platform/upstream/v8.git
9 years agoMake snapshot.h usable without objects-inl.h header.
mstarzinger [Thu, 20 Aug 2015 13:05:58 +0000 (06:05 -0700)]
Make snapshot.h usable without objects-inl.h header.

This CL us a pure refactoring that makes an empty compilation unit
including just "snapshot.h" but not "objects-inl.h" compile without
warnings or errors. This is needed to further reduce the header
dependency tangle.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/1287113010

Cr-Commit-Position: refs/heads/master@{#30270}

9 years ago- remove the Backing-Store speficic code from builtins.cc and put it in elements.cc.
cbruni [Thu, 20 Aug 2015 12:54:40 +0000 (05:54 -0700)]
- remove the Backing-Store speficic code from builtins.cc and put it in elements.cc.
- adding tests to improve coverage of the splice method

BUG=

Review URL: https://codereview.chromium.org/1293683005

Cr-Commit-Position: refs/heads/master@{#30269}

9 years agoSeparate UnicodeCache out into an own file.
mstarzinger [Thu, 20 Aug 2015 11:46:54 +0000 (04:46 -0700)]
Separate UnicodeCache out into an own file.

This make inclusion of unicode-inl.h in object.h absolete. Now most
compilation units don't require that header. It also breaks a cycle
within declarations of the scanner.h header.

This tries to remove includes of "-inl.h" headers from normal ".h"
headers, thereby reducing the chance of any cyclic dependencies and
decreasing the average size of our compilation units.

Note that this change still leaves 3 violations of that rule in the
code, checked with the "tools/check-inline-includes.sh" tool.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/1287893006

Cr-Commit-Position: refs/heads/master@{#30268}

9 years ago[heap] Remove unflattened_string_length
mlippautz [Thu, 20 Aug 2015 10:00:12 +0000 (03:00 -0700)]
[heap] Remove unflattened_string_length

BUG=

Review URL: https://codereview.chromium.org/1303443005

Cr-Commit-Position: refs/heads/master@{#30267}

9 years agoAdd deserialized scripts to script list.
yangguo [Thu, 20 Aug 2015 09:26:21 +0000 (02:26 -0700)]
Add deserialized scripts to script list.

TBR=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1296793003

Cr-Commit-Position: refs/heads/master@{#30266}

9 years agoTurn v8.h into a normal header.
mstarzinger [Thu, 20 Aug 2015 09:24:24 +0000 (02:24 -0700)]
Turn v8.h into a normal header.

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/1293593005

Cr-Commit-Position: refs/heads/master@{#30265}

9 years agoKeep track of script objects in a weak fixed array.
yangguo [Thu, 20 Aug 2015 08:04:41 +0000 (01:04 -0700)]
Keep track of script objects in a weak fixed array.

We need this for the debugger and for future changes that need to
find all shared function infos (through scripts).

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1297273005

Cr-Commit-Position: refs/heads/master@{#30264}

9 years agoRemove grab-bag includes of v8.h from everywhere.
mstarzinger [Thu, 20 Aug 2015 07:44:00 +0000 (00:44 -0700)]
Remove grab-bag includes of v8.h from everywhere.

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/1285183010

Cr-Commit-Position: refs/heads/master@{#30263}

9 years agoMake FlushICache NOP for Nvidia Denver 1.0 only
sbonda [Thu, 20 Aug 2015 04:44:36 +0000 (21:44 -0700)]
Make FlushICache NOP for Nvidia Denver 1.0 only

FlushICache should be NOP for Denver with part numbers 0x0, 0x1 and 0x2 only.
Instruction cache needs to flushed for future versions of denver.

Review URL: https://codereview.chromium.org/1287173004

Cr-Commit-Position: refs/heads/master@{#30262}

9 years agoUpdate V8 DEPS.
v8-autoroll [Thu, 20 Aug 2015 03:22:34 +0000 (20:22 -0700)]
Update V8 DEPS.

Rolling v8/third_party/android_tools to 4238a28593b7e6178c95431f91ca8c24e45fa7eb

Rolling v8/tools/clang to 1c7f9147c834d78b36787f31ecfc5c47f3c98da8

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/1293873004

Cr-Commit-Position: refs/heads/master@{#30261}

9 years agoAdd experimental, non-snapshotted V8 extras
domenic [Thu, 20 Aug 2015 00:01:07 +0000 (17:01 -0700)]
Add experimental, non-snapshotted V8 extras

Embedders would use these for features which must be able to be turned
off at runtime, despite being compiled into V8. They can be turned on
and off by the embedder using the --experimental_extras flag, e.g. via
v8::SetFlagsFromString.

R=yangguo@chromium.org, mlippautz@chromium.org, hpayer@chromium.org
BUG=chromium:507137
LOG=Y

Review URL: https://codereview.chromium.org/1284413002

Cr-Commit-Position: refs/heads/master@{#30260}

9 years ago[api] Relax CHECK for ArrayBuffer API abuse
adamk [Wed, 19 Aug 2015 21:53:01 +0000 (14:53 -0700)]
[api] Relax CHECK for ArrayBuffer API abuse

Zero-length ArrayBuffers are allowed to have NULL backing stores.

BUG=522496
LOG=n

Review URL: https://codereview.chromium.org/1302803003

Cr-Commit-Position: refs/heads/master@{#30259}

9 years agoRe-enable regress-crbug-501711 and regress-4279 for --isolates tests
Ben Smith [Wed, 19 Aug 2015 20:59:46 +0000 (13:59 -0700)]
Re-enable regress-crbug-501711 and regress-4279 for --isolates tests

BUG=v8:4378
TBR=yangguo@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/1305583002 .

Cr-Commit-Position: refs/heads/master@{#30258}

9 years agoRemove regress-crbug-518748. It's too hard to make this non-flaky
Ben Smith [Wed, 19 Aug 2015 18:58:02 +0000 (11:58 -0700)]
Remove regress-crbug-518748. It's too hard to make this non-flaky

BUG=chromium:518748
R=mstarzinger@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/1297403002 .

Cr-Commit-Position: refs/heads/master@{#30257}

9 years agoDisable regress-crbug-518748. It is failing/flaking on many bots
Ben Smith [Wed, 19 Aug 2015 17:48:37 +0000 (10:48 -0700)]
Disable regress-crbug-518748. It is failing/flaking on many bots

BUG=chromium:518748
TBR=mstarzinger@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/1306443002 .

Cr-Commit-Position: refs/heads/master@{#30256}

9 years ago[heap] Fix compilation of LargeObjectSpace on Windows.
Michael Starzinger [Wed, 19 Aug 2015 17:39:09 +0000 (19:39 +0200)]
[heap] Fix compilation of LargeObjectSpace on Windows.

R=thakis@chromium.org, titzer@chromium.org

Review URL: https://codereview.chromium.org/1288723005 .

Cr-Commit-Position: refs/heads/master@{#30255}

9 years agoRename ParserInfo::function() and CompilationInfo::function() to literal().
titzer [Wed, 19 Aug 2015 16:51:37 +0000 (09:51 -0700)]
Rename ParserInfo::function() and CompilationInfo::function() to literal().

R=rossberg@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1301583005

Cr-Commit-Position: refs/heads/master@{#30254}

9 years ago[heap] Remove obsolete Heap::sweep_generation field.
mstarzinger [Wed, 19 Aug 2015 16:09:40 +0000 (09:09 -0700)]
[heap] Remove obsolete Heap::sweep_generation field.

R=mlippautz@chromium.org

Review URL: https://codereview.chromium.org/1289963006

Cr-Commit-Position: refs/heads/master@{#30253}

9 years ago[heap] Fix comment in tracer.
mlippautz [Wed, 19 Aug 2015 15:38:44 +0000 (08:38 -0700)]
[heap] Fix comment in tracer.

BUG=

Review URL: https://codereview.chromium.org/1292583004

Cr-Commit-Position: refs/heads/master@{#30252}

9 years agoDo not compact weak fixed array when re-allocating new backing store.
yangguo [Wed, 19 Aug 2015 14:29:35 +0000 (07:29 -0700)]
Do not compact weak fixed array when re-allocating new backing store.

JSObject::UnregisterPrototypeUser relies on the index of the weak
fixed array entry not to change.

See commit 1e65e2018.

There is no bug currently, because we would only re-allocate if there is no empty slot left. In which case there is no actual compaction going on during re-allocation.

R=jkummerow@chromium.org, mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1294883004

Cr-Commit-Position: refs/heads/master@{#30251}

9 years agoCleanup: Remove unncessary leave_frame parameter from stub cache.
mvstanton [Wed, 19 Aug 2015 13:51:19 +0000 (06:51 -0700)]
Cleanup: Remove unncessary leave_frame parameter from stub cache.

BUG=

Review URL: https://codereview.chromium.org/1299213002

Cr-Commit-Position: refs/heads/master@{#30250}

9 years agoFix variable decl register collision on ARM.
jarin [Wed, 19 Aug 2015 12:50:03 +0000 (05:50 -0700)]
Fix variable decl register collision on ARM.

BUG=chromium:455207
LOG=n
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/1040703003

Cr-Commit-Position: refs/heads/master@{#30249}

9 years agoNative context: install JS builtins via container object.
yangguo [Wed, 19 Aug 2015 12:48:57 +0000 (05:48 -0700)]
Native context: install JS builtins via container object.

Instead of installing them on the JS builtins object and later
grab them from there.

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1296163003

Cr-Commit-Position: refs/heads/master@{#30248}

9 years agoRealize IWYU pattern for fdlibm.cc file.
mstarzinger [Wed, 19 Aug 2015 12:47:49 +0000 (05:47 -0700)]
Realize IWYU pattern for fdlibm.cc file.

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/1292753004

Cr-Commit-Position: refs/heads/master@{#30247}

9 years agoNative context: do not hold onto helper functions on the utils object.
yangguo [Wed, 19 Aug 2015 11:30:21 +0000 (04:30 -0700)]
Native context: do not hold onto helper functions on the utils object.

This is to avoid holding onto the function context of prologue.js.

R=cbruni@chromium.org

Review URL: https://codereview.chromium.org/1285133009

Cr-Commit-Position: refs/heads/master@{#30246}

9 years agoIgnore test failure for mjsunit/for-in-opt in gc stress.
yangguo [Wed, 19 Aug 2015 10:03:32 +0000 (03:03 -0700)]
Ignore test failure for mjsunit/for-in-opt in gc stress.

TBR=hablich@chromium.org
BUG=v8:4381

Review URL: https://codereview.chromium.org/1295513004

Cr-Commit-Position: refs/heads/master@{#30245}

9 years agoNative context: debug.js does not load from js builtins object anymore.
yangguo [Wed, 19 Aug 2015 08:34:21 +0000 (01:34 -0700)]
Native context: debug.js does not load from js builtins object anymore.

This mainly changes how we share ToBoolean, ToNumber and ToString between
native scripts. Instead of putting them on the js builtins object, we now
explicitly export and import those functions.

I also had to change the import/export mechanism slightly. Previously,
exports and imports are hooked up after all native scripts have been
executed. This means that imported functions cannot be called at the time
the native script is executed.

However, since ToBoolean (and also e.g. ObjectDefineProperties) is called
in v8natives.js, I changed the mechanism so that exports from previous
native scripts (runtime.js for ToBoolean) is imported immediately and can
be called.

R=cbruni@chromium.org

Review URL: https://codereview.chromium.org/1302533002

Cr-Commit-Position: refs/heads/master@{#30244}

9 years ago[API] Check for NULL external data pointer in ArrayBuffer::New()
jkummerow [Wed, 19 Aug 2015 06:30:41 +0000 (23:30 -0700)]
[API] Check for NULL external data pointer in ArrayBuffer::New()

Embedders must not provide invalid pointers for external backing stores.

BUG=chromium:522128
LOG=n
R=cbruni@chromium.org

Review URL: https://codereview.chromium.org/1289373004

Cr-Commit-Position: refs/heads/master@{#30243}

9 years agoUpdate V8 DEPS.
v8-autoroll [Wed, 19 Aug 2015 03:24:16 +0000 (20:24 -0700)]
Update V8 DEPS.

Rolling v8/third_party/android_tools to 9e9b6169a098bc19986e44fbbf65e4c29031e4bd

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/1303573003

Cr-Commit-Position: refs/heads/master@{#30242}

9 years agoX87: [simd.js] Single SIMD128_VALUE_TYPE for all Simd128Values.
chunyang.dai [Wed, 19 Aug 2015 03:18:48 +0000 (20:18 -0700)]
X87: [simd.js] Single SIMD128_VALUE_TYPE for all Simd128Values.

port f4c079d450a5990639b295d40a3d1663d70412d6 (r30107).

original commit message:

    There's no need to have one InstanceType per SIMD primitive type (this
    will not scale long-term).  Also reduce the amount of code duplication
    and make it more robust wrt adding new SIMD types.

BUG=

Review URL: https://codereview.chromium.org/1286313003

Cr-Commit-Position: refs/heads/master@{#30241}

9 years agoOnly evaluate length once in %TypedArray%.prototype.set
ejcaruso [Wed, 19 Aug 2015 01:32:56 +0000 (18:32 -0700)]
Only evaluate length once in %TypedArray%.prototype.set

The ES6 spec for this function declares that ToLength
should only be called once. We were evaluating it multiple
times, so if length was an object with a valueOf method,
we could see effects take place multiple times.

R=littledan@chromium.org
LOG=N
BUG=v8:4218

Review URL: https://codereview.chromium.org/1237583005

Cr-Commit-Position: refs/heads/master@{#30240}

9 years ago[simd.js] Macro-ize more SIMD code.
bbudge [Tue, 18 Aug 2015 20:39:43 +0000 (13:39 -0700)]
[simd.js] Macro-ize more SIMD code.

Use macros for SIMD get/set_lane, and print fns.

LOG=N
BUG=v8:4124

Review URL: https://codereview.chromium.org/1302513004

Cr-Commit-Position: refs/heads/master@{#30239}

9 years agoDisable regress-crbug-518748 on ARM
Ben Smith [Tue, 18 Aug 2015 18:55:44 +0000 (11:55 -0700)]
Disable regress-crbug-518748 on ARM

BUG=chromium:518748
TBR=mstarzinger@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/1290683004 .

Cr-Commit-Position: refs/heads/master@{#30238}

9 years ago[test] Silence array bounds warning. GCC, I think you are wrong in this case.
Ben L. Titzer [Tue, 18 Aug 2015 18:51:20 +0000 (20:51 +0200)]
[test] Silence array bounds warning. GCC, I think you are wrong in this case.

TBR=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1295323003 .

Cr-Commit-Position: refs/heads/master@{#30237}

9 years agoAdd a makefile option for wasm prototype.
bradnelson [Tue, 18 Aug 2015 18:25:18 +0000 (11:25 -0700)]
Add a makefile option for wasm prototype.

Adding wasm=on when invoking make will build with the wasm prototype, Ex:

make x64.debug wasm=on V=1

BUG=None
TEST=manual
R=titzer@chromium.org,ncbray@chromium.org
LOG=N

Review URL: https://codereview.chromium.org/1293073004

Cr-Commit-Position: refs/heads/master@{#30236}

9 years ago[turbofan] Fix stack->stack double moves for pushing on ia32 and x64.
titzer [Tue, 18 Aug 2015 18:21:57 +0000 (11:21 -0700)]
[turbofan] Fix stack->stack double moves for pushing on ia32 and x64.

R=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1299023002

Cr-Commit-Position: refs/heads/master@{#30235}

9 years agoRemove empty string-search.cc file.
mstarzinger [Tue, 18 Aug 2015 18:18:11 +0000 (11:18 -0700)]
Remove empty string-search.cc file.

R=titzer@chromium.org

Review URL: https://codereview.chromium.org/1295333002

Cr-Commit-Position: refs/heads/master@{#30234}

9 years agoDisable regress-crbug-518748 on debug
Ben Smith [Tue, 18 Aug 2015 18:16:18 +0000 (11:16 -0700)]
Disable regress-crbug-518748 on debug

Seems to fail only on V8 Linux - debug - code serializer
(See http://build.chromium.org/p/client.v8/builders/V8%20Linux%20-%20debug%20-%20code%20serializer/builds/3762)

BUG=chromium:518748
TBR=mstarzinger@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/1294163003 .

Cr-Commit-Position: refs/heads/master@{#30233}

9 years agoRemove grab-bag includes of v8.h from runtime entries.
mstarzinger [Tue, 18 Aug 2015 17:55:19 +0000 (10:55 -0700)]
Remove grab-bag includes of v8.h from runtime entries.

R=titzer@chromium.org

Review URL: https://codereview.chromium.org/1293053004

Cr-Commit-Position: refs/heads/master@{#30232}

9 years ago[d8] Fix compile failure due to kMaxWorkers
Ben Smith [Tue, 18 Aug 2015 17:30:57 +0000 (10:30 -0700)]
[d8] Fix compile failure due to kMaxWorkers

TBR=adamk@chromium.org

Review URL: https://codereview.chromium.org/1302593002 .

Cr-Commit-Position: refs/heads/master@{#30231}

9 years ago[d8 Workers] Add max worker count, throw an exception if too many.
binji [Tue, 18 Aug 2015 17:12:03 +0000 (10:12 -0700)]
[d8 Workers] Add max worker count, throw an exception if too many.

BUG=chromium:518748
R=mstarzinger@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/1284683004

Cr-Commit-Position: refs/heads/master@{#30230}

9 years ago[d8] Fix flakiness when calling quit() with isolates
binji [Tue, 18 Aug 2015 16:47:19 +0000 (09:47 -0700)]
[d8] Fix flakiness when calling quit() with isolates

Don't use exit(), use Shell::Exit() (which calls _exit() instead). This won't
run C++ static destructors, atexit() functions, etc., which can occasionally
cause flaky failures.

BUG=v8:4279
R=machenbach@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/1294913005

Cr-Commit-Position: refs/heads/master@{#30229}

9 years agoAdd presubmit check for header inclusion violation.
mstarzinger [Tue, 18 Aug 2015 16:41:02 +0000 (09:41 -0700)]
Add presubmit check for header inclusion violation.

This warns about include directives of inline headers within normal
header files. Note that this warning should not close the tree or
prevent the CQ from landing the patch.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/1293273005

Cr-Commit-Position: refs/heads/master@{#30228}

9 years agoMake heap.h usable without objects-inl.h header.
mstarzinger [Tue, 18 Aug 2015 16:21:31 +0000 (09:21 -0700)]
Make heap.h usable without objects-inl.h header.

This CL us a pure refactoring that makes an empty compilation unit
including just "heap.h" but not "objects-inl.h" compile without
warnings or errors. This is needed to further reduce the header
dependency tangle.

R=mlippautz@chromium.org

Review URL: https://codereview.chromium.org/1301583003

Cr-Commit-Position: refs/heads/master@{#30227}

9 years ago[Interpreter] Add implementations for load immediate bytecodes.
rmcilroy [Tue, 18 Aug 2015 15:29:21 +0000 (08:29 -0700)]
[Interpreter] Add implementations for load immediate bytecodes.

Adds implementations and tests for the following bytecodes:
  - LdaZero
  - LdaSmi8
  - LdaUndefined
  - LdaNull
  - LdaTheHole
  - LdaTrue
  - LdaFalse
  - LdaLdar
  - LdaStar

Also adds  Smi tagging / untagging and OperandType typed
BytecodeOperand operations to InterpreterAssembler.

BUG=v8:4280
LOG=N

Review URL: https://codereview.chromium.org/1294793002

Cr-Commit-Position: refs/heads/master@{#30226}

9 years agoAllowing optional build of a WASM prototype behind a gyp define.
bradnelson [Tue, 18 Aug 2015 15:04:02 +0000 (08:04 -0700)]
Allowing optional build of a WASM prototype behind a gyp define.

Place a copy of the v8-native-prototype in third_party/wasm.
GYP_DEFINES='v8_wasm=1' gclient runhooks

BUG=None
TEST=None
R=titzer@chromium.org
LOG=N

Review URL: https://codereview.chromium.org/1294543006

Cr-Commit-Position: refs/heads/master@{#30225}

9 years ago[turbofan] Unify referencing of stack slots
danno [Tue, 18 Aug 2015 14:47:56 +0000 (07:47 -0700)]
[turbofan] Unify referencing of stack slots

Previously, it was not possible to specify StackSlotOperands for all
slots in both the caller and callee stacks. Specifically, the region
of the callee's stack including the saved return address, frame
pointer, function pointer and context pointer could not be addressed
by the register allocator/gap resolver.

In preparation for better tail call support, which will use the gap
resolver to reconcile outgoing parameters, this change makes it
possible to address all slots on the stack, because slots in the
previously inaccessible dead zone may become parameter slots for
outgoing tail calls. All caller stack slots are accessible as they
were before, with slot -1 corresponding to the last stack
parameter. Stack slot indices >= 0 access the callee stack, with slot
0 corresponding to the callee's saved return address, 1 corresponding
to the saved frame pointer, 2 corresponding to the current function
context, 3 corresponding to the frame marker/JSFunction, and slots 4
and above corresponding to spill slots.

The following changes were specifically needed:

* Frame has been changed to explicitly manage three areas of the
  callee frame, the fixed header, the spill slot area, and the
  callee-saved register area.
* Conversions from stack slot indices to fp offsets all now go through
  a common bottleneck: OptimizedFrame::StackSlotOffsetRelativeToFp
* The generation of deoptimization translation tables has been changed
  to support the new stack slot indexing scheme. Crankshaft, which
  doesn't support the new slot numbering in its register allocator,
  must adapt the indexes when creating translation tables.
* Callee-saved parameters are now kept below spill slots, not above,
  to support saving only the optimal set of used registers, which is
  only known after register allocation is finished and spill slots
  have been allocated.

Review URL: https://codereview.chromium.org/1261923007

Cr-Commit-Position: refs/heads/master@{#30224}

9 years agoSkip regress-4279 for --isolates tests.
yangguo [Tue, 18 Aug 2015 14:35:59 +0000 (07:35 -0700)]
Skip regress-4279 for --isolates tests.

R=hablich@chromium.org
BUG=v8:4378
LOG=N

Review URL: https://codereview.chromium.org/1290383004

Cr-Commit-Position: refs/heads/master@{#30223}

9 years agoRemove inline header includes from natives.h header.
mstarzinger [Tue, 18 Aug 2015 14:31:26 +0000 (07:31 -0700)]
Remove inline header includes from natives.h header.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/1303463002

Cr-Commit-Position: refs/heads/master@{#30222}

9 years ago[Interpreter] Minimal bytecode generator.
oth [Tue, 18 Aug 2015 13:46:43 +0000 (06:46 -0700)]
[Interpreter] Minimal bytecode generator.

Bytecode generator for local assignment and basic binary operations.

Command-line flag for printing bytecodes.

BUG=v8:4280
LOG=N

Review URL: https://codereview.chromium.org/1294543002

Cr-Commit-Position: refs/heads/master@{#30221}

9 years agoNative context: Fix issue when running prologue.js before runtime.js
yangguo [Tue, 18 Aug 2015 13:08:05 +0000 (06:08 -0700)]
Native context: Fix issue when running prologue.js before runtime.js

%InstallFunctionsFromArray is not entirely equivalent to the old
InstallFunctions implementation, which causes gc stress failures.

TBR=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1287203006

Cr-Commit-Position: refs/heads/master@{#30220}

9 years ago[interpreter]: Changes to interpreter builtins for accumulator and register file...
rmcilroy [Tue, 18 Aug 2015 12:41:41 +0000 (05:41 -0700)]
[interpreter]: Changes to interpreter builtins for accumulator and register file registers.

Makes the following modifications to the interpreter builtins and
InterpreterAssembler:
 - Adds an accumulator register and initializes it to undefined()
 - Adds a register file pointer register and use it instead of FramePointer to
   access registers
 - Modifies builtin to support functions with 0 regiters in the register file
 - Modifies builtin to Call rather than TailCall to first bytecode handler.

BUG=v8:4280
LOG=N

Review URL: https://codereview.chromium.org/1289863003

Cr-Commit-Position: refs/heads/master@{#30219}

9 years ago[api,heap] Fix external GC callbacks.
mlippautz [Tue, 18 Aug 2015 12:03:45 +0000 (05:03 -0700)]
[api,heap] Fix external GC callbacks.

* Add types to properly report what has been executed in the GC
* Unify GCPrologueCallback and GCEpilogueCallback into GCCallback
* Report processing of second round weak handels, either synchronously or asynchronously

BUG=chromium:521946
LOG=N

Review URL: https://codereview.chromium.org/1298113003

Cr-Commit-Position: refs/heads/master@{#30218}

9 years agoNative context: run prologue.js before runtime.js
yangguo [Tue, 18 Aug 2015 11:15:34 +0000 (04:15 -0700)]
Native context: run prologue.js before runtime.js

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1294803004

Cr-Commit-Position: refs/heads/master@{#30217}

9 years agoSkip regress-crbug-501711 for --isolates tests.
yangguo [Tue, 18 Aug 2015 11:14:26 +0000 (04:14 -0700)]
Skip regress-crbug-501711 for --isolates tests.

TBR=hablich@chromium.org
BUG=v8:4378
LOG=N

Review URL: https://codereview.chromium.org/1295813006

Cr-Commit-Position: refs/heads/master@{#30216}

9 years agoReland of move property loads from js builtins objects from runtime. (patchset #1...
yangguo [Tue, 18 Aug 2015 09:55:40 +0000 (02:55 -0700)]
Reland of move property loads from js builtins objects from runtime. (patchset #1 id:1 of https://codereview.chromium.org/1297803003/ )

Reason for revert:
Debug isolate failure has nothing to do with this CL.

Original issue's description:
> Revert of Remove property loads from js builtins objects from runtime. (patchset #2 id:20001 of https://codereview.chromium.org/1293113002/ )
>
> Reason for revert:
> Still failures in debug-isolates tests
>
> Original issue's description:
> > Remove property loads from js builtins objects from runtime.
> >
> > R=cbruni@chromium.org
> >
> > Committed: https://crrev.com/40f6e80d22d2e146b781aa661b76087ab9a492c4
> > Cr-Commit-Position: refs/heads/master@{#30199}
> >
> > Committed: https://crrev.com/f22d0f205031054a5f3116e052c81ae85741e8e0
> > Cr-Commit-Position: refs/heads/master@{#30209}
>
> TBR=cbruni@chromium.org,hpayer@chromium.org
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
>
> Committed: https://crrev.com/4106a4cbb701b5fe7d0b639e28a4ebfca5c05630
> Cr-Commit-Position: refs/heads/master@{#30213}

TBR=cbruni@chromium.org,hpayer@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/1298733003

Cr-Commit-Position: refs/heads/master@{#30215}

9 years agoMIPS: Fix bug in disassembler for JALR
Djordje.Pesic [Tue, 18 Aug 2015 09:38:33 +0000 (02:38 -0700)]
MIPS: Fix bug in disassembler for JALR

Adding missing rd register in JALR disassembly, for mips and mipps64

Review URL: https://codereview.chromium.org/1297183002

Cr-Commit-Position: refs/heads/master@{#30214}

9 years agoRevert of Remove property loads from js builtins objects from runtime. (patchset...
yangguo [Tue, 18 Aug 2015 09:37:25 +0000 (02:37 -0700)]
Revert of Remove property loads from js builtins objects from runtime. (patchset #2 id:20001 of https://codereview.chromium.org/1293113002/ )

Reason for revert:
Still failures in debug-isolates tests

Original issue's description:
> Remove property loads from js builtins objects from runtime.
>
> R=cbruni@chromium.org
>
> Committed: https://crrev.com/40f6e80d22d2e146b781aa661b76087ab9a492c4
> Cr-Commit-Position: refs/heads/master@{#30199}
>
> Committed: https://crrev.com/f22d0f205031054a5f3116e052c81ae85741e8e0
> Cr-Commit-Position: refs/heads/master@{#30209}

TBR=cbruni@chromium.org,hpayer@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/1297803003

Cr-Commit-Position: refs/heads/master@{#30213}

9 years ago[turbofan] Remove the output_index field that was unused in Node::Use.
jarin [Tue, 18 Aug 2015 09:10:19 +0000 (02:10 -0700)]
[turbofan] Remove the output_index field that was unused in Node::Use.

BUG=

Review URL: https://codereview.chromium.org/1294913003

Cr-Commit-Position: refs/heads/master@{#30212}

9 years agoOnly evacuation candidate pages have a slots buffer, just visit these pages when...
hpayer [Tue, 18 Aug 2015 09:02:27 +0000 (02:02 -0700)]
Only evacuation candidate pages have a slots buffer, just visit these pages when filtering slots.

BUG=

Review URL: https://codereview.chromium.org/1299623003

Cr-Commit-Position: refs/heads/master@{#30211}

9 years agoReenable code recompilation verification.
yangguo [Tue, 18 Aug 2015 07:53:45 +0000 (00:53 -0700)]
Reenable code recompilation verification.

BUG=chromium:516304
LOG=N

Review URL: https://codereview.chromium.org/1267953002

Cr-Commit-Position: refs/heads/master@{#30210}

9 years agoRemove property loads from js builtins objects from runtime.
yangguo [Tue, 18 Aug 2015 07:48:13 +0000 (00:48 -0700)]
Remove property loads from js builtins objects from runtime.

R=cbruni@chromium.org

Committed: https://crrev.com/40f6e80d22d2e146b781aa661b76087ab9a492c4
Cr-Commit-Position: refs/heads/master@{#30199}

Review URL: https://codereview.chromium.org/1293113002

Cr-Commit-Position: refs/heads/master@{#30209}

9 years agoUpdate V8 DEPS.
v8-autoroll [Tue, 18 Aug 2015 03:24:59 +0000 (20:24 -0700)]
Update V8 DEPS.

Rolling v8/third_party/icu to 89dcdec16381883782b9cc9cff38e00f047a0f46

Rolling v8/tools/clang to 5b12e334ec0e571a8e1f68d028dc5427b58c17ec

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/1294523005

Cr-Commit-Position: refs/heads/master@{#30208}

9 years agoPoint @@isConcatSpreadable test failure line at the correct bug
Adam Klein [Mon, 17 Aug 2015 21:46:44 +0000 (14:46 -0700)]
Point @@isConcatSpreadable test failure line at the correct bug

BUG=v8:4317
LOG=n
TBR=littledan@chromium.org

Review URL: https://codereview.chromium.org/1285983004 .

Cr-Commit-Position: refs/heads/master@{#30207}

9 years ago[simd.js] Macro-ize more SIMD code.
bbudge [Mon, 17 Aug 2015 18:58:38 +0000 (11:58 -0700)]
[simd.js] Macro-ize more SIMD code.

Use macros for factory functions, and in hydrogen code stubs.

LOG=N
BUG=v8:4124

Review URL: https://codereview.chromium.org/1293533003

Cr-Commit-Position: refs/heads/master@{#30206}

9 years agoRevert "Regularly check hash set addresses to verify memory integrity."
hpayer [Mon, 17 Aug 2015 18:25:11 +0000 (11:25 -0700)]
Revert "Regularly check hash set addresses to verify memory integrity."

This debug check is not needed anymore.

This reverts commit ce311ed00123be1ffefdfbee39bd67cec63599ed.

BUG=

Review URL: https://codereview.chromium.org/1300663002

Cr-Commit-Position: refs/heads/master@{#30205}

9 years ago[heap] Get rid of unused regexp includes.
mlippautz [Mon, 17 Aug 2015 18:19:43 +0000 (11:19 -0700)]
[heap] Get rid of unused regexp includes.

BUG=

Review URL: https://codereview.chromium.org/1290403002

Cr-Commit-Position: refs/heads/master@{#30204}

9 years ago[turbofan] Support unboxed float and double stack parameters and add tests.
titzer [Mon, 17 Aug 2015 17:35:40 +0000 (10:35 -0700)]
[turbofan] Support unboxed float and double stack parameters and add tests.

R=jarin@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1291113003

Cr-Commit-Position: refs/heads/master@{#30203}

9 years ago[heap] Unify MarkingDeque push and unshift operations.
mstarzinger [Mon, 17 Aug 2015 16:58:16 +0000 (09:58 -0700)]
[heap] Unify MarkingDeque push and unshift operations.

R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/1294093003

Cr-Commit-Position: refs/heads/master@{#30202}

9 years agoRevert of Remove property loads from js builtins objects from runtime. (patchset...
hablich [Mon, 17 Aug 2015 15:31:46 +0000 (08:31 -0700)]
Revert of Remove property loads from js builtins objects from runtime. (patchset #1 id:1 of https://codereview.chromium.org/1293113002/ )

Reason for revert:
Reverting because it made the waterfall red http://build.chromium.org/p/client.v8/builders/V8%20Win32%20-%20nosnap%20-%20shared/builds/8390

Original issue's description:
> Remove property loads from js builtins objects from runtime.
>
> R=cbruni@chromium.org
>
> Committed: https://crrev.com/40f6e80d22d2e146b781aa661b76087ab9a492c4
> Cr-Commit-Position: refs/heads/master@{#30199}

TBR=cbruni@chromium.org,yangguo@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/1294123002

Cr-Commit-Position: refs/heads/master@{#30201}

9 years agoFilter out slot buffer slots, that point to SMIs in dead objects.
hpayer [Mon, 17 Aug 2015 15:24:13 +0000 (08:24 -0700)]
Filter out slot buffer slots, that point to SMIs in dead objects.

The following situation may happen which reproduces this bug:
(1) We allocate JSObject A on an evacuation candidate.
(2) We allocate JSObject B on a non-evacuation candidate.
(3) Incremental marking starts and marks object A and B.
(4) We create a reference from B.field = A; which records the slot B.field since A is on an evacuation candidate.
(5) After that we write a SMI into B.field.
(6) After that B goes into dictionary mode and shrinks its original size. B.field is now outside of the JSObject, i.e B.field is in memory that will be freed by the sweeper threads.
(7) GC is triggered.
(8) BUG: Slots buffer filtering walks over the slots buffer, SMIs are not filtered out because we assumed that SMIs are just ignored when the slots get updated later. However, recorded SMI slots of dead objects may be overwritten by double values at evacuation time.
(9) During evacuation, a heap number that looks like a valid pointer is moved over B.field.
(10) The slots buffer is scanned for updates, follows B.field since it looks like a pointer (the double value looks like a pointer), and crashes.

BUG=chromium:519577,chromium:454297
LOG=y

Review URL: https://codereview.chromium.org/1286343004

Cr-Commit-Position: refs/heads/master@{#30200}

9 years agoRemove property loads from js builtins objects from runtime.
yangguo [Mon, 17 Aug 2015 15:08:20 +0000 (08:08 -0700)]
Remove property loads from js builtins objects from runtime.

R=cbruni@chromium.org

Review URL: https://codereview.chromium.org/1293113002

Cr-Commit-Position: refs/heads/master@{#30199}

9 years agoDefault-enable external startup snapshot for, like, everywhere.
vogelheim [Mon, 17 Aug 2015 13:59:40 +0000 (06:59 -0700)]
Default-enable external startup snapshot for, like, everywhere.

BUG=

Review URL: https://codereview.chromium.org/1209223005

Cr-Commit-Position: refs/heads/master@{#30198}

9 years ago[turbofan] Handle void return in simplified-lowering.cc.
titzer [Mon, 17 Aug 2015 13:56:11 +0000 (06:56 -0700)]
[turbofan] Handle void return in simplified-lowering.cc.

R=jarin@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1296933002

Cr-Commit-Position: refs/heads/master@{#30197}

9 years ago[test] Remove FLAG_always_opt special case in NotifyDeoptimized
jkummerow [Mon, 17 Aug 2015 13:26:28 +0000 (06:26 -0700)]
[test] Remove FLAG_always_opt special case in NotifyDeoptimized

Always unlink optimized code on deopt, even when FLAG_always_opt is present, because assumptions that the code made could have become invalid.
BUG=v8:4375
LOG=n
R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/1284103006

Cr-Commit-Position: refs/heads/master@{#30196}

9 years agoRemove grab-bag includes of v8.h from heap.
mstarzinger [Mon, 17 Aug 2015 13:13:32 +0000 (06:13 -0700)]
Remove grab-bag includes of v8.h from heap.

R=mlippautz@chromium.org

Review URL: https://codereview.chromium.org/1291833003

Cr-Commit-Position: refs/heads/master@{#30195}

9 years ago[heap] Simplify MarkingDeque implementation.
mstarzinger [Mon, 17 Aug 2015 13:09:29 +0000 (06:09 -0700)]
[heap] Simplify MarkingDeque implementation.

This removes the dependency of the MarkingDeque on the underlying
collector by moving out the live bytes accounting and object color
mutations into a wrapper. The MarkingDeque can now be split into a
separate file.

R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/1293773002

Cr-Commit-Position: refs/heads/master@{#30194}

9 years ago[turbofan]: Fix bug in register hinting
danno [Mon, 17 Aug 2015 12:43:59 +0000 (05:43 -0700)]
[turbofan]: Fix bug in register hinting

Previously, UsePositions for Phis that are constant LiveRanges which are also used
as hints for other USePositions would always force-allocate the first register in
the allocatable register file to that range.

Review URL: https://codereview.chromium.org/1289313003

Cr-Commit-Position: refs/heads/master@{#30193}

9 years ago[es6] Implement default parameters
rossberg [Mon, 17 Aug 2015 12:01:55 +0000 (05:01 -0700)]
[es6] Implement default parameters

Based on caitp's https://codereview.chromium.org/1127063003/

R=adamk@chromium.org, littledan@chromium.org
BUG=v8:2160
LOG=N

Review URL: https://codereview.chromium.org/1287063004

Cr-Commit-Position: refs/heads/master@{#30192}

9 years agofix StrDup memory leak in CcTest
jianghua.yjh [Mon, 17 Aug 2015 11:30:30 +0000 (04:30 -0700)]
fix StrDup memory leak in CcTest

BUG=

Review URL: https://codereview.chromium.org/1287023002

Cr-Commit-Position: refs/heads/master@{#30191}

9 years ago[Interpreter] Move interpreter initialization until after snapshot deserialization.
rmcilroy [Mon, 17 Aug 2015 11:24:34 +0000 (04:24 -0700)]
[Interpreter] Move interpreter initialization until after snapshot deserialization.

The interpreter needs to be initialized after the snapshot has been deserialized.

BUG=v8:4280
LOG=N

Review URL: https://codereview.chromium.org/1290883004

Cr-Commit-Position: refs/heads/master@{#30190}

9 years agoDo not use js builtins object when constructing an error.
yangguo [Mon, 17 Aug 2015 11:02:31 +0000 (04:02 -0700)]
Do not use js builtins object when constructing an error.

R=cbruni@chromium.org

Review URL: https://codereview.chromium.org/1295093002

Cr-Commit-Position: refs/heads/master@{#30189}

9 years agoAdd DCHECK that the script context table do not contain native scripts.
yangguo [Mon, 17 Aug 2015 11:00:43 +0000 (04:00 -0700)]
Add DCHECK that the script context table do not contain native scripts.

This is to prevent unnecessary script contexts to be created for
native scripts, which could end up with a script context table
containing twenty script contexts for a newly created native context.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/1301533002

Cr-Commit-Position: refs/heads/master@{#30188}

9 years agoRemove grab-bag includes of v8.h from architecture ports.
mstarzinger [Mon, 17 Aug 2015 09:42:26 +0000 (02:42 -0700)]
Remove grab-bag includes of v8.h from architecture ports.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/1299563003

Cr-Commit-Position: refs/heads/master@{#30187}

9 years ago[runtime] Unify and fix the strict equality comparison.
bmeurer [Mon, 17 Aug 2015 08:01:55 +0000 (01:01 -0700)]
[runtime] Unify and fix the strict equality comparison.

Add Object::StrictEquals to unify the implementation of strict equality
comparison in the runtime and the api (the api was already missing a
case for SIMD).  Now we (almost) have a single bottleneck for strict
equality, we just need to reduce the amount of unnecessary complexity
for the code stub.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/1298603002

Cr-Commit-Position: refs/heads/master@{#30186}

9 years agoX87: Realize IWYU pattern for frames-inl.h header.
chunyang.dai [Mon, 17 Aug 2015 00:40:28 +0000 (17:40 -0700)]
X87: Realize IWYU pattern for frames-inl.h header.

port 19a49abf02dfac722dc8c5079aba9d73fcd60725 (r30127).

original commit message:

BUG=

Review URL: https://codereview.chromium.org/1286143009

Cr-Commit-Position: refs/heads/master@{#30185}

9 years ago[parser] make kInvalidLhsInFor a SyntaxError
caitpotter88 [Sat, 15 Aug 2015 17:57:01 +0000 (10:57 -0700)]
[parser] make kInvalidLhsInFor a SyntaxError

Second item in section 13.7.5.1 states that the error should be a
SyntaxError, when previously CheckAndRewriteReferenceExpression
would always emit a ReferenceError.

BUG=v8:4373
R=adamk, rossberg
LOG=N
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_chromium_rel_ng;tryserver.blink:linux_blink_rel

Review URL: https://codereview.chromium.org/1292393002

Cr-Commit-Position: refs/heads/master@{#30184}

9 years agoPut V8 extras into the snapshot
domenic [Fri, 14 Aug 2015 18:47:46 +0000 (11:47 -0700)]
Put V8 extras into the snapshot

Previously, all extras were "experimental" and left out of the snapshot. This
patch moves them to the snapshot, so now all extras are non-experimental. A
future patch will re-introduce experimental extras as part of the linked bug.

R=yangguo@chromium.org
BUG=https://code.google.com/p/chromium/issues/detail?id=507137
LOG=Y

Review URL: https://codereview.chromium.org/1289603002

Cr-Commit-Position: refs/heads/master@{#30183}

9 years agoAdd per-file OWNERS for x87-specific cctests.
chunyang.dai [Fri, 14 Aug 2015 16:02:22 +0000 (09:02 -0700)]
Add per-file OWNERS for x87-specific cctests.

BUG=

Review URL: https://codereview.chromium.org/1290963005

Cr-Commit-Position: refs/heads/master@{#30182}

9 years agoClean up native context slots and add new ones.
yangguo [Fri, 14 Aug 2015 15:12:34 +0000 (08:12 -0700)]
Clean up native context slots and add new ones.

Newly added native context slots are used to avoid accessing
the js builtins object to get native functions.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/1294583006

Cr-Commit-Position: refs/heads/master@{#30181}

9 years agoRemove grab-bag includes of v8.h from regexp engine.
mstarzinger [Fri, 14 Aug 2015 15:09:27 +0000 (08:09 -0700)]
Remove grab-bag includes of v8.h from regexp engine.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/1294783002

Cr-Commit-Position: refs/heads/master@{#30180}

9 years agoNew flag --perf_basic_prof_only_functions
ofrobots [Fri, 14 Aug 2015 14:51:51 +0000 (07:51 -0700)]
New flag --perf_basic_prof_only_functions

Restricts linux perf-event code range reporting to functions only (i.e. on
stubs.) While this makes the gathered ticks less accurate, it reduces the
growth of the /tmp/perf-${pid}.map file.

BUG=v8:3453
R=hablich@chromium.org,danno@chromium.org
LOG=N

Review URL: https://codereview.chromium.org/1292743002

Cr-Commit-Position: refs/heads/master@{#30179}

9 years agoRemove grab-bag includes of v8.h from IC subsystem.
mstarzinger [Fri, 14 Aug 2015 13:33:32 +0000 (06:33 -0700)]
Remove grab-bag includes of v8.h from IC subsystem.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/1293793002

Cr-Commit-Position: refs/heads/master@{#30178}

9 years ago[serializer] Move WeakFixedArray compaction to separate heap walk phase
jkummerow [Fri, 14 Aug 2015 13:09:36 +0000 (06:09 -0700)]
[serializer] Move WeakFixedArray compaction to separate heap walk phase

This avoids discovering and compacting the same array repeatedly. No change in resulting snapshots intended.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/1290393002

Cr-Commit-Position: refs/heads/master@{#30177}

9 years ago[turbofan] Gracefully handle missing info()->context() in CodeGenerator::IsMaterializ...
titzer [Fri, 14 Aug 2015 13:07:27 +0000 (06:07 -0700)]
[turbofan] Gracefully handle missing info()->context() in CodeGenerator::IsMaterializableFromFrame()

R=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1292233004

Cr-Commit-Position: refs/heads/master@{#30176}

9 years agoDo not export natives to runtime via js builtins object.
yangguo [Fri, 14 Aug 2015 12:34:45 +0000 (05:34 -0700)]
Do not export natives to runtime via js builtins object.

Using a temporary container object instead.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/1293493003

Cr-Commit-Position: refs/heads/master@{#30175}

9 years agoRework startup-data-util.
vogelheim [Fri, 14 Aug 2015 12:11:29 +0000 (05:11 -0700)]
Rework startup-data-util.

- Make the API look like v8::V8::InitializeICU.
  (That is: A static method call, not an object to be created on the stack.)
- Fix path separator on Windows, by calling base::OS::isPathSeparator.
- Move into API, so that it can be called by hello-world & friends.
- Actually call it from hello-world and friends.

R=jochen@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1292053002

Cr-Commit-Position: refs/heads/master@{#30174}

9 years agoNo longer use js builtins object as receiver for calls into JS.
yangguo [Fri, 14 Aug 2015 10:22:16 +0000 (03:22 -0700)]
No longer use js builtins object as receiver for calls into JS.

This is no longer necessary. Everything the callee needs should
be in its context chain.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/1289203003

Cr-Commit-Position: refs/heads/master@{#30173}

9 years agoRemove grab-bag includes of v8.h from several files.
mstarzinger [Fri, 14 Aug 2015 09:41:32 +0000 (02:41 -0700)]
Remove grab-bag includes of v8.h from several files.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/1297583002

Cr-Commit-Position: refs/heads/master@{#30172}

9 years agoMake some foo.h headers usable without foo-inl.h header.
mstarzinger [Fri, 14 Aug 2015 08:48:11 +0000 (01:48 -0700)]
Make some foo.h headers usable without foo-inl.h header.

This CL is a pure refactoring that makes an empty compilation unit
including just "foo.h" but not "foo-inl.h" compile without warnings or
errors. This is needed to further reduce the header dependency tangle.

This realizes above state for "spaces.h" and "mark-compact.h".

R=hpayer@chromium.org

Review URL: https://codereview.chromium.org/1288413002

Cr-Commit-Position: refs/heads/master@{#30171}