X87: Remove frame pointer from StackHandler.

port 36e69a916f2a88664887d2b62f0eaf8400edb182 (r27115)

original commit message:

 This reduces the size of the StackHandler by yet another word. We no
 longer need to keep track of the frame pointer, as the stack walk will
 be able to recalculate it.

BUG=

Review URL: https://codereview.chromium.org/1030563002

Cr-Commit-Position: refs/heads/master@{#27357}

X87:  [es6] Throw TypeError for computed static prototype property name

port 8d946b9c3f6ea42dd5232c0529be4d47798b06aa (r27106).

original commit message:

  [es6] Throw TypeError for computed static prototype property name

  The prototype of a class constructor function is read only. When we set
  computed property names we were ignoring this and we were overriding the
  property.

  Since the prototype is the only possible own read only property on the
  constructor function object we special case this so we do not have to
  check this for every property in the class literal.

BUG=

Review URL: https://codereview.chromium.org/1028983002

Cr-Commit-Position: refs/heads/master@{#27356}

Use TO_OBJECT_INLINE a bit more

We were using ToObject in a lot of places where the common case is that
we already have an object. By changing to TO_OBJECT_INLINE we don't
have to go through 5 different if statements before falling through.

We were also calling ToObject too many times in DefineObjectProperty
where we already know that obj is an object.

BUG=None
LOG=N

Review URL: https://codereview.chromium.org/1019413002

Cr-Commit-Position: refs/heads/master@{#27355}

[es6] Object.getPrototypeOf should work with values

The final spec for Object.getPrototypeOf calls ToObject on the
parameter, which means that it should only throw for null and
undefined. For other non object values the prototype of the wrapper
should be used.

BUG=v8:3964
LOG=N
R=adamk, rossberg@chromium.org

Review URL: https://codereview.chromium.org/1014813003

Cr-Commit-Position: refs/heads/master@{#27354}

X87: Remove code object from StackHandler.

port e0aa8ebf933b4c659e1c4a01ba244c5de452ff1c (r27103).

original commit message:

  This reduces the size of the StackHandler by one word. We no longer
  need to keep track of the code object, as the stack walk finds it.

BUG=
R=weiliang.lin@intel.com

Review URL: https://codereview.chromium.org/1022403002

Cr-Commit-Position: refs/heads/master@{#27353}

[es6] remove --harmony-templates flag

BUG=v8:3230
R=dslomov@chromium.org, arv@chromium.org
LOG=N

Review URL: https://codereview.chromium.org/1027593005

Cr-Commit-Position: refs/heads/master@{#27352}

MIPS: Fix 'MIPS: Serializer: serialize internal references via object visitor.'

BUG=

Review URL: https://codereview.chromium.org/1025453003

Cr-Commit-Position: refs/heads/master@{#27351}

Update V8 DEPS.

Rolling v8/third_party/icu to 7c81740601355556e630da515b74d889ba2f8d08

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/1030513002

Cr-Commit-Position: refs/heads/master@{#27350}

MIPS: [es6] generate rest parameters correctly for subclass constructors

Port bef80fcfd7e89cadc215f7d10a016a375e346490

BUG=v8:3977
LOG=N

Review URL: https://codereview.chromium.org/1028703002

Cr-Commit-Position: refs/heads/master@{#27349}

MAP_NORESERVE was never implemented in FreeBSD as has been retired
as of https://reviews.freebsd.org/D848. This makes V8 build on
FreeBSD-current (and the forthcoming 11).

BUG=none
LOG=n

Review URL: https://codereview.chromium.org/1025823003

Cr-Commit-Position: refs/heads/master@{#27348}

MIPS: Fix 'MIPS: [es6] implement Reflect.apply() & Reflect.construct()'

Port d21fd15467e16f185e511dbfbaeef7caddfe804a

TEST=mjsunit/harmony/reflect-construct
BUG=

Review URL: https://codereview.chromium.org/1022053002

Cr-Commit-Position: refs/heads/master@{#27347}

Ensure we don't overflow in BCE

BUG=chromium:469148
LOG=y
R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/1023123003

Cr-Commit-Position: refs/heads/master@{#27346}

Revert of Allow compaction when incremental marking is on. (patchset #1 id:1 of https://codereview.chromium.org/1014263002/)

Reason for revert:
It seems to cause crbug/469146.

Original issue's description:
> Allow compaction when incremental marking is on.
>
> BUG=chromium:450824
> LOG=NO
>
> Committed: https://crrev.com/92f96e4e9a527fcb085b68f81ee14b26acdd4719
> Cr-Commit-Position: refs/heads/master@{#27267}

TBR=hpayer@chromium.org,ulan@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:450824

Review URL: https://codereview.chromium.org/1026813002

Cr-Commit-Position: refs/heads/master@{#27345}

[es6] generate rest parameters correctly for subclass constructors

BUG=v8:3977
R=dslomov@chromium.org, arv@chromium.org
LOG=N

Review URL: https://codereview.chromium.org/1018043003

Cr-Commit-Position: refs/heads/master@{#27344}

MIPS: [turbofan] Turn Math.clz32 into an inlinable builtin.

Port 3aa206b86560da94f895625186295bf07a0301d8

BUG=v8:3952
LOG=n

Review URL: https://codereview.chromium.org/1020223002

Cr-Commit-Position: refs/heads/master@{#27343}

X87: [turbofan] Implement throwing exceptions into TurboFan code.

port 1382879f29bd71c36aeda2421e867c4cdd232a0d (r27016).

oringinal commit message:

    [turbofan] Implement throwing exceptions into TurboFan code.

    This extends the stack unwinding logic to respect optimized frames
    and perform a lookup in the handler table to find handlers. It also
    contains fixes to the API call stubs to allow a stack walk while
    promoting scheduled exceptions.

BUG=

Review URL: https://codereview.chromium.org/1023943002

Cr-Commit-Position: refs/heads/master@{#27342}

[turbofan] Fix lowering of Math.max for integral inputs.

R=jarin@chromium.org
BUG=chromium:468162
LOG=y

Review URL: https://codereview.chromium.org/1027753002

Cr-Commit-Position: refs/heads/master@{#27341}

Serializer: cache hashmaps on the isolate.

This speeds up multiple uses of the serializer quite a bit.

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1003363003

Cr-Commit-Position: refs/heads/master@{#27340}

[harmony] Move some regression tests to the right place

TBR=dslomov@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1027693003

Cr-Commit-Position: refs/heads/master@{#27339}

[strong] Less pleading error messages

R=dslomov@chromium.org
BUG=v8:3956
LOG=N

Review URL: https://codereview.chromium.org/1021213002

Cr-Commit-Position: refs/heads/master@{#27338}

Remove harmony-strings flag.

String functions are shipping since 4.1/Chrome M41, it is time to unflag.

R=yanngguo@chromium.org,rossberg@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1024813002

Cr-Commit-Position: refs/heads/master@{#27337}

[turbofan] More inlinable Math builtins.

Turn Math.acos, Math.asin, Math.atan, Math.atan2, Math.sign and Math.trunc
into inlinable builtins.

R=yangguo@chromium.org
BUG=v8:3952
LOG=n

Review URL: https://codereview.chromium.org/1027713002

Cr-Commit-Position: refs/heads/master@{#27336}

Don't DCHECK that the weak closure wasn't yet overapproximated

Maybe it was overapproximated via the idle notification already, so just
don't retry in that case

BUG=chromium:469023
R=hpayer@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/1025603003

Cr-Commit-Position: refs/heads/master@{#27335}

Check for GC interrupt in JSON parser.

BUG=v8:3974
LOG=NO

Review URL: https://codereview.chromium.org/1021523002

Cr-Commit-Position: refs/heads/master@{#27334}

[turbofan] Work-around untagged result of CompareIC in pointer maps.

BUG=chromium:469089
LOG=n
R=jarin@chromium.org

Review URL: https://codereview.chromium.org/1026683002

Cr-Commit-Position: refs/heads/master@{#27333}

X87: Fix exception for assignment to uninitialised const.

port 2ecdf736cf0409b52832f78a4439ace855d6739b (r27014).

original commit message:

  Fix exception for assignment to uninitialised const.

BUG=

Review URL: https://codereview.chromium.org/1028533002

Cr-Commit-Position: refs/heads/master@{#27332}

[turbofan] Add explicit set for nodes with weakened types.

To ensure termination, we need to be sure that once we start weakening
a node, we keep weakening that node in subsequent re-typings. Until now,
we were guessing that we previously weakened from the type. This
change introduces a set of nodes that have already be weakened, so
that we have a reliable way to detect previous weakening.

BUG=chromium:468799
LOG=n
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/1019883002

Cr-Commit-Position: refs/heads/master@{#27331}

[turbofan] Improve typing for string::length property.

R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/1028563002

Cr-Commit-Position: refs/heads/master@{#27330}

[turbofan] Turn Math.clz32 into an inlinable builtin.

R=dcarney@chromium.org, yangguo@chromium.org
BUG=v8:3952
LOG=n

Review URL: https://codereview.chromium.org/1021183002

Cr-Commit-Position: refs/heads/master@{#27329}

[turbofan] Fix unit test coverage for JSIntrinsicLowering.

R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/1028553002

Cr-Commit-Position: refs/heads/master@{#27328}

Update V8 DEPS.

Rolling v8/third_party/icu to d319ad9e25e8045f051dcd30bd5835f99936af6c

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/1021953002

Cr-Commit-Position: refs/heads/master@{#27327}

MIPS: [es6] implement Reflect.apply() & Reflect.construct()

Port d21fd15467e16f185e511dbfbaeef7caddfe804a

Known issue: the mjsunit/harmony/reflect-construct test fails, it will be addressed in a follow-up CL.

BUG=

Review URL: https://codereview.chromium.org/1021863002

Cr-Commit-Position: refs/heads/master@{#27326}

Parser: Fix crash on stack overflow when lazy-parsing arrow functions

The problem manifests itself when parsing manages to return something
meaningful in the presence of a stack overflow. This happens because
calling ParserBase::Next() will still return one valid token on stack
overflow, before starting to return invalid tokens.

Take the following input as example:

        a.map(v => v + 1);
              | |
       already   next token
        parsed   (which will be an invalid token
  (identifier)   because of a stack overflow)

The "v" may have been already parsed into a VariableProxy, then if a
stack overflow occurs, next token will be an invalid token (instead
of Token::ARROW), but the parser will return the VariableProxy.

This always happens when lazy-parsing arrow functions, so the position
in the input stream where the the arrow function code ends is known.
This fix adds a check that ensures that parsing ended at the end
position of the arrow function.

BUG=465671
LOG=N

Review URL: https://codereview.chromium.org/1023483003

Cr-Commit-Position: refs/heads/master@{#27325}

Cannot use Handle<T>::cast in Unique<T>::cast since it will try to do a T::cast (and its typecheck) concurrently, which is unsafe concurrently on moving values

BUG=

Review URL: https://codereview.chromium.org/1022943002

Cr-Commit-Position: refs/heads/master@{#27324}

Revert "Merge old data and pointer space."

TBR=verwaest@chromium.org,ulan@chromium.org,ishell@chromium.org
NOTRY=true

Review URL: https://codereview.chromium.org/1027463002

Cr-Commit-Position: refs/heads/master@{#27323}

[turbofan] Emit constant pool before blocking it in AssembleArchTableSwitch.

BUG=chromium:468749
LOG=NO
TEST=Emscripten/NBodyJava

Review URL: https://codereview.chromium.org/1014093005

Cr-Commit-Position: refs/heads/master@{#27322}

Rename kMaxFrameRenderingIdleTime to kMaxActiveIdleTime and adjust it to the currently maximum value of 50ms passed in by the Blink scheduler.

BUG=468692
LOG=n

Review URL: https://codereview.chromium.org/1024543002

Cr-Commit-Position: refs/heads/master@{#27321}

[strong] Check constructor returns

R=dslomov@chromium.org
BUG=v8:3956
LOG=N

Review URL: https://codereview.chromium.org/1019283002

Cr-Commit-Position: refs/heads/master@{#27320}

[turbofan] Remove last_use_ field from Node.

R=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1022783002

Cr-Commit-Position: refs/heads/master@{#27319}

Use libdl to get symbols for backtraces

With this patch, it'll look like this:

$ out/x64.optdebug/d8 --expose-trigger-failure test/mjsunit/mjsunit.js test/mjsunit/verify-assert-false.js

==== C stack trace ===============================

 1: V8_Fatal
 2: v8::internal::FunctionCallbackArguments::Call(void (*)(v8::FunctionCallbackInfo<v8::Value> const&))
 3: 0x727ced
 4: 0x72b6ba
 5: 0x188c7f607f9b

BUG=none
R=svenpanne@chromium.org
LOG=y

Review URL: https://codereview.chromium.org/1018313003

Cr-Commit-Position: refs/heads/master@{#27318}

Remove early bail-out in VisitWeakList to investigate chrasher.

BUG=468601
LOG=n

Review URL: https://codereview.chromium.org/1016353002

Cr-Commit-Position: refs/heads/master@{#27317}

[es6] implement Reflect.apply() & Reflect.construct()

BUG=v8:3900
LOG=N
R=dslomov@chromium.org

Review URL: https://codereview.chromium.org/913073003

Cr-Commit-Position: refs/heads/master@{#27316}

[x86] Atom check includes AMT

BUG=

Review URL: https://codereview.chromium.org/998883002

Cr-Commit-Position: refs/heads/master@{#27315}

Add debug checks to catch crashes with WeakCell::cast().

BUG=

Review URL: https://codereview.chromium.org/1009603003

Cr-Commit-Position: refs/heads/master@{#27314}

Add a counter for legacy const.
Only adds a counter id, does not count anything.

R=jochen@chromium.org
BUG=v8:3942
LOG=N

Review URL: https://codereview.chromium.org/1020883002

Cr-Commit-Position: refs/heads/master@{#27313}

Serializer: another small reorder of root objects.

I was mistaken, cons strings only occur a lot for custom heap snapshots,
not for the default start-up snapshot.

R=erik.corry@gmail.com, mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1023673003

Cr-Commit-Position: refs/heads/master@{#27312}

[arm64] serialize internal references via object visitor

BUG=

Review URL: https://codereview.chromium.org/1020863003

Cr-Commit-Position: refs/heads/master@{#27311}

[turbofan] Remember types for deoptimization during simplified lowering.

With this change, we remember the types of frame state inputs (in a new
operator, called TypedStateValues). Instead of inferring the value types
when building translations, we used the recorded types.

The original approach was not reliable because the passes after
simplified lowering can change node types, and this in turn confuses
the translation builder.

BUG=chromium:468727
LOG=n
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/1015423002

Cr-Commit-Position: refs/heads/master@{#27310}

Add debug checks to catch PropertyCell::cast failures.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/1025433002

Cr-Commit-Position: refs/heads/master@{#27309}

[turbofan] Use proper parameter representation for stub calls.

The CallInterfaceDescriptor already provides information about the
expected representation of parameters, so we can use that instead of
hardcoding tagged representation for all parameters.

R=jarin@chromium.org

Review URL: https://codereview.chromium.org/1019293002

Cr-Commit-Position: refs/heads/master@{#27308}

MIPS: Tweak constants to be in favor of table switch.

In prevous set-up table switch is rarely generated.

TEST=cctest/test-serialize/SerializeInternalReference
BUG=

Review URL: https://codereview.chromium.org/1019463003

Cr-Commit-Position: refs/heads/master@{#27307}

Revert of Cleanup and unify Isolate::ReportPendingMessages. (patchset #4 id:60001 of https://codereview.chromium.org/999923004/)

Reason for revert:
Failing Blink tests on Win and Mac:
- plugins/netscape-plugin-property-access-exception.html
- http/tests/plugins/cross-frame-object-access.html

Original issue's description:
> Cleanup and unify Isolate::ReportPendingMessages.
>
> Note that this is a pure cleanup CL and shouldn't have an observable
> impact on the functional behavior of message reporting.
>
> R=yangguo@chromium.org

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

Review URL: https://codereview.chromium.org/1016323002

Cr-Commit-Position: refs/heads/master@{#27306}

[stubs] Add missing interface descriptor for the CompareIC.

R=jarin@chromium.org

Review URL: https://codereview.chromium.org/1006173003

Cr-Commit-Position: refs/heads/master@{#27305}

[turbofan] Fix GVN of projections and add verification for projection uniqueness.

R=bmeurer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1020873002

Cr-Commit-Position: refs/heads/master@{#27304}

[ARM64] [turbofan] Support uxtb/uxth for add/sub.

Add support for appending extend modes uxtb or uxth to add and subtract
instructions, and using them in the instruction selector.

BUG=

Review URL: https://codereview.chromium.org/1021533002

Cr-Commit-Position: refs/heads/master@{#27303}

Fix a stray reference to 'build_os' in BUILD.gn.

This should be 'host_os'; the build_os variable is going away.

R=jochen@chromium.org
BUG=344767
LOG=n

Review URL: https://codereview.chromium.org/1010683003

Cr-Commit-Position: refs/heads/master@{#27302}

MIPS64: Fix bug in jump table test.

TEST=test-assembler-mips64/jump_tables3
BUG=

Review URL: https://codereview.chromium.org/1011383004

Cr-Commit-Position: refs/heads/master@{#27301}

[turbofan] Remove stale TODO from FunctionTester.

R=titzer@chromium.org
TEST=cctest/test-run-jsexceptions

Review URL: https://codereview.chromium.org/1022463003

Cr-Commit-Position: refs/heads/master@{#27300}

Some CompilationInfo-related cleanup.

Use a delegating constructor for CompilationInfo, reducing duplicated
code. Simplified handling of InlinedFunctionInfos on the way: When we
start compiling, we have bigger things to worry about than a default
vector.

Reduced the usage of a SharedFunctionInfo for compiling, this is a
slighty strange concept.

Review URL: https://codereview.chromium.org/1018853004

Cr-Commit-Position: refs/heads/master@{#27299}

Clarify what APIs return Maybe and MaybeLocal values

If the Maybe is nothing or the MaybeLocal is empty, it means that the
API call either threw an exception or an exception was already pending.

In that case, the embedder needs to handle the exception or otherwise
react to the failed API call.

BUG=v8:3929
R=svenpanne@chromium.org
LOG=y

Review URL: https://codereview.chromium.org/1022803002

Cr-Commit-Position: refs/heads/master@{#27298}

[turbofan] Turn weaken min/max limits for typer into constants.

The weaken limits are constants, there's no point in having a separate
copy of those constants per typer that is initialized per Typer
instance.

R=jarin@chromium.org

Review URL: https://codereview.chromium.org/1008923004

Cr-Commit-Position: refs/heads/master@{#27297}

[arm64] Assembler support for internal references.

BUG=

Review URL: https://codereview.chromium.org/1016073002

Cr-Commit-Position: refs/heads/master@{#27296}

[turbofan] Remove indirection in JSToBoolean/JSUnaryNot lowering.

This reduces the overhead of typed lowering, because we lower
JSToBoolean/JSUnaryNot directly if possible, instead of first lowering
to AnyToBoolean, and then letting the SimplifiedOperatorReducer do the
further lowering.

Also remove some obsolete tests from the cctest suite that have since
been removed by proper unittests. And improve unitttest coverage for the
typed lowering cases.

R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/999173003

Cr-Commit-Position: refs/heads/master@{#27295}

Do not test debug-scripts-request with TurboFan.

This test has been failing flakily on gc-stress, only with --turbo-filter=*

R=jarin@chromium.org

Review URL: https://codereview.chromium.org/1024483002

Cr-Commit-Position: refs/heads/master@{#27294}

Serializer: reorder root list items.

The first 32 root list items can be encoded very efficiently.
By reordering, we can better encode some often referenced root objects.

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1019243002

Cr-Commit-Position: refs/heads/master@{#27293}

Serializer: clean up opcodes.

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1018263002

Cr-Commit-Position: refs/heads/master@{#27292}

Add regression test for dependency to field type tracked weak map.

TBR=jkummerow@chromium.org
BUG=v8:3969
LOG=N

Review URL: https://codereview.chromium.org/1019223002

Cr-Commit-Position: refs/heads/master@{#27291}

Introduce explicit constant for per Context debug data set by embedder

It's value is hardwired in gin[1] and is referenced in Blink[2]. Since it is treated specially by v8 debugger it should be defined in v8 API and referenced in gin and blink..

[1] https://src.chromium.org/viewvc/chrome/trunk/src/gin/public/context_holder.h?r1=239099&r2=239098&pathrev=239099
[2] https://code.google.com/p/chromium/codesearch#chromium/src/third_party/WebKit/Source/bindings/core/v8/V8PerContextData.cpp&q=kDebugIdIndex%20f:V8PerContextData&sq=package:chromium&type=cs

BUG=chromium:466631
LOG=Y

Review URL: https://codereview.chromium.org/1020453002

Cr-Commit-Position: refs/heads/master@{#27290}

Vector ICs: Clear the vector ic slots appropriately during debugging.

When the debugger is active, a host of connected functions get an associated
DebugInfo structure, and their code is copied. The debugger uses the original
code to occasionally patch ICs with their initial value. Although IC learning
can occur, it's thrown away often, depending on the constellation of
breakpoints active or deactivating. Finally, feedback is discarded when the
debugger is turned off.

The type feedback vector needs to be brought into line with this behavior, so
now the debugger clears it's IC slots at appropriate bottlenecks in debug.cc.

R=yangguo@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1014373002

Cr-Commit-Position: refs/heads/master@{#27289}

[turbofan] Eliminatable JS/call nodes should not have a control input.

The control input is only relevant for operations that may "write" (to
prevent hoisting) or "throw" (because they are part of the control chain).

R=jarin@chromium.org

Review URL: https://codereview.chromium.org/1015353004

Cr-Commit-Position: refs/heads/master@{#27288}

PPC: cleanup excludes

Updated excludes to remove SKIPs where possible. Either
removing if no longer needed or replacing with SLOW where
tests just need more time to run

modified:   test/cctest/cctest.status
modified:   test/mjsunit/mjsunit.status

R=danno@chromium.org, svenpanne@chromium.org

BUG=

Review URL: https://codereview.chromium.org/1017073003

Cr-Commit-Position: refs/heads/master@{#27287}

Revert of Reland of "Fix memory leak caused by field type in descriptor array."

TBR=ulan@chromium.org
BUG=v8:3969
LOG=N

Review URL: https://codereview.chromium.org/1020843002

Cr-Commit-Position: refs/heads/master@{#27286}

[turbofan] Support lazy deopt for truncating store to a typed array.

The change introduces a second frame state (for the state before
the operation) for the StoreProperty nodes. If the store writes
into a typed array, the frame state is used for lazy deopt from
the to-number conversion that is performed by the store.

BUG=v8:3963
LOG=n
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/997983004

Cr-Commit-Position: refs/heads/master@{#27285}

Update V8 DEPS.

Rolling v8/buildtools to 3b302fef93f7cc58d9b8168466905237484b2772

Rolling v8/tools/clang to 14dcc71cf53d4d97d0e6e8745089ff81de7b8a94

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/1019133003

Cr-Commit-Position: refs/heads/master@{#27284}

MIPS: Serializer: serialize internal references via object visitor.

Port 7c149afb6c875b1c53723384459dc14a0e961927

BUG=

Review URL: https://codereview.chromium.org/1019083002

Cr-Commit-Position: refs/heads/master@{#27283}

MIPS: Load from PropertyCells using PropertyCell::kValueOffset rather than Cell::kValueOffset

Port dda2bd6f4f952c604ab836f04052684722480849

BUG=

Review URL: https://codereview.chromium.org/1015613003

Cr-Commit-Position: refs/heads/master@{#27282}

MIPS: Disinherit PropertyCell from Cell.

Port 8c0d289772649217a55011133c0e481741618330

BUG=

Review URL: https://codereview.chromium.org/1017103003

Cr-Commit-Position: refs/heads/master@{#27281}

Load from PropertyCells using PropertyCell::kValueOffset rather than Cell::kValueOffset

BUG=

Review URL: https://codereview.chromium.org/1019563002

Cr-Commit-Position: refs/heads/master@{#27280}

Skip flaky mjsunit/debug-references in gc-stress.

BUG=v8:3969
LOG=NO
TBR=hpayer@chromium.org

Review URL: https://codereview.chromium.org/1016063002

Cr-Commit-Position: refs/heads/master@{#27279}

Add IBM people to PPC owners.

Review URL: https://codereview.chromium.org/1014063002

Cr-Commit-Position: refs/heads/master@{#27278}

MIPS: Remove PropertyCell space.

Port 16c8485a35efc36cf6ccd15185282d65412e1502

Original commit message:
Replaces StoreGlobalCell / LoadGlobalCell with NamedField variants that use write barriers.

BUG=

Review URL: https://codereview.chromium.org/1014293002

Cr-Commit-Position: refs/heads/master@{#27277}

Cleanup and unify Isolate::ReportPendingMessages.

Note that this is a pure cleanup CL and shouldn't have an observable
impact on the functional behavior of message reporting.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/999923004

Cr-Commit-Position: refs/heads/master@{#27276}

Serializer: serialize internal references via object visitor.

Review URL: https://codereview.chromium.org/1005183006

Cr-Commit-Position: refs/heads/master@{#27275}

Disinherit PropertyCell from Cell

BUG=

Review URL: https://codereview.chromium.org/1015983003

Cr-Commit-Position: refs/heads/master@{#27274}

MIPS: Remove kind field from StackHandler.

Port 15f8213809a57c2a163b500a732c9ffe5160a41a

Original commit message:
This relands commit 96f79568a926966ebcf0685bf9adc947f4e1fbff.

This makes the Isolate::Throw logic not depend on a prediction of
whether an exception is caught or uncaught. Such a prediction is
inherently undecidable because a finally block can decide between
consuming or re-throwing an exception depending on arbitray control
flow.

There still is a conservative prediction mechanism in place that
components like the debugger or tracing can use for reporting.

With this change we can get rid of the StackHandler::kind field, a
pre-requisite to do table-based lookups of exception handlers.

BUG=

Review URL: https://codereview.chromium.org/1015993002

Cr-Commit-Position: refs/heads/master@{#27273}

Adjust key behaviour for weak collections

R=dslomov@chromium.org
BUG=460083,v8:3970,v8:3971
LOG=Y

Review URL: https://codereview.chromium.org/1018923002

Cr-Commit-Position: refs/heads/master@{#27272}

add interceptors which do not mask existing properties

R=verwaest@chromium.org

BUG=

Review URL: https://codereview.chromium.org/962613002

Cr-Commit-Position: refs/heads/master@{#27271}

convert last api functions which try to handle exceptions to maybes

BUG=v8:3929
LOG=n

Review URL: https://codereview.chromium.org/1015083002

Cr-Commit-Position: refs/heads/master@{#27270}

Remove PropertyCell space
Replaces StoreGlobalCell / LoadGlobalCell with NamedField variants that use write barriers.
BUG=

Review URL: https://codereview.chromium.org/1016803002

Cr-Commit-Position: refs/heads/master@{#27269}

Tweak the TurboFan pipeline for stub compilation.

Review URL: https://codereview.chromium.org/1010673004

Cr-Commit-Position: refs/heads/master@{#27268}

Allow compaction when incremental marking is on.

BUG=chromium:450824
LOG=NO

Review URL: https://codereview.chromium.org/1014263002

Cr-Commit-Position: refs/heads/master@{#27267}

[turbofan] Don't run value numbering with typed lowering.

R=jarin@chromium.org

Review URL: https://codereview.chromium.org/1019443002

Cr-Commit-Position: refs/heads/master@{#27266}

Turn on overapproximation of the weak closure

BUG=v8:3862
R=hpayer@chromium.org
LOG=y

Review URL: https://codereview.chromium.org/1013133002

Cr-Commit-Position: refs/heads/master@{#27265}

CodeCleanup: eliminate unnecessary base class and make the children unvirtual.

I found some strange split in deopt entry points generator.
The code for table entry generator had two classes.
It is safe to join these classes together and drop virtual.

BUG=
LOG=n

Review URL: https://codereview.chromium.org/1010413003

Cr-Commit-Position: refs/heads/master@{#27264}

Remove kind field from StackHandler.

This relands commit 96f79568a926966ebcf0685bf9adc947f4e1fbff.

This makes the Isolate::Throw logic not depend on a prediction of
whether an exception is caught or uncaught. Such a prediction is
inherently undecidable because a finally block can decide between
consuming or re-throwing an exception depending on arbitray control
flow.

There still is a conservative prediction mechanism in place that
components like the debugger or tracing can use for reporting.

With this change we can get rid of the StackHandler::kind field, a
pre-requisite to do table-based lookups of exception handlers.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/997213003

Cr-Commit-Position: refs/heads/master@{#27263}

Make counter and histogram related callbacks part of the Isolate::CreateParams.

Some native counters (e.g. KeyedLoadGenericSlow) are referenced from stubs that are generated very early in the Isolate lifecycle before v8::Isolate::New returns. Thus counter lookup callback also needs to be installed early prior to v8::internal::Isolate::Init call. Otherwise assembler will just assume that the counter is not enabled and produce no code from IncrementCounter - because address of the counter is not yet available.

Histogram related callbacks are moved for consistency to make them able to collect samples which occur at isolate initialization time.

BUG=

Review URL: https://codereview.chromium.org/1010233002

Cr-Commit-Position: refs/heads/master@{#27262}

Make tests pass with weak closure overapproximation enabled

BUG=v8:3862
R=hpayer@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/1014873002

Cr-Commit-Position: refs/heads/master@{#27261}

[turbofan] Cache more common operators.

R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/1015053002

Cr-Commit-Position: refs/heads/master@{#27260}

Merge old data and pointer space.

BUG=

Review URL: https://codereview.chromium.org/1012023002

Cr-Commit-Position: refs/heads/master@{#27259}

CpuProfiler: log pc offset for deopts.

This is the fifth part of https://codereview.chromium.org/1012633002
In this part we collect the offsets of deopt calls and save it into
an inlined function info.

On the Next:
Later when deopt happens we will get the offset of deopt call and
search it among inlined infos.

BUG=chromium:452067
LOG=n

Review URL: https://codereview.chromium.org/1011113004

Cr-Commit-Position: refs/heads/master@{#27258}