AuroraMiddleware/v8
1
0
Fork 0
Code Issues 2 Pull Requests Releases Wiki Activity
7,514 Commits 1 Branch 0 Tags 839 MiB
4e4a901d96
Commit Graph

7514 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
kmillikin@chromium.org
4e4a901d96 Remove the forward-bailout stack from the non-optimizing compiler. 
This was pretty heavyweight.  It was kept in just for a few corner cases
that assumed it was there.  We can work around them by making sure that the
expression in a reified test context is always really the expression that
was visited in that context; and by inspecting the context manually and
consing up a pair of extra AST IDs for the unusual case of unary not in a
value AST context.

R=fschneider@chromium.org
BUG=
TEST=

Review URL: http://codereview.chromium.org/8386037

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9863 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-11-02 16:53:32 +00:00
ulan@chromium.org
307b589e61 Fix error introduced in r9431, which leads to redundant lookup when setting property with handler. 
BUG=
TEST=

Review URL: http://codereview.chromium.org/8386030

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9862 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-11-02 13:04:59 +00:00
danno@chromium.org
5c40d7a55d Enable VFP instructions for Android. 
R=jkummerow@chromium.org
BUG=none
TEST=none

Review URL: http://codereview.chromium.org/8437047

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9861 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-11-02 10:57:03 +00:00
fschneider@chromium.org
f3f16dc17a Make non-templatized versions of LIR printing functions. 
This avoid duplicating the code for each template instance.

Also remove dead code from different places in our code base.

Removed some verification code from release builds.
Review URL: http://codereview.chromium.org/8387070

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9860 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-11-02 08:32:40 +00:00
fschneider@chromium.org
acf8fd28bd MIPS: Merge IR classes for different bitwise operations AND, OR and XOR into one class. 
Port r9846 (4cd055a).

Original commit message:

Since we already have only one LIR class, it does not make much sense to separate them at the HIR level.

BUG=
TEST=
Review URL: http://codereview.chromium.org/8432010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9859 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-11-01 16:13:05 +00:00
fschneider@chromium.org
aa34b432de Remove one-line helper used in two places. 
Review URL: http://codereview.chromium.org/8387067

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9858 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-11-01 12:25:18 +00:00
whesse@chromium.org
7a25c43a07 Remove stack height tracking from ia32 non-optimizing code generator. 
Review URL: http://codereview.chromium.org/8340023

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9857 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-11-01 12:20:46 +00:00
fschneider@chromium.org
d418a11fdd Use placement-new for zone-allocation in the Lithium classes. 
This avoid calling Isolate::Current on each zone-allocation of
a Lithium object.
Review URL: http://codereview.chromium.org/8429001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9856 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-11-01 11:49:31 +00:00
fschneider@chromium.org
97c3750fd8 Fix bug introduced by my refactoring of bitwise HIR instructions. 
DataEquals was not overridden properly.
Review URL: http://codereview.chromium.org/8387066

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9855 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-11-01 11:35:54 +00:00
lrn@chromium.org
65b9ab93af Merged Scanner and JavaScriptScanner. 
JavaScriptScanner had become the only concrete subclass of Scanner, so there
was no longer a need for the distinction.

Also fixed up comments.

Review URL: http://codereview.chromium.org/8384003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9854 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-11-01 07:47:15 +00:00
vegorov@chromium.org
858320725b Split incremental marking write barrier into fast and slow paths. 
Force inlining of the fast path.

Force inlining LiteralBuffer::AddChar and Scanner::AddLiteralChar.

R=erik.corry@gmail.com

Review URL: http://codereview.chromium.org/8431010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9853 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 20:59:28 +00:00
ulan@chromium.org
da49e4d83d Convert array index to integer when computing pause distribution for spinning-balls. 
BUG=
TEST=

Review URL: http://codereview.chromium.org/8430006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9852 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 17:26:59 +00:00
ulan@chromium.org
e67096aded Allow to set a time limit for spinning-balls and output the pause distribution at the end. 
BUG=
TEST=

Review URL: http://codereview.chromium.org/8428002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9851 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 17:05:37 +00:00
keuchel@chromium.org
6b533c315a Fix unwanted implicit conversion from bool to int. 
We have the following signatures:
VariableProxy* Scope::NewUnresolved(Handle<String> name,
                                    int position = RelocInfo::kNoPosition);

bool Parser::inside_with() const;

Review URL: http://codereview.chromium.org/8427001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9850 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 15:04:27 +00:00
danno@chromium.org
75585d9643 Create stub and runtime function for ia32 full-codegen array literal element initialization. 
BUG=none
TEST=none

Review URL: http://codereview.chromium.org/8359014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9849 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 14:42:06 +00:00
jkummerow@chromium.org
669bfc0363 Optimize JS date implementation for smi-only arrays 
Review URL: http://codereview.chromium.org/8409010

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9848 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 14:39:21 +00:00
danno@chromium.org
0766a138a6 Add and use ElementsKind side effect 
Also partition side effects into observable and not observable, with only observable requiring Simulates and non-observable changes able to participate in GVN and code hoisting.

BUG=none
TEST=none

Review URL: http://codereview.chromium.org/8380017

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9847 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 14:15:10 +00:00
fschneider@chromium.org
0df9569fe5 Merge IR classes for different bitwise operations AND, OR and XOR into one class. 
Since we already have only one LIR class, it does not make much sense to separate
them at the HIR level.
Review URL: http://codereview.chromium.org/8426005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9846 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 13:06:26 +00:00
kmillikin@chromium.org
d00b82ff91 MIPS: updated atomic operations. 
Specifically:
-fixed a bug in CompareAndSwap that caused randomly occuring timeouts on MIPS boards.
-added gcc inline asm control push/pop instructions to correctly save/restore the reorder setting instead of simply assuming it should be enabled.
-reordered/optimized some instructions to utilize MIPS-specific pipelining features (branch delay slot).
-fixed improper usage of write barriers

BUG=
TEST=

Review URL: http://codereview.chromium.org/8413073
Patch from Gergely Kis <gergely@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9845 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 12:47:02 +00:00
kmillikin@chromium.org
ee3197d309 MIPS: Handlify CompileConstructStub and the remaining CallStubCompiler functions. 
Port r9837 (38061af).
Port r9834 (95ac04).

Original commit message (r9834):
Also, handlify functions for loading with interceptors and callbacks.
Remove some unneeded code.  Rename Foreign::address() because it
confusingly shadows HeapObject::address() which does something quite
different.

BUG=
TEST=

Review URL: http://codereview.chromium.org/8400087
Patch from Gergely Kis <gergely@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9844 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 12:36:11 +00:00
kmillikin@chromium.org
3ba063197d MIPS: Fixed wrong register usage in StringDictionaryLookupStub::GenerateNegativeLookup. 
This fixes a lot of test failures triggered by r9834, for example mjsunit/const-redecl.js.

BUG=
TEST=

Review URL: http://codereview.chromium.org/8417050
Patch from Gergely Kis <gergely@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9843 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 12:34:13 +00:00
mstarzinger@chromium.org
358d7c2078 Adapt date test to be timezone independant. 
R=yangguo@chromium.org
BUG=v8:1792
TEST=mjsunit/date

Review URL: http://codereview.chromium.org/8423004

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9842 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 11:15:23 +00:00
fschneider@chromium.org
a2240fc52b Eliminate some virtual function from AST classes. 
1. Remove unused dead functions.

2. Replace the virtual As-* type cast functions with non-virtual version
that uses node_type().

Result is around 13K reduction in binary size.
Review URL: http://codereview.chromium.org/8335006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9841 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 11:11:26 +00:00
lrn@chromium.org
30465596e6 Make eval consider anything on the form eval(args...) a potential direct cal 
Previously we omitted all cases where the global eval property was shadowed,
even if by a variable holding the same value. ES5 requires us to treat these
as direct calls.

We still throw if calling indirect eval with a detached global object.

BUG=v8:994
TEST=mjsunit/eval.js

Review URL: http://codereview.chromium.org/8343054

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9838 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-31 09:38:52 +00:00
ulan@chromium.org
fa8a71df9a Handlify CompileConstructStub. Based on 8391045. 
R=kmillikin@chromium.org
BUG=
TEST=

Review URL: http://codereview.chromium.org/8399032

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9837 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-28 14:08:43 +00:00
mstarzinger@chromium.org
00bafbbc9d Mark Test262 test cases for known issue 1772. 
R=svenpanne@chromium.org
BUG=v8:1772
TEST=test262

Review URL: http://codereview.chromium.org/8341111

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9836 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-28 13:03:55 +00:00
vegorov@chromium.org
c933e28935 Make C++ to JS transition faster by avoiding JavaScriptFrameIterator in SaveContext. 
R=kmillikin@chromium.org
BUG=v8:1730

Review URL: http://codereview.chromium.org/8403037

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9835 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-28 12:49:09 +00:00
kmillikin@chromium.org
0df252b38b Handlify the remaining CallStubCompiler functions. 
Also, handlify functions for loading with interceptors and callbacks.
Remove some unneeded code.  Rename Foreign::address() because it
confusingly shadows HeapObject::address() which does something quite
different.

R=vegorov@chromium.org,ulan@chromium.org
BUG=
TEST=

Review URL: http://codereview.chromium.org/8391045

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9834 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-28 12:37:29 +00:00
danno@chromium.org
6d7d6d4e4e Force transition to FAST_ELEMENTS on out-of-bounds KeyedLoads. 
Proactively ensure that that objects don't get FAST_DOUBLE_ELEMENTS to reduce the number of double boxing operations when generated code calls the runtime frequently to satisfy KeyedLoad requests.

Review URL: http://codereview.chromium.org/8416014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9833 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-28 10:35:38 +00:00
keuchel@chromium.org
a9a97d010e Remove kInvalidStrictFlag and make gcc-4.5 happy again. 
Review URL: http://codereview.chromium.org/8403036

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9832 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-28 09:10:29 +00:00
mstarzinger@chromium.org
6d950a748f Fix assertSame for unit testing harness. 
Using isNaN() here is bogus because it performs an implicit toNumber()
conversion, hence something like assertSame(undefined, {}) would not
throw an exception. These are not the NaNs you are looking for.

R=rossberg@chromium.org
TEST=mjsunit

Review URL: http://codereview.chromium.org/8400056

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9831 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-28 09:09:51 +00:00
kmillikin@chromium.org
9fd8930a70 Fix presubmit error. 
git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9830 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-28 08:51:08 +00:00
mstarzinger@chromium.org
cd9bc6c3a6 Fix error handling in Date.prototype.toISOString. 
This fixes Date.prototyoe.toISOString to throw a RangeError exception
for invalid time values. It also includes a fix to removes the arbitrary
(and completely bogus) range limit on the date value during construction
of a Date object. Note that we still have bogus range limits on the year
and month values.

R=lrn@chromium.org
BUG=v8:1792
TEST=mjsunit/date,test262/15.9.5.43-0-*

Review URL: http://codereview.chromium.org/8392036

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9829 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-28 08:45:04 +00:00
kmillikin@chromium.org
f077a41b12 MIPS: crankshaft implementation 
BUG=
TEST=

Review URL: http://codereview.chromium.org/7934002
Patch from Paul Lind <plind44@gmail.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9828 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-28 08:14:46 +00:00
yangguo@chromium.org
b5b3345799 MIPS: Fixing performance regression in issue 1787. 
Port r9812 (c1644a9d)

Note: MIPS already had the fix for this regression.
This commit only adds the extra assertions.

BUG=
TEST=

Review URL: http://codereview.chromium.org/8343030
Patch from Gergely Kis <gergely@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9827 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-28 08:03:13 +00:00
erik.corry@gmail.com
31834e0f9d Prepare push to trunk. Now working on 3.7.4. 
Review URL: http://codereview.chromium.org/8417020

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9824 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-28 07:38:27 +00:00
fschneider@chromium.org
8b1494b819 Temporarily disable target recording CallFunctionStub until issue 1789 is fixed. 
Review URL: http://codereview.chromium.org/8340029

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9823 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-27 15:46:25 +00:00
yangguo@chromium.org
8624fb85d2 Tiny improvements on fast element conversion loops in ia32 and x64. 
Review URL: http://codereview.chromium.org/8403035

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9820 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-27 14:19:22 +00:00
erik.corry@gmail.com
1dd598387e Prepare push to trunk. Now working on 3.7.3 
Review URL: http://codereview.chromium.org/8407001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9819 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-27 14:03:32 +00:00
keuchel@chromium.org
6e767e3f2d Use StrictModeFlag in preparser and preparse data. 
Review URL: http://codereview.chromium.org/8396040

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9818 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-27 13:08:51 +00:00
fschneider@chromium.org
a5b40e27b8 Revert r9805. 
It did not fix the original problem, but instead introduced new ones.

R=vegorov@chromium.org
Review URL: http://codereview.chromium.org/8404037

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9817 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-27 12:12:53 +00:00
yangguo@chromium.org
b56466f4da Fixing dead code in empty array init. 
TEST=set JSArray::kPreallocatedArrayElements to larger than 4.

Review URL: http://codereview.chromium.org/8381014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9816 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-27 11:11:59 +00:00
kmillikin@chromium.org
9cc75769bc MIPS: Handlify CompileLoad and CompileStore 
Port the following 3 stub-cache commits:
r9788   (67147b7)  Handlify CompileLoadGlobal, CompileLoadElement, CompileLoadPolymorphic.
r9791   (b19c694)  Handlify simple functions of [keyed] store stub compiler.
r9803   (e0c3c89)  Handlify CompileStoreCallback, CompileStoreInterceptor.

BUG=
TEST=

Review URL: http://codereview.chromium.org/8404016
Patch from Paul Lind <plind44@gmail.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9815 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-27 09:21:15 +00:00
yangguo@chromium.org
2923136f95 Fixing performance regression in issue 1787. 
BUG=v8:1787

Review URL: http://codereview.chromium.org/8390050

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9812 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-27 07:59:12 +00:00
mstarzinger@chromium.org
973383f4e9 Mark Test262 test cases for known issue 1790. 
R=lrn@chromium.org
BUG=v8:1790
TEST=test262/15.4.4.22-9-9

Review URL: http://codereview.chromium.org/8396042

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9811 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-27 07:50:40 +00:00
ricow@chromium.org
d9597f0086 Skip live edit debug tests, these are flaky because in the case of osr we will get wrong frame heights. 
Review URL: http://codereview.chromium.org/8401029

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9808 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-27 07:10:25 +00:00
erik.corry@gmail.com
678200bc11 Prepare push to trunk. Now working on 3.7.2 
Review URL: http://codereview.chromium.org/8390049

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9807 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-27 05:27:38 +00:00
yangguo@chromium.org
fdb13b2e56 MIPS: port Implement for-in loop for proxies. 
Port r9760 (a1c4dc80)

BUG=
TEST=

Review URL: http://codereview.chromium.org/8396029
Patch from Paul Lind <plind44@gmail.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9806 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-26 15:02:00 +00:00
fschneider@chromium.org
4ded3ccf7c Fix bug in lazy deoptimization after HInvokeFunction. 
The full code generator expects the function expression on top
of the stack until after the all. We have to emit the call + the
corresponding simulate before dropping it in the hydrogen code.

BUG=v8:1789
Review URL: http://codereview.chromium.org/8391042

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9805 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-26 14:05:18 +00:00
lrn@chromium.org
a9edfefa58 Remove special-casing of calls to RegExp test and exec methods with no argument. 
Matches new JSC behavior. Fix issue 75740.

BUG=75740
TEST=mjsunit/regexp-static

Review URL: http://codereview.chromium.org/6677020

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@9804 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
 2011-10-26 12:51:07 +00:00