Add unit test for regression in GVN caused by field type tracking.

BUG=v8:3347
LOG=n
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/333273004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21858 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "GVN fix, preventing loads hoisting above stores to the same field when HObjectAccess's representation is not the same."

This reverts commit r21830 for tanking performance on Deltablue.

TBR=ishell@chromium.org

Review URL: https://codereview.chromium.org/336223002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21857 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Reuse AstValueFactory when optimizing.

HOptimizedGraphBuilder::TryInline creates a temporary CompilationInfo and
destroys it, but we don't want the AstValueFactory to be destroyed at the same
time. Reuse the upper CompilationInfo's AstValueFactory.

BUG=
R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/334173003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21851 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix representation of Phis for mutable-heapnumber-in-object-literal properties

BUG=v8:3392
LOG=y
R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/328343004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21850 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix x87 compile

TBR=marja@chormium.org
LOG=n
BUG=none

Review URL: https://codereview.chromium.org/337713003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21847 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove dependency on Vector from platform files

Add wrappers to utils.h instead.

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

Review URL: https://codereview.chromium.org/328343003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21846 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix windows build (size_t vs. int) some more.

TBR=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/336803002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21845 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix windows build (size_t vs. int).

Sadly, Vector ctor takes an argument "int length".

TBR=dcarney@chromium.org, mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/334903002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21844 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix gn build.

TBR=mstarzinger@chromium.org
BUG=

Review URL: https://codereview.chromium.org/338553002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21843 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix clang build after r21839

R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/330793003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21842 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Parser: Delay internalizing strings and values.

This is needed so that we can run Parser on a non-main thread (independent
of the Isolate and the V8 heap).

BUG=
R=rossberg@chromium.org

Review URL: https://codereview.chromium.org/314603004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21841 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Optimize Function.prototype.call

- May inline the function, or call it directly, instead of going through call
- Supports arguments object escaping when it escapes to builtins (preparation for slice.call(arguments, ...) optimization)
- Both .call and .apply now support inlining when calling builtins indirectly

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

Review URL: https://codereview.chromium.org/335683002

Patch from Petka Antonov <p.antonov@partner.samsung.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21840 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Have one, long-lived map for bound functions.

This avoids creating a new map for every bound function. Bonus: some cleanup in Runtime_FunctionBindArguments.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/335653002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21839 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Keep maps when resetting elements.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/335553004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21838 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Reland "Rename kIs64BitArch with kRequiresCodeRange."

R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/331823002

Patch from Weiliang Lin <weiliang.lin@intel.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21837 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove one more arch-specific target name.

One more target had the architecture added to its name since r14209
originally landed and was missed when that change was reverted. Remove
the architecture from v8_libbase as well.

BUG=
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/334793004

Patch from Richard Coles <torne@chromium.org>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21836 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Emulate MLS on pre-ARMv6T2. Cleaned up thumbee vs. thumb2 confusion.

This should unbreak things on the Raspberry Pi.

LOG=y
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/331803003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21835 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Allow all Names to be fast property names

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/329393005

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21834 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Use generic representation when normalizing boilerplate properties

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/323403008

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21833 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Mark timed range in the profile plot.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/333643006

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21832 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Disable regress/regress-2653 until deopt_every_n_garbage_collections is fixed.

BUG=3389
LOG=N
R=marja@chromium.org

Review URL: https://codereview.chromium.org/331823003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21831 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

GVN fix, preventing loads hoisting above stores to the same field when HObjectAccess's representation is not the same.

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/331493006

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21830 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

X87: Fixed flooring division by a power of 2, once again...

port r21769

original message:
    Avoid right shifts by zero bits: On ARM it actually means shifting by
    32 bits (correctness issue) and on other platforms they are useless
    (performance issue). This is fix for the fix in r20544.

    BUG=v8:3259
    LOG=y
R=weiliang.lin@intel.com

Review URL: https://codereview.chromium.org/330133004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21829 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fixed undefined behavior in RNG.

We're basically trading undefined behavior for implementation defined
behavior, which should be OK for UBSan. :-) The generated code should
be identical, at least I checked that for GCC 4.6.3 on x64.

BUG=377790
LOG=y
R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/332733002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21828 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Update Lithium AddI, SubI, MulI, BitI, ShiftI, MathMinMax to support x32 port.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/321373002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21827 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Specially handle the key of the LoadKeyed and StoreKeyed instruction for x32 port.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/324913002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21826 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Fix register usage after r21822.

The at register is no longer in usage, so changed it to "scratch".

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/335753003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21825 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Fix r21780 - “Fixed flooring division by a power of 2, once again...”

Initialize scratch register for all possible code paths.

TEST=mjsunit/compiler/division-by-constant.js
BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/333713002

Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21822 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Make timestamp of fuzz harness archives finer grained.

R=machenbach@chromium.org

Review URL: https://codereview.chromium.org/329303004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21821 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

For-of calls [Symbol.iterator]() on RHS to get iterator

R=rossberg@chromium.org
BUG=http://code.google.com/p/v8/issues/detail?id=2735
LOG=N

Review URL: https://codereview.chromium.org/332663004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21820 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Drop dependency on Isolate* from platform.h

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

Review URL: https://codereview.chromium.org/328993003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21819 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove forced type changes when they can't deopt

Hydrogen attempts to force representation changes on certain operations in order
to deoptimise on the change rather than the operation. However, these forced
changes are often unnecessary on 64-bit platforms, and cause poor code
generation, so this patch makes some of them conditional on whether it's
possible for deoptimisation to occur in the change.

On ARM64, this prevents sequences like:
                  ;;; <@46,#89> smi-tag
0x7ff282c7f050   144  lsl x4, x4, #32
                  ;;; <@48,#90> smi-untag
0x7ff282c7f054   148  asr x5, x4, #32
                  ;;; <@50,#31> mul-const-i-s
0x7ff282c7f058   152  lsl w6, w5, #3

BUG=
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/303263010

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21818 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Optimize prototype chain when creating initial maps for functions used as constructors

Review URL: https://codereview.chromium.org/332783002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21817 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Add non-miss slow path to LoadIC_Normal.

This avoids endless IC patching cycles between "normal" and "nonexistent" handlers when objects having and not having the property are seen alternatingly

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/328353002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21816 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Fix unsigned comparison.

TEST=mjsunit/regress/regress-3380
BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/329223006

Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21815 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Rewrite GetPropertyAttribute to use the LookupIterator

BUG=
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/321543004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21814 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

X87: Reland 21774: Generate KeyedLoadGeneric with Hydrogen

Port r21781 (4869c528)

BUG=
R=danno@chromium.org

Review URL: https://codereview.chromium.org/336543002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21813 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "Active use of Hydrogen-generated generic KeyedLoad IC"

Due to mirror-object test failure (although I'm pretty sure I was framed)

R=mstarzinger@chromium.org
TBR=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/328333002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21812 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Active use of Hydrogen-generated generic KeyedLoad IC

R=dcarney@chromium.org

Review URL: https://codereview.chromium.org/330023002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21811 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Don't use LOG() from platform.

All places that use OS::Allocate either CHECK() that the result is non-NULL,
or use a reasonable fallback if they can't mmap memory.

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

Review URL: https://codereview.chromium.org/326323002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21810 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

build fix after r21807

TBR=mstarzinger@chromium.org

BUG=

Review URL: https://codereview.chromium.org/333503005

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21809 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Grow big old generation slower.

BUG=
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/324403008

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21808 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

remove this == null

R=danno@chromium.org

BUG=chromium:381910

Review URL: https://codereview.chromium.org/336483002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21807 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Sign extend the dehoisted key at the definition point for x64 port only; for x32 port, we need to sign extend the dehoisted key at the use points.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/328553005

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21806 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Add v8::Promise::Then.

Blink needs v8::Promise::Then to implement ScriptPromise::then.
Blink-side CL: https://codereview.chromium.org/316453002

BUG=371288
LOG=Y
R=rossberg@chromium.org

Review URL: https://codereview.chromium.org/314553002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21805 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Do away with variable length memcpy to Set/Get registers in simulator

About a 32% boost.

Before - 5:31

Richards: 84.5
DeltaBlue: 128
Crypto: 65.3
RayTrace: 203
EarleyBoyer: 149
RegExp: 23.4
Splay: 121
NavierStokes: 98.9
----
Score (version 7): 93.8

After - 4:10

Richards: 107
DeltaBlue: 175
Crypto: 93.9
RayTrace: 258
EarleyBoyer: 186
RegExp: 32.7
Splay: 165
NavierStokes: 124
----
Score (version 7): 124

R=jacob.bramley@arm.com, svenpanne@chromium.org

Committed: https://code.google.com/p/v8/source/detail?r=21448

Review URL: https://codereview.chromium.org/213943002

Patch from Fritz Koenig <frkoenig@google.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21804 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "Revert "Reland 21774: Generate KeyedLoadGeneric with Hydrogen""

This CL tickled an unrelated arm64 bug which was is fixed separately.

The MIPS port (originally landed 21784) is also included.

TBR=verwaest@chromium.org

Review URL: https://codereview.chromium.org/331633002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21803 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove duplicate code in SetPropertyPostInterceptor

BUG=
R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/314673002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21802 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "Rename kIs64BitArch with kRequiresCodeRange."

TBR=haitao.feng@intel.com

Review URL: https://codereview.chromium.org/325343003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21801 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Rename kIs64BitArch with kRequiresCodeRange.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/335473003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21800 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Remove unnecessary HCompareNumericAndBranch when generating HShr instruction.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/334573003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21799 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix register mixup in arm64's Crankshafted typeof() == 'object'

R=ulan@chromium.org

Review URL: https://codereview.chromium.org/330593003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21798 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Add @@iterator for generator objects

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

Review URL: https://codereview.chromium.org/328093002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21797 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "MIPS: Reland 21774: Generate KeyedLoadGeneric with Hydrogen"

This reverts commit r21784 as the original CL also got reverted.

BUG=
TBR=danno@chromium.org

Review URL: https://codereview.chromium.org/329973002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21794 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

filter cross context eval

R=verwaest@chromium.org

BUG=

Review URL: https://codereview.chromium.org/294073002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21793 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "Reland 21774: Generate KeyedLoadGeneric with Hydrogen"

Due to lingering arm64 failures in Test262

TBR=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/332663003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21790 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix detection of VFP3D16 on Galaxy Tab 10.1.

BUG=v8:3387
LOG=y
R=svenpanne@chromium.org

Review URL: https://codereview.chromium.org/329963002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21789 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Drop unused v8threads include from platform files

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

Review URL: https://codereview.chromium.org/323303002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21788 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

X87: add OWNERS file

BUG=
R=danno@chromium.org

Review URL: https://codereview.chromium.org/323413002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21785 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Reland 21774: Generate KeyedLoadGeneric with Hydrogen

Port r21781 (4869c528)

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/327373002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21784 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Whitespace change to trigger bots.

BUG=
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/330573002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21783 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Whitespace change to trigger bots.

BUG=
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/330563002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21782 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Reland 21774: Generate KeyedLoadGeneric with Hydrogen

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/57123002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21781 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Fixed flooring division by a power of 2, once again...

Port r21769 (52e191b)

Original commit message:
Avoid right shifts by zero bits: On ARM it actually means shifting by
32 bits (correctness issue) and on other platforms they are useless
(performance issue). This is fix for the fix in r20544.

BUG=v8:3259
LOG=y
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/322403006

Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21780 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Drop globals.h include from platform.h

Also delete dead cpu-features-implied-by-platform code.

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

Review URL: https://codereview.chromium.org/326333002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21779 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Rossberg's suggested changes to the LookupIterator.

BUG=
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/324383005

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21778 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Revert 21774: "Generate KeyedLoadGeneric with Hydrogen"

Port r21776 (10a4872)

Original commit message:
Due to arm64 and GCMole failures

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/329723007

Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21777 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert 21774: "Generate KeyedLoadGeneric with Hydrogen"

Due to arm64 and GCMole failures

TBR=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/329253003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21776 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Generate KeyedLoadGeneric with Hydrogen.

Port r21774 (f0462b8)

BUG=
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/328173002

Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21775 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Generate KeyedLoadGeneric with Hydrogen

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/57123002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21774 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

MIPS: Simplify string copy in SubStringStub.

Port r21756 (3d58e58)

Original commit message:
The optimizations are unnecessary since the maximum
string length they operate on is currently 12.

BUG=352155
LOG=N
R=plind44@gmail.com

Review URL: https://codereview.chromium.org/326413002

Patch from Balazs Kilvady <kilvadyb@homejinni.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21773 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

X87: Simplify string copy in SubStringStub.

port r21756.

original commit message:
  The optimizations are unnecessary since the maximum
  string length they operate on is currently 12.

R=yangguo@chromium.org, bmeurer@chromium.org
  BUG=352155
  LOG=N

Review URL: https://codereview.chromium.org/326353002

Patch from Chunyang Dai <chunyang.dai@intel.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21772 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Correctly lookup starting at the holder

BUG=
R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/329153002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21771 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Ignore live_edit_ flag when when dealing with LiveEdit in a debug break.

LiveEdit maybe disabled when we enter the break and again when we
leave it, but enabled in between.

TEST=https://codereview.chromium.org/329533002
R=ulan@chromium.org

Review URL: https://codereview.chromium.org/325183003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21770 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fixed flooring division by a power of 2, once again...

Avoid right shifts by zero bits: On ARM it actually means shifting by
32 bits (correctness issue) and on other platforms they are useless
(performance issue). This is fix for the fix in r20544.

BUG=v8:3259
LOG=y
R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/324403003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21769 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Whitespace change to trigger bots.

BUG=
TBR=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/324373002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21768 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Implement LookupIterator designed to replace LookupResult

BUG=
R=ishell@chromium.org

Review URL: https://codereview.chromium.org/314953006

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21767 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Set SAHF flag correctly for ia32

sahf flag will not be set for ia32 on some old platform because some old processors does not support CPUID's extended features.
This also avoids redundant cpuid check in ia32 for sahf.

BUG=
R=danno@chromium.org

Review URL: https://codereview.chromium.org/298823008

Patch from Weiliang Lin <weiliang.lin@intel.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21766 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix typos in BUILD.gn source lists

"foo,cc" is not the same as "foo.cc"

R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/326193002

Patch from James Robinson <jamesr@chromium.org>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21765 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Unbreak build.

TBR=ulan@chromium.org

Review URL: https://codereview.chromium.org/329053005

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21764 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Removed dead API entries.

R=ulan@chromium.org

Review URL: https://codereview.chromium.org/322293003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21763 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix unsigned comparisons.

Instead of marking the comparison instruction itself as Uint32, we
look at its arguments. This is more consistent what HChange does.

BUG=v8:3380
TEST=mjsunit/regress/regress-3380
LOG=y
R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/325133004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21762 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix GN build.

R=machenbach@chromium.org

Review URL: https://codereview.chromium.org/328883005

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21759 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Simplify string copy in SubStringStub.

The optimizations are unnecessary since the maximum
string length they operate on is currently 12.

R=bmeurer@chromium.org
BUG=352155
LOG=N

Review URL: https://codereview.chromium.org/326943002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21756 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

V8 can clear exception pending message, when should not do this.

The case:
v8::TryCatch try_catch;
CompileRun(try { CEvaluate('throw 1;'); } finally {});
CHECK(try_catch.HasCaught());
CHECK(!try_catch.Message().IsEmpty());

CEvaluate is native call. Last check is not passed without patch. Patch contains test TryCatchFinallyStoresMessageUsingTryCatchHandler with more details.

R=yangguo@chromium.org, mstarzinger@chromium.org, vsevik@chromium.org

Review URL: https://codereview.chromium.org/321763002

Patch from Alexey Kozyatinskiy <kozyatinskiy@google.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21755 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Temporarily disable uint32 mode for comparisons

BUG=v8:3380
LOG=n
R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/323103002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21750 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "Fix gn build".

This was reverted due to GN build failures while rolling V8 into
Chromium. Locally reproducing it verified that this makes the GN
build pass again.

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

Review URL: https://codereview.chromium.org/324013006

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21749 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Add arity checks to mjsunit's assertEquals and assertSame

BUG=None
LOG=n
R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/309173003

Patch from Erik Arvidsson <arv@chromium.org>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21748 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Do not merge adjourning ranges when calculating percentages in plot.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/319703009

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21747 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Reland 21720: Introduce FieldIndex to unify and abstract property/field offset

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/300283002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21746 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Temporary disabled failing tests on ARM64 no-snap.

R=machenbach@chromium.org
TEST=cctest/test-serialize
BUG=v8:3385
LOG=N

Review URL: https://codereview.chromium.org/324943002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21745 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Relax assertion in StoreBuffer::FindPointersToNewSpaceInMapsRegion

Since r21232, maps are now a non-power-of-two size and thus don't fit
evenly into a page.  No test case, as it is somewhat difficult to get
this condition to trigger.  It is possible with this case, however:

    var result;

    (function() {
        for (var i = 0; i < 1e7; i++) {
            result = (function*() { yield 1; })();
            result.foo = 1;
        }
    })();

R=hpayer@chromium.org
BUG=

Review URL: https://codereview.chromium.org/291913002

Patch from Andy Wingo <wingo@igalia.com>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21744 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Fix invalid attributes when generalizing because of incompatible map change.

BUG=382143
LOG=y
TEST=mjsunit/regress/regress-382143
R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/324933003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21743 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "Support external startup data in V8."

This reverts commit r21696 for breaking chromium windows compilation in the chromium cq.

Conflicts:
src/d8.cc

BUG=
R=mstarzinger@chromium.org

Review URL: https://codereview.chromium.org/328693003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21740 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Revert "Port 'external startup data' flag from gyp to gn."

This cause compile failures on the GN buildbot about 'gen/v8/libraries.bin'
missing and no known rule to make it.

TBR=vogelheim@chromium.org

Review URL: https://codereview.chromium.org/327703004

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21739 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Update DecodeFieldToSmi to support x32 port.

R=verwaest@chromium.org

Review URL: https://codereview.chromium.org/310803003

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21738 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Do not clear mirror cache when fetching loaded scripts.

R=yurys@chromium.org
BUG=376534
LOG=N

Review URL: https://codereview.chromium.org/309313002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21737 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Log IC misses as timer events.

R=jkummerow@chromium.org

Review URL: https://codereview.chromium.org/318983005

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21736 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Make presubmit script happy again.

No clue why this seems to be an issue only for me, though...

R=bmeurer@chromium.org

Review URL: https://codereview.chromium.org/321123002

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21735 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

Reuse RelocInfo's IsPatchedDebugBreakSlotSequence() implementation.

RelocInfo::IsPatchedDebugBreakSlotSequence() already does this assembly
check, so just reuse it in BreakLocationIterator::IsDebugBreakAtSlot(),
like the other debug ports do.

R=yangguo@chromium.org

Review URL: https://codereview.chromium.org/321743002

Patch from Thiago Farina <tfarina@chromium.org>.

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@21734 ce2b1a6d-e550-0410-aec6-3dcde31c8c00