack test on Windows

Review URL: http://codereview.chromium.org/543115

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3653 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd conditional counter update to IA-32
sgjesse@chromium.org [Tue, 19 Jan 2010 14:17:34 +0000 (14:17 +0000)]
Add conditional counter update to IA-32
Review URL: http://codereview.chromium.org/554001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3652 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix some usage of "this" in builtins
sgjesse@chromium.org [Tue, 19 Jan 2010 14:15:47 +0000 (14:15 +0000)]
Fix some usage of "this" in builtins

The implementation of Object.prototype.valueOf and Object.prototype.toString now calls ToObject on "this" as mandated by the spec.
Review URL: http://codereview.chromium.org/542112

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3651 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoMake tests pass.
antonm@chromium.org [Tue, 19 Jan 2010 13:26:40 +0000 (13:26 +0000)]
Make tests pass.

TBR=kmillikin@chromium.org

Review URL: http://codereview.chromium.org/543114

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3648 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix a build.
antonm@chromium.org [Tue, 19 Jan 2010 13:15:52 +0000 (13:15 +0000)]
Fix a build.

Some tests still fail.

TBR=ager@chromium.org

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3647 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdded Extensible property to objects and made methods for extracting and setting it.
ricow@chromium.org [Tue, 19 Jan 2010 12:56:36 +0000 (12:56 +0000)]
Added Extensible property to objects and made methods for extracting and setting it.
Also added one method to runtime to get the extensible value
Additionally, added a check on the number of arguments in the start of GetOwnProperty.

Review URL: http://codereview.chromium.org/545116

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3646 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFirst step of refactoring expression contexts in the toplevel code
kmillikin@chromium.org [Tue, 19 Jan 2010 12:41:02 +0000 (12:41 +0000)]
First step of refactoring expression contexts in the toplevel code
generator.

Contexts are no longer stored in the AST but in the code generator's
state.  This means that the running the code generator selector is not
required to use the toplevel code generator (for instance, if we
already know that we can and should use it).

Review URL: http://codereview.chromium.org/546075

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3645 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoX64: Fix bug that reads from rax instead of correct register.
lrn@chromium.org [Tue, 19 Jan 2010 12:27:04 +0000 (12:27 +0000)]
X64: Fix bug that reads from rax instead of correct register.
Crashes if rax doesn't point to readable memory.
Recent change made rax contain garbage due to write-barrier computation.
(The fact that it hasn't crashed before highly suggests that the line does nothing.)

Review URL: http://codereview.chromium.org/545117

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3644 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFixed issue 582: set the right construct stub for native functions.
vitalyr@chromium.org [Tue, 19 Jan 2010 10:32:20 +0000 (10:32 +0000)]
Fixed issue 582: set the right construct stub for native functions.

TEST=cctest/test-api/NativeFunctionConstructCall
BUG=582

Review URL: http://codereview.chromium.org/551063

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3643 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdded ES5 conform tests for getOwnPropertyNames.
ricow@chromium.org [Tue, 19 Jan 2010 08:23:29 +0000 (08:23 +0000)]
Added ES5 conform tests for getOwnPropertyNames.

Most of the test cases fail as the different objects according to the tests
can not have additional properties attached to them. I will file a bug report
on the es5 conform site as they should allow this. Some of the test fails
because we still miss some of the es5 features used.

Review URL: http://codereview.chromium.org/545109

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3642 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix Windows build by using OS::StrNCpy.
ager@chromium.org [Tue, 19 Jan 2010 08:03:26 +0000 (08:03 +0000)]
Fix Windows build by using OS::StrNCpy.

Fix lint issue in lineprocessor.cc.

TBR=peter.rybin
Review URL: http://codereview.chromium.org/549081

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3641 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd missing break in continue.
kmillikin@chromium.org [Mon, 18 Jan 2010 23:18:09 +0000 (23:18 +0000)]
Add missing break in continue.

TBR=fschneider@chromium.org

Review URL: http://codereview.chromium.org/542107

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3640 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd another test on debug global evaluate
peter.rybin@gmail.com [Mon, 18 Jan 2010 19:14:27 +0000 (19:14 +0000)]
Add another test on debug global evaluate

Review URL: http://codereview.chromium.org/553009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3639 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix test-serialize.
mikhail.naganov@gmail.com [Mon, 18 Jan 2010 16:56:17 +0000 (16:56 +0000)]
Fix test-serialize.

Sorry for a breakage...

TBR=erik.corry@gmail.com

Review URL: http://codereview.chromium.org/536090

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3638 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd "lineprocessor" sample program
peter.rybin@gmail.com [Mon, 18 Jan 2010 16:45:13 +0000 (16:45 +0000)]
Add "lineprocessor" sample program

Review URL: http://codereview.chromium.org/491006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3637 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix a bug in the short-circuit logical operations in the toplevel
kmillikin@chromium.org [Mon, 18 Jan 2010 16:23:24 +0000 (16:23 +0000)]
Fix a bug in the short-circuit logical operations in the toplevel
code generator.

Review URL: http://codereview.chromium.org/542105

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3636 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix issue 571: display descriptive names for code objects from snapshot.
mikhail.naganov@gmail.com [Mon, 18 Jan 2010 16:04:25 +0000 (16:04 +0000)]
Fix issue 571: display descriptive names for code objects from snapshot.

As this is only needed for internal profiling (not for DevTools),
the following approach had been chosen:

 - during snapshot creation, positions of serialized objects inside
   a snapshot are logged;

 - then during V8 initialization, positions of deserealized objects
   are logged;

 - those positions are used for retrieving code objects names from
   snapshot creation log, which needs to be supplied to tick processor
   script.

Positions logging is controlled with the new flag: --log_snapshot_positions.
This flag is turned off by default, and this adds no startup penalty.

To plug this fix to Golem, the following actions are needed:

 - logs created using 'mksnapshot' need to be stored along with VM images;

 - tick processor script needs to be run with '--snapshot-log=...' cmdline
   argument.

BUG=571

Review URL: http://codereview.chromium.org/551062

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3635 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoImplement issue 549 Make V8 call DebugMessageDispatchHandler with Locker locked
peter.rybin@gmail.com [Mon, 18 Jan 2010 15:48:41 +0000 (15:48 +0000)]
Implement issue 549  Make V8 call DebugMessageDispatchHandler with Locker locked

Review URL: http://codereview.chromium.org/505025

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3634 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix GC bug
sgjesse@chromium.org [Mon, 18 Jan 2010 14:13:58 +0000 (14:13 +0000)]
Fix GC bug

The patching of the receiver added in r3616 was not GC-safe and could leave a failure object in place of the receiver if allocation of the wrapper JS object failed.
Review URL: http://codereview.chromium.org/546068

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3633 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoTrace command in ARM simulator debugger
sgjesse@chromium.org [Mon, 18 Jan 2010 14:11:37 +0000 (14:11 +0000)]
Trace command in ARM simulator debugger

The ARM simulator debugger now has a 'trace' command which is used to toggle the trace of executed instructions.
Review URL: http://codereview.chromium.org/548068

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3632 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix some of my earlier typos.
kmillikin@chromium.org [Mon, 18 Jan 2010 14:09:36 +0000 (14:09 +0000)]
Fix some of my earlier typos.

TBR=fschneider@chromium.org
Review URL: http://codereview.chromium.org/543106

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3631 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoIntroduce 'top-of-stack caching' to the toplevel code generator by
kmillikin@chromium.org [Mon, 18 Jan 2010 14:04:55 +0000 (14:04 +0000)]
Introduce 'top-of-stack caching' to the toplevel code generator by
allowing the value of the rightmost subexpressions to be evaluated
into an accumulator register (eg, eax, rax, or r0) rather than onto
the stack.

Review URL: http://codereview.chromium.org/541047

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3630 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoLift filtering of the test higher as otherwise we can get
antonm@chromium.org [Mon, 18 Jan 2010 13:17:38 +0000 (13:17 +0000)]
Lift filtering of the test higher as otherwise we can get
empty list of tests to run after the filtering and
would actually die with division by 0.

And there was a type (|c| instead of |case|) which is fixed too.

Review URL: http://codereview.chromium.org/545080

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3629 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix windows tests. In the context of fopen the opposite of 'b' for binary
erik.corry@gmail.com [Mon, 18 Jan 2010 12:02:53 +0000 (12:02 +0000)]
Fix windows tests.  In the context of fopen the opposite of 'b' for binary
is '' for ascii, not 'a' for ascii.
Review URL: http://codereview.chromium.org/542104

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3628 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoX64 implementation of native ascii string compare.
lrn@chromium.org [Mon, 18 Jan 2010 11:22:03 +0000 (11:22 +0000)]
X64 implementation of native ascii string compare.

Review URL: http://codereview.chromium.org/545108

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3627 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRegExp bitmap test for word character.
lrn@chromium.org [Mon, 18 Jan 2010 09:49:50 +0000 (09:49 +0000)]
RegExp bitmap test for word character.

Review URL: http://codereview.chromium.org/547024

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3626 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoSmall optimization of ARM compare stub. Reverse all references to
erik.corry@gmail.com [Mon, 18 Jan 2010 08:36:06 +0000 (08:36 +0000)]
Small optimization of ARM compare stub.  Reverse all references to
left and right sides of the comparison to reflect reality.  Don't
check explicitly for NaNs when using VFP3 since the compare
operation can signal this case with the v flag.  Use cmp instead
of tst in the fast compilers since tst leaves the v flag unchanged
and thus can only work by accident on non-equality comparisons.
Review URL: http://codereview.chromium.org/551048

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3625 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoMake debugger unloading deferred operation
peter.rybin@gmail.com [Fri, 15 Jan 2010 22:40:57 +0000 (22:40 +0000)]
Make debugger unloading deferred operation

This CL should make debugger handler setting an asynchronous operation for real.

Review URL: http://codereview.chromium.org/548045

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3624 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoImplement issue 554 Add "ProcessDebuggerRequests" call to Debug Agent API
peter.rybin@gmail.com [Fri, 15 Jan 2010 21:14:56 +0000 (21:14 +0000)]
Implement issue 554 Add "ProcessDebuggerRequests" call to Debug Agent API

Review URL: http://codereview.chromium.org/549057

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3623 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix build problems.
mikhail.naganov@gmail.com [Fri, 15 Jan 2010 20:15:47 +0000 (20:15 +0000)]
Fix build problems.

Review URL: http://codereview.chromium.org/543087

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3622 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd diagnostic print to debug agent if it fails to open port
peter.rybin@gmail.com [Fri, 15 Jan 2010 18:11:12 +0000 (18:11 +0000)]
Add diagnostic print to debug agent if it fails to open port

Review URL: http://codereview.chromium.org/549061

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3621 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoSubmit Object.getOwnPropertyNames patch by Pavel Feldman. See http://codereview.chrom...
mikhail.naganov@gmail.com [Fri, 15 Jan 2010 15:34:32 +0000 (15:34 +0000)]
Submit Object.getOwnPropertyNames patch by Pavel Feldman. See codereview.chromium.org/549050.

Add copyright to regression test to fix build broken by r3619.

TBR=sgjesse@chromium.org

Review URL: http://codereview.chromium.org/542092

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3620 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoImplement enough of the partial snapshots that we can deserialize
erik.corry@gmail.com [Fri, 15 Jan 2010 14:20:31 +0000 (14:20 +0000)]
Implement enough of the partial snapshots that we can deserialize
a single string.
Review URL: http://codereview.chromium.org/536077

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3619 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix ARM build.
vitalyr@chromium.org [Fri, 15 Jan 2010 13:43:28 +0000 (13:43 +0000)]
Fix ARM build.

TBR=ager@chromium.org

Review URL: http://codereview.chromium.org/551047

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3618 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoEnsure correct boxing of values when calling functions on them
sgjesse@chromium.org [Fri, 15 Jan 2010 13:42:32 +0000 (13:42 +0000)]
Ensure correct boxing of values when calling functions on them

When a function is called with a value type as the receiver this is now boxed as an object.

This is a low-impact solution where the receiver is only boxed when required. For IC calls to the V8 builtins values are not boxed and as most of the functions on String.prototype, Number.prototype and Boolean.prototype are sitting there most IC calls on values will not need any boxing of the receiver.

For calls which are not IC calls but calls through the CallFunctionStub a flag is used to determine whether the receiver might be a value and only when that is the case will the receiver be boxed.

No changtes to Function.call and Function.apply - they already boxed values. According to the ES5 spec the receiver should not be boxed for these functions, but current browsers have not adopted that change yet.

BUG=223
TEST=test/mjsunit/value-wrapper.js
TEST=test/mjsunit/regress/regress-crbug-3184.js
Review URL: http://codereview.chromium.org/542087

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3617 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoUpdated README in es5conform so the description says to get revision
ricow@chromium.org [Fri, 15 Jan 2010 13:40:13 +0000 (13:40 +0000)]
Updated README in es5conform so the description says to get revision
62998 which corrects some errors in the tests that will otherwise make us
(wrongly) fail.

Review URL: http://codereview.chromium.org/545079

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3616 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoOptimized calling of C++ builtins (and HandleApiCall
vitalyr@chromium.org [Fri, 15 Jan 2010 12:25:24 +0000 (12:25 +0000)]
Optimized calling of C++ builtins (and HandleApiCall
in particular).

  * Called function is passed on the stack instead of
    using a static variable.

  * Builtins that don't need the called function don't
    get it.

  * Made is_construct statically known to HandleApiCall
    by setting custom construct stub for API functions.

Review URL: http://codereview.chromium.org/536065

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3613 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix issue 541 and some refactoring of the top-level compiler.
fschneider@chromium.org [Fri, 15 Jan 2010 12:00:59 +0000 (12:00 +0000)]
Fix issue 541 and some refactoring of the top-level compiler.

* Refactor VisitProperty to use the platform-specific methods for emitting the IC calls.
* Refactor recording of source positions in the top-level compiler.
* Correct the recorded source positions for assignments and property loads.
* Fix bug on x64 where source positions were not recorded before a calling a call-IC.
* Correct some inconsistencies between IA-32 and X64 top-level code generator.

We now pass all regression tests with
--always-fast-compiler.

Review URL: http://codereview.chromium.org/550043

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3612 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix bug in keyed load stub for strings.
fschneider@chromium.org [Fri, 15 Jan 2010 12:00:21 +0000 (12:00 +0000)]
Fix bug in keyed load stub for strings.

Instead of returning the empty string when indexing
a string out of bounds we now correctly return undefined.

Review URL: http://codereview.chromium.org/542089

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3611 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix a problem when compiling built-ins with the top-level compiler.
fschneider@chromium.org [Thu, 14 Jan 2010 17:22:59 +0000 (17:22 +0000)]
Fix a problem when compiling built-ins with the top-level compiler.

Replace runtime call to NumberAdd with call to binary op stub.

Until now the top-level compiler always called a runtime function
for count operations.

In some places we expected in the JS builtins smis as arguments.
If we perform a count operation before all smis would get converted into
heap numbers by the runtime number add function and result in a runtime
assert.

Also: Add missing debugger information in the top-level compiler for
do-while loops.

Review URL: http://codereview.chromium.org/548029

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3610 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPush to trunk. Now working on 2.0.7.
fschneider@chromium.org [Thu, 14 Jan 2010 15:09:43 +0000 (15:09 +0000)]
Push to trunk. Now working on 2.0.7.

Review URL: http://codereview.chromium.org/548041

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3607 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years ago* Improve the interface to the memory-reservation functionality.
erik.corry@gmail.com [Thu, 14 Jan 2010 14:46:31 +0000 (14:46 +0000)]
* Improve the interface to the memory-reservation functionality.
* Add a test case that generates a serialization of a single flat string.
Review URL: http://codereview.chromium.org/542073

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3606 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix crash when using Object.getOwnPropertyDescriptor on an API defined
ager@chromium.org [Thu, 14 Jan 2010 14:32:19 +0000 (14:32 +0000)]
Fix crash when using Object.getOwnPropertyDescriptor on an API defined
callback property.

Return undefined for Object.getOwnPropertyDescriptor on an API defined
callback properties for now.

Review URL: http://codereview.chromium.org/542074

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3605 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoChange return type of MapSpace::NeedsCompaction from int to bool.
ager@chromium.org [Thu, 14 Jan 2010 12:51:38 +0000 (12:51 +0000)]
Change return type of MapSpace::NeedsCompaction from int to bool.
Fixes compiler warnings on Windows.

TBR=antonm@chromium.org

Review URL: http://codereview.chromium.org/549049

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3604 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRemoved check on element in case it is undefined in ArrayIndexOf - according to spec...
ricow@chromium.org [Thu, 14 Jan 2010 12:05:48 +0000 (12:05 +0000)]
Removed check on element in case it is undefined in ArrayIndexOf - according to spec this is OK.

Review URL: http://codereview.chromium.org/545056

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3603 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix the tests that change because of change to toISOString format.
ager@chromium.org [Thu, 14 Jan 2010 09:05:52 +0000 (09:05 +0000)]
Fix the tests that change because of change to toISOString format.
Review URL: http://codereview.chromium.org/543057

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3602 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix Date.prototype.toISOString for NaN dates and add milliseconds for
ager@chromium.org [Thu, 14 Jan 2010 08:55:15 +0000 (08:55 +0000)]
Fix Date.prototype.toISOString for NaN dates and add milliseconds for
compatibility with Safari.

Review URL: http://codereview.chromium.org/543056

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3601 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoCompact map space when doing mark-sweep if after collection size of map space would
antonm@chromium.org [Wed, 13 Jan 2010 19:16:07 +0000 (19:16 +0000)]
Compact map space when doing mark-sweep if after collection size of map space would
drop below threshold.

Review URL: http://codereview.chromium.org/509035

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3600 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRemove an inlined ASSERT made obsolete by r3598.
kmillikin@chromium.org [Wed, 13 Jan 2010 16:21:06 +0000 (16:21 +0000)]
Remove an inlined ASSERT made obsolete by r3598.

Review URL: http://codereview.chromium.org/549035

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3599 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRemove a pair of problematic uses of the Reference utility class from
kmillikin@chromium.org [Wed, 13 Jan 2010 16:01:15 +0000 (16:01 +0000)]
Remove a pair of problematic uses of the Reference utility class from
the code generators.

These uses broke the rules of the class because it was safe to do so,
but there was no real reason to do it that way.
Review URL: http://codereview.chromium.org/543041

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3598 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRevert r3583 which seems to degrade performance.
ager@chromium.org [Wed, 13 Jan 2010 13:26:31 +0000 (13:26 +0000)]
Revert r3583 which seems to degrade performance.

Review URL: http://codereview.chromium.org/536042

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3597 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix bug in top-level code generator with global receiver.
fschneider@chromium.org [Wed, 13 Jan 2010 12:13:37 +0000 (12:13 +0000)]
Fix bug in top-level code generator with global receiver.

We did not correcty pass the global receiver object in some cases.
This change fixes this bug on all three platforms.

Review URL: http://codereview.chromium.org/542053

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3596 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoEnabled es5conform tests for new array methods and corrected errors that was discover...
ricow@chromium.org [Wed, 13 Jan 2010 12:10:57 +0000 (12:10 +0000)]
Enabled es5conform tests for new array methods and corrected errors that was discovered by enabling these.

Added new es5 methods:
GetOwnPropertyDescriptor
GetOwnProperty
FromPropertyDescriptor

Review URL: http://codereview.chromium.org/546032

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3595 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoCleanup the handling of control flow in the toplevel code generator.
kmillikin@chromium.org [Wed, 13 Jan 2010 11:29:08 +0000 (11:29 +0000)]
Cleanup the handling of control flow in the toplevel code generator.

Do abstract the setting and restoring of 'argument' state into a
function that takes arguments.

Do not set the argument state in the code generator unless it
represents arguments to a recursive call to Visit.

Review URL: http://codereview.chromium.org/550010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3594 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix bug in the ARM port of FastNewContextStub. Load the function from
ager@chromium.org [Wed, 13 Jan 2010 10:27:54 +0000 (10:27 +0000)]
Fix bug in the ARM port of FastNewContextStub.  Load the function from
the stack instead of popping it - it is needed by the slow case.

Review URL: http://codereview.chromium.org/546031

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3593 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPort FastNewContextStub to x64 and arm.
ager@chromium.org [Wed, 13 Jan 2010 08:16:02 +0000 (08:16 +0000)]
Port FastNewContextStub to x64 and arm.

BUG=551
Review URL: http://codereview.chromium.org/541027

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3592 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix ARM thumb/debug build
sgjesse@chromium.org [Wed, 13 Jan 2010 07:45:53 +0000 (07:45 +0000)]
Fix ARM thumb/debug build

Landing http://codereview.chromium.org/548007

TBR=piman@chromium.org
Review URL: http://codereview.chromium.org/549033

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3591 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix ARM cross compiles generated from gyp.
sgjesse@chromium.org [Wed, 13 Jan 2010 07:35:31 +0000 (07:35 +0000)]
Fix ARM cross compiles generated from gyp.

Landing of http://codereview.chromium.org/549018.

TBR=piman@chromium.org

Review URL: http://codereview.chromium.org/546030

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3590 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoReverting temporary reverts 3586 and 3588 back.
antonm@chromium.org [Tue, 12 Jan 2010 23:42:36 +0000 (23:42 +0000)]
Reverting temporary reverts 3586 and 3588 back.

TBR=kasperl@chromium.org

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3589 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoTemporary backing out r3536 to see impact on DOM benchmarks.
antonm@chromium.org [Tue, 12 Jan 2010 19:36:09 +0000 (19:36 +0000)]
Temporary backing out r3536 to see impact on DOM benchmarks.

TBR=kasperl@chromium.org

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3588 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix a problem with const initialization in the top-level code generator.
fschneider@chromium.org [Tue, 12 Jan 2010 17:22:57 +0000 (17:22 +0000)]
Fix a problem with const initialization in the top-level code generator.

When initializing the special local variable containing the reference to the enclosing
function in named functions we now (correctly) emit an INIT_CONST instead of INIT_VAR,
and we correctly bail out in the top-level code generator.

Also part of this change is adding missing statement position information
for some statements in the top-level code generator.

Review URL: http://codereview.chromium.org/536029

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3587 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoTemporary backing out r3538 to see impact on DOM benchmarks.
antonm@chromium.org [Tue, 12 Jan 2010 16:57:18 +0000 (16:57 +0000)]
Temporary backing out r3538 to see impact on DOM benchmarks.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3586 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix float conversion warning from some gcc versions.
erik.corry@gmail.com [Tue, 12 Jan 2010 15:36:43 +0000 (15:36 +0000)]
Fix float conversion warning from some gcc versions.
Review URL: http://codereview.chromium.org/541026

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3585 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd some interfaces to the GC that allow us to reserve space. This is needed
erik.corry@gmail.com [Tue, 12 Jan 2010 15:16:23 +0000 (15:16 +0000)]
Add some interfaces to the GC that allow us to reserve space.  This is needed
for partial snapshots.  After reserving space we can be sure that allocations
will happen linearly (no GCs and no free-list allocation).  This change also
contains the start of the partial snapshot support, which, however is not yet
completed or tested.
Review URL: http://codereview.chromium.org/545026

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3584 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoReapply the parts of revision 3545 that did not work because of broken
ager@chromium.org [Tue, 12 Jan 2010 11:56:19 +0000 (11:56 +0000)]
Reapply the parts of revision 3545 that did not work because of broken
fixups code on ARM.
Review URL: http://codereview.chromium.org/545025

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3583 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoChange the ARM fixup code to handle the use of the following
ager@chromium.org [Tue, 12 Jan 2010 11:54:19 +0000 (11:54 +0000)]
Change the ARM fixup code to handle the use of the following
instruction sequence for jumps:

  mov(ip, Operand(target, rmode), LeaveCC, cond);
  bx(ip, cond)

Changed a JS call in the compare stub to a tail call to avoid GC
problems where the pushed return address is not updated on GC.
Review URL: http://codereview.chromium.org/549022

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3582 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix lint error. I must have been following the Standard ML style
kmillikin@chromium.org [Tue, 12 Jan 2010 09:58:50 +0000 (09:58 +0000)]
Fix lint error.  I must have been following the Standard ML style
guide.

Review URL: http://codereview.chromium.org/541024

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3581 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoSome cleanup of the toplevel code generator:
kmillikin@chromium.org [Tue, 12 Jan 2010 08:48:26 +0000 (08:48 +0000)]
Some cleanup of the toplevel code generator:

* Rename the operation of filling a context with a value.  Formerly it
  was 'Move', now it's 'Apply' so as to avoid confusion with various
  other Moves (eg, in the toplevel codegen, in the macro assemblers).

* Use the abstraction Drop rather than math on the stack pointer.

* Add a predicate on AST expression nodes to abstract a repeated test
  whether we should us a regular (named) IC or a keyed IC.

Review URL: http://codereview.chromium.org/546006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3580 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix variable name when build with oprofile agent.
ager@chromium.org [Tue, 12 Jan 2010 08:31:00 +0000 (08:31 +0000)]
Fix variable name when build with oprofile agent.

BUG=572
Review URL: http://codereview.chromium.org/549020

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3579 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPort FastNewClosureStub introduced for ia32 in r3477 to x64 and arm.
ager@chromium.org [Tue, 12 Jan 2010 07:16:49 +0000 (07:16 +0000)]
Port FastNewClosureStub introduced for ia32 in r3477 to x64 and arm.

BUG=552
Review URL: http://codereview.chromium.org/543010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3578 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd missing instructions to the IA-32 disasembler
sgjesse@chromium.org [Mon, 11 Jan 2010 15:19:53 +0000 (15:19 +0000)]
Add missing instructions to the IA-32 disasembler

Added newly added instructions to test-disasem-ia32.cc and implemented the missi
ng ones in the disasembler.

Added some asserts to 8-bit instructions which only work with eax, ebx, ecx and
edx (al, bl, cl and dl).

Removed the loope instruction.
Review URL: http://codereview.chromium.org/548002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3577 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdded two missing instructions to disassembler.
fschneider@chromium.org [Mon, 11 Jan 2010 12:59:17 +0000 (12:59 +0000)]
Added two missing instructions to disassembler.

The subb and the loop instructions are also added to the IA-32 disassembler.

Review URL: http://codereview.chromium.org/541008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3576 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdded ScriptData::HasError.
christian.plesner.hansen@gmail.com [Mon, 11 Jan 2010 12:13:24 +0000 (12:13 +0000)]
Added ScriptData::HasError.

Review URL: http://codereview.chromium.org/542010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3575 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix building of D8
sgjesse@chromium.org [Mon, 11 Jan 2010 10:00:07 +0000 (10:00 +0000)]
Fix building of D8

This fixes building D8 when SCons cannot figure the build environment and it is passed using the env argument. This is usually only used when building with Visual Studio 2008.
Review URL: http://codereview.chromium.org/542009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3574 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix wrong assert in compare stub minor key computation.
ager@chromium.org [Mon, 11 Jan 2010 07:35:32 +0000 (07:35 +0000)]
Fix wrong assert in compare stub minor key computation.

TBR=erik.corry
Review URL: http://codereview.chromium.org/536010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3573 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFixed wrong cmov operand order
sgjesse@chromium.org [Fri, 8 Jan 2010 14:17:26 +0000 (14:17 +0000)]
Fixed wrong cmov operand order

TBR=erik.corry@gmail.com
Review URL: http://codereview.chromium.org/518082

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3572 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix same Windows compiler error in another place.
erik.corry@gmail.com [Fri, 8 Jan 2010 13:48:27 +0000 (13:48 +0000)]
Fix same Windows compiler error in another place.
Review URL: http://codereview.chromium.org/518081

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3571 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix Windows build.
erik.corry@gmail.com [Fri, 8 Jan 2010 12:42:58 +0000 (12:42 +0000)]
Fix Windows build.
Review URL: http://codereview.chromium.org/527031

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3570 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd generated code for ascii string comparison
sgjesse@chromium.org [Fri, 8 Jan 2010 11:58:15 +0000 (11:58 +0000)]
Add generated code for ascii string comparison

Careted a stub for string comparison and used part of the code from that to inline string comparison in the compare stub.
Review URL: http://codereview.chromium.org/525115

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3569 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix potential length-miscalculation in %StringBuilderConcat.
lrn@chromium.org [Fri, 8 Jan 2010 11:27:37 +0000 (11:27 +0000)]
Fix potential length-miscalculation in %StringBuilderConcat.

Review URL: http://codereview.chromium.org/521074

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3568 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoLint, schmlint!
erik.corry@gmail.com [Fri, 8 Jan 2010 10:41:25 +0000 (10:41 +0000)]
Lint, schmlint!
Review URL: http://codereview.chromium.org/525114

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3567 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoSpeed up compares with characters ie single-character strings.
erik.corry@gmail.com [Fri, 8 Jan 2010 09:54:11 +0000 (09:54 +0000)]
Speed up compares with characters ie single-character strings.
Make use of it when we know that something can't be a NaN.
Review URL: http://codereview.chromium.org/524059

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3566 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoInline a number of simple date computations.
ager@chromium.org [Fri, 8 Jan 2010 07:49:07 +0000 (07:49 +0000)]
Inline a number of simple date computations.

The minifier has been updated to have better recognition of regular
expressions.

Review URL: http://codereview.chromium.org/521057

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3565 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoMake sure to use the CHECK_PARSE_ERROR macro (and thereby
kasperl@chromium.org [Fri, 8 Jan 2010 06:40:09 +0000 (06:40 +0000)]
Make sure to use the CHECK_PARSE_ERROR macro (and thereby
the static CheckParse function) even in release builds.

TBR=lrn@chromium.org
Review URL: http://codereview.chromium.org/518079

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3564 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAttempt to make \b\w+ faster. Slight performance increase on, e.g., string unpacking.
lrn@chromium.org [Thu, 7 Jan 2010 19:01:23 +0000 (19:01 +0000)]
Attempt to make \b\w+ faster. Slight performance increase on, e.g., string unpacking.

Review URL: http://codereview.chromium.org/507051

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3563 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoForgot patch to make it compiler on Linux.
lrn@chromium.org [Thu, 7 Jan 2010 14:04:56 +0000 (14:04 +0000)]
Forgot patch to make it compiler on Linux.

Review URL: http://codereview.chromium.org/525087

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3562 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix wrong asserts from r3554
sgjesse@chromium.org [Thu, 7 Jan 2010 13:17:37 +0000 (13:17 +0000)]
Fix wrong asserts from r3554
Review URL: http://codereview.chromium.org/523128

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3561 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFixed potential length miscalculations by limiting max size of arrays and strings.
lrn@chromium.org [Thu, 7 Jan 2010 13:17:18 +0000 (13:17 +0000)]
Fixed potential length miscalculations by limiting max size of arrays and strings.

Review URL: http://codereview.chromium.org/525064

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3560 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix runtime function StringCharAt where an argument was not checked.
fschneider@chromium.org [Thu, 7 Jan 2010 12:31:45 +0000 (12:31 +0000)]
Fix runtime function StringCharAt where an argument was not checked.

Review URL: http://codereview.chromium.org/523126

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3559 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoMake String.prototype.replace a tiny bit faster by avoiding
kasperl@chromium.org [Thu, 7 Jan 2010 12:18:56 +0000 (12:18 +0000)]
Make String.prototype.replace a tiny bit faster by avoiding
ToString conversions for strings.
Review URL: http://codereview.chromium.org/518059

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3558 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix x64 and arm build.
fschneider@chromium.org [Thu, 7 Jan 2010 11:38:43 +0000 (11:38 +0000)]
Fix x64 and arm build.

Review URL: http://codereview.chromium.org/524058

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3557 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoImprove keyed loads on strings by using a new stub.
fschneider@chromium.org [Thu, 7 Jan 2010 10:25:20 +0000 (10:25 +0000)]
Improve keyed loads on strings by using a new stub.

Instead of going through a runtime function for keyed loads
on strings we invoke a separate specialized stub that
assumes string as receiver type and the key to be a number.

The stub calls a JS builtin function to return the corresponding one-character string.

Review URL: http://codereview.chromium.org/521041

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3556 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdded ES5 15.2.3.2 Object.getPrototypeOf.
ricow@chromium.org [Thu, 7 Jan 2010 10:01:24 +0000 (10:01 +0000)]
Added ES5 15.2.3.2 Object.getPrototypeOf.

Review URL: http://codereview.chromium.org/523124

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3555 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoUse generated code to create sub strings.
sgjesse@chromium.org [Thu, 7 Jan 2010 09:59:37 +0000 (09:59 +0000)]
Use generated code to create sub strings.

Added a stub to allocate and fill a string object with a substring from another string.

Use the rep movs instruction to copy the string data as it turned out to be the fastest way.

While preparing this I experimented with some SSE2 instructions, so the instructions movdqa and movdqu are still in the IA-32 assembler even though they are not used.
Review URL: http://codereview.chromium.org/525085

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3554 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoImprove the performance of String.prototype.concat and the slow-case
kasperl@chromium.org [Thu, 7 Jan 2010 09:40:50 +0000 (09:40 +0000)]
Improve the performance of String.prototype.concat and the slow-case
for compare operations.
Review URL: http://codereview.chromium.org/521054

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3553 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoQuick fix to ARM fixup issue introduced in r3545. The fixup code
kasperl@chromium.org [Wed, 6 Jan 2010 15:50:34 +0000 (15:50 +0000)]
Quick fix to ARM fixup issue introduced in r3545. The fixup code
seems somewhat broken on ARM, but this solves the issue for now.
I'll look into the code patching tomorrow.
Review URL: http://codereview.chromium.org/525068

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3546 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoImprove performance of Array.prototype.join and String.prototype.substring
kasperl@chromium.org [Wed, 6 Jan 2010 14:40:21 +0000 (14:40 +0000)]
Improve performance of Array.prototype.join and String.prototype.substring
by tweaking the JavaScript implementation of these functions.
Review URL: http://codereview.chromium.org/519061

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3545 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoMake the direct call to native RegExp work
sgjesse@chromium.org [Wed, 6 Jan 2010 12:21:28 +0000 (12:21 +0000)]
Make the direct call to native RegExp work

Single glitch in r3542 which caused the direct call to native regexp code to always bail out to the runtime system.

TBR=erik.corry@gmail.com
Review URL: http://codereview.chromium.org/523091

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3544 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years ago- Adjust the number to string cache based on the max semispace size.
bak@chromium.org [Wed, 6 Jan 2010 11:19:28 +0000 (11:19 +0000)]
- Adjust the number to string cache based on the max semispace size.
  Flushed at compacting mark sweep.
- Simplified FindEntry by eliminating the counter.

Review URL: http://codereview.chromium.org/527006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3543 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoDirect call to native RegExp code from JavaScript.
sgjesse@chromium.org [Wed, 6 Jan 2010 11:09:30 +0000 (11:09 +0000)]
Direct call to native RegExp code from JavaScript.

Calls to RegExp no longer have to be via a call to the runtime system. A new stub have been added which can handle this call in generated code. The stub checks all the parameters and creates RegExp entry frame in the same way as it is created by the runtime system. Bailout to the runtime system is done whenever an uncommon situation is encountered or when the static data used is not initialized. After running the native RegExp code the last match info is updated like in the runtime system.

Currently only ASCII strings are handled.

Added another argument to the RegExp entry frame. It indicated whether the call is direct from JavaScript code or through the runtime system. This information is used when RegExp execution is interrupted. If an interruption happens when RegExp code is called directly a retry is issued causing the interruption to be handled via the runtime system. The reason for this is that the direct call to RegExp code does not support garbage collection.
Review URL: http://codereview.chromium.org/521028

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3542 ce2b1a6d-e550-0410-aec6-3dcde31c8c00