platform/upstream/v8.git
15 years agoLog ticks when the prof flag is set. Avoid letting --prof
kasperl@chromium.org [Fri, 3 Oct 2008 10:51:08 +0000 (10:51 +0000)]
Log ticks when the prof flag is set. Avoid letting --prof
imply --log.
Review URL: http://codereview.chromium.org/6444

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@423 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoChanged so that --log-code implies --log.
christian.plesner.hansen@gmail.com [Fri, 3 Oct 2008 10:42:53 +0000 (10:42 +0000)]
Changed so that --log-code implies --log.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@422 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRefactored the logic for entering the debugger into one abstraction EnterDebugger.
sgjesse@chromium.org [Fri, 3 Oct 2008 09:57:18 +0000 (09:57 +0000)]
Refactored the logic for entering the debugger into one abstraction EnterDebugger.

Removed the static initializer for Top::break_access_.
Review URL: http://codereview.chromium.org/6234

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@421 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years ago- Added fast case for extending the JSObject properties storage.
bak@chromium.org [Fri, 3 Oct 2008 09:16:12 +0000 (09:16 +0000)]
- Added fast case for extending the JSObject properties storage.
- Changed date to compute local_time_offset at load time.

Review URL: http://codereview.chromium.org/6441

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@420 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoUpdated the snapshot builder to reflect the removed extensions in http://codereview...
sgjesse@chromium.org [Fri, 3 Oct 2008 09:05:58 +0000 (09:05 +0000)]
Updated the snapshot builder to reflect the removed extensions in codereview.chromium.org/6233.
Review URL: http://codereview.chromium.org/6440

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@419 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFixed the path for include files v8.h and v8-debug.h in Visual Studio project.
sgjesse@chromium.org [Fri, 3 Oct 2008 08:31:19 +0000 (08:31 +0000)]
Fixed the path for include files v8.h and v8-debug.h in Visual Studio project.
Review URL: http://codereview.chromium.org/6601

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@418 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRemoved the print, load, quit and version extensions from the VM. Moved the print...
sgjesse@chromium.org [Fri, 3 Oct 2008 08:30:22 +0000 (08:30 +0000)]
Removed the print, load, quit and version extensions from the VM. Moved the print extension to the compiler test as it was used there. Changed the serializer tests to only use the gc extension.

This also gets rid of 8 static initializers.
Review URL: http://codereview.chromium.org/6233

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@417 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoCleanup string builtins.
ager@chromium.org [Fri, 3 Oct 2008 07:58:18 +0000 (07:58 +0000)]
Cleanup string builtins.
Review URL: http://codereview.chromium.org/6602

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@416 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoMinor cleanups to macros.py.
kasperl@chromium.org [Fri, 3 Oct 2008 07:36:03 +0000 (07:36 +0000)]
Minor cleanups to macros.py.
Review URL: http://codereview.chromium.org/6232

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@415 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoMake sure that the name accessor on functions return the expected
ager@chromium.org [Fri, 3 Oct 2008 07:14:31 +0000 (07:14 +0000)]
Make sure that the name accessor on functions return the expected
names.

- Set the correct name of library functions.
- Set the name of C++ callback functions.
- Clean up a couple of out-dated comments related to literal creation.

Review URL: http://codereview.chromium.org/6223

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@414 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years ago- Case-sensitive atomic regular expressions now use the same code as
christian.plesner.hansen@gmail.com [Thu, 2 Oct 2008 15:35:28 +0000 (15:35 +0000)]
- Case-sensitive atomic regular expressions now use the same code as
  String.indexOf to do matching.
- The --log option is no longer automatically enabled by the other log
  options.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@413 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoPrint one line when pushing an immediate followed by a pop is
kmillikin@chromium.org [Thu, 2 Oct 2008 14:23:20 +0000 (14:23 +0000)]
Print one line when pushing an immediate followed by a pop is
eliminated.
Review URL: http://codereview.chromium.org/6078

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@412 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoReplaces two non-private uses of AddProperty with IgnoreAttributesAndSetLocalProperty...
whessev8 [Thu, 2 Oct 2008 13:45:21 +0000 (13:45 +0000)]
Replaces two non-private uses of AddProperty with IgnoreAttributesAndSetLocalProperty.  Adds attributes parameter to IgnoreAtt..Property().  Makes IgnoreAtt..Property() an exact clone of SetProperty(), with explicit changes.

Review URL: http://codereview.chromium.org/5665

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@411 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoIn the code generator, avoid loading the arguments object to the
kmillikin@chromium.org [Thu, 2 Oct 2008 12:44:47 +0000 (12:44 +0000)]
In the code generator, avoid loading the arguments object to the
expression stack when it is already there.  Also, cleanup up the
(two!) extra copies of the arguments object left on the stack.
Review URL: http://codereview.chromium.org/5667

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@410 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoMove code for code generator static member functions, code generation
kmillikin@chromium.org [Thu, 2 Oct 2008 11:54:32 +0000 (11:54 +0000)]
Move code for code generator static member functions, code generation
functions on AST nodes, and helper class member functions (eg, stub
classes) toward the bottom of the codegen*.cc files.  The macro __ is
now #define'd exactly twice and #undef'd twice.
Review URL: http://codereview.chromium.org/6075

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@409 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoSmall cleanup of the code generator: make the static code gen
kmillikin@chromium.org [Thu, 2 Oct 2008 11:14:50 +0000 (11:14 +0000)]
Small cleanup of the code generator: make the static code gen
functions and the ones defined on the AST nodes take a code generator,
rather than a macro assembler and (sometimes) scope.

Uniformly use the __ macro for masm_/masm in the codegen*.cc files.
Review URL: http://codereview.chromium.org/5663

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@408 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoChanged optimization from /Ox to the (more) commonly used /O2 (which includes
sgjesse@chromium.org [Thu, 2 Oct 2008 09:41:01 +0000 (09:41 +0000)]
Changed optimization from /Ox to the (more) commonly used /O2 (which includes
/GF) in both SCons and Visual Studio build. There is not mesurable difference
neither in speed nor in code size.

Added whole program optimization to the SCons release build. This clutters the
linking of the samples a little with the option /LTCG. This option is not
strictly needed for linking, but if it is not specified the linker complaints
as it can see object files compiled with /GL and then restarts itself with
/LTCG which dosen't look very pretty.
Review URL: http://codereview.chromium.org/5664

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@407 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFixed use of undefined in ArraySort.
olehougaard [Thu, 2 Oct 2008 08:58:03 +0000 (08:58 +0000)]
Fixed use of undefined in ArraySort.

Changed 'undefined' in ArraySort to 'void 0'. Also added regression test to catch the error.
Review URL: http://codereview.chromium.org/6073

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@406 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoVarious minor improvements of sort.
olehougaard [Thu, 2 Oct 2008 08:15:20 +0000 (08:15 +0000)]
Various minor improvements of sort.
Review URL: http://codereview.chromium.org/6035

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@405 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRemoved the use of Math.random() and new Date() for building
kasperl@chromium.org [Thu, 2 Oct 2008 07:43:46 +0000 (07:43 +0000)]
Removed the use of Math.random() and new Date() for building
the RNG pool in the crypto benchmark.
Review URL: http://codereview.chromium.org/6071

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@404 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFollow up to change 402:
iposva@chromium.org [Wed, 1 Oct 2008 20:21:34 +0000 (20:21 +0000)]
Follow up to change 402:
- Remove unused symbol finally_state_symbol.

Review URL: http://codereview.chromium.org/5640

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@403 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoGet rid of the local variable we use to keep the state
kasperl@chromium.org [Wed, 1 Oct 2008 10:57:37 +0000 (10:57 +0000)]
Get rid of the local variable we use to keep the state
during the execution of a finally block by just pushing
the state on the execution stack instead.
Review URL: http://codereview.chromium.org/5626

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@402 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAdded test timing.
christian.plesner.hansen@gmail.com [Wed, 1 Oct 2008 09:07:45 +0000 (09:07 +0000)]
Added test timing.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@401 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFix issue 86 by keeping track of the fact that finally blocks
kasperl@chromium.org [Wed, 1 Oct 2008 07:43:00 +0000 (07:43 +0000)]
Fix issue 86 by keeping track of the fact that finally blocks
are evaluated with an extra element on the stack, which needs
to be taken into account when breaking and continuing.

I'll clean up the code and add an abstraction for manipulating
the break stack height in a future CL -- I want to try to get
rid of the separate local variable we keep around for the "state"
when running in a finally block.
Review URL: http://codereview.chromium.org/5625

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@400 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAdded option to parallelize test runs.
christian.plesner.hansen@gmail.com [Tue, 30 Sep 2008 15:32:07 +0000 (15:32 +0000)]
Added option to parallelize test runs.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@399 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoDisallow copy and assigning a few public objects.
deanm@chromium.org [Tue, 30 Sep 2008 11:43:18 +0000 (11:43 +0000)]
Disallow copy and assigning a few public objects.

BUG=97

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@398 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRemove some craaaazy trailing whitespace.
deanm@chromium.org [Tue, 30 Sep 2008 11:41:58 +0000 (11:41 +0000)]
Remove some craaaazy trailing whitespace.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@397 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRemove an ARM jump elimination test.
deanm@chromium.org [Tue, 30 Sep 2008 10:20:58 +0000 (10:20 +0000)]
Remove an ARM jump elimination test.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@396 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFaster sort.
olehougaard [Tue, 30 Sep 2008 09:58:22 +0000 (09:58 +0000)]
Faster sort.

Using insertion sort below a certain threshold to give faster sorting of arrays (esp. short ones).
Review URL: http://codereview.chromium.org/6006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@395 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRemove the ARM jump elimination. This completely removes jump elimination.
deanm@chromium.org [Tue, 30 Sep 2008 09:42:08 +0000 (09:42 +0000)]
Remove the ARM jump elimination.  This completely removes jump elimination.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@394 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRemove x86 jump elimination. This was just complicated enough to make it annoying...
deanm@chromium.org [Mon, 29 Sep 2008 07:33:58 +0000 (07:33 +0000)]
Remove x86 jump elimination.  This was just complicated enough to make it annoying to support short jump encoding.  I measured a code size increase of 5 bytes on the V8 benchmark, from missing one jump to next elimination possibility.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@393 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoClean up some naming to be consistent between runtime and JS code.
deanm@chromium.org [Sun, 28 Sep 2008 13:21:41 +0000 (13:21 +0000)]
Clean up some naming to be consistent between runtime and JS code.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@392 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRemove two cases of needlessly jumping to the next instruction. The jump eliminator...
deanm@chromium.org [Sun, 28 Sep 2008 12:21:12 +0000 (12:21 +0000)]
Remove two cases of needlessly jumping to the next instruction.  The jump eliminator will remove these cases, but it's easy enough to just do statically in the code, and not require the eliminator to do it at runtime.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@391 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRemove ComparisonDeferred and inline the non-smi case. ARM is doing it's own thing...
deanm@chromium.org [Sun, 28 Sep 2008 09:38:09 +0000 (09:38 +0000)]
Remove ComparisonDeferred and inline the non-smi case.  ARM is doing it's own thing here.  This should cut down on code size, and open up two possiblities for short jump encoding.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@390 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFix failing test caused by extra precision in register allocated fp values.
erik.corry@gmail.com [Fri, 26 Sep 2008 23:14:43 +0000 (23:14 +0000)]
Fix failing test caused by extra precision in register allocated fp values.
Review URL: http://codereview.chromium.org/5019

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@389 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFix build breakage on ARM by removing IA32 flags
erik.corry@gmail.com [Fri, 26 Sep 2008 16:14:22 +0000 (16:14 +0000)]
Fix build breakage on ARM by removing IA32 flags

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@388 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoUse -O9, tune for nocoma, assume at least a Pentium when generating code.
erik.corry@gmail.com [Fri, 26 Sep 2008 15:53:42 +0000 (15:53 +0000)]
Use -O9, tune for nocoma, assume at least a Pentium when generating code.
Review URL: http://codereview.chromium.org/4298

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@387 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAdded new test type, message, to test error reporting.
christian.plesner.hansen@gmail.com [Fri, 26 Sep 2008 13:06:49 +0000 (13:06 +0000)]
Added new test type, message, to test error reporting.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@386 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFix stack check wraparound problem for ARM simulator.
erik.corry@gmail.com [Fri, 26 Sep 2008 12:11:35 +0000 (12:11 +0000)]
Fix stack check wraparound problem for ARM simulator.
Review URL: http://codereview.chromium.org/2926

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@385 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoStack checks in generated code for function apply is now controlled
sgjesse@chromium.org [Fri, 26 Sep 2008 10:27:39 +0000 (10:27 +0000)]
Stack checks in generated code for function apply is now controlled
by the check-stack flag. Changed the condition code from greater to
above_equal as the SP should be unsigned (this matches the stack
check in function entry).
Review URL: http://codereview.chromium.org/4296

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@384 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAdding comment to the fix of issue 95.
olehougaard [Fri, 26 Sep 2008 10:25:14 +0000 (10:25 +0000)]
Adding comment to the fix of issue 95.
Review URL: http://codereview.chromium.org/5003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@383 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFix for issue 95.
olehougaard [Fri, 26 Sep 2008 09:15:02 +0000 (09:15 +0000)]
Fix for issue 95.

Fixed QuickSort so it doesn't overflow the stack with non-reflexsive comparison functions.
Review URL: http://codereview.chromium.org/4297

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@382 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoMark test test-debug/ThreadedDebugging as flaky.
sgjesse@chromium.org [Fri, 26 Sep 2008 07:57:11 +0000 (07:57 +0000)]
Mark test test-debug/ThreadedDebugging as flaky.
Review URL: http://codereview.chromium.org/5001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@381 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agowhich it now does.
christian.plesner.hansen@gmail.com [Fri, 26 Sep 2008 03:44:09 +0000 (03:44 +0000)]
which it now does.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@380 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoIntroduce some checks to indicate that the benchmarks have run correctly.
erik.corry@gmail.com [Thu, 25 Sep 2008 13:48:27 +0000 (13:48 +0000)]
Introduce some checks to indicate that the benchmarks have run correctly.
This change affects the results by around 2% so the version number has been
bumped to 2 to avoid confusion.
Review URL: http://codereview.chromium.org/4401

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@379 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoTuning quick sort.
olehougaard [Thu, 25 Sep 2008 13:21:56 +0000 (13:21 +0000)]
Tuning quick sort.

Tuning the quick sort algorithm to avoid degenerating to an n^2 algorithm when all elements are the same.
Review URL: http://codereview.chromium.org/4083

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@378 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAdded copyright headers to tests.
christian.plesner.hansen@gmail.com [Thu, 25 Sep 2008 13:14:23 +0000 (13:14 +0000)]
Added copyright headers to tests.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@377 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFixed problem where the test framework would assume that .svn was a
christian.plesner.hansen@gmail.com [Thu, 25 Sep 2008 13:01:26 +0000 (13:01 +0000)]
Fixed problem where the test framework would assume that .svn was a
test suite and then fail.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@376 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years ago- Added support for warnings on unused test rules.
christian.plesner.hansen@gmail.com [Thu, 25 Sep 2008 12:38:34 +0000 (12:38 +0000)]
- Added support for warnings on unused test rules.
- Added automatic loading of test suites

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@375 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoUsing quick sort for arrays.
olehougaard [Thu, 25 Sep 2008 11:28:02 +0000 (11:28 +0000)]
Using quick sort for arrays.

Using quick sort in ArraySort instead of heap sort for better performance.
Review URL: http://codereview.chromium.org/4065

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@374 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoUpdate version string and ChangeLog. We are now working on version
ager@chromium.org [Thu, 25 Sep 2008 08:45:38 +0000 (08:45 +0000)]
Update version string and ChangeLog.  We are now working on version
0.3.4.
Review URL: http://codereview.chromium.org/4272

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@372 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years ago- Added a map cache for literal objects. This will
bak@chromium.org [Thu, 25 Sep 2008 07:46:07 +0000 (07:46 +0000)]
- Added a map cache for literal objects. This will
  canonicalize maps for object literals. JSON objects
  with the same set of properties names will then
  share the same map.
  This reduces the amount of generated code associated
  with object literals.
- Added a flag canonicalize_object_literal_maps.
  (default true)
- Changed the format of a function's literal array.
  Only the global context is now stored in the literal prefix.

Review URL: http://codereview.chromium.org/4078

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@371 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoMake sure that the body of the function created by calling Function is
ager@chromium.org [Thu, 25 Sep 2008 07:35:45 +0000 (07:35 +0000)]
Make sure that the body of the function created by calling Function is
on a line of its own.  This allows the body to be terminated by a
single-line comment.

Also, make sure to set the name of the function to anonymous after the
fact so that recursion through the name anonymous is not allowed and
so that global variables called anonymous are not shadowed.

This is a fix for http://code.google.com/p/v8/issues/detail?id=85
Review URL: http://codereview.chromium.org/4248

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@370 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoDon't defer the stack check failure code. It is a CallStub, which will be a single...
deanm@chromium.org [Wed, 24 Sep 2008 16:34:00 +0000 (16:34 +0000)]
Don't defer the stack check failure code.  It is a CallStub, which will be a single 5 byte call instruction.  This should cause equivalent code size now, but opens up the opportunity to make one of the most common jcc's to use short encoding in the future.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@369 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoThis is a fix of issue http://b/issue?id=1381845.
feng@chromium.org [Wed, 24 Sep 2008 15:47:34 +0000 (15:47 +0000)]
This is a fix of issue http://b/issue?id=1381845.

Check domain security on prototypes in for-in loop.

Review URL: http://codereview.chromium.org/4236

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@368 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAdded a profiling option (prof) to the SCons build. Currently
sgjesse@chromium.org [Wed, 24 Sep 2008 08:47:39 +0000 (08:47 +0000)]
Added a profiling option (prof) to the SCons build. Currently
it just makes the linker generate a MAP file for the shell
sample on Windows.
Review URL: http://codereview.chromium.org/4064

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@367 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAdded some peephole optimizaitions regarding push of immediate followed
sgjesse@chromium.org [Tue, 23 Sep 2008 13:31:14 +0000 (13:31 +0000)]
Added some peephole optimizaitions regarding push of immediate followed
by pop eax.
Review URL: http://codereview.chromium.org/4212

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@366 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoEnable string pooling in the VS release build. This saves 46k for a v8_shell_sample...
deanm@chromium.org [Tue, 23 Sep 2008 12:32:43 +0000 (12:32 +0000)]
Enable string pooling in the VS release build.  This saves 46k for a v8_shell_sample.exe, I didn't measure performance or the overall change to a Chromium build.  I think this option should always make sense anyway, unless we're using some string constants incorrectly.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@365 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFixed lint error.
christian.plesner.hansen@gmail.com [Tue, 23 Sep 2008 12:23:20 +0000 (12:23 +0000)]
Fixed lint error.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@364 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoMove more functionality from CEntryStub to the helper
kasperl@chromium.org [Tue, 23 Sep 2008 12:21:54 +0000 (12:21 +0000)]
Move more functionality from CEntryStub to the helper
functions in the macro assembler.
Review URL: http://codereview.chromium.org/4402

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@363 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years ago- Added new object type, RegExp, for regular expressions.
christian.plesner.hansen@gmail.com [Tue, 23 Sep 2008 11:45:43 +0000 (11:45 +0000)]
- Added new object type, RegExp, for regular expressions.
- Renamed some ::kSizes to ::kHeaderSizes to avoid confusion.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@362 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAllow platforms (linux and win32) to not force 16-byte alignment
kasperl@chromium.org [Tue, 23 Sep 2008 10:06:58 +0000 (10:06 +0000)]
Allow platforms (linux and win32) to not force 16-byte alignment
of activation frames (needed on Mac OS X).
Review URL: http://codereview.chromium.org/4211

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@361 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoParse /proc/self/maps lines better to handle variations between Linux kernels.
erik.corry@gmail.com [Tue, 23 Sep 2008 08:56:12 +0000 (08:56 +0000)]
Parse /proc/self/maps lines better to handle variations between Linux kernels.
It seems noone has had time to file a bug on this.
Review URL: http://codereview.chromium.org/4210

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@360 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRefactored the code for entering and leaving exit frames (calls
kasperl@chromium.org [Tue, 23 Sep 2008 08:19:26 +0000 (08:19 +0000)]
Refactored the code for entering and leaving exit frames (calls
from JavaScript to C++). Includes a few slight optimizations
like keeping argv in a callee-saved register.
Review URL: http://codereview.chromium.org/4035

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@359 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoChange the logging information for generated stubs to be
kasperl@chromium.org [Tue, 23 Sep 2008 07:33:08 +0000 (07:33 +0000)]
Change the logging information for generated stubs to be
consistent with the other log events for generated code.
Review URL: http://codereview.chromium.org/4033

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@358 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRemove debugging code that snuck in with the ARM calling convention changes.
iposva@chromium.org [Mon, 22 Sep 2008 16:00:56 +0000 (16:00 +0000)]
Remove debugging code that snuck in with the ARM calling convention changes.
Thanks to Erik Corry for spotting this.

Review URL: http://codereview.chromium.org/4015

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@357 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoWork around issue 80 by splitting strings from the environment
kasperl@chromium.org [Mon, 22 Sep 2008 15:50:30 +0000 (15:50 +0000)]
Work around issue 80 by splitting strings from the environment
variables into lists before appending to them.
Review URL: http://codereview.chromium.org/3185

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@356 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRefactor the enum RelocMode changing the naming scheme from lower case to
sgjesse@chromium.org [Mon, 22 Sep 2008 13:57:03 +0000 (13:57 +0000)]
Refactor the enum RelocMode changing the naming scheme from lower case to
upper case. Moved it into the RelocInfo class together with the associated
is_xxx functions. Renamed is_xxx to IsXxx in the process.

Removed the exit_js_frame mode as it was no longer used.

Patch Set 2 renames RELOC_MODE_COUNT to NUMBER_OF_MODES and fixes a couple of lint errors.
Review URL: http://codereview.chromium.org/3186

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@354 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoIn release, disable UNREACHABLE and make UNIMPLEMENTED and FATAL simpler, removing...
deanm@chromium.org [Mon, 22 Sep 2008 09:22:18 +0000 (09:22 +0000)]
In release, disable UNREACHABLE and make UNIMPLEMENTED and FATAL simpler, removing the file name and line numbers, so we carry around less data and can share more code.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@353 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAdd compilation flags for optimizing samples built in release
kasperl@chromium.org [Mon, 22 Sep 2008 08:27:54 +0000 (08:27 +0000)]
Add compilation flags for optimizing samples built in release
mode. Remove /Ob2 /Oi /Oy from VS build because they are
already enabled by /Ox.
Review URL: http://codereview.chromium.org/4012

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@352 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoMade the ARM port with simulator build and run on Windows.
sgjesse@chromium.org [Fri, 19 Sep 2008 12:35:15 +0000 (12:35 +0000)]
Made the ARM port with simulator build and run on Windows.

The reason for the "ugly" definition of SScanF is that the Windows CRT does not have a vsscanf function making it difficult to add OS::SSprintF to platform.h.
Review URL: http://codereview.chromium.org/2988

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@350 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAvoid the KMP overhead for simple indexOf() operations. Will look into evaluating...
deanm@chromium.org [Fri, 19 Sep 2008 12:30:14 +0000 (12:30 +0000)]
Avoid the KMP overhead for simple indexOf() operations.  Will look into evaluating the best cutoff between a simple search and KMP in the future.  This improves some simple operations ~1.5x.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@349 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoDon't try an indexOf() when the search string is bigger than the string. The current...
deanm@chromium.org [Fri, 19 Sep 2008 11:31:02 +0000 (11:31 +0000)]
Don't try an indexOf() when the search string is bigger than the string.  The current code will spend a bunch of time trying to match, even though we should know a match is impossible.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@348 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoTryFlatten is inlined, while Flatten is not. Make an optimization to avoid the call...
deanm@chromium.org [Fri, 19 Sep 2008 11:06:35 +0000 (11:06 +0000)]
TryFlatten is inlined, while Flatten is not.  Make an optimization to avoid the call to Flatten when we're already flat.  This gives me 5% on some simple indexOf experiments.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@347 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAdd assembler.h the the Visual Studio project file.
sgjesse@chromium.org [Fri, 19 Sep 2008 10:10:21 +0000 (10:10 +0000)]
Add assembler.h the the Visual Studio project file.
Review URL: http://codereview.chromium.org/2985

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@346 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFix lint issues.
kasperl@chromium.org [Thu, 18 Sep 2008 13:42:39 +0000 (13:42 +0000)]
Fix lint issues.
Review URL: http://codereview.chromium.org/3144

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@344 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAdded fast-case for switch statement where all lables are constant Smi's in a limited...
lrn@chromium.org [Thu, 18 Sep 2008 11:59:55 +0000 (11:59 +0000)]
Added fast-case for switch statement where all lables are constant Smi's in a limited range (IA32 only so far).
Implemented using a jump-table, for constant time lookup.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@343 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoUse null instead of undefined for deleted elements in code caches.
ager@chromium.org [Thu, 18 Sep 2008 11:18:27 +0000 (11:18 +0000)]
Use null instead of undefined for deleted elements in code caches.
Update the lookup and update code for code caches to deal with deleted
elements.

Do not clear the code cache for the builtins object.  If there was a
matching element in the code cache, we would have hit the monomorphic
prototype failure case and removed it.
Review URL: http://codereview.chromium.org/3140

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@340 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFix ARM build by adding a dummy last_statement_position() accessor
kasperl@chromium.org [Thu, 18 Sep 2008 10:40:25 +0000 (10:40 +0000)]
Fix ARM build by adding a dummy last_statement_position() accessor
to the assembler.
Review URL: http://codereview.chromium.org/2961

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@339 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRevert revision 331. The propagation of monomorphic prototype failure
ager@chromium.org [Thu, 18 Sep 2008 10:22:46 +0000 (10:22 +0000)]
Revert revision 331.  The propagation of monomorphic prototype failure
information does not work.  In certains situations, it will keep
alternating between unrelated monomorphic states instead of going
megamorphic.
Review URL: http://codereview.chromium.org/2959

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@338 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRevert changes to src\assembler-arm.h which slipped into a previous checkin breaking...
sgjesse@chromium.org [Thu, 18 Sep 2008 10:15:17 +0000 (10:15 +0000)]
Revert changes to src\assembler-arm.h which slipped into a previous checkin breaking the ARM build.
Review URL: http://codereview.chromium.org/2958

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@337 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoDefer the writing of the source position data to the relocation information
sgjesse@chromium.org [Thu, 18 Sep 2008 08:51:43 +0000 (08:51 +0000)]
Defer the writing of the source position data to the relocation information
until a possible debug break location is reached. Currently this is call sites
with calls to code objects and JS return. Source position information in the
code therefore no longer refers to the "first" instruction generated for a
given source position (which was not the case defered code anyway) but to the
first break location after that source position was passed (again defered code
always start with source position information). This doesn't make a difference
for the debugger as it will always be stopped only at debug break locations.
However, this makes the life of the peep-hole optimizer much easier as many
oportunities for posh/pop eliminations where previosly blocked by relocation
information already written to the code object.

Two types of source positions are still collected. Statement positions indicate
the position of the start of the statement leading to this code and (plain)
positions indicate other places typically call sites to help indicate current
position in backtraces. The two different types of positions are also used to
distinguish between step next and step in.

Runs all the tests (including debugger tests) as before.

Moved the checking for the FLAG_debug_info to one place.

I will do the same changes to the ARM codegenerator in a seperate changelist.

Review URL: http://codereview.chromium.org/2957

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@335 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRemove unused codereview.settings file.
kasperl@chromium.org [Thu, 18 Sep 2008 08:29:06 +0000 (08:29 +0000)]
Remove unused codereview.settings file.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@334 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFlush the runtime %DebugPrint output. Should have no performance impact,
whessev8 [Thu, 18 Sep 2008 07:31:37 +0000 (07:31 +0000)]
Flush the runtime %DebugPrint output.  Should have no performance impact,
since %DebugPrint is not often used.  Needed for some performance testing
statistics outputting.

Fix the declaration of descriptor enumerated constants to be more readable.
Review URL: http://codereview.chromium.org/3100

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@332 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoPropagate the information that a monomorphic prototype failure has
ager@chromium.org [Thu, 18 Sep 2008 06:55:14 +0000 (06:55 +0000)]
Propagate the information that a monomorphic prototype failure has
occured.  After a monomorphic prototype failure has been detected,
other monomorphic inline caches for the same type will have a code
object in the map's code cache which is not the current target.  We
recognize this case and use the code object that is in the cache
instead of going megamorphic.

Review URL: http://codereview.chromium.org/2928

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@331 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoUpdate version string and ChangeLog; we're now working on
kasperl@chromium.org [Wed, 17 Sep 2008 13:56:12 +0000 (13:56 +0000)]
Update version string and ChangeLog; we're now working on
version 0.3.3.
Review URL: http://codereview.chromium.org/3101

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@328 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoChanged the disassembler formatting
sgjesse@chromium.org [Wed, 17 Sep 2008 12:38:50 +0000 (12:38 +0000)]
Changed the disassembler formatting
Changed the formatting of the comment in the disassembler output to contain
more information on code targets.
Review URL: http://codereview.chromium.org/3099

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@327 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoMove the initial code space to be before the young space instead of after it.
erik.corry@gmail.com [Wed, 17 Sep 2008 10:59:19 +0000 (10:59 +0000)]
Move the initial code space to be before the young space instead of after it.
The old arrangement meant that the alignment requirements of young space were
likely to nullify ASLR.
Review URL: http://codereview.chromium.org/2925

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@326 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoThe OutOfMemory and OutOfMemoryNested tests set a heap limit that was too low.
erik.corry@gmail.com [Wed, 17 Sep 2008 10:58:45 +0000 (10:58 +0000)]
The OutOfMemory and OutOfMemoryNested tests set a heap limit that was too low.
A change in the way heap was allocated meant that the initial heap size was
sometimes not higher than the limit, which mean that the limit started having
and effect and prevented V8 from booting up.

Fixes http://code.google.com/p/v8/issues/detail?id=25
Review URL: http://codereview.chromium.org/2924

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@325 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoMore thorough tests of sorting integers in lexicographic order.
erik.corry@gmail.com [Wed, 17 Sep 2008 10:25:05 +0000 (10:25 +0000)]
More thorough tests of sorting integers in lexicographic order.
Review URL: http://codereview.chromium.org/2923

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@324 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAdded missing static to returned buffer.
sgjesse@chromium.org [Wed, 17 Sep 2008 09:05:36 +0000 (09:05 +0000)]
Added missing static to returned buffer.
Review URL: http://codereview.chromium.org/3096

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@323 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years ago- Rename flags.defs to flag-definitions.h.
iposva@chromium.org [Wed, 17 Sep 2008 08:47:50 +0000 (08:47 +0000)]
- Rename flags.defs to flag-definitions.h.
- Add better support for ignoring files in the presubmit tool.

Review URL: http://codereview.chromium.org/3082

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@322 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoAdded presubmit.py to the tools tab for easier access.
iposva@chromium.org [Tue, 16 Sep 2008 14:37:24 +0000 (14:37 +0000)]
Added presubmit.py to the tools tab for easier access.

TBR=mark
Review URL: http://codereview.chromium.org/2892

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@319 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoOnly remove the code object that caused the monomorphic prototype
ager@chromium.org [Tue, 16 Sep 2008 12:41:36 +0000 (12:41 +0000)]
Only remove the code object that caused the monomorphic prototype
failure instead of clearing the cache.  Clearing the cache makes us
miss subsequent monomorphic prototype failures.

Review URL: http://codereview.chromium.org/2889

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@318 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFix issue 67 by copying the receiver function one slot
kasperl@chromium.org [Tue, 16 Sep 2008 11:40:21 +0000 (11:40 +0000)]
Fix issue 67 by copying the receiver function one slot
down too. Also fix lint issue in codegen-arm.cc.

Reviewed offline by Ivan (iposva@chromium.org).
Review URL: http://codereview.chromium.org/3080

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@316 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoFix http://code.google.com/p/v8/issues/detail?id=69 :
iposva@chromium.org [Tue, 16 Sep 2008 11:23:02 +0000 (11:23 +0000)]
Fix code.google.com/p/v8/issues/detail?id=69 :
- Simplify the switch statement code generation.
- Ensure that the switch value is always popped from the stack.

Credit goes to Feng for isolating the issue and proposing a fix.
Review URL: http://codereview.chromium.org/2888

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@315 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoRemove the old builtins framework with the huge, nasty
kasperl@chromium.org [Tue, 16 Sep 2008 10:12:32 +0000 (10:12 +0000)]
Remove the old builtins framework with the huge, nasty
macro.
Review URL: http://codereview.chromium.org/3079

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@314 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoKeep the inline caches for the builtins object monomorphic when
ager@chromium.org [Tue, 16 Sep 2008 09:21:22 +0000 (09:21 +0000)]
Keep the inline caches for the builtins object monomorphic when
libraries are loaded lazily.

Review URL: http://codereview.chromium.org/2885

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@313 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoStop adapting the arguments passed to the builtin implementations
kasperl@chromium.org [Tue, 16 Sep 2008 07:24:46 +0000 (07:24 +0000)]
Stop adapting the arguments passed to the builtin implementations
of Array.prototype.push and Array.prototype.pop. Avoid going
through the arguments adaptor trampoline for call ICs that end
up calling a builtin that does not need arguments adaption.
Review URL: http://codereview.chromium.org/2884

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@312 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

15 years agoGeneralize the Function.prototype.call hooks in the
kasperl@chromium.org [Mon, 15 Sep 2008 15:02:38 +0000 (15:02 +0000)]
Generalize the Function.prototype.call hooks in the
arguments adaptor code to allow builtins to work without
argument adaptor frames. Get rid of unused JavaScript
implementation of call and apply and the associated
code generation hooks.
Review URL: http://codereview.chromium.org/2850

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@311 ce2b1a6d-e550-0410-aec6-3dcde31c8c00