Commits · a07e129e9a70838ad7e1b2912e4e18cb8d87ca65 · Linshizhi / V8

16 Feb, 2012 12 commits

Relax TransitionElementsKind DependsOn/Changes dependencies. · a07e129e

danno@chromium.org authored Feb 16, 2012

Ensure that GVN eliminates all transitions that are dominated by an equivalent transition, even if there is a DependsOn-changing instruction in between.

R=fschneider@chromium.org

Review URL: https://chromiumcodereview.appspot.com/9365057

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10731 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

a07e129e

Prepare push to trunk. Now working on version 3.9.8. · 0ee19e40

yangguo@chromium.org authored Feb 16, 2012

R=jkummerow@chromium.org
BUG=
TEST=

Review URL: https://chromiumcodereview.appspot.com/9416016

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10728 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

0ee19e40

Temporarily disable known failing test to make builder cycle green. · 781427a1

fschneider@chromium.org authored Feb 16, 2012

BUG=v8:1952
Review URL: https://chromiumcodereview.appspot.com/9414006

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10727 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

781427a1

Increase ARM/MIPS simulators stack safety margin to 1024 bytes · b6551b83

vegorov@chromium.org authored Feb 16, 2012

Based on r9686 (10e1ea5).

BUG=
TEST=mozilla regress-355497.js

Review URL: https://chromiumcodereview.appspot.com/9309030
Patch from Daniel Kalmar <kalmard@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10722 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

b6551b83

Allow inlining of functions containing function literals. · acb83c06

mstarzinger@chromium.org authored Feb 16, 2012

R=fschneider@chromium.org
BUG=v8:1322
TEST=mjsunit/compiler/inline-literals

Review URL: https://chromiumcodereview.appspot.com/9419005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10721 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

acb83c06

MIPS: Initial support for count-based profiling · 0199492d

jkummerow@chromium.org authored Feb 16, 2012

Port r10699 (18a0fc0).

Original commit message:

(behind FLAG_count_based_interrupts; only on ia32)

BUG=
TEST=

Review URL: https://chromiumcodereview.appspot.com/9420005
Patch from Daniel Kalmar <kalmard@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10720 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

0199492d

Ensure using byte registers for byte instructions on ia32 and x64. · cc278040

yangguo@chromium.org authored Feb 16, 2012

BUG=v8:1945
TEST=regress-1945.js

Review URL: https://chromiumcodereview.appspot.com/9418005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10719 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

cc278040

Cast NaN to float in ExternalFloatArray::SetValue. · dd95fee0

danno@chromium.org authored Feb 16, 2012

TBR=mstarzinger@chromium.org

Review URL: https://chromiumcodereview.appspot.com/9419003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10718 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

dd95fee0

MIPS: Re-worked the deopt entry table. · 596a0fa7

fschneider@chromium.org authored Feb 16, 2012

This method works around the Branch offset and relocinfo issues by emulating a pc-relative jump.
This allows us to generate larger entry tables. The theoretical limit is 2^16 (number of entries)
but even that can be extended by allowing a larger instruction count.

Also reverted the mips-specific constant (kNumberOfEntries) in deoptimizer.h

BUG=
TEST=
Review URL: https://chromiumcodereview.appspot.com/9347016

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10717 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

596a0fa7

Fix GCC-4.7 warnings · 1cd5f2c7

jkummerow@chromium.org authored Feb 16, 2012

which were introduced in r10700.

Review URL: https://chromiumcodereview.appspot.com/9401005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10716 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

1cd5f2c7

Prefix usage of kExternalFloatArray and kExternalDoubleArray with v8. · 2fa1cc03

danno@chromium.org authored Feb 16, 2012

TBR=mstarzinger@chromium.org

Review URL: https://chromiumcodereview.appspot.com/9417002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10715 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

2fa1cc03

Uniformly handle 'undefined' store to Float64Array and Float32Array. · bd79e299

danno@chromium.org authored Feb 16, 2012

Previous behavior diverged in ICs and Crankshaft. When storing to a Float32Array or Float64Array, the ICs treated undefined as zero while Crankshaft treated it as NaN. Now both ICs and Crankshaft treat it as NaN, which is consistent with the WebGL & ECMAScript spec.

R=mstarzinger@chromium.org

Review URL: https://chromiumcodereview.appspot.com/9402008

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10714 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

bd79e299

15 Feb, 2012 4 commits

Cleanup idle notification tests. · bc7e0153

mstarzinger@chromium.org authored Feb 15, 2012

R=ulan@chromium.org
TEST=cctest/test-api/IdleNotification

Review URL: https://chromiumcodereview.appspot.com/9403014

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10713 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

bc7e0153

Initialize internal arrays with the correct map. · 01e46b95

yangguo@chromium.org authored Feb 15, 2012

BUG=v8:1878
TEST=regress-1878.js

Review URL: https://chromiumcodereview.appspot.com/9402009

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10712 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

01e46b95

Remove unnecessary elements type check when allocating array in runtime. · b21ada9a

yangguo@chromium.org authored Feb 15, 2012

BUG=
TEST=--smi-only-arrays should be perf-neutral to allocating big arrays in a tight loop.

Review URL: https://chromiumcodereview.appspot.com/9356002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10711 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

b21ada9a

MIPS: Enabled mips in tools/disasm.py. · 941d884a

yangguo@chromium.org authored Feb 15, 2012

BUG=
TEST=

Review URL: https://chromiumcodereview.appspot.com/9307118
Patch from Daniel Kalmar <kalmard@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10710 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

941d884a

14 Feb, 2012 17 commits

MIPS: Split AST Declaration class, in preparation for new module declaration forms. · f2715fe7

rossberg@chromium.org authored Feb 14, 2012

Port r10662 (c8054f6).

Original commit message:

Turns Declaration into an abstract class, and introduces VariableDeclaration as a concrete subclass.

BUG=
TEST=

Review URL: https://chromiumcodereview.appspot.com/9395015

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10709 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

f2715fe7

Properly disable count-based profiler on non-ia32 · 8ac0c1cd

jkummerow@chromium.org authored Feb 14, 2012

TEST=buildbot redness gone

Review URL: https://chromiumcodereview.appspot.com/9360043

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10708 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

8ac0c1cd

MIPS: Implement KeyedStoreICs to grow arrays on out-of-bound stores. · b098a2bd

danno@chromium.org authored Feb 14, 2012

Port r10673 (18d3af5).

Original commit message:
Supports growing non-COW JSArray by a single element if the backing store has room, and initial allocation of a backing store for the store to index zero of an empty array to kPreallocatedArrayElements elements (e.g. the [] array literal).

BUG=
TEST=

Review URL: https://chromiumcodereview.appspot.com/9378005
Patch from Daniel Kalmar <kalmard@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10707 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

b098a2bd

Fix crashing bugs in store-and-grow IC for double values. · 71cd77e2

danno@chromium.org authored Feb 14, 2012

R=jkummerow@chromium.org
BUG=chromium:113924
TEST=test/mjsunit/regress/regress-113924.js

Review URL: https://chromiumcodereview.appspot.com/9365055

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10706 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

71cd77e2

Add Navier-Stokes benchmark. · 579ca743

ulan@chromium.org authored Feb 14, 2012

R=danno@chromium.org,kasperl@chromium.com,stefanoc@chromium.org,sandholm@chromium.org

Review URL: https://chromiumcodereview.appspot.com/9359033

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10705 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

579ca743

Reapply r10690 ("Temporarily disable inlining of Math.floor called as a function") · 82a1adcb

jkummerow@chromium.org authored Feb 14, 2012

This reverts r10692.

Review URL: https://chromiumcodereview.appspot.com/9359035

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10702 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

82a1adcb

Small update to experimental profiler flags · 866e0eee

jkummerow@chromium.org authored Feb 14, 2012

Review URL: https://chromiumcodereview.appspot.com/9360041

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10701 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

866e0eee

Enable optimization of top-level code and generate deoptimization support lazily. · 0be449d6

fschneider@chromium.org authored Feb 14, 2012

This change enables optimization of top-level and eval-code. For this to work, it adds
support for declaring global variables in optimized code.

At the same time it disables the eager generation of deoptimization support data
in the full code generator (originally introduced in
r10040). This speeds up initial compilation and saves
memory for functions that won't be optimized. It requires
recompiling the function with deoptimization
support when we decide to optimize it.

Review URL: https://chromiumcodereview.appspot.com/9187005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10700 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

0be449d6

Initial support for count-based profiling · 4233bf83

jkummerow@chromium.org authored Feb 14, 2012

(behind FLAG_count_based_interrupts; only on ia32)

Review URL: https://chromiumcodereview.appspot.com/9373028

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10699 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

4233bf83

Don't treat function parameters as let-bound variables in Harmony mode. · 559f5eec

rossberg@chromium.org authored Feb 14, 2012

R=ulan@chromium.org
BUG=v8:1942
TEST=

Review URL: https://chromiumcodereview.appspot.com/9365054

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10698 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

559f5eec

Prepare push to trunk. Now working on version 3.9.7. · 9da06ece

ulan@chromium.org authored Feb 14, 2012

R=jkummerow@chromium.org
BUG=
TEST=

Review URL: https://chromiumcodereview.appspot.com/9395007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10695 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

9da06ece

Revert r10690. Enable inlining of Math.floor again. · d4625504

fschneider@chromium.org authored Feb 14, 2012

The test failures reported in issue v8:1947 are unrelated to this change since it is not reproducible anymore.

BUG=v8:1947
Review URL: https://chromiumcodereview.appspot.com/9365048

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10692 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

d4625504

objects: fix template-related linker error · 7a79e354

danno@chromium.org authored Feb 14, 2012

R=erik.corry@gmail.com
BUG=1936
TEST=

Review URL: https://chromiumcodereview.appspot.com/9382033
Patch from Fedor Indutny <fedor@indutny.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10691 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

7a79e354

Temporarily disable inlining of Math.floor called as a function because of test failures. · fd4ceab2

fschneider@chromium.org authored Feb 14, 2012

BUG=v8:1947
Review URL: https://chromiumcodereview.appspot.com/9359032

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10690 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

fd4ceab2

Allow inlining of functions containing object literals. · 0db91d40

mstarzinger@chromium.org authored Feb 14, 2012

R=fschneider@chromium.org
BUG=v8:1322
TEST=mjsunit/compiler/inline-literals

Review URL: https://chromiumcodereview.appspot.com/9388007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10689 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

0db91d40

Port r10674 to MIPS. · ed54dd20

yangguo@chromium.org authored Feb 14, 2012

BUG=
TEST=

Review URL: https://chromiumcodereview.appspot.com/9395003
Patch from Daniel Kalmar <kalmard@homejinni.com>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10688 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

ed54dd20

Fix crash when using --smi-only-arrays and --trace-elements-transitions · 36141ddf

yangguo@chromium.org authored Feb 14, 2012

BUG=
TEST=3d-cube on debug mode with --smi-only-arrays and --trace-elements-transitions

Review URL: https://chromiumcodereview.appspot.com/9361054

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10687 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

36141ddf

13 Feb, 2012 7 commits

Clear map transitions in CALLBACKS when normalizing properties. · 0a9e97e9

svenpanne@chromium.org authored Feb 13, 2012

As a bonus, simplify the surrounding code a bit by using the templatized To method.

Review URL: https://chromiumcodereview.appspot.com/9392002

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10686 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

0a9e97e9

DescriptorArray::CopyFrom should always drop transitions for CALLBACKS. · 2a572579

svenpanne@chromium.org authored Feb 13, 2012

Review URL: https://chromiumcodereview.appspot.com/9389005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10685 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

2a572579

Fix crash in CALLBACK KeyedIC stub. · da2e84fc

danno@chromium.org authored Feb 13, 2012

TBR=jkummerow@chromium.org

Review URL: https://chromiumcodereview.appspot.com/9390005

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10684 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

da2e84fc

Optimize call sites that only ever see a single elements transition · ad41b9eb

danno@chromium.org authored Feb 13, 2012

R=jkummerow@chromium.org
TEST=imaging-gaussian-blur performance improves

Review URL: https://chromiumcodereview.appspot.com/9389001

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10683 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

ad41b9eb

Fix test expectations for the tickprocessor. · fff8eba0

yangguo@chromium.org authored Feb 13, 2012

Review URL: https://chromiumcodereview.appspot.com/9388003

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10682 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

fff8eba0

Ensure that we do a non-incremental GC before relying on weak handle callbacks. · 8a7f1568

svenpanne@chromium.org authored Feb 13, 2012

Previously, some tests failed when being run with a low GC interval. This was
caused by a switch of the GC to incremental marking mode, which in turn did not
fire any callbacks for weak global handles. Now we make sure that we run in
non-incremental mode, although via a slightly misleading GC flag.

We should probably review the uses of PerformScavenge() and gc() in our tests to
see if they actually mean "make sure our callbacks fired".

Review URL: https://chromiumcodereview.appspot.com/9378007

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10681 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

8a7f1568

Add --call-graph-size option to tickprocessor. · 80326e08

yangguo@chromium.org authored Feb 13, 2012

BUG=v8:1937

Review URL: https://chromiumcodereview.appspot.com/9386007
Patch from Ben Noordhuis <info@bnoordhuis.nl>.

git-svn-id: http://v8.googlecode.com/svn/branches/bleeding_edge@10680 ce2b1a6d-e550-0410-aec6-3dcde31c8c00

80326e08