Add ObjectTemplate::New() taking FunctionTemplate.

I know the bug has been closed but this seems like a simple addition that may
be useful in other ways as well.

BUG=v8:2180
LOG=N

Review URL: https://codereview.chromium.org/1128553002

Cr-Commit-Position: refs/heads/master@{#28261}

Optimize the typeof operator.

typeof was implemented as a runtime function. Calling it in
optimized code with a non-constant input becomes burdensome.

BUG=

Review URL: https://codereview.chromium.org/1114563003

Cr-Commit-Position: refs/heads/master@{#28260}

[turbofan] Connect non-terminating loops via Terminate.

This revives the Terminate operator and removes the weird Always
operator. As a first step we let the ControlReducer connect non
terminating loops via Terminate. The next step will be to change the
graph builder to insert Terminate nodes into every loop.

Review URL: https://codereview.chromium.org/1123213002

Cr-Commit-Position: refs/heads/master@{#28259}

Manage size of lists used in global handles

Make sure all lists are either regularly cleared or trimmed

R=hpayer@chromium.org
BUG=none
LOG=n

Review URL: https://codereview.chromium.org/1123213003

Cr-Commit-Position: refs/heads/master@{#28258}

Implement IdentityMap<V>, a robust, GC-safe object-identity HashMap.

R=hpayer@chromium.org, erikcorry@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1105693002

Cr-Commit-Position: refs/heads/master@{#28257}

Ship string unicode escapes

R=yangguo@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1130723002

Cr-Commit-Position: refs/heads/master@{#28256}

Remove GetDefaultReceiver, pass in undefined to sloppy-mode functions instead.

Review URL: https://codereview.chromium.org/1116003005

Cr-Commit-Position: refs/heads/master@{#28255}

Fix FreeBSD build.

Fix the following build error:

    ../deps/v8/src/compiler/graph-visualizer.h:23:1:
    error: 'FILE' does not name a type

See https://github.com/iojs/io.js/issues/1637 for background.

BUG=v8:4079
LOG=N

Review URL: https://codereview.chromium.org/1128053002

Cr-Commit-Position: refs/heads/master@{#28254}

[strong] Fix inlining issue

The Hydrogen representation for binops was never changed to care about the
language mode. We thought this was ok, but it turns out we need to keep track
of it to make sure inlining doesn't mess with the "strongness" of binops.

Also added more rigorous inlining testing.

BUG=v8:3956
LOG=N

Review URL: https://codereview.chromium.org/1123043002

Cr-Commit-Position: refs/heads/master@{#28253}

Implement a 'trial parse' step, that will abort pre-parsing excessively
long and trivial functions, so that they can be eagerly compiled after all.
This essentially allows the parser to renege on its earlier decision to
lazy-parse, if additional information suggests it was a bad decision.

BUG=chromium:470930
LOG=Y

Review URL: https://codereview.chromium.org/1102523003

Cr-Commit-Position: refs/heads/master@{#28252}

[turbofan] Add support for advanced reducers.

An AdvancedReducer is basically a regular Reducer with an editor
that can perform graph editing operations beyond changing or
replacing the node that is currently being reduced. The GraphReducer
is the default implementation of the AdvancedReducer::Editor interface.

The ControlReducerImpl is now just an AdvancedReducer, which
temporarily requires a Finish method in the reducer to implement
the dead node trimming until we move that to the GraphReducer
(which in turn requires that all loops are connected to End).

Review URL: https://codereview.chromium.org/1122423003

Cr-Commit-Position: refs/heads/master@{#28251}

Migrate error messages, part 8.

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1118273004

Cr-Commit-Position: refs/heads/master@{#28250}

[test] Mark test as flaky.

TBR=yangguo@chromium.org
NOTRY=true

Review URL: https://codereview.chromium.org/1125123002

Cr-Commit-Position: refs/heads/master@{#28249}

Mark mjsunit/allocation-site-info.js as NO_VARIANTS.

NOTRY=true
TBR=machenbach@chromium.org
BUG=4078
LOG=N

Review URL: https://codereview.chromium.org/1123183003

Cr-Commit-Position: refs/heads/master@{#28248}

Revert of [V8] Reland https://codereview.chromium.org/1100993003/ (patchset #1 id:1 of https://codereview.chromium.org/1121833003/)

Reason for revert:
[Sheriff] Speculative revert for breaking layout tests - will reland if it doesn't help:
http://build.chromium.org/p/client.v8/builders/V8-Blink%20Linux%2064/builds/3178

Original issue's description:
> [V8] Reland https://codereview.chromium.org/1100993003/
>
> Chromium tests were prepared in https://codereview.chromium.org/1118743003/
> Blink tests in https://codereview.chromium.org/1115193002/
>
> LOG=Y
> BUG=chromium:480652
> R=yangguo@chromium.org
>
> Committed: https://crrev.com/19b62e211d1829108e5c85c5043dd755afa17178
> Cr-Commit-Position: refs/heads/master@{#28229}

TBR=yangguo@chromium.org,kozyatinskiy@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:480652

Review URL: https://codereview.chromium.org/1125113003

Cr-Commit-Position: refs/heads/master@{#28247}

X87: Handle the case when derived constructor is [[Call]]ed with 0 args.

port cf53fed972896bf23c037ce7ac9f8e1512463c62 (r28242).

original commit message:

    ArgumentsAdaptorStub for derived constructor (the one that needs
    new.target) works in this way:
     - If the constructor is invoked via the Construct stub, we know that
       actual arguments always include new.target. ``arguments`` object
       however should not include a new.target, therefore we remove it.
       We achieve this by decrementing the argument count.
     - If the constructor is invoked as a call, we do not care for a correct
       ``arguments`` array since the constructor will immediately throw on
       entrance.
    The bug is that the call could actually pass 0 actual arguments, but I
    decrement unconditionally :(. The fix is to detect this case and avoid
    decrementing. ``arguments`` is bogus, but it is ok as constructor
    throws.

    Long-term we should just remove mucking about with arguments for
    new.target and just get it from the stack.

BUG=

Review URL: https://codereview.chromium.org/1124063002

Cr-Commit-Position: refs/heads/master@{#28246}

Update V8 DEPS.

Rolling v8/tools/clang to 8ed88fc35e20182d6ecc3dde107606f609c46b23

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/1130663002

Cr-Commit-Position: refs/heads/master@{#28245}

Disable RunAllocate test case for turbofan unsupported platform.

  "RunAllocate" test case is added in d6945db7. it invokes Linkage::GetStubCallDescriptor
   which is unimplemented for turbofan unsupported platform.

BUG=

Review URL: https://codereview.chromium.org/1129483003

Cr-Commit-Position: refs/heads/master@{#28244}

Adjust the visibility of the standalone targets in the GN build.

As part of the migration from GYP->GN, we want to make sure that we
can track when new targets are added to either the GYP or GN builds
and that we are building everything we expect to build.

In GN, unlike GYP, if a build file gets referenced from other files,
building 'all' will cause every target to be built in it. This means in
particular, that we can end up trying to build targets that are not
necessarily intended to be visible to the rest of the build. To get
around this, any target that is defined but hidden (like 'v8_snapshot',
in V8's case) should still be visible to a top-level target called
"//:gn_visibility".

R=brettw@chromium.org, machenbach@chromium.org
BUG=461019
LOG=N

Review URL: https://codereview.chromium.org/1120093005

Cr-Commit-Position: refs/heads/master@{#28243}

Handle the case when derived constructor is [[Call]]ed with 0 args.

ArgumentsAdaptorStub for derived constructor (the one that needs
new.target) works in this way:
 - If the constructor is invoked via the Construct stub, we know that
   actual arguments always include new.target. ``arguments`` object
   however should not include a new.target, therefore we remove it.
   We achieve this by decrementing the argument count.
 - If the constructor is invoked as a call, we do not care for a correct
   ``arguments`` array since the constructor will immediately throw on
   entrance.
The bug is that the call could actually pass 0 actual arguments, but I
decrement unconditionally :(. The fix is to detect this case and avoid
decrementing. ``arguments`` is bogus, but it is ok as constructor
throws.

Long-term we should just remove mucking about with arguments for
new.target and just get it from the stack.

R=arv@chromium.org,rossberg@chromium.org
BUG=chromium:474783
LOG=Y

Review URL: https://codereview.chromium.org/1126783003

Cr-Commit-Position: refs/heads/master@{#28242}

MIPS: Fix long branch mode and FPU branches.

- Add long branche mode for FPU branches.
- Fix FPU branches for unordered conditions.
- Provide FPU [un]ordered condition negation schema.

TEST=mjsunit/miror-objects, constant-folding-2, external-array
BUG=

Review URL: https://codereview.chromium.org/1120753010

Cr-Commit-Position: refs/heads/master@{#28241}

MIPS: Improve '[es6] When comparing two symbols we may need to throw a TypeError'.

Improve d26f5d392317dcf1eef3693eadfa1b95a8850df8

Original commit message:
When comparing a symbol to istself using <, <=, > or >= we need to
throw a TypeError. This is correctly handled in the runtime function
so if we are comparing a symbol fall back to use the runtime.

BUG=

Review URL: https://codereview.chromium.org/1124863004

Cr-Commit-Position: refs/heads/master@{#28240}

PPC: [turbofan] Fix tail call optimization.

Port d21de2a48bc63823876d3f9bf52753f9da0aea6b

Original commit message:
Tail calls are matched on the graph, with a dedicated tail call
optimization that is actually testable. The instruction selection can
still fall back to a regular if the platform constraints don't allow to
emit a tail call (i.e. the return locations of caller and callee differ
or the callee takes non-register parameters, which is a restriction that
will be removed in the future).

Also explicitly limit tail call optimization to stubs for now and drop
the global flag.

R=dstence@us.ibm.com, michael_dawson@ca.ibm.com
BUG=

Review URL: https://codereview.chromium.org/1114193003

Cr-Commit-Position: refs/heads/master@{#28239}

Revert of Resolve references to "this" the same way as normal variables (patchset #11 id:240001 of https://codereview.chromium.org/1097283003/)

Reason for revert:
nosnap failures

Original issue's description:
> Resolve references to "this" the same way as normal variables
>
> Make the parser handle references to "this" as unresolved variables, so the
> same logic as for the rest of function parameters is used for the receiver.
> Minor additions to the code generation handle copying the receiver to the
> context, along with the rest of the function parameters.
>
> Based on work by Adrian Perez de Castro <aperez@igalia.com>.
>
> BUG=
> LOG=N
>
> Committed: https://crrev.com/18619d355192e2699203d12d9ebb9caea107b693
> Cr-Commit-Position: refs/heads/master@{#28236}

TBR=rossberg@chromium.org,mstarzinger@chromium.org,dslomov@chromium.org,adamk@chromium.org,arv@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=

Review URL: https://codereview.chromium.org/1113133006

Cr-Commit-Position: refs/heads/master@{#28238}

Revert of Collect type feedback on result of Math.[round|ceil|floor] (patchset #13 id:230001 of https://codereview.chromium.org/1053143005/)

Reason for revert:
All sorts of performance regressions

Original issue's description:
> Collect type feedback on result of Math.[round|ceil|floor]
>
> By recording invocations of these builtins that can return -0, we now learn to not emit Crankshaft code that only handles integer results, avoiding deopt loops.
>
> Committed: https://crrev.com/f36ecaf3a4d61568ca50a20718acce7dd5da9a5f
> Cr-Commit-Position: refs/heads/master@{#28215}

TBR=mvstanton@chromium.org,bmeurer@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review URL: https://codereview.chromium.org/1115973005

Cr-Commit-Position: refs/heads/master@{#28237}

Resolve references to "this" the same way as normal variables

Make the parser handle references to "this" as unresolved variables, so the
same logic as for the rest of function parameters is used for the receiver.
Minor additions to the code generation handle copying the receiver to the
context, along with the rest of the function parameters.

Based on work by Adrian Perez de Castro <aperez@igalia.com>.

BUG=
LOG=N

Review URL: https://codereview.chromium.org/1097283003

Cr-Commit-Position: refs/heads/master@{#28236}

Revert of [handles] Sanitize Handle and friends. (patchset #4 id:50001 of https://codereview.chromium.org/1128533002/)

Reason for revert:
[Sheriff] Breaks mac asan:
http://build.chromium.org/p/client.v8/builders/V8%20Mac64%20ASAN/builds/1198

Original issue's description:
> [handles] Sanitize Handle and friends.
>
> This almost allows us to get rid of handles-inl.h, once we fix the
> cyclic dependencies (follow up CL).
>
> R=yangguo@chromium.org
>
> Committed: https://crrev.com/3283195d0408333cce552cf4087577e6f41054e5
> Cr-Commit-Position: refs/heads/master@{#28222}

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

Review URL: https://codereview.chromium.org/1128543004

Cr-Commit-Position: refs/heads/master@{#28235}

freed_nodes in global-handles should be addititive

otherwise phantom nodes won't ever be taken into account

BUG=chromium:484671,chromium:479796
R=hpayer@chromium.org
LOG=y

Review URL: https://codereview.chromium.org/1125673010

Cr-Commit-Position: refs/heads/master@{#28234}

Be sure to abort incremental GC when trying to reserve space

R=hpayer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1124553007

Cr-Commit-Position: refs/heads/master@{#28233}

GrowArrayElementsStub must save caller doubles for double ElementsKind.

BUG=
R=danno@chromium.org

Review URL: https://codereview.chromium.org/1124653004

Cr-Commit-Position: refs/heads/master@{#28232}

[grokdump] Update v8heapconst.py

Review URL: https://codereview.chromium.org/1125543004

Cr-Commit-Position: refs/heads/master@{#28231}

[d8] Make Realm.dispose() trigger a ContextDisposedNotification

Review URL: https://codereview.chromium.org/1129543002

Cr-Commit-Position: refs/heads/master@{#28230}

[V8] Reland https://codereview.chromium.org/1100993003/

Chromium tests were prepared in https://codereview.chromium.org/1118743003/
Blink tests in https://codereview.chromium.org/1115193002/

LOG=Y
BUG=chromium:480652
R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/1121833003

Cr-Commit-Position: refs/heads/master@{#28229}

[tick processor] Introduce --pairwise-timed-range processing mode

Review URL: https://codereview.chromium.org/1123883002

Cr-Commit-Position: refs/heads/master@{#28228}

[test] Add arm perf trybot to runner script.

BUG=chromium:484589
LOG=n
NOTRY=true
TBR=hpayer@chromium.org

Review URL: https://codereview.chromium.org/1123473007

Cr-Commit-Position: refs/heads/master@{#28227}

[es6] When comparing two symbols we may need to throw a TypeError

When comparing a symbol to istself using <, <=, > or >= we need to
throw a TypeError. This is correctly handled in the runtime function
so if we are comparing a symbol fall back to use the runtime.

BUG=v8:4073
LOG=Y
R=rossberg@chromium.org

Review URL: https://codereview.chromium.org/1125783002

Cr-Commit-Position: refs/heads/master@{#28226}

Reduce the size of initial JSON string buffer.

BUG=v8:3925
LOG=NO

Review URL: https://codereview.chromium.org/1123013002

Cr-Commit-Position: refs/heads/master@{#28225}

Do not cache object literal maps during bootstrapping.

We do not expect to share maps for object literals created by builtin
Javascript code.

R=ishell@chromium.org

Review URL: https://codereview.chromium.org/1127693009

Cr-Commit-Position: refs/heads/master@{#28224}

Move more parts of stack trace formatting to runtime.

R=ishell@chromium.org

Review URL: https://codereview.chromium.org/1122973002

Cr-Commit-Position: refs/heads/master@{#28223}

[handles] Sanitize Handle and friends.

This almost allows us to get rid of handles-inl.h, once we fix the
cyclic dependencies (follow up CL).

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/1128533002

Cr-Commit-Position: refs/heads/master@{#28222}

When adding constant string with something unknown, assume it's a string.

Review URL: https://codereview.chromium.org/1122823002

Cr-Commit-Position: refs/heads/master@{#28221}

Fix bug in "Migrate error messages, part 7".

TBR=mvstanton@chromium.org
NOTRY=true

Review URL: https://codereview.chromium.org/1124583006

Cr-Commit-Position: refs/heads/master@{#28220}

[turbofan] Fix tail call optimization.

Tail calls are matched on the graph, with a dedicated tail call
optimization that is actually testable. The instruction selection can
still fall back to a regular if the platform constraints don't allow to
emit a tail call (i.e. the return locations of caller and callee differ
or the callee takes non-register parameters, which is a restriction that
will be removed in the future).

Also explicitly limit tail call optimization to stubs for now and drop
the global flag.

BUG=v8:4076
LOG=n

Review URL: https://codereview.chromium.org/1114163005

Cr-Commit-Position: refs/heads/master@{#28219}

Add heap growing strategy details to --trace-gc-verbose.

BUG=

Review URL: https://codereview.chromium.org/1128483003

Cr-Commit-Position: refs/heads/master@{#28218}

Reland #2 "Wrap v8natives.js into a function."

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/1127693006

Cr-Commit-Position: refs/heads/master@{#28217}

Migrate error messages, part 7.

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1122033002

Cr-Commit-Position: refs/heads/master@{#28216}

Collect type feedback on result of Math.[round|ceil|floor]

By recording invocations of these builtins that can return -0, we now learn to not emit Crankshaft code that only handles integer results, avoiding deopt loops.

Review URL: https://codereview.chromium.org/1053143005

Cr-Commit-Position: refs/heads/master@{#28215}

[c++11] Remove remnants of pre-C++11 delete, final and override.

These macros are not needed anymore, so there's no point in supporting
them.

Review URL: https://codereview.chromium.org/1123723003

Cr-Commit-Position: refs/heads/master@{#28214}

Correctly accept already preprocessed stack traces.

When we preprocess stack traces, we turn code pointer and offset to
source position, and store it in place of code pointer as smi.

Preprocessing stack traces is currently disabled due to issue 4065.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/1125723002

Cr-Commit-Position: refs/heads/master@{#28213}

Fix another may not be init compile error on AIX

AIX compiler is complaining that variable may not be initialized

modified:   src/heap/heap.cc

R=danno@chromium.org, svenpanne@chromium.org, mbrandy@us.ibm.com

BUG=

Review URL: https://codereview.chromium.org/1123743002

Cr-Commit-Position: refs/heads/master@{#28212}

Make CPU profiler do not hog 100% of CPU.

Tick event processor should not stay in a tight loop
when there's nothing to do. It can go sleep until next sample event.

LOG=N
BUG=v8:3967

Review URL: https://codereview.chromium.org/1118533003

Cr-Commit-Position: refs/heads/master@{#28211}

comment typo fix

Review URL: https://codereview.chromium.org/1120373003

Cr-Commit-Position: refs/heads/master@{#28210}

PPC: Fix 'Remove materialized objects on stack unwind.'

R=mstarzinger@chromium.org, jarin@chromium.org, dstence@us.ibm.com, michael_dawson@ca.ibm.com
BUG=

Review URL: https://codereview.chromium.org/1123793002

Cr-Commit-Position: refs/heads/master@{#28209}

Revert of Reland "Wrap v8natives.js into a function." (patchset #2 id:20001 of https://codereview.chromium.org/1123703002/)

Reason for revert:
[Sheriff] Speculative revert for braking arm64 nosnap:
http://build.chromium.org/p/client.v8/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20nosnap%20-%20debug%20-%202/builds/2314

(reverted already titzer's CL which didn't help)

Original issue's description:
> Reland "Wrap v8natives.js into a function."
>
> Committed: https://crrev.com/72ab42172979b60a1b784ea0c6a495d7ee2bba67
> Cr-Commit-Position: refs/heads/master@{#28193}

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

Review URL: https://codereview.chromium.org/1127543003

Cr-Commit-Position: refs/heads/master@{#28208}

Revert of Revert of [turbofan] Small fixes to PipelineStatistics. Collect statistics in tests. (patchset #1 id:1 of https://codereview.chromium.org/1119963003/)

Reason for revert:
Revert didn't fix the problem. Maybe it was another CL.

Original issue's description:
> Revert of [turbofan] Small fixes to PipelineStatistics. Collect statistics in tests. (patchset #1 id:1 of https://codereview.chromium.org/1122753002/)
>
> Reason for revert:
> [Sheriff] Speculative revert for breaking arm64 nosnap:
> http://build.chromium.org/p/client.v8/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20nosnap%20-%20debug%20-%202/builds/2314
>
> Original issue's description:
> > [turbofan] Small fixes to PipelineStatistics. Collect statistics in tests.
> >
> > R=bmeurer@chromium.org
> > BUG=
> >
> > Committed: https://crrev.com/9833844268357590243732a53e2a7fc0dc5fcc9e
> > Cr-Commit-Position: refs/heads/master@{#28192}
>
> TBR=bmeurer@chromium.org,titzer@chromium.org
> NOPRESUBMIT=true
> NOTREECHECKS=true
> NOTRY=true
> BUG=
>
> Committed: https://crrev.com/d0ed8940fc6c995eba57149b8e8cb214001e8381
> Cr-Commit-Position: refs/heads/master@{#28205}

TBR=bmeurer@chromium.org,titzer@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=

Review URL: https://codereview.chromium.org/1126603002

Cr-Commit-Position: refs/heads/master@{#28207}

Remove materialized objects on stack unwind.

BUG=v8:3985
LOG=n
R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/1122083002

Cr-Commit-Position: refs/heads/master@{#28206}

Revert of [turbofan] Small fixes to PipelineStatistics. Collect statistics in tests. (patchset #1 id:1 of https://codereview.chromium.org/1122753002/)

Reason for revert:
[Sheriff] Speculative revert for breaking arm64 nosnap:
http://build.chromium.org/p/client.v8/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20nosnap%20-%20debug%20-%202/builds/2314

Original issue's description:
> [turbofan] Small fixes to PipelineStatistics. Collect statistics in tests.
>
> R=bmeurer@chromium.org
> BUG=
>
> Committed: https://crrev.com/9833844268357590243732a53e2a7fc0dc5fcc9e
> Cr-Commit-Position: refs/heads/master@{#28192}

TBR=bmeurer@chromium.org,titzer@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=

Review URL: https://codereview.chromium.org/1119963003

Cr-Commit-Position: refs/heads/master@{#28205}

[test] Skip some flaky layout tests.

NOTRY=true
TBR=jochen@chromium.org

Review URL: https://codereview.chromium.org/1126433004

Cr-Commit-Position: refs/heads/master@{#28204}

PPC: Only swap undefined for the global object if necessary in the prologue

Port fab3508062396c2768b5dcfa0fd7d40eba527d62

R=verwaest@chromium.org, dstence@us.ibm.com, michael_dawson@ca.ibm.com
BUG=

Review URL: https://codereview.chromium.org/1122773002

Cr-Commit-Position: refs/heads/master@{#28203}

Fix smi scanning

BUG=chromium:483176
LOG=n

Review URL: https://codereview.chromium.org/1114073003

Cr-Commit-Position: refs/heads/master@{#28202}

MIPS: Skip test-simplified-lowering/RunNumberDivide_2_TruncatingToUint32.

This test fails on board, temporarily skip failing test until we resolve this issue.

NOTRY=true

Review URL: https://codereview.chromium.org/1127573002

Cr-Commit-Position: refs/heads/master@{#28201}

Only swap undefined for the global object if necessary in the prologue

BUG=

Review URL: https://codereview.chromium.org/1120093002

Cr-Commit-Position: refs/heads/master@{#28200}

New insertion write barrier.

BUG=

Review URL: https://codereview.chromium.org/1073953006

Cr-Commit-Position: refs/heads/master@{#28199}

[turbofan] Free TurboFan from the claws of ZoneList.

This removes the last occurence of the hand-written ZoneList from the
compiler directory and uses ZoneVector everywhere instead.

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/1120083004

Cr-Commit-Position: refs/heads/master@{#28198}

[release-tools] Tool to find related commits

usage: This tool analyzes the commit range between <of> and <until>. It finds commits which belong together e.g. Implement/Revert pairs and Implement/Port/Revert triples. All supplied hashes need to be from the same branch e.g. master.

Example for M42: ./search_related_commits.py --prettyprint --separator e0110920d6f98f0ba2ac0d680f635ae3f094a04e b856e8785933a2a9cd884ab8966fee0e7098927e b1c2a3495624a9776c7df865d972886f2d078c10

BUG=
NOTRY=true

Review URL: https://codereview.chromium.org/1098123002

Cr-Commit-Position: refs/heads/master@{#28197}

Don't perform marking barrier for weak cell values.

BUG=

Review URL: https://codereview.chromium.org/1124583003

Cr-Commit-Position: refs/heads/master@{#28196}

[turbofan] Add SimplifiedOperator::Allocate operator.

This introduces a simplified allocation operator which can be used to
model inline allocations in TurboFan. It is currently used for context
allocations, but still disabled because change lowering introduces
floating allocations outside the effect chain that interfere.

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/1109773002

Cr-Commit-Position: refs/heads/master@{#28195}

MIPS: Do not use the 64-bit floor and truncate instructions in fp32 mode.

These instructions have unpredictable result when the processor is in fp32 mode.

BUG=

TEST=mjsunit/math-floor-global,math-floor-local,math-floor-part1

Review URL: https://codereview.chromium.org/1116073002

Cr-Commit-Position: refs/heads/master@{#28194}

Reland "Wrap v8natives.js into a function."

Review URL: https://codereview.chromium.org/1123703002

Cr-Commit-Position: refs/heads/master@{#28193}

[turbofan] Small fixes to PipelineStatistics. Collect statistics in tests.

R=bmeurer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1122753002

Cr-Commit-Position: refs/heads/master@{#28192}

Extract Signature from src/compiler/machine-type.h to src/signature.h

R=bmeurer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1118823003

Cr-Commit-Position: refs/heads/master@{#28191}

Set inferred name of bound function to empty string.

Otherwise it's whatever the js minifier assigns it to.

R=jkummerow@chromium.org
BUG=chromium:484077
LOG=N

Review URL: https://codereview.chromium.org/1122733002

Cr-Commit-Position: refs/heads/master@{#28190}

X87: VectorICs: built-in function apply should use an IC.

port 83a0af55009548238a2ff6ef71c3387a49da0a88 (r28165).

original commit message:

   VectorICs: built-in function apply should use an IC.

   Handled a TODO that sent builtin function apply to the runtime on property get.

BUG=

Review URL: https://codereview.chromium.org/1119263002

Cr-Commit-Position: refs/heads/master@{#28189}

X87: Use a stub in crankshaft for grow store arrays.

port fb8e6136387339e8089bbf61d4ffee50e4cd957e (r28163).

original commit message:

 We were deopting without learning anything.

 This is a rebase/reland of https://codereview.chromium.org/368263003

BUG=

Review URL: https://codereview.chromium.org/1125623002

Cr-Commit-Position: refs/heads/master@{#28188}

Revert of [V8] Use previous token location as EOS token location (patchset #2 id:20001 of https://codereview.chromium.org/1100993003/)

Reason for revert:
[Sheriff] Speculative revert. This seems to block the current roll:
https://codereview.chromium.org/1124463003/

This bisect also points at this CL:
https://codereview.chromium.org/1124523002/

Please prepare the chromium side tests before a reland.

Original issue's description:
> [V8] Use previous token location as EOS token location
>
> EOS token location is useless for users and messages.js are not ready for its location.
> With this CL we use location of token before EOS for it.
>
> LOG=Y
> BUG=chromium:480652
> R=yurys@chromium.org,yangguo@chromium.org
>
> Committed: https://crrev.com/81afc9313ce84350bcba9f84b255a77e97cd3726
> Cr-Commit-Position: refs/heads/master@{#28164}

TBR=yangguo@chromium.org,yurys@chromium.org,kozyatinskiy@chromium.org
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true
BUG=chromium:480652

Review URL: https://codereview.chromium.org/1116233004

Cr-Commit-Position: refs/heads/master@{#28187}

Update V8 DEPS.

Rolling v8/tools/clang to a8549b6dd82c3b35a06f9894fa7abdcbbf0bd6f5

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/1120243002

Cr-Commit-Position: refs/heads/master@{#28186}

Most of the issues so far encountered with the greedy allocator rest with the splitting mechanism. This change has:

- all RegisterAllocatorTest unit tests passing, when unittest has the flags: --turbo-filter=* --always-opt --turbo-deoptimization --turbo-verify-allocation --turbo_greedy_regalloc

./tools/run-tests.py passing when providing --turbo_greedy_regalloc, but still having some failing (more than the "normal" 43) when passing all the flags. I realize just passing --turbo_greedy_regalloc does not provide full coverage, but it did uncover some bugs, still

The CL centralizes the computing of split points (for "losing" intervals) with the determination of whether an interval is irreducible, and, therefore, of infinite spill weight (if an interval can't be split or spilled, it can't lose in weight comparison because there's nothing we can do to it and make progress).

There are allocator efficiency opportunities I haven't yet taken (this refers to the code itself, not the generated code). For example, the above split/spill computation can be cached. My plan is to defer this to at least after we see numbers showing the value of the algorithm, and potentially do at the same time we remove the linear algorithm, if that is still the plan. At that time, we can look where some APIs best belong (e.g. weight computation may fully live and cache itself on LiveRange)

In addition, the CL adds a few debug-time-only Print APIs I found very useful:  on demand (while in GDB) dump of the code, live range info (use positions & operand description, and intervals), etc.

BUG=

Review URL: https://codereview.chromium.org/1105793002

Cr-Commit-Position: refs/heads/master@{#28185}

Update V8 DEPS.

Rolling v8/third_party/icu to f8c0e585b0a046d83d72b5d37356cb50d5b2031a

Rolling v8/tools/clang to 15ea3e6cc23fd6ff5545bd939ee153b97807668d

TBR=machenbach@chromium.org

Review URL: https://codereview.chromium.org/1117163003

Cr-Commit-Position: refs/heads/master@{#28184}

MIPS: Fix 'Detect simple tail calls'.

4b122b7504ab458faa4132c338809b40147751bf

BUG=

Review URL: https://codereview.chromium.org/1116913002

Cr-Commit-Position: refs/heads/master@{#28183}

Revert of Wrap v8natives.js into a function. (patchset #2 id:20001 of https://codereview.chromium.org/1109343004/)

Reason for revert:
[Sheriff] Speculative revert for breaking layout tests, e.g.:
http://build.chromium.org/p/client.v8/builders/V8-Blink%20Linux%2064%20%28dbg%29/builds/2682

See. e.g.:
https://storage.googleapis.com/chromium-layout-test-archives/V8-Blink_Win/3130/layout-test-results/http/tests/websocket/workers/worker-reload-diff.txt

Original issue's description:
> Wrap v8natives.js into a function.
>
> R=jkummerow@chromium.org
>
> Committed: https://crrev.com/ee1b39b4303829e6c6805fe8b2f2602b13f6463a
> Cr-Commit-Position: refs/heads/master@{#28174}

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

Review URL: https://codereview.chromium.org/1121453003

Cr-Commit-Position: refs/heads/master@{#28182}

MIPS: Add min/max suffixed variants.

TEST=
BUG=

Review URL: https://codereview.chromium.org/1118693002

Cr-Commit-Position: refs/heads/master@{#28181}

Migrate error messages, part 6. (string.js and date.js)

R=mvstanton@chromium.org

Review URL: https://codereview.chromium.org/1116953002

Cr-Commit-Position: refs/heads/master@{#28180}

Remove unused Module-related AST nodes and associated codegen

BUG=v8:1569
LOG=n

Review URL: https://codereview.chromium.org/1106383008

Cr-Commit-Position: refs/heads/master@{#28179}

PPC: VectorICs: built-in function apply should use an IC.

Port 83a0af55009548238a2ff6ef71c3387a49da0a88

Original commit message:
Handled a TODO that sent builtin function apply to the runtime on property get.

R=mvstanton@chromium.org, dstence@us.ibm.com, michael_dawson@ca.ibm.com
BUG=

Review URL: https://codereview.chromium.org/1118713002

Cr-Commit-Position: refs/heads/master@{#28178}

PPC: Use a stub in crankshaft for grow store arrays.

Port fb8e6136387339e8089bbf61d4ffee50e4cd957e

Original commit message:
We were deopting without learning anything.

This is a rebase/reland of https://codereview.chromium.org/368263003

BUG=
R=mvstanton@chromium.org, dstence@us.ibm.com, michael_dawson@ca.ibm.com

Review URL: https://codereview.chromium.org/1113063002

Cr-Commit-Position: refs/heads/master@{#28177}

Cache experimental natives sources as external strings.

R=ulan@chromium.org
BUG=v8:4054
LOG=N

Review URL: https://codereview.chromium.org/1114043002

Cr-Commit-Position: refs/heads/master@{#28176}

Disable stack trace preprocessing.

This has to wait until issue 4065 has been fixed.

R=ulan@chromium.org
BUG=v8:2340, v8:4065, chromium:482312
LOG=N

Review URL: https://codereview.chromium.org/1110393002

Cr-Commit-Position: refs/heads/master@{#28175}

Wrap v8natives.js into a function.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/1109343004

Cr-Commit-Position: refs/heads/master@{#28174}

[test] Remove pesky MachineCallHelper from inheritance chain.

R=titzer@chromium.org

Review URL: https://codereview.chromium.org/1109853003

Cr-Commit-Position: refs/heads/master@{#28173}

Allow TurboFan to compile more methods.

Reorganize some bailout conditions to be after the decision to use TurboFan.

R=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1110573002

Cr-Commit-Position: refs/heads/master@{#28172}

Mark instruction blocks with spills (for frame elision).

Review URL: https://codereview.chromium.org/1119683002

Cr-Commit-Position: refs/heads/master@{#28171}

Initialize typed array content for rempio2result during bootstrapping.

This is to pacify MSAN.

R=mvstanton@chromium.org
BUG=465759
LOG=N

Review URL: https://codereview.chromium.org/1119693004

Cr-Commit-Position: refs/heads/master@{#28170}

Fix typo in builtins-x87, introduced in crrev.com/1107233004.

BUG=470930
LOG=N
R=jochen@chromium.org

Review URL: https://codereview.chromium.org/1116933002

Cr-Commit-Position: refs/heads/master@{#28169}

Unify internal and external typed arrays a bit

Just give internal ones an ArrayBuffer with a NULL backing store. This
simplifies the access checks a lot.

BUG=v8:3996
R=hpayer@chromium.org,verwaest@chromium.org
LOG=n

Review URL: https://codereview.chromium.org/1109353003

Cr-Commit-Position: refs/heads/master@{#28168}

[test] make instruction sequence test emit cfgs more like the scheduler

BUG=

Review URL: https://codereview.chromium.org/1116923002

Cr-Commit-Position: refs/heads/master@{#28167}

[test] Mark test as flaky.

BUG=v8:3803
LOG=n
TBR=mstarzinger@chromium.org
NOTRY=true

Review URL: https://codereview.chromium.org/1112373002

Cr-Commit-Position: refs/heads/master@{#28166}

VectorICs: built-in function apply should use an IC.

Handled a TODO that sent builtin function apply to the runtime on property get.

R=yangguo@chromium.org
BUG=

Review URL: https://codereview.chromium.org/1120613002

Cr-Commit-Position: refs/heads/master@{#28165}

[V8] Use previous token location as EOS token location

EOS token location is useless for users and messages.js are not ready for its location.
With this CL we use location of token before EOS for it.

LOG=Y
BUG=chromium:480652
R=yurys@chromium.org,yangguo@chromium.org

Review URL: https://codereview.chromium.org/1100993003

Cr-Commit-Position: refs/heads/master@{#28164}

Use a stub in crankshaft for grow store arrays.

We were deopting without learning anything.

This is a rebase/reland of https://codereview.chromium.org/368263003

BUG=v8:3417
LOG=N

Review URL: https://codereview.chromium.org/1109333003

Cr-Commit-Position: refs/heads/master@{#28163}

Implement kToBeExecutedOnceCodeAge.

An initial 'code age' state that will turn into a 'pre-aging' code age only after it was executed the first time.

BUG=470930
LOG=Y

Review URL: https://codereview.chromium.org/1107233004

Cr-Commit-Position: refs/heads/master@{#28162}