Commit Graph

2164 Commits

Author SHA1 Message Date
fschneider@chromium.org
c76856c292 Introduce flag for using the fast compiler where possible.
We use the fast compiler only for top-level code right now.
When always_fast_compiler is set to true, we compile with
the fast compiler whereever possible.

By default this flag is set to false. 

Review URL: http://codereview.chromium.org/449012

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3381 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-30 13:35:59 +00:00
kasperl@chromium.org
402e356e82 Clear the compilation cache just before starting to do mark-sweep
garbage collections when receiving idle notifications. This allows
us to get rid of source code strings and generated code in the
heap that would otherwise be kept around in idle V8 instances.
Review URL: http://codereview.chromium.org/450007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3375 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-30 07:57:32 +00:00
sgjesse@chromium.org
6d163d9fd0 Remove usage of JSArray in Script object
Storing a JSArray in the Script object could cause an indirect reference from the compilation cache to a global object to be created. Now the line ends are only stored as a FixedArrya and when that is needed in JavaScript a JSArray copy is created. Changed some of the JavaScript code to cache the line ends in a local variable for better performance.

BUG=http://code.google.com/p/v8/issues/detail?id=528
TEST=test/test-api/Bug528
Review URL: http://codereview.chromium.org/434117

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3374 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-27 14:10:48 +00:00
sgjesse@chromium.org
5764362d6f Add test to expose bug 528
BUG=http://code.google.com/p/v8/issues/detail?id=528
TEST=cctest/test-api/Bug528
Review URL: http://codereview.chromium.org/443021

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3372 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-27 07:57:45 +00:00
fschneider@chromium.org
3a21aafa7c Added fast compiler support for calling JS runtime functions.
Also added a simple test that invokes a JS runtime function
in top-level code.

Review URL: http://codereview.chromium.org/437081

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3371 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-27 00:28:06 +00:00
fschneider@chromium.org
1c90793581 Fix bug in the fast compiler's object literal code
Fixes issue 526:

BUG=http://code.google.com/p/v8/issues/detail?id=526

The object literals code in the fast compiler returned an incorrect result 
when getter or setters are defined together with computed properties.

Added a regression test that captures the most reduced version of this
problem. 

Also added a test for object literals with getters/setters and 
prototype properties.


Review URL: http://codereview.chromium.org/444001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3370 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-26 21:13:20 +00:00
lrn@chromium.org
eded148caf Fast-codegen: Arguments object working on all platforms.
This time it's true.

Review URL: http://codereview.chromium.org/405033


git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3369 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-26 10:28:32 +00:00
kbr@chromium.org
c135f2de07 Fixed incorrect instruction usage in KeyedLoadIC for byte and word
external array types. Added regression test based on real-world
failing code and verified that it would have caught this error.

Review URL: http://codereview.chromium.org/437052

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3366 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-25 20:29:11 +00:00
ager@chromium.org
7a6ebbdd6d Do not crash if a V8 extension fails to compile or throws an exception
when the code is run.

Instead, return an empty context handle so the failure to create a
context can be handled.

BUG=http://crbug.com/28486
Review URL: http://codereview.chromium.org/442005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3364 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-25 16:46:56 +00:00
mikhail.naganov@gmail.com
2af151ee63 Include getters and setters callbacks invocations in CPU profiler log.
Logging getters and setters from DOM API is extremely useful for web
developers as setting (and getting!) several properties can cause
page relayouts which take significant time.

Review URL: http://codereview.chromium.org/434074

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3363 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-25 16:39:18 +00:00
antonm@chromium.org
93d6199de0 When processing global vars initialization account for the case of splitted
global object (using hidden prototypes):

1) setters might be not on the global object itself, but on its prototypes;
2) if property on one of prototypes is readonly, we could shadow it.
Review URL: http://codereview.chromium.org/434035

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3362 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-25 15:45:37 +00:00
erik.corry@gmail.com
c5052399e4 Fix 64 bit build. TBR=ager.
Review URL: http://codereview.chromium.org/442003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3361 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-25 13:27:15 +00:00
erik.corry@gmail.com
10e183d353 Make heap serialization nondestructive.
Review URL: http://codereview.chromium.org/441017

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3360 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-25 12:55:33 +00:00
erik.corry@gmail.com
059b9bbc5e Update with a hopefully complete list of macros
used by gcc to indicate ARM architectures.
Review URL: http://codereview.chromium.org/440017

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3359 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-25 11:23:48 +00:00
mikhail.naganov@gmail.com
3267107040 Change 'debugger-auto-break' flag status.
I think we can promote this flag to stable status and enable it by default.

Review URL: http://codereview.chromium.org/441016

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3357 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-25 10:11:34 +00:00
erik.corry@gmail.com
b0b5dbadbd Fix bug 516 by recognizing yet another gcc macro that
indicates ARMv6.
Review URL: http://codereview.chromium.org/437056

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3356 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-25 09:55:49 +00:00
lrn@chromium.org
d4d2277e2e Fast-compiler: Add stack limit checks to back edges of while, do-while and for.
A few other tweaks.

Review URL: http://codereview.chromium.org/435020


git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3355 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-25 09:05:30 +00:00
sgjesse@chromium.org
8dbedbb30e Added an ASSERT to code from r3350
Review URL: http://codereview.chromium.org/441014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3354 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-25 08:14:23 +00:00
sgjesse@chromium.org
4bc5aee624 Fix compilation error in r3350
TBR=ager@chromium.org
Review URL: http://codereview.chromium.org/435021

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3352 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-24 14:45:27 +00:00
sgjesse@chromium.org
eb4d261e24 Remove the different length string types
The different length string types was used to encode the string length and the hash in one field. This is now split into two fields one for length and one for hash. The hash field still encodes the array index of the string if it has one. If an array index is encoded in the hash field the string length is added to the top bits of the hash field to avoid a hash value of zero.

On 32-bit this causes an additional 4 bytes to be used for all string objects. On 64-bit this will be half on average dur to pointer alignment.
Review URL: http://codereview.chromium.org/436001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3350 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-24 14:10:06 +00:00
ager@chromium.org
56074f783a Prepare push to trunk. Now working on version 2.0.3.
Review URL: http://codereview.chromium.org/435018

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3349 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-24 13:30:15 +00:00
ager@chromium.org
6ce4b10608 Re-enable all declarations in fast top-level compiler.
Disable fast top-level compiler for now because of issues 525 and 526.

Add regression test for issue 525.

Review URL: http://codereview.chromium.org/438017

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3348 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-24 12:23:29 +00:00
mikhail.naganov@gmail.com
9d440fdd36 Store API callback entry address prior to entering a callback.
Callback entry address is stored in VMState and is later retrieved by
profiler stack sampler. This makes possible relating API entry to JS
stack, and this is simpler than trying to unwind native stack.

Review URL: http://codereview.chromium.org/437004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3344 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-23 14:43:00 +00:00
mikhail.naganov@gmail.com
c2f12a7293 Reimplement logging of API callbacks entry points in an easier way.
Now they are logging during "LogCompiledFunctions" cycle. API functions
are detected by examining SFI's "function_data" field.

Review URL: http://codereview.chromium.org/414036

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3343 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-23 09:19:24 +00:00
ager@chromium.org
1fe8655fbc Change the signature for StrNDup to match the implementation. This
mismatch causes problems if attempting to build d8 with readline
support.

Review URL: http://codereview.chromium.org/414033

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3342 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-20 13:08:44 +00:00
mikhail.naganov@gmail.com
72d7453dc2 Fix issue 520: wrap Logger calls into LOG macro
BUG=520

Review URL: http://codereview.chromium.org/422001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3341 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-20 12:38:59 +00:00
mikhail.naganov@gmail.com
a9d7c378d8 Add logging of callbacks in prof-lazy mode.
This is needed to show calls to DOM in CPU profiles. I can think
of a better approach like adding specific functions into V8 API
for explicitly providing callback names and modifying bindings codegen
appropriately. My plan is as follows:
 - submit this CL;
 - implement anything I need to process log data and display DOM
   calls in profiles;
 - think again about adding specific functions and modifying bindings
   codegen.

BUG=http://code.google.com/p/chromium/issues/detail?id=27613

Review URL: http://codereview.chromium.org/402100

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3340 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-20 12:15:46 +00:00
ager@chromium.org
11ba1bd163 Prepare push to trunk. Now working on version 2.0.2.
Review URL: http://codereview.chromium.org/421003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3338 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-20 12:03:44 +00:00
ager@chromium.org
157e4f993b Disable the fast top-level compiler for non-global declarations.
Reintroduce the bailout that was put in in revision 3224.

It causes interactive ui test failures in Chromium.

TBR=fschneider@chromium.org
Review URL: http://codereview.chromium.org/421002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3337 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-20 11:45:18 +00:00
erik.corry@gmail.com
43e8d5fa7f Some optimizations for packer.js.
Review URL: http://codereview.chromium.org/409007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3336 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-20 10:11:45 +00:00
fschneider@chromium.org
d9d18b30f5 Implement IS_OBJECT and IS_FUNCTION as inlined runtime functions.
Summary:
This change fixes a performance regression introduced by the special
handling of regular expressions in typeof expressions.
As a result we regain ~8% speedup on 3d-raytrace and ~13% on boyer
(vs bleeding edge)

Description:
The macros IS_OBJECT and IS_FUNCTION are frequently used in the 
JS runtime functions. 
By introducing new inlined runtime functions %_IsFunction and %_IsObject
we avoid invoking the more expensive %_ClassOf function plus comparing
its result to a string. 


Review URL: http://codereview.chromium.org/399111

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3335 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-19 07:41:32 +00:00
ager@chromium.org
3cf9ce4736 Fix crash in string replace with regexp. If the suffix of the subject
string is larger than 2047 chars we will encode the slice as two Smis
instead of one.  The calculation of the max size of the string builder
did not take this into account.

BUG=http://code.google.com/p/v8/issues/detail?id=515
Review URL: http://codereview.chromium.org/402056

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3334 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-18 18:48:04 +00:00
ager@chromium.org
981480c3bd Prepare push of version 2.0.0 to trunk. Now working on version 2.0.1.
Review URL: http://codereview.chromium.org/405018

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3332 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-18 12:24:46 +00:00
erik.corry@gmail.com
446b1f4443 Correct dumb assert.
Review URL: http://codereview.chromium.org/402048

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3331 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-18 12:14:21 +00:00
ager@chromium.org
ce20b5b461 Fixed bug in pixel array inline cache on x64. The value was not
zero-extended as it should be.  Therefore, the index into the pixel
array could influence the value on reads.

BUG=http://code.google.com/p/chromium/issues/detail?id=26337
Review URL: http://codereview.chromium.org/399067

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3328 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-18 11:36:29 +00:00
erik.corry@gmail.com
701c00f8b4 Accelerate charCodeAt on ARM.
Review URL: http://codereview.chromium.org/402012

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3327 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-18 10:20:24 +00:00
sgjesse@chromium.org
d55d3ce89a Add DebugMessageDispatchHandler
A callback on the debugger thread when a message is received. Allows the 
embedding application to wake up the main thread in order to handle the 
message. Useful when the embedding application is idle and sitting in a 
select() call.

Patch by Ryan Dahl <coldredlemur@gmail.com>

Review URL: http://codereview.chromium.org/395013


git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3326 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-18 08:59:28 +00:00
kasperl@chromium.org
d5c355e367 Make sure to retry SetHiddenPropertiesObject if it
returns a failure.
Review URL: http://codereview.chromium.org/402034

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3323 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-18 00:09:28 +00:00
lrn@chromium.org
1bc9a21e9a Sorry, this should have gone to http://codereview.chromium.org/391014
I added some tests and swapped the lhs and rhs of the && operator in v8natives.js as requested.

Review URL: http://codereview.chromium.org/384132


git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3319 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-17 13:54:05 +00:00
antonm@chromium.org
9a545949a4 Restore invariant (next of first deallocated must point to the head) before calling into weak
callbacks.

Otherwise if callback allocates a new handle, it could orphan some global handles (with disastorous
consequences if those global handles are cached).

Review URL: http://codereview.chromium.org/395024

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3318 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-17 13:50:07 +00:00
erik.corry@gmail.com
41749529dd Speed up charCodeAt on very large cons strings, by insisting on
flattening the strings and not trying too hard to traverse a big
cons tree from generated code.
Review URL: http://codereview.chromium.org/402008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3317 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-17 10:28:04 +00:00
sgjesse@chromium.org
63925e5013 Add missing case for shr in IA-32 disassembler
Review URL: http://codereview.chromium.org/397024

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3316 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-17 10:24:27 +00:00
sgjesse@chromium.org
459e4c6b0c Changes to Intel shift functions
Change name of shifts picking the shift count from cl to sal_cl, shl_cl and shr_cl.

Add special encoding of shift by one for shr which was missing it.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3314 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-17 08:35:43 +00:00
fschneider@chromium.org
53bee811ad Re-enable using push instructions for syncing the virtual frame.
This change fixes the problem with the original version of this approach
(r3032) that may lead to a corrupted stack if we would invoke spilling during 
syncing a large SMI constant (unsafe SMIs) in the virtual frame.

The new code for storing unsafe SMI constants does not use an extra temporary 
register. This prevents the compiler from ever having to spill during a 
virutal frame sync operation.

For storing a large SMI constant we previously generated:

  mov ecx, (large_smi & 0x0000ffff)
  xor ecx, (large_smi & 0xffff0000)
  push ecx

we now generate:

  push (large_smi & 0x0000ffff)
  or   [esp], (large_smi & 0xffff0000)

Not using a temporary register avoids spilling within an nvocation 
of VirtualFrame::SyncRange.

Review URL: http://codereview.chromium.org/391079

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3313 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-16 23:11:19 +00:00
yurys@chromium.org
9bafc32932 Step next should respect do/while condition block.
Condition block of do/while statements is a valid break location so it should have its own position. The block is represented by a regular Expression node so we cannot store the position in it, instead the position is stored in a separate field in DoWhileStatement AST node.

BUG=514
Review URL: http://codereview.chromium.org/385136

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3312 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-16 21:59:31 +00:00
yurys@chromium.org
b6ce7ce8a3 Fix some presubmit errors.
TBR=sgjesse@chromium.org
Review URL: http://codereview.chromium.org/400001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3311 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-16 16:58:09 +00:00
yurys@chromium.org
0a384a708d Skip debugger frames when looking for calling context.
BUG=509

Review URL: http://codereview.chromium.org/398002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3310 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-16 14:41:55 +00:00
erik.corry@gmail.com
cc3896dfac Fix bug 503: undefined <= undefined should return false on ARM.
Review URL: http://codereview.chromium.org/399001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3309 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-16 14:12:27 +00:00
erik.corry@gmail.com
64941f1cf9 * Remove old snapshot implementation
Review URL: http://codereview.chromium.org/394007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3307 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-16 12:08:40 +00:00
fschneider@chromium.org
5820f951c1 Improve the allocation and initialization of locals on IA32 in the top-level compiler.
This optimization is already done on x64 and ARM. 

Until now we used a push immediate for each local variable on IA32:

  push $undefined
  push $undefined
  ...

to initialize each local variable. This change does:

  mov eax, $undefined
  push eax
  push eax
  ...



Review URL: http://codereview.chromium.org/393009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@3304 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
2009-11-13 13:59:07 +00:00