platform/upstream/v8.git
14 years agoMake the FastCloneShallowArrayStub a bit prettier.
kasperl@chromium.org [Tue, 22 Dec 2009 13:10:24 +0000 (13:10 +0000)]
Make the FastCloneShallowArrayStub a bit prettier.

TBR=fschneider@chromium.org
Review URL: http://codereview.chromium.org/507069

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3518 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRevert r3514 and r3515. The new cache is too large for some tests
ager@chromium.org [Tue, 22 Dec 2009 13:07:27 +0000 (13:07 +0000)]
Revert r3514 and r3515.  The new cache is too large for some tests
that attempt to run with a small heap.  Additionally, it can
potentially keep a lot of string data alive and it is never flushed.
Can we make it grow dynamically if used so that we can still start the
VM with a small heap size?
Review URL: http://codereview.chromium.org/503081

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3517 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoUse one runtime call for creating object/array literals in
fschneider@chromium.org [Tue, 22 Dec 2009 12:41:45 +0000 (12:41 +0000)]
Use one runtime call for creating object/array literals in
the code generator.

The runtime function checks if it needs to create a boilerplate
object or if it can clone from an existing boilerplate.

This is already done in the top-level compiler.

Review URL: http://codereview.chromium.org/507036

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3516 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix linto.
kasperl@chromium.org [Tue, 22 Dec 2009 12:39:56 +0000 (12:39 +0000)]
Fix linto.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3515 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years ago- Increased size of number string cache.
bak@chromium.org [Tue, 22 Dec 2009 11:35:05 +0000 (11:35 +0000)]
- Increased size of number string cache.
- Change the instruction order for inlined allocation.

Review URL: http://codereview.chromium.org/501170

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3514 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd fast case stub for BIT_NOT.
kasperl@chromium.org [Tue, 22 Dec 2009 10:16:27 +0000 (10:16 +0000)]
Add fast case stub for BIT_NOT.
Review URL: http://codereview.chromium.org/503079

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3513 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoCheck for undefined in the binary operation stub when convertion to
ager@chromium.org [Tue, 22 Dec 2009 09:48:55 +0000 (09:48 +0000)]
Check for undefined in the binary operation stub when convertion to
int32 for bitops.  undefined converts to zero in ToInt32 conversions.

Review URL: http://codereview.chromium.org/508020

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3512 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoVery experimental fix for issue 555.
kasperl@chromium.org [Mon, 21 Dec 2009 15:09:26 +0000 (15:09 +0000)]
Very experimental fix for issue 555.
Review URL: http://codereview.chromium.org/508006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3509 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoOptimize implementation of Math.floor a little by special casing
erik.corry@gmail.com [Mon, 21 Dec 2009 15:04:00 +0000 (15:04 +0000)]
Optimize implementation of Math.floor a little by special casing
the comparison it uses in the code generator.  Use Math.floor for
date operations.
Review URL: http://codereview.chromium.org/509007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3508 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRemove complicated Math.sin and Math.cos optimizations that do not buy
ager@chromium.org [Mon, 21 Dec 2009 13:30:10 +0000 (13:30 +0000)]
Remove complicated Math.sin and Math.cos optimizations that do not buy
us much.
Review URL: http://codereview.chromium.org/509006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3507 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoOptimize sine and cosine by checking up front if the fsin or fcos
ager@chromium.org [Mon, 21 Dec 2009 12:08:59 +0000 (12:08 +0000)]
Optimize sine and cosine by checking up front if the fsin or fcos
operation can throw an exception.

Review URL: http://codereview.chromium.org/504073

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3506 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoThe number of heap slots stored in a scope includes the
kasperl@chromium.org [Mon, 21 Dec 2009 10:24:11 +0000 (10:24 +0000)]
The number of heap slots stored in a scope includes the
fixed contexts slots. Take this into account when using
the new, fast context creation path to avoid allocating
too many slots (wasteful).
Review URL: http://codereview.chromium.org/501148

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3505 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoBring back the fisttp instruction on machines with SSE3, but check the
erik.corry@gmail.com [Mon, 21 Dec 2009 08:32:20 +0000 (08:32 +0000)]
Bring back the fisttp instruction on machines with SSE3, but check the
input so we don't have to check the exception flags afterwards.
Review URL: http://codereview.chromium.org/509001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3504 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoDisallow garbage collection at another site in the LoadCallback ICs.
kmillikin@chromium.org [Sun, 20 Dec 2009 08:40:13 +0000 (08:40 +0000)]
Disallow garbage collection at another site in the LoadCallback ICs.
MacroAssembler::PopHandleScope emits a runtime call (through a stub),
which should not be allowed to perform a GC but return a failure
instead.

BUG=30790
TEST=none

Review URL: http://codereview.chromium.org/504071

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3501 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoReapply "wait for connection" feature implementation
peter.rybin@gmail.com [Fri, 18 Dec 2009 20:30:29 +0000 (20:30 +0000)]
Reapply "wait for connection" feature implementation

Review URL: http://codereview.chromium.org/491079

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3500 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRecommit r3494 after fixing stupid mistake where the deferred code stub didn't know...
kasperl@chromium.org [Fri, 18 Dec 2009 13:47:58 +0000 (13:47 +0000)]
Recommit r3494 after fixing stupid mistake where the deferred code stub didn't know it was reversed.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3499 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFast-codegen: Adding support for try/catch and throw.
lrn@chromium.org [Fri, 18 Dec 2009 13:38:28 +0000 (13:38 +0000)]
Fast-codegen: Adding support for try/catch and throw.
Still no support for lookup-variables, so we bailout if using the catch variable.

Review URL: http://codereview.chromium.org/501076

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3498 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoForce mark sweep instead of compcation if size of map space is too big to allow forwa...
antonm@chromium.org [Fri, 18 Dec 2009 13:38:09 +0000 (13:38 +0000)]
Force mark sweep instead of compcation if size of map space is too big to allow forward pointers encoding.

Review URL: http://codereview.chromium.org/507025

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3497 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRevert r3495 - some weird corner case is broken.
kasperl@chromium.org [Fri, 18 Dec 2009 13:14:37 +0000 (13:14 +0000)]
Revert r3495 - some weird corner case is broken.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3496 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoImprove code generated for left shifts where the left side
kasperl@chromium.org [Fri, 18 Dec 2009 13:07:02 +0000 (13:07 +0000)]
Improve code generated for left shifts where the left side
is a constant by avoiding a few checks.
Review URL: http://codereview.chromium.org/504057

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3495 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoConvert the rest of the sar(xxx, kSmiTagSize) to SmiUntag(xxx).
kasperl@chromium.org [Fri, 18 Dec 2009 11:20:25 +0000 (11:20 +0000)]
Convert the rest of the sar(xxx, kSmiTagSize) to SmiUntag(xxx).

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3494 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRefactor smi tagging and untagging on IA-32.
kasperl@chromium.org [Fri, 18 Dec 2009 11:13:33 +0000 (11:13 +0000)]
Refactor smi tagging and untagging on IA-32.
Review URL: http://codereview.chromium.org/501113

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3493 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoOptimize bitops with non-Smi inputs. Instead of converting both inputs
erik.corry@gmail.com [Fri, 18 Dec 2009 09:33:24 +0000 (09:33 +0000)]
Optimize bitops with non-Smi inputs.  Instead of converting both inputs
to floating point and then converting back we convert directly to a
32 bit integer.  In addition the bit twiddling implementation of float-
to-integer conversion has been ported from ARM.  Testing has shown that
this runs faster than the x87 or SSE3 rounding instructions.  This change
is IA32 only.  There may be a smaller benefit from doing the same on x64.
Review URL: http://codereview.chromium.org/506052

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3492 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPrepare push to trunk. We are now working on version 2.0.6
sgjesse@chromium.org [Fri, 18 Dec 2009 08:21:21 +0000 (08:21 +0000)]
Prepare push to trunk. We are now working on version 2.0.6
Review URL: http://codereview.chromium.org/504056

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3489 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoadded else
bak@chromium.org [Fri, 18 Dec 2009 06:58:45 +0000 (06:58 +0000)]
added else

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3488 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years ago-Inlined double variant of compare iff one of the sides is a constant smi and it...
bak@chromium.org [Fri, 18 Dec 2009 06:38:12 +0000 (06:38 +0000)]
-Inlined double variant of compare iff one of the sides is a constant smi and it is not a for loop condition.

Review URL: http://codereview.chromium.org/507040

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3487 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years ago- Handle HeapNumbers in deferred code instead of jumping to stub
iposva@chromium.org [Fri, 18 Dec 2009 01:43:50 +0000 (01:43 +0000)]
- Handle HeapNumbers in deferred code instead of jumping to stub
  for DeferredInlineBinaryOperations.

Review URL: http://codereview.chromium.org/502067

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3486 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoOptimize the allocation of small, non-nested literal
kasperl@chromium.org [Thu, 17 Dec 2009 15:35:15 +0000 (15:35 +0000)]
Optimize the allocation of small, non-nested literal
arrays and argument objects on IA-32.
Review URL: http://codereview.chromium.org/503042

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3485 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoStreamline the calling convention of the call ICs by passing the
kmillikin@chromium.org [Thu, 17 Dec 2009 10:23:20 +0000 (10:23 +0000)]
Streamline the calling convention of the call ICs by passing the
property name in a register rather than on the stack below the
receiver and arguments.

Implemented only for IA32, passing the name in the ecx register to
match the calling convention of the load ICs.

Review URL: http://codereview.chromium.org/502028

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3484 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoDon't bother callling into the ProducerHeapProfile methods
kasperl@chromium.org [Thu, 17 Dec 2009 09:16:44 +0000 (09:16 +0000)]
Don't bother callling into the ProducerHeapProfile methods
if logging producers is turned off.
Review URL: http://codereview.chromium.org/500092

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3483 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoEnable fast allocation of context objects on IA-32.
kasperl@chromium.org [Thu, 17 Dec 2009 08:53:19 +0000 (08:53 +0000)]
Enable fast allocation of context objects on IA-32.
Review URL: http://codereview.chromium.org/507035

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3482 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoExtend the maximum size map space
sgjesse@chromium.org [Thu, 17 Dec 2009 08:53:18 +0000 (08:53 +0000)]
Extend the maximum size map space

On 32-bit the maps are now aligned on a 32-byte boundary in order to encode more maps during compacting GC. The actual size of a map on 32-bit is 28 bytes making this change waste 4 bytes per map.

On 64-bit the encoding for compacting GC is now using more than 32-bits and the maps here are still pointer size aligned. The actual size of a map on 64-bit is 48 bytes and this change does not intruduce any waste.

My choice of 16 bits for kMapPageIndexBits for 64-bit should give the same maximum number of pages (8K) for map space. As maps on 64-bit are larger than on 32-bit the total number of maps on 64-bit will be smaller than on 32-bit. We could consider raising this to 17 or 18.

I moved the kPageSizeBits to globals.h as the calculation of the encoding really depended on this.

There are still an #ifdef/#endif in objects.h and this constant could be moved to globaks.h as well, but I kept it together with the related constants.

All the tests run in debug mode with additional options --gc-global --always-compact as well (except for a few tests on which also fails before this change when run with --gc-global --always-compact).

BUG=http://code.google.com/p/v8/issues/detail?id=524
BUG=http://crbug.com/29428
TEST=test/mjsunit/regress/regress-524.js

Review URL: http://codereview.chromium.org/504026

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3481 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix ARM and x64 tests in debug mode after r3477.
kasperl@chromium.org [Thu, 17 Dec 2009 07:35:12 +0000 (07:35 +0000)]
Fix ARM and x64 tests in debug mode after r3477.

TBR=ager@chromium.org
Review URL: http://codereview.chromium.org/500090

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3480 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years ago- Fix x64 build.
iposva@chromium.org [Thu, 17 Dec 2009 07:18:05 +0000 (07:18 +0000)]
- Fix x64 build.

Review URL: http://codereview.chromium.org/500089

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3479 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years ago- Undo unfortunate renaming of StaticType.
iposva@chromium.org [Thu, 17 Dec 2009 06:21:50 +0000 (06:21 +0000)]
- Undo unfortunate renaming of StaticType.

Review URL: http://codereview.chromium.org/506050

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3478 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoImprove performance of allocating closures for nested
kasperl@chromium.org [Wed, 16 Dec 2009 15:43:20 +0000 (15:43 +0000)]
Improve performance of allocating closures for nested
functions by allocating them in new space without
entering the runtime system.
Review URL: http://codereview.chromium.org/506037

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3477 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPrepare push to trunk. We are now working on version 2.0.5.
kmillikin@chromium.org [Wed, 16 Dec 2009 15:23:25 +0000 (15:23 +0000)]
Prepare push to trunk.  We are now working on version 2.0.5.

Review URL: http://codereview.chromium.org/502031

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3474 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFast codegen: Implement with.
lrn@chromium.org [Wed, 16 Dec 2009 09:51:07 +0000 (09:51 +0000)]
Fast codegen: Implement with.

Review URL: http://codereview.chromium.org/491077

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3473 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix presubmit error: remove whitespace
yurys@chromium.org [Tue, 15 Dec 2009 09:34:00 +0000 (09:34 +0000)]
Fix presubmit error: remove whitespace

TBR=sgjesse
Review URL: http://codereview.chromium.org/504012

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3466 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoIgnore debug break events when bootstrapper is active. Collecting debug data when...
yurys@chromium.org [Tue, 15 Dec 2009 09:17:15 +0000 (09:17 +0000)]
Ignore debug break events when bootstrapper is active. Collecting debug data when the context is not yet setup may lead to subtle errors like in the following Chromium bug: crbug.com/28933
Review URL: http://codereview.chromium.org/497006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3465 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agostats-viewer.py: support passing test_shell pid.
vitalyr@chromium.org [Mon, 14 Dec 2009 17:05:38 +0000 (17:05 +0000)]
stats-viewer.py: support passing test_shell pid.

Review URL: http://codereview.chromium.org/496010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3464 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoUpdated stats-viewer.py to work with chromium.
vitalyr@chromium.org [Mon, 14 Dec 2009 16:13:52 +0000 (16:13 +0000)]
Updated stats-viewer.py to work with chromium.

Review URL: http://codereview.chromium.org/473002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3463 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoDo a fast check for undefined values when comparing.
kasperl@chromium.org [Mon, 14 Dec 2009 12:18:20 +0000 (12:18 +0000)]
Do a fast check for undefined values when comparing.
Review URL: http://codereview.chromium.org/496008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3462 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFast-codegen: Implementing try/finally on top of nesting context.
lrn@chromium.org [Mon, 14 Dec 2009 11:09:25 +0000 (11:09 +0000)]
Fast-codegen: Implementing try/finally on top of nesting context.

Review URL: http://codereview.chromium.org/492002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3461 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoModified SConstruct to enable building of shared library
erik.corry@gmail.com [Mon, 14 Dec 2009 10:20:35 +0000 (10:20 +0000)]
Modified SConstruct to enable building of shared library
with hidden visibility on Mac OS X.  Patch from
Abdulla Kamar. http://codereview.chromium.org/475002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3460 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoTry to make the contextuality check in the IC code faster
kasperl@chromium.org [Mon, 14 Dec 2009 09:54:13 +0000 (09:54 +0000)]
Try to make the contextuality check in the IC code faster
by checking if the receiver is a global object before diving
into traversing reloc info.
Review URL: http://codereview.chromium.org/491075

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3459 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoReverted r3457
sgjesse@chromium.org [Mon, 14 Dec 2009 07:51:59 +0000 (07:51 +0000)]
Reverted r3457

Caused compilation errors in cctests and on ARM platform.

TBR=peter.rybin@gmail.com
Review URL: http://codereview.chromium.org/491074

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3458 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoImplement "wait for connection" feature
peter.rybin@gmail.com [Sun, 13 Dec 2009 21:15:02 +0000 (21:15 +0000)]
Implement "wait for connection" feature

Review URL: http://codereview.chromium.org/489005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3457 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix lint error.
fschneider@chromium.org [Fri, 11 Dec 2009 22:48:14 +0000 (22:48 +0000)]
Fix lint error.

TBR=kmillikin@chromium.org
Review URL: http://codereview.chromium.org/491063

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3456 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdding compound assignments to the top-level compiler.
fschneider@chromium.org [Fri, 11 Dec 2009 16:09:16 +0000 (16:09 +0000)]
Adding compound assignments to the top-level compiler.

Review URL: http://codereview.chromium.org/486008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3455 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoGive the binary op stubs better names to make profiles more informative.
erik.corry@gmail.com [Fri, 11 Dec 2009 10:40:01 +0000 (10:40 +0000)]
Give the binary op stubs better names to make profiles more informative.
Review URL: http://codereview.chromium.org/488017

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3454 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoCorrect mistyped file name for svn:ignore property.
mikhail.naganov@gmail.com [Fri, 11 Dec 2009 09:12:56 +0000 (09:12 +0000)]
Correct mistyped file name for svn:ignore property.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3453 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd '.cpplint-cache' to ignore lists for SVN and Git. This change also sets 'ignore...
mikhail.naganov@gmail.com [Fri, 11 Dec 2009 09:01:38 +0000 (09:01 +0000)]
Add '.cpplint-cache' to ignore lists for SVN and Git. This change also sets 'ignore' property on '.', although this change isn't visible in Rietveld.

TBR=sgjesse@chromium.org

Review URL: http://codereview.chromium.org/489011

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3452 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years ago- Ensure the stack trace limit is not negative.
iposva@chromium.org [Thu, 10 Dec 2009 18:33:34 +0000 (18:33 +0000)]
- Ensure the stack trace limit is not negative.
- Enable fuzzing of the CollectStackTrace runtime call.

Review URL: http://codereview.chromium.org/491005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3451 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoUse uint64_t to keep 64 MacOS happy.
antonm@chromium.org [Thu, 10 Dec 2009 17:46:45 +0000 (17:46 +0000)]
Use uint64_t to keep 64 MacOS happy.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3450 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoReapply keyed load cache probing in generated code. I introduced a
ager@chromium.org [Thu, 10 Dec 2009 15:10:50 +0000 (15:10 +0000)]
Reapply keyed load cache probing in generated code.  I introduced a
bug just before committing which broke snapshot builds.  The code is
nearly identical to the previous submit.

TBR=erik.corry@gmail.com
Review URL: http://codereview.chromium.org/491004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3449 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFast codegen: Working break and continue.
lrn@chromium.org [Thu, 10 Dec 2009 14:06:08 +0000 (14:06 +0000)]
Fast codegen: Working break and continue.
Started framework for all intra-functional outward control transfers,
including handling of try/finally.

Review URL: http://codereview.chromium.org/466033

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3448 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoDisible the toplevel compiler for code containing for loops.
kmillikin@chromium.org [Thu, 10 Dec 2009 13:12:13 +0000 (13:12 +0000)]
Disible the toplevel compiler for code containing for loops.
Review URL: http://codereview.chromium.org/492003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3447 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRevert keyed load cache probing in generated code.
ager@chromium.org [Thu, 10 Dec 2009 12:52:28 +0000 (12:52 +0000)]
Revert keyed load cache probing in generated code.

Crashes on Windows.

TBR=erik.corry@gmail.com
Review URL: http://codereview.chromium.org/488006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3446 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoProbe keyed load cache in generic keyed load stub.
ager@chromium.org [Thu, 10 Dec 2009 09:21:23 +0000 (09:21 +0000)]
Probe keyed load cache in generic keyed load stub.

Only implemented on ia32 and x64 for now.  The generic keyed load stub
on arm is falling behind and it is time to fix that, but that will be
a separate change.

Review URL: http://codereview.chromium.org/460142

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3445 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix extra slash typo in comment.
lrn@chromium.org [Thu, 10 Dec 2009 08:33:09 +0000 (08:33 +0000)]
Fix extra slash typo in comment.

Review URL: http://codereview.chromium.org/487007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3444 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years ago- Drop the executable bit of SConstruct.
iposva@chromium.org [Wed, 9 Dec 2009 17:20:51 +0000 (17:20 +0000)]
- Drop the executable bit of SConstruct.

Review URL: http://codereview.chromium.org/467057

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3443 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix Win64 build problem.
lrn@chromium.org [Wed, 9 Dec 2009 16:40:54 +0000 (16:40 +0000)]
Fix Win64 build problem.

Review URL: http://codereview.chromium.org/475003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3442 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix arm and x64 build.
kmillikin@chromium.org [Wed, 9 Dec 2009 15:34:57 +0000 (15:34 +0000)]
Fix arm and x64 build.

The ApiGetterEntryStub is not implemented on arm or x64, but the
platform-specific implementations need to change to match the change
to the platform-independent signatures.

TBR=ager@chromium.org

Review URL: http://codereview.chromium.org/471009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3441 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix a crash caused by garbage collection during generation of a
kmillikin@chromium.org [Wed, 9 Dec 2009 14:54:34 +0000 (14:54 +0000)]
Fix a crash caused by garbage collection during generation of a
callback load (or keyed load) IC.

The problem was that the IC code calls a stub, which can allocate and
thus trigger a GC if the stub is not already generated.  Problem is
solved by adding the ability to "try" to call a stub, trying to
generate the stub code if necessary but signaling an allocation
failure if generating the code is not possible.

Review URL: http://codereview.chromium.org/472002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3440 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoExternal string table.
vitalyr@chromium.org [Wed, 9 Dec 2009 14:32:45 +0000 (14:32 +0000)]
External string table.

Instead of weak handles external strings use a separate table.  This
table uses 5 times less memory than weak handles.  Moreover, since we
don't have to follow the weak handle callback protocol we can collect
the strings faster and even on scavenge collections.

Review URL: http://codereview.chromium.org/467037

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3439 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd Object.create from ECMAScript5. Supports value, writable, enumerable, get
erik.corry@gmail.com [Wed, 9 Dec 2009 13:56:58 +0000 (13:56 +0000)]
Add Object.create from ECMAScript5.  Supports value, writable, enumerable, get
and set.  Doesn't support configurable yet.  See
http://code.google.com/p/v8/issues/detail?id=460
Review URL: http://codereview.chromium.org/463040

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3438 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoCreate literal boilerplate as part of cloning in the top-level compiler.
fschneider@chromium.org [Wed, 9 Dec 2009 13:06:08 +0000 (13:06 +0000)]
Create literal boilerplate as part of cloning in the top-level compiler.

When generating code for object and array literals we performed
the check if the a boilerplate already exists in generated code.

In the top-level compiler we now do this check in a new runtime
function. This makes the generated code more compact for top-level code.

Review URL: http://codereview.chromium.org/465148

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3437 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoCreate the framework for adding a thumb2 backend for ARM. See http://codereview...
erik.corry@gmail.com [Wed, 9 Dec 2009 12:56:42 +0000 (12:56 +0000)]
Create the framework for adding a thumb2 backend for ARM.  See codereview.chromium.org/464070

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3436 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix for bug 512 from Subrato De, CodeAurora.
erik.corry@gmail.com [Wed, 9 Dec 2009 11:14:45 +0000 (11:14 +0000)]
Fix for bug 512 from Subrato De, CodeAurora.
Review URL: http://codereview.chromium.org/464016

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3435 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPerform string add in generated code on X64 platform
sgjesse@chromium.org [Wed, 9 Dec 2009 09:35:41 +0000 (09:35 +0000)]
Perform string add in generated code on X64 platform

This is a port of the IA-32 version from r3400 (http://code.google.com/p/v8/source/detail?r=3400). In the X64 version the additional registers are used to avoid loading the instance type and arguments several times.
Review URL: http://codereview.chromium.org/460109

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3434 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix subtle bug in Math.min and Math.max with non-Smi zero.
erik.corry@gmail.com [Tue, 8 Dec 2009 10:18:28 +0000 (10:18 +0000)]
Fix subtle bug in Math.min and Math.max with non-Smi zero.
See http://codereview.chromium.org/470001
From sra.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3433 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix for issue 545: don't reuse this VariableProxy.
lrn@chromium.org [Tue, 8 Dec 2009 09:43:51 +0000 (09:43 +0000)]
Fix for issue 545: don't reuse this VariableProxy.

Review URL: http://codereview.chromium.org/464069

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3432 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdding ThisFunction expression to top-level compiler.
fschneider@chromium.org [Mon, 7 Dec 2009 15:10:39 +0000 (15:10 +0000)]
Adding ThisFunction expression to top-level compiler.

Review URL: http://codereview.chromium.org/466056

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3430 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoThe toplevel code generator assumed that declarations did not shadow
kmillikin@chromium.org [Mon, 7 Dec 2009 13:31:47 +0000 (13:31 +0000)]
The toplevel code generator assumed that declarations did not shadow
parameters.  This could case the initial value to be lost or worse, a
crash.

Fix by handling the case of a declaration shadowing both
stack-allocated parameters and those in the arguments object.

This is related to V8 issue 540.
http://code.google.com/p/v8/issues/detail?id=540

BUG=29565
Review URL: http://codereview.chromium.org/469006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3429 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPerformance improvement for Math.max and Math.min
erik.corry@gmail.com [Mon, 7 Dec 2009 08:38:20 +0000 (08:38 +0000)]
Performance improvement for Math.max and Math.min
Patch from sra.
http://codereview.chromium.org/470001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3427 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFixed operator precedence bug in heap stat recording.
christian.plesner.hansen@gmail.com [Fri, 4 Dec 2009 14:35:33 +0000 (14:35 +0000)]
Fixed operator precedence bug in heap stat recording.

Review URL: http://codereview.chromium.org/465055

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3426 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRefactor code for generating assignments in the fast compiler.
fschneider@chromium.org [Fri, 4 Dec 2009 14:30:27 +0000 (14:30 +0000)]
Refactor code for generating assignments in the fast compiler.

1. Refactor the structure of VisitAssignment: The existing code is not ideal
to be extended with support for compound assignments.

2. Reuse common code for keyed property assigments: Now variables rewritten
to a property (.arguments access) are treated like normal keyed property
assignments. This allows us to remove some code duplication.

Review URL: http://codereview.chromium.org/456024

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3425 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRemove passing of argc in r0 when calling builtins on ARM as it is not used
sgjesse@chromium.org [Fri, 4 Dec 2009 12:17:02 +0000 (12:17 +0000)]
Remove passing of argc in r0 when calling builtins on ARM as it is not used
Review URL: http://codereview.chromium.org/466031

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3422 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix issue 540 by handling the case that a declaration is in the
kmillikin@chromium.org [Fri, 4 Dec 2009 11:59:09 +0000 (11:59 +0000)]
Fix issue 540 by handling the case that a declaration is in the
arguments object.

See http://code.google.com/p/v8/issues/detail?id=540
Review URL: http://codereview.chromium.org/460070

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3421 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPrepare push to trunk. Now working on 2.0.4.
sgjesse@chromium.org [Fri, 4 Dec 2009 11:50:27 +0000 (11:50 +0000)]
Prepare push to trunk. Now working on 2.0.4.
Review URL: http://codereview.chromium.org/462034

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3420 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd a test that was left off my last submit.
kmillikin@chromium.org [Fri, 4 Dec 2009 11:46:20 +0000 (11:46 +0000)]
Add a test that was left off my last submit.
Review URL: http://codereview.chromium.org/466003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3419 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd alignment check to object allocated in generated code for x64 and ARM
sgjesse@chromium.org [Fri, 4 Dec 2009 10:22:41 +0000 (10:22 +0000)]
Add alignment check to object allocated in generated code for x64 and ARM
Review URL: http://codereview.chromium.org/460068

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3418 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoChanged recording of heap stats on OOM to store data directly in local
christian.plesner.hansen@gmail.com [Fri, 4 Dec 2009 10:18:30 +0000 (10:18 +0000)]
Changed recording of heap stats on OOM to store data directly in local
variables rather than in a stack-allocated struct.  The struct field
values turned out not to be available in minidumps.

Review URL: http://codereview.chromium.org/460069

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3417 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoString check for binary add on x64 and ARM
sgjesse@chromium.org [Fri, 4 Dec 2009 09:58:05 +0000 (09:58 +0000)]
String check for binary add on x64 and ARM

The checking for strings when performing binary add was missing on x64 and ARM. This is a prerequisite for adding string add in generated code for these two platforms.
Review URL: http://codereview.chromium.org/465028

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3416 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoBetter handling of allocation alignment in generated code
sgjesse@chromium.org [Fri, 4 Dec 2009 07:43:40 +0000 (07:43 +0000)]
Better handling of allocation alignment in generated code

Calculate the aligned size of strings when allocating from generated code using kObjectAlignmentMask

Add native code assert to make sure allocation from generated code keeps alignment.
Review URL: http://codereview.chromium.org/462025

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3415 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix Windows build.
ager@chromium.org [Thu, 3 Dec 2009 11:34:45 +0000 (11:34 +0000)]
Fix Windows build.

TBR=sgjesse@chromium.org
Review URL: http://codereview.chromium.org/466018

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3410 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoMove for-in cache validity check to generated code.
ager@chromium.org [Thu, 3 Dec 2009 10:24:16 +0000 (10:24 +0000)]
Move for-in cache validity check to generated code.
Review URL: http://codereview.chromium.org/464002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3407 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdded recording of heap and global handle stats in a stack-allocated
christian.plesner.hansen@gmail.com [Thu, 3 Dec 2009 10:16:37 +0000 (10:16 +0000)]
Added recording of heap and global handle stats in a stack-allocated
struct on fatal out of memory.  This should cause the information to
be included in minidumps so we can get a better idea of the state of
v8 on OOMs.

Review URL: http://codereview.chromium.org/462019

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3406 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoMove function nan_value to POSIX platform file
sgjesse@chromium.org [Thu, 3 Dec 2009 09:29:21 +0000 (09:29 +0000)]
Move function nan_value to POSIX platform file

Strictly speaking NAN from math.h is not in a POSIX standard but in C99. However it seems that all our POSIX platforms have it.
Review URL: http://codereview.chromium.org/464015

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3405 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoCheck return code from fwrite
sgjesse@chromium.org [Thu, 3 Dec 2009 08:48:32 +0000 (08:48 +0000)]
Check return code from fwrite

BUG=http://code.google.com/p/v8/issues/detail?id=453
TEST=none
Review URL: http://codereview.chromium.org/458016

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3404 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoWhen executing 'profile' request, delegate to api function, not to internal one.
mikhail.naganov@gmail.com [Thu, 3 Dec 2009 08:44:33 +0000 (08:44 +0000)]
When executing 'profile' request, delegate to api function, not to internal one.

Review URL: http://codereview.chromium.org/466006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3403 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix compilation errors on some compilers
sgjesse@chromium.org [Thu, 3 Dec 2009 08:32:04 +0000 (08:32 +0000)]
Fix compilation errors on some compilers

TBR=ager@chromium.org
Review URL: http://codereview.chromium.org/462018

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3402 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoAdd missing function CpuFeaturesImpliedByPlatform to OpenBSD platform file
sgjesse@chromium.org [Thu, 3 Dec 2009 08:23:10 +0000 (08:23 +0000)]
Add missing function CpuFeaturesImpliedByPlatform to OpenBSD platform file
Review URL: http://codereview.chromium.org/465025

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3401 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoPerform string add in generated code on IA-32 platforms
sgjesse@chromium.org [Thu, 3 Dec 2009 07:56:21 +0000 (07:56 +0000)]
Perform string add in generated code on IA-32 platforms

This adds a code stub which can do most of what Heap::AllocateConsString can do. It bails out if the result cannot fit in new space or if the result is a short (flat) string and one argument is an ascii string and the other a two byte string. It also bails out if adding two one character strings as Heap::AllocateConsString has special handling of this utilizing the symbol table. The stub is used both for the binary add operation and for StringAdd calls from runtime JavaScript files. Extended the string add test to cover all sizes of flat result stings.
Review URL: http://codereview.chromium.org/442024

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3400 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFixed register usage in 64-bit version of
kbr@chromium.org [Wed, 2 Dec 2009 19:26:16 +0000 (19:26 +0000)]
Fixed register usage in 64-bit version of
KeyedLoadIC::GenerateExternalArray which was causing it to go to the
slow case all the time.

Review URL: http://codereview.chromium.org/458008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3399 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoInitial OpenBSD support
sgjesse@chromium.org [Wed, 2 Dec 2009 13:38:50 +0000 (13:38 +0000)]
Initial OpenBSD support

Patch by Peter Valchev <pvalchev@gmail.com>.

TBR=ager@chromium.org
Review URL: http://codereview.chromium.org/465002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3398 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix adding short external ascii strings
sgjesse@chromium.org [Wed, 2 Dec 2009 12:58:10 +0000 (12:58 +0000)]
Fix adding short external ascii strings

BUG=http://code.google.com/p/v8/issues/detail?id=536
TEST=cctest/test-strings/ExternalShortStringAdd
Review URL: http://codereview.chromium.org/466001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3397 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoFix debug build
sgjesse@chromium.org [Tue, 1 Dec 2009 15:52:02 +0000 (15:52 +0000)]
Fix debug build

Missed out on reflecting changes to Script object layout in objects-debug.cc.

Also fixed http://codereview.chromium.org/450034/diff/2006/2010.

TBR=ager@chromium.org
Review URL: http://codereview.chromium.org/456021

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3394 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

14 years agoRemove the last context dependent reference from the Script object
sgjesse@chromium.org [Tue, 1 Dec 2009 14:36:45 +0000 (14:36 +0000)]
Remove the last context dependent reference from the Script object

For scripts originating from a call to eval the Script object used to hold a reference to the function from where the eval was called together with the code offset within that function of the eval call. This is used by the stack trace and is part of the debugger protocol. In order to avoid storing the function the script, the position within the script and the name of the function calling eval is stored instead. This avoids holding context dependent objects in the script object.

The calculation of the position of the eval in the script holding the eval is now done when the eval script is compiled as it is not possible to postpone this unless a reference is kept to the generated code for the function calling eval.

BUG=http://code.google.com/p/v8/issues/detail?id=528
TEST=cctest/test-api/Regress528
Review URL: http://codereview.chromium.org/450034

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3393 ce2b1a6d-e550-0410-aec6-3dcde31c8c00