AuroraMiddleware/v8
1
0
Fork 0
Code Issues 2 Pull Requests Releases Wiki Activity
506 Commits 1 Branch 0 Tags 839 MiB
b14a2e4cb7
Commit Graph

59 Commits

Author SHA1 Message Date
feng@chromium.org
47ac6ff1e0 fix lint, approved offline by iposva 
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@721 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-11-07 23:00:09 +00:00
feng@chromium.org
fc8a0e3fa5 Fix OOM handler code. Old code didn't pop up the state slot. 
Review URL: http://codereview.chromium.org/10223

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@720 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-11-07 22:56:41 +00:00
kmillikin@chromium.org
6edea51f89 Reporting -1 as the size of an ILLEGAL reference which actually has 
size 0 was too cute.
Review URL: http://codereview.chromium.org/9689

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@709 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-11-07 08:58:23 +00:00
kmillikin@chromium.org
bdb896cc40 Emit pushes and pops through the virtual frame on ARM. Merging of 
frames is not yet handled.  The ARM code generator should be back in
line with the IA32 one.
Review URL: http://codereview.chromium.org/9182

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@706 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-11-07 08:21:07 +00:00
kmillikin@chromium.org
47e55a77a7 Initial port of VirtualFrame to the ARM architecture. To reduce the 
size of the change, the initial implementation provides read access to
the frame (eg, locals, parameters, and expression stack elements) but
no mutation.
Review URL: http://codereview.chromium.org/7302

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@694 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-11-05 08:44:54 +00:00
ager@chromium.org
439f3e6708 Do not put failures in handles. 
Assert in the Handle constructor that the object is not a failure.

I have run our own tests in debug mode and the WebKit layout tests in
debug mode and there are no regressions.
Review URL: http://codereview.chromium.org/9114

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@691 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-11-04 16:23:56 +00:00
ager@chromium.org
240d818bb8 Smi ranges are not symmetrical. -kMinSmiValue is not a Smi. 
Review URL: http://codereview.chromium.org/9041

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@680 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-11-03 15:21:46 +00:00
lrn@chromium.org
4e3bbd8247 If a HeapNumber is the incoming value, it must be converted to Smi before 
checking. This is not done in a fast way.


git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@679 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-11-03 13:33:13 +00:00
kmillikin@chromium.org
ec3c4ced39 Reverted 671. 
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@673 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-31 13:18:52 +00:00
kmillikin@chromium.org
67ae89e417 Simplify the way we materialize boolean values that are not yet pushed 
on the stack frame.
Review URL: http://codereview.chromium.org/8764

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@671 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-31 11:55:06 +00:00
kasperl@chromium.org
a2be3b6f84 Make sure that allocations through CALL_HEAP_FUNCTION 
and runtime calls from JavaScript will always succeed
eventually if we have enough memory.
Review URL: http://codereview.chromium.org/8700

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@646 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-30 09:15:58 +00:00
kasperl@chromium.org
a26b3535de Allow three runtime call attempts before throwing an out of 
memory exception. Still needs work in Runtime_PerformGC to
make sure we'll allow future allocations.
Review URL: http://codereview.chromium.org/8873

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@638 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-29 12:16:34 +00:00
kasperl@chromium.org
2dcb4d2c59 Fix issue 120 by patching the on-stack receiver in the 
IC stubs for calls just before invoking the target
function instead of doing it before resolving the
function.
Review URL: http://codereview.chromium.org/8192

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@607 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-27 14:12:02 +00:00
kasperl@chromium.org
4e4fc50842 Get ready for fixing issue 120: Pin point the places 
where the receiver needs to be patched with the proxy
and get ready of unnecessary function patching on ARM.


git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@591 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-24 12:58:17 +00:00
kmillikin@chromium.org
681ae19bd4 Change a few runtime functions that took and returned a dummy argument 
to instead take no arguments and return the undefined value.


git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@584 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-24 10:13:22 +00:00
erik.corry@gmail.com
dbc6dd66e4 Fix some style issues. 
Review URL: http://codereview.chromium.org/8055

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@563 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-23 07:28:21 +00:00
kmillikin@chromium.org
65821f321c Rename member variables and functions and clarify comments to try to 
make the operation of LabelShadows more understandable.
Review URL: http://codereview.chromium.org/7868

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@556 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-22 12:45:33 +00:00
bak@chromium.org
bf948c8313 - Optimized CopyFixedArray and CopyJSObject. 
- Refactored block copying.

Review URL: http://codereview.chromium.org/7863

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@548 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-22 08:21:18 +00:00
feng@chromium.org
202dedbdb3 No more failures than before. It is ready to be reviewed. 
Review URL: http://codereview.chromium.org/7420

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@544 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-21 20:11:50 +00:00
iposva@chromium.org
89c762edf4 Simplify CodeGenerator hierarchy by not using a base class. 
There is nothing virtual about a CodeGenerator since we
either generate code for one platform or for the other.

Review URL: http://codereview.chromium.org/6334

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@480 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-10 00:00:52 +00:00
kasperl@chromium.org
77643dbef6 Improve the generated code for the instanceof operator, 
and extended the instanceof test case.
Review URL: http://codereview.chromium.org/6341

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@470 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-08 13:33:16 +00:00
kmillikin@chromium.org
5d7f354384 Moved the function GetValue from the code generator to the Reference 
helper class.

As a consequence, also removed the optional Reference from the code
generator state and got rid of some state-dependent behavior in the
code generator.
Review URL: http://codereview.chromium.org/6538

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@466 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-08 07:24:06 +00:00
kmillikin@chromium.org
6afa5c64b7 Move code generation for storing to a reference out of the AST nodes, and 
onto the platform-specific Reference class defined in codegen-*.cc.  This
removes all of the static code generator functions.
Review URL: http://codereview.chromium.org/6527

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@460 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-07 11:19:44 +00:00
kmillikin@chromium.org
5c80e6a83a Document (and assert) some of the safe-but-brittle implicit assumptions 
about references in the code generators.
Review URL: http://codereview.chromium.org/6301

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@453 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-07 08:47:15 +00:00
kmillikin@chromium.org
ca0924349d DeclareContextSlot took an extra completely random argument from the 
stack which was returned and always pushed back.

Removed it.
Review URL: http://codereview.chromium.org/6802

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@448 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-06 16:57:38 +00:00
kasperl@chromium.org
892b0cac3c Refactor the arguments access code to make it easier to read. 
Review URL: http://codereview.chromium.org/6491

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@442 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-06 11:04:03 +00:00
kasperl@chromium.org
5458eac183 Improve performance of arguments object allocation by taking 
care of arguments adaptor frames in the generated code.
Review URL: http://codereview.chromium.org/6262

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@434 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-06 06:08:15 +00:00
lrn@chromium.org
17dc503149 Added fast-case switch to ARM. 
Refactored to reuse code between ia32 and ARM.
Added tests for switch.


git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@424 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-10-03 12:04:59 +00:00
kmillikin@chromium.org
a30916cf27 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
 2008-10-02 12:44:47 +00:00
kmillikin@chromium.org
eca6c6ae65 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
 2008-10-02 11:54:32 +00:00
kmillikin@chromium.org
36e050fa6b 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
 2008-10-02 11:14:50 +00:00
kasperl@chromium.org
68103cee8e 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
 2008-10-01 10:57:37 +00:00
kasperl@chromium.org
016fcd4aff 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
 2008-10-01 07:43:00 +00:00
kasperl@chromium.org
e4a82f2440 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
 2008-09-23 12:21:54 +00:00
kasperl@chromium.org
9d30045d06 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
 2008-09-23 08:19:26 +00:00
sgjesse@chromium.org
8391b19b07 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
 2008-09-22 13:57:03 +00:00
kasperl@chromium.org
6b4f5aba96 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
 2008-09-16 11:40:21 +00:00
iposva@chromium.org
88192fc01c Fix http://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
 2008-09-16 11:23:02 +00:00
kasperl@chromium.org
64923a6a83 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
 2008-09-15 15:02:38 +00:00
deanm@chromium.org
77196c5771 This is a new static flag system, designed to have all flags in a central place, and compiled into the binary without requiring static constructors for registration. All flags are moved out of the specific modules and into flags.defs, with different sections for debug, release, etc. The flag variables are always defined. For example, a debug flag in release mode still exists, but is read only and set to the default value. 
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@296 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-09-12 10:19:33 +00:00
kmillikin@chromium.org
b5d8866ced Flexo is tired, but this should perk him up. Single-argument constructors 
should be marked explicit.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@295 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-09-12 08:52:22 +00:00
kmillikin@chromium.org
0c4fd55c57 Change the code generator state constructor to implicitly push the state on 
stack, rather than explicitly saving and restoring it.
Review URL: http://codereview.chromium.org/3002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@294 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-09-12 08:24:57 +00:00
iposva@chromium.org
c5ee961882 Adapt to new calling convention on ARM: 
- Simplified frame entry and frame exit code.
- Added ArgumentsAdaptorTrampoline and check for matching argument counts in the InvokePrologue.
- Removed definition and uses of USE_OLD_CALLING_CONVENTIONS.
- Changed MacroAssembler::InvokeBuiltin to match ia32 version.
- Start introducing convenience instructions in the ARM assembler as needed. These instructions take all Register parameters to avoid extra typing of "Operand(reg)".


To keep the architectures in sync these changes have been made to the ia32 files:
- Changed MacroAssembler::EnterFrame(StackFrame::Type type) to MacroAssembler::EnterInternalFrame().


These parts are still missing:
- unimplemented: Builtins::Generate_FunctionApply - large limit
- unimplemented: Builtins::Generate_ArgumentsAdaptorTrampoline - non-function call
- The files have not been lint'd yet.


Review URL: http://codereview.chromium.org/1930

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@289 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-09-12 03:29:06 +00:00
kmillikin@chromium.org
4afbecf50e Fix performace regression due to missed peephole optimization 
opportunity.
Review URL: http://codereview.chromium.org/2002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@286 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-09-11 15:16:08 +00:00
kmillikin@chromium.org
c2022ec76e Remove some of the state-dependent behavior from the code generator. 
Simplify the code generator by eliminating the access types STORE and
INIT_CONST and delegating code generation for stores to the appropriate AST
nodes.
Review URL: http://codereview.chromium.org/1889

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@265 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-09-11 08:44:32 +00:00
christian.plesner.hansen@gmail.com
9bed566bdb Changed copyright header from google inc. to v8 project authors. 
Added presubmit step to check copyright.



git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@242 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-09-09 20:08:45 +00:00
ager@chromium.org
e0b50dde0e Avoid string conversion when comparing Smis during sorting. 
Avoid runtime calls for trivial object equality checks.

Minor style cleanups.



git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@185 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-09-08 06:17:38 +00:00
ager@chromium.org
d295ddd922 Fix the 'in' operator so it works correctly for negative indices. 
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@117 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-09-03 07:34:21 +00:00
feng.qian.v8
194baea8ca Move JS_VALUE_TYPE ahead of JS_OBJECT_TYPE to save one comparison in 
KeyedLoadIC::GenerateGeneric (on IA32), and the same on ARM.
Re-assignmed FIRST_JS_OBJECT_TYPE to JS_VALUE_TYPE.

Also changed JS_OBJECT_TYPE to FIRST_JS_OBJECT_TYPE in several places where
FIRST_JS_OBJECT_TYPE is intended.



git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@113 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-09-02 23:52:50 +00:00
feng.qian.v8
6287c675d8 Change __ mov(eax, TOS) to 
__ pop(eax);
...
__ push(eax);
to enable push/pop eliminations.

__ push(eax) must happen before RecordWrite because RecordWrite may destroy 
eax value. To be safe, also moved __ push(r0) on ARM to above RecordWrite.

This only affects the case where a context variable is used in a inner scope.

Create a tests for it. It fails if __ push(eax) is after RecordWrite.



git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@111 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2008-09-02 21:47:45 +00:00