GitLab

1) The results were skewed by scripts from the same resource being ran several
times (so we saw times > 10 s).

2) Script run time was included in the measurement, and it shouldn't be.

WebCore.Scripts.Deferred.TimeBetweenLoadedAndCompiled is removed for now (to be
fixed later, when it's really needed). Tracking whether a deferred script was
already loaded when requested is more complicated.

BUG=

Review URL: https://codereview.chromium.org/551333002

git-svn-id: svn://svn.chromium.org/blink/trunk@181641 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Reason: As discussed with @haraken over IRC, we need to migrate to using chromium
resource, grd, system. In that we may need to think over how to use the include directive.

One of the harmony modules, [1], describes about using the 'import "module-name"' directive.
We could consider using that as well. As this needs some discussion, we decided to revert
this patch first and reconsider landing this once the formal decision is made about it.

R=haraken@chromium.org, yosin@chromium.org

[1] https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Statements/import

Review URL: https://codereview.chromium.org/559493002

git-svn-id: svn://svn.chromium.org/blink/trunk@181640 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=391593

Review URL: https://codereview.chromium.org/554243002

git-svn-id: svn://svn.chromium.org/blink/trunk@181639 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We poison the heap object headers because only our code should ever be
able to access them and only from a handful of methods. Poisoning the
headers lets us catch stray read/writes that end up in our headers and we
use the NO_SANITIZE_ADDRESS annotation on the handful of methods that
operate on the headers.

The ASan NO_SANITIZE_ADDRESS annotation does not propagate to acquireLoad.
Our first attempt to fix that was to unpoison the address accessed.
However, that does not work because we are using this code from multiple
threads without locking (which is the reason for using atomic ops).
Therefore, the threads will have races when it comes to poisoning.

This change fixes the issue by introducing asan aware
asanAcquireLoad/asanReleaseStore which will work on poisoned memory
when you know what you are doing.

Kostya, do you have any alternative suggestions?

R=erik.corry@gmail.com, kcc@chromium.org, oilpan-reviews@chromium.org, zerny@chromium.org
BUG=411712

Review URL: https://codereview.chromium.org/556443003

git-svn-id: svn://svn.chromium.org/blink/trunk@181638 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The row index may be bogus or even unset in that case.

Rows may be inserted with an initially unknown row index. This happens
when the row isn't inserted at the end of its table section, and also
when splitting an anonymous row. This would lead to an assertion failure
for row()->rowIndexWasSet() in RenderTableCell::styleDidChange().
rowIndex() is used there in a call to
RenderTableSection::rowLogicalHeightChanged(). rowIndex() is also used
in a similar way in RenderTableRow::styleDidChange(), albeit without
asserting first.

styleDidChange() may be called while needsCellRecalc() is set, which
may mean that row index hasn't been calculated yet (or is bogus). We
could of course just recalculate the cells (which also involves
calculating all row indices) in styleDidChange(), but it's not really
necessary, since the row index was just needed in order to call
rowLogicalHeightChanged(), which just bails anyway if cells need recalc.

So the solution is to pass a RenderTableRow instead of the row index to
rowLogicalHeightChanged(). This way we don't call rowIndex() when
row index is unknown.

BUG=397442

Review URL: https://codereview.chromium.org/532513002

git-svn-id: svn://svn.chromium.org/blink/trunk@181637 bbb929c8-8fbe-4397-9dbb-9b2b20218538

InlineFlowBox, sets word spacing first if in LTR scenario. In RTL scenario first logicalLeft should be set prior to setting word-spacing length.

RenderBlockLineLayout, sets needsWordSpacing as true to not ignore space if first character.

BUG=344873
R=behdad,eae, aharon
TEST=fast/text/international/bidi-word-spacing.html

Review URL: https://codereview.chromium.org/529753002

git-svn-id: svn://svn.chromium.org/blink/trunk@181636 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Every time background parser sends chunk, tree is flushed.

If page contains very large script, then script node content is updated
many times. Every update is causes string concatenation.

Solution: do not flush pending text until it is mandatory.

Test: https://codereview.chromium.org/500363002
Test depends on: https://codereview.chromium.org/544453004/

BUG=410790

Review URL: https://codereview.chromium.org/494993002

git-svn-id: svn://svn.chromium.org/blink/trunk@181635 bbb929c8-8fbe-4397-9dbb-9b2b20218538

r179621 added support to handle infinite sizes to the
sortByGridTrackGrowthPotential() sorting function. The problem is that it broke
the strict weak ordering required by std::sort when the compared items had both
infinite maximum sizes, so the algorithm was going nuts.

BUG=401983, 402508, 402757

Review URL: https://codereview.chromium.org/493093002

git-svn-id: svn://svn.chromium.org/blink/trunk@181634 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Right now if you add a positioned block item after an inline item, the
positioned block item is inserted inside the anonymous block wrapping
the inline item.

This was causing an assert in RenderGrid::addChildToIndexesMap() while
navigating the sibling boxes, as its previous sibling is an inline.

As the positioned block is now not direct child of the grid, we can
ignore it, as it's not considered a grid item. The grid item would
be the anonymous block. So the proper check is added in
RenderGrid::addChild to do it.

BUG=401463
TEST=fast/css-grid-layout/grid-add-positioned-block-item-after-inline-item.html

Review URL: https://codereview.chromium.org/450783002

git-svn-id: svn://svn.chromium.org/blink/trunk@181633 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch shouldn't have any practical effect on CSP, but it will make
it straightforward to separate side-effects from parsing in a future
patch.

BUG=411889

Review URL: https://codereview.chromium.org/550223002

git-svn-id: svn://svn.chromium.org/blink/trunk@181632 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This is an attempt to re-land r181610. The CL was reverted due to build breakage, but I couldn't reproduce the link error.

TBR=haraken@chromium.org,keishi@chromium.org

The original description:

Add FontFaceDescriptors IDL dictionary

To follow the spec[1]. This is the first place to use IDL dictionary
execept for testing, so this CL also includes build related changes.
Existing tests(e.g. fast/css/fontface-constructor-error.html) should
cover this change.

[1] http://dev.w3.org/csswg/css-font-loading/#fontface-interface

BUG=403150
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=181610

Review URL: https://codereview.chromium.org/556543003

git-svn-id: svn://svn.chromium.org/blink/trunk@181631 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Moved all test idl files into Source/bindings/tests/idls/core.
Added one new idl file (TestPartialInterface3) into Source/bindings/tests/idls/modules.

BUG=358074

Review URL: https://codereview.chromium.org/556453005

git-svn-id: svn://svn.chromium.org/blink/trunk@181630 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Currently, when CSPSourceList encounters "'self'" in a source list, it
creates a CSPSource using the currently available SecurityOrigin. If
we want to be able to create CSP objects before documents are available
(to support `frame-ancestors` for instance), we need to break this
dependency.

This patch adjusts CSPSourceList to ask the policy object whether a URL
matches self, and the policy object to store a replacable CSPSource
object that could be swapped out if the SecurityOrigin changes.

BUG=411889

Review URL: https://codereview.chromium.org/549163003

git-svn-id: svn://svn.chromium.org/blink/trunk@181629 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Goal: PendingScript should know how to watch for its resource to load, and upper
layers (such as HTMLScriptRunner) should delegate the "how to watch for load"
logic to PendingScript, instead of accessing its resource directly.

In addition, a function to extract source code from PendingScript should be in
PendingScript, not in HTMLScriptRunner.

The refactoring is split off from https://codereview.chromium.org/368283002
(script streaming), but it makes sense as standalone too.

BUG=

Review URL: https://codereview.chromium.org/551293002

git-svn-id: svn://svn.chromium.org/blink/trunk@181627 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=410419

Review URL: https://codereview.chromium.org/535283003

git-svn-id: svn://svn.chromium.org/blink/trunk@181624 bbb929c8-8fbe-4397-9dbb-9b2b20218538

border-padding-pagination.html exists for both these writing modes, and it won't be
used in the new implementation, so better make sure that we don't lose coverage.

R=jchaffraix@chromium.org

Review URL: https://codereview.chromium.org/510123002

git-svn-id: svn://svn.chromium.org/blink/trunk@181622 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=391593

Review URL: https://codereview.chromium.org/548213004

git-svn-id: svn://svn.chromium.org/blink/trunk@181620 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=arv@chromium.org
NOTRY=true
BUG=412273

Review URL: https://codereview.chromium.org/539383005

git-svn-id: svn://svn.chromium.org/blink/trunk@181618 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL adds no new tests. Will add tests after a Chromium change for
string resource.

BUG=123896

Review URL: https://codereview.chromium.org/552163002

git-svn-id: svn://svn.chromium.org/blink/trunk@181617 bbb929c8-8fbe-4397-9dbb-9b2b20218538

ES6 Map returns `entries' method when [@@iterator] property is required[1].
MIDIInputMap and MIDIOutputMap should do the same.

1: https://people.mozilla.org/~jorendorff/es6-draft.html#sec-map.prototype-@@iterator

BUG=341423
R=toyoshim@chromium.org

Review URL: https://codereview.chromium.org/555683003

git-svn-id: svn://svn.chromium.org/blink/trunk@181616 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Heap::collectGarbage is needed in media_stream_impl_unittest.

BUG=340522

Review URL: https://codereview.chromium.org/543603003

git-svn-id: svn://svn.chromium.org/blink/trunk@181615 bbb929c8-8fbe-4397-9dbb-9b2b20218538

They are same and very confusing.
isText is used outside of HTMLInputElement.  It can be replaced with
|isTextField() && type() != InputTypeNames::number|.
isTextType is used only in HTMLInputElement. Its callsites should be
replaced with virtual function dispatches.

BUG=

Review URL: https://codereview.chromium.org/551283002

git-svn-id: svn://svn.chromium.org/blink/trunk@181614 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Review URL: https://codereview.chromium.org/551183003

git-svn-id: svn://svn.chromium.org/blink/trunk@181613 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Revert of Add FontFaceDescriptors IDL dictionary (patchset #4 id:60001 of https://codereview.chromium.org/481453003/)

Reason for revert:
Broke build

http://build.chromium.org/p/chromium.webkit/builders/GPU%20Mac%20Builder/builds/19956

Original issue's description:
> Add FontFaceDescriptors IDL dictionary
> 
> To follow the spec[1]. This is the first place to use IDL dictionary
> execept for testing, so this CL also includes build related changes.
> Existing tests(e.g. fast/css/fontface-constructor-error.html) should
> cover this change.
> 
> [1] http://dev.w3.org/csswg/css-font-loading/#fontface-interface
> 
> BUG=403150
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=181610

TBR=haraken@chromium.org,ksakamoto@chromium.org,bashi@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=403150

Review URL: https://codereview.chromium.org/551873004

git-svn-id: svn://svn.chromium.org/blink/trunk@181612 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=oilpan-reviews
BUG=321462
NOTRY=true

Review URL: https://codereview.chromium.org/553993004

git-svn-id: svn://svn.chromium.org/blink/trunk@181611 bbb929c8-8fbe-4397-9dbb-9b2b20218538

To follow the spec[1]. This is the first place to use IDL dictionary
execept for testing, so this CL also includes build related changes.
Existing tests(e.g. fast/css/fontface-constructor-error.html) should
cover this change.

[1] http://dev.w3.org/csswg/css-font-loading/#fontface-interface

BUG=403150

Review URL: https://codereview.chromium.org/481453003

git-svn-id: svn://svn.chromium.org/blink/trunk@181610 bbb929c8-8fbe-4397-9dbb-9b2b20218538

As the tests in KeyframeEffectModelTest are unit tests, it doesn't make
sense for them to use the parser to construct timing functions when
they can be constructed directly.

Review URL: https://codereview.chromium.org/552123003

git-svn-id: svn://svn.chromium.org/blink/trunk@181609 bbb929c8-8fbe-4397-9dbb-9b2b20218538

See: http://dev.w3.org/fxtf/web-animations/#dfn-zero-time

Review URL: https://codereview.chromium.org/551253002

git-svn-id: svn://svn.chromium.org/blink/trunk@181608 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Revert of Go back to old underline position and thickness logic (patchset #2 id:20001 of https://codereview.chromium.org/550163003/)

Reason for revert:
Many tests failed on WinXP because of different underline position.

Original issue's description:
> Go back to old underline position and thickness logic
> 
> In r174431 from May this year we changed the underline position and
> thickness logic to respect the underline data specified in the font.
> While this improved rendering for certain fonts and scripts it did cause
> a regression for others. This change reverts to the old behavior while
> we figure out how to handle fonts with inaccurate or undesirable
> metrics.
> 
> BUG=408075
> R=scottmg@chromium.org, cpu@chromium.org
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=181591

TBR=scottmg@chromium.org,cpu@chromium.org,eae@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=408075

Review URL: https://codereview.chromium.org/557653002

git-svn-id: svn://svn.chromium.org/blink/trunk@181607 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The method setProperty(CSSPropertyID, CSSPropertyID identifier, bool important = false) is not used anywhere in the code, so removing it.

Review URL: https://codereview.chromium.org/547333002

git-svn-id: svn://svn.chromium.org/blink/trunk@181606 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This reverts commit f98363c805c098bf6f5b69c90ea7e5ae120a32a3.

NOTRY=true
TBR=arv@chromium.org
BUG=None

Review URL: https://codereview.chromium.org/555033002

git-svn-id: svn://svn.chromium.org/blink/trunk@181605 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Add DictionaryHelper::getWithUndefinedOrNullCheck() to fix a bug that
newly added tests found. When IDL dictionary member is undefined or null,
the member should be treated as 'not present' and default values should
be used when specified. Since DictionaryHelper::get() converts
null/undefined into a valid value, we can't use DictionaryHelper::get()
in IDL dictionary implementation directly.

BUG=321462

Review URL: https://codereview.chromium.org/554893004

git-svn-id: svn://svn.chromium.org/blink/trunk@181604 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Currently standard JS errors thrown by private scripts are treated
as real errors of the private scripts and crash the renderer.
However, we need to special-case a stackoverflow error because
user's script can create code that causes a stackoverflow error
in private scripts.

BUG=412143
TEST=fast/dom/private_script_unittest.html

Review URL: https://codereview.chromium.org/553983007

git-svn-id: svn://svn.chromium.org/blink/trunk@181603 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=367689

Review URL: https://codereview.chromium.org/552673002

git-svn-id: svn://svn.chromium.org/blink/trunk@181602 bbb929c8-8fbe-4397-9dbb-9b2b20218538

TBR=arv@chromium.org
NOTRY=true
BUG=None

Review URL: https://codereview.chromium.org/549413002

git-svn-id: svn://svn.chromium.org/blink/trunk@181601 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Needed in order to pair up constructor and destructor calls. Should be
generally useful for debugging.

BUG=393331,407976

Review URL: https://codereview.chromium.org/556703002

git-svn-id: svn://svn.chromium.org/blink/trunk@181600 bbb929c8-8fbe-4397-9dbb-9b2b20218538

My last change to it made it not work if there were no failures
for a tree because failures[tree] would be undefined. Make it
less brittle, add tests for this code and while here add an
explanatory message that this page means there are no failures.

BUG=407496
NOTRY=true
R=jyasskin@chromium.org

Review URL: https://codereview.chromium.org/530613002

git-svn-id: svn://svn.chromium.org/blink/trunk@181599 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Use |type() == InputTypeNames:foo| instead.
These predicates are inconsistent and confusing.

BUG=
TEST=none; No behavior changes.

Review URL: https://codereview.chromium.org/557613002

git-svn-id: svn://svn.chromium.org/blink/trunk@181598 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This isn't an ideal solution, but it hides the failure category labels when there are
no failure cards. Also adding a bit of styling in the meantime.

NOTRY=true
TBR=ojan@chromium.org

Review URL: https://codereview.chromium.org/555493005

git-svn-id: svn://svn.chromium.org/blink/trunk@181597 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Oilpan: blink_heap_unittests: asan check failed. HeapObjectHeader::finalize zap a wrong memory for object without VTable

1. In the HeapObjectHeader::finalize, we zap the primary vTable entry, but we have not checked if the object has vTable.

R=ager@chromium.org, oilpan-reviews@chromium.org
BUG=411204

Review URL: https://codereview.chromium.org/547903002

git-svn-id: svn://svn.chromium.org/blink/trunk@181596 bbb929c8-8fbe-4397-9dbb-9b2b20218538