GitLab

Revert of Explicit notification and removal of lifecycle observers. (https://codereview.chromium.org/194923007/)

Reason for revert:
http://test-results.appspot.com/dashboards/flakiness_dashboard.html#group=%40ToT%20Blink&tests=fast/workers/storage/interrupt-database.html,http/tests/w3c/webperf/submission/Google/resource-timing/html/test_resource_dedicated_worker.html,http/tests/xmlhttprequest/workers/referer.html,http/tests/xmlhttprequest/workers/xmlhttprequest-response-type-blob.html

Original issue's description:
> Explicit notification and removal of lifecycle observers.
> 
> To help make disposing of execution contexts cleaner in a setting
> where both a context and its lifetime observers are being finalized
> by a garbage collector, add support for explicitly clearing out
> all lifecycle observers:
> 
> protected:
>   void LifecycleNotifier<T>::removeAndNotifyAllObservers()
> 
> Notifying them that the context has been destroyed while doing so.
> An operation that was previously (only) done by the LifecycleNotifier
> destructor.
> 
> Also provide a convenience method over ExecutionContext for clearing
> out its observers:
> 
>   void ExecutionContext::removeAllLifecycleObservers()
> 
> The operation is used here when shutting down and finalizing a worker
> thread and its global scope -- explicitly severing the connection
> of all observers to its execution context before performing the
> final garbage collection.
> 
> Doing so ensures that all observer objects being finalized during that
> garbage collection will not have access to its execution context object.
> A heap object that may or may not have been finalized by the garbage
> collector already.
> 
> R=haraken@chromium.org,ager@chromium.org,abarth@chromium.org
> BUG=340522
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=169435

TBR=oilpan-reviews@chromium.org,abarth@chromium.org,haraken@chromium.org,ager@chromium.org,eseidel@chromium.org,tkent@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=340522

Review URL: https://codereview.chromium.org/203393002

git-svn-id: svn://svn.chromium.org/blink/trunk@169444 bbb929c8-8fbe-4397-9dbb-9b2b20218538

DevTools: once migrated to document-based stylesheet processing there is no need to track changes to frames.

Review URL: https://codereview.chromium.org/203153002

git-svn-id: svn://svn.chromium.org/blink/trunk@169443 bbb929c8-8fbe-4397-9dbb-9b2b20218538

R=lushnikov, pfeldman

Review URL: https://codereview.chromium.org/201293007

git-svn-id: svn://svn.chromium.org/blink/trunk@169442 bbb929c8-8fbe-4397-9dbb-9b2b20218538

R=apavlov
BUG=353016

Review URL: https://codereview.chromium.org/197283031

git-svn-id: svn://svn.chromium.org/blink/trunk@169441 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This makes run-bindings-tests faster by doing more work in
Python, instead of going out to the OS.
Notably: calling Python compiler directly (reusing a compiler object),
instead of launching separate processes, and
avoiding a temporary file for interfaces_info.

(Earlier versions had many refactoring changes,
which improve support for Dart CG and other backends,
which Nils broke off and committed separately.)

R=nbarth

Review URL: https://codereview.chromium.org/169743005

git-svn-id: svn://svn.chromium.org/blink/trunk@169440 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Display MathML fallback content when it is specified and take into account inline/display modes for mathematical formulas. This is important for Web authors because Blink does not support MathML yet.

R=ojan@chromium.org,eseidel@chromium.org,jchaffraix@chromium.org
BUG=chromium:324764,chromium:152430

Review URL: https://codereview.chromium.org/138383003

git-svn-id: svn://svn.chromium.org/blink/trunk@169439 bbb929c8-8fbe-4397-9dbb-9b2b20218538

it from being reused as-is from the binary. This changes it to make
the storage more compact by replacing pointers with offsets into a
shared string. Also switching to more compact data types.

The new database will look like:

static const LChar staticEntityStringStorage[] = {
'A', 'E', 'l', 'i', 'g',
';',
'A', 'M', 'P',
';',
'A', 'a', 'c', 'u', 't', 'e',
';',
...

static const HTMLEntityTableEntry staticEntityTable[2231] = {
    { 0x000C6, 0, 0, 5 }, // &AElig
    { 0x000C6, 0, 0, 6 }, // Æ
    { 0x00026, 0, 6, 3 }, // &AMP
    { 0x00026, 0, 6, 4 }, // &
    { 0x000C1, 0, 10, 6 }, // &Aacute
...

The binary will shrink by about 20-35 KB (depending on architecture),
the memory usage should shrink by similar numbers for every
renderer (since the memory now can be shared between renderers) and
there will be a tiny startup time improvement due to
fewer relocation blocks and less to read from disk.

It accomplishes the savings by avoiding pointers in the data (resolving pointers
during startup is expensive), using one string instead of 2000+, not storing the
trailing NUL byte, ordering the members for minimum padding, using minimal size
data types and reusing the same string space for things like "amp;" and "amp".

BUG=352141

R=

Review URL: https://codereview.chromium.org/199103002

git-svn-id: svn://svn.chromium.org/blink/trunk@169438 bbb929c8-8fbe-4397-9dbb-9b2b20218538

FontBuilder must be properly initialised before font properties can be
applied to a RenderStyle.

BUG=352505,352506

Review URL: https://codereview.chromium.org/203193002

git-svn-id: svn://svn.chromium.org/blink/trunk@169437 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Since test return will be aync, introduced function to test query string return value.

This test fixes isuse introduced by fix for bug196640

R=keishi1, tkent
BUG=353066
TEST=Automated

Review URL: https://codereview.chromium.org/200613004

git-svn-id: svn://svn.chromium.org/blink/trunk@169436 bbb929c8-8fbe-4397-9dbb-9b2b20218538

To help make disposing of execution contexts cleaner in a setting
where both a context and its lifetime observers are being finalized
by a garbage collector, add support for explicitly clearing out
all lifecycle observers:

protected:
  void LifecycleNotifier<T>::removeAndNotifyAllObservers()

Notifying them that the context has been destroyed while doing so.
An operation that was previously (only) done by the LifecycleNotifier
destructor.

Also provide a convenience method over ExecutionContext for clearing
out its observers:

  void ExecutionContext::removeAllLifecycleObservers()

The operation is used here when shutting down and finalizing a worker
thread and its global scope -- explicitly severing the connection
of all observers to its execution context before performing the
final garbage collection.

Doing so ensures that all observer objects being finalized during that
garbage collection will not have access to its execution context object.
A heap object that may or may not have been finalized by the garbage
collector already.

R=haraken@chromium.org,ager@chromium.org,abarth@chromium.org
BUG=340522

Review URL: https://codereview.chromium.org/194923007

git-svn-id: svn://svn.chromium.org/blink/trunk@169435 bbb929c8-8fbe-4397-9dbb-9b2b20218538

As --enable-memory-info flag and web preference were removed from Chrome,
remove the corresponding WebSetting.

The internal setting is used by PerfTestRunner, so leave it, but rename to
"preciseMemoryInfoEnabled" for clarity.

BUG=350339

Review URL: https://codereview.chromium.org/196533018

git-svn-id: svn://svn.chromium.org/blink/trunk@169434 bbb929c8-8fbe-4397-9dbb-9b2b20218538

R=vsevik@chromium.org, aandrey, eustas, vsevik

Review URL: https://codereview.chromium.org/202623004

git-svn-id: svn://svn.chromium.org/blink/trunk@169433 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Similar to the CPU profiler there are top-down and bottom up allocation profile views. Data structures are renamed to use correponding naming scheme.

BUG=277984
R=alph@chromium.org

Review URL: https://codereview.chromium.org/202953003

git-svn-id: svn://svn.chromium.org/blink/trunk@169432 bbb929c8-8fbe-4397-9dbb-9b2b20218538

ReplaceSelectionCommand::makeInsertedContentRoundTrippableWithHTMLTreeBuilder should take a const reference to InsertedNodes

This method doesn't modify InsertedNodes object so that it can take a const reference.

BUG=

Review URL: https://codereview.chromium.org/200613003

git-svn-id: svn://svn.chromium.org/blink/trunk@169431 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=353087

Review URL: https://codereview.chromium.org/201683003

git-svn-id: svn://svn.chromium.org/blink/trunk@169430 bbb929c8-8fbe-4397-9dbb-9b2b20218538

There were three problems:

 1. When fullScreenEnabled() was false, HTMLMediaElement's
    enterFullscreen() was used as a fallback, but that function checks
    the setting and does nothing. This has been the case since cleanup
    from April 2013, r149489: https://codereview.chromium.org/14670004

 2. MediaController's hasVideo() returned true if any of the slave
    elements had video, which could show the button for a video element
    with no video.

 3. The fullscreen button didn't do anything for slaved media elements.

The added tests verify that (1) no fullscreen button is shown when the
fullscreen feature is disabled or (2) when there is no video track, and
(3) that clicking the fullscreen button actually enters fullscreen.

BUG=351661

Review URL: https://codereview.chromium.org/196793021

git-svn-id: svn://svn.chromium.org/blink/trunk@169429 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We need to rollback a transaction if SQLTransactionBackendSync is dying and its
transaction is in progress. However Oilpan architecture doesn't allow us to
touch associated DatabaseSync in ~SQLTransactionBackendSync.

This CL adds a PersistentHeapHashMap<WeakMember<T>, U> to
DatabaseSync to manage SQLTransactionSync, and it makes possible
to call SQLTransactionSync::rollbackIfInProgress before the destruction.

Note:
- m_database->executionContext() can be null in some functions because on-heap
  objects can live after dissociation of ExecutionContext.
- Change the argument of DatabaseSync::rollbackTransaction to a reference
  because PassRefPtr is unnecessary here and the argument is always non-null.

BUG=347902

Review URL: https://codereview.chromium.org/198793003

git-svn-id: svn://svn.chromium.org/blink/trunk@169428 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=v8:3213

Review URL: https://codereview.chromium.org/196423009

git-svn-id: svn://svn.chromium.org/blink/trunk@169427 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Instead of exposing tons of getters and setters on Element we can use an
enum of flags and treat the bits inside ElementRareData as just a flag
field like we do in Node. This removes a lot of boilerplate code and also
makes the check for style sharing a lot simpler since now we can just
OR the flags that disable sharing together. That's good since that check
is super hot in profiles.

This also lets us make all the other "flag reading" methods into simple
inline getters that call hasElementFlag() which early outs on not having
any rare data.

Review URL: https://codereview.chromium.org/202633006

git-svn-id: svn://svn.chromium.org/blink/trunk@169426 bbb929c8-8fbe-4397-9dbb-9b2b20218538

1.5% of the time in the peacekeeper benchmark (and 2% of the recalcStyle
time) is spent inside adjustRenderStyle. This patch is the start of
trying to make adjustRenderStyle do less work in the common cases.

This patch moves the logic for adjusting the style based on the display
property out into a function and adds an early return for when the
element is display: block and is not floating since none of the other
sections of the function are needed then.

Review URL: https://codereview.chromium.org/202283003

git-svn-id: svn://svn.chromium.org/blink/trunk@169425 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch fixes a bug where resources in element.animate() calls were
not being fetched. This resulted in StylePendingImages leaking out of
the style resolving stage and into rendering which causes a crash.

This patch fixes the bug by ensuring StylePendingImages introduced in
applyAnimatedProperties() will be resolved as non-pending StyleImages
through an additional call to loadPendingResources().
This patch also modifies AnimatableImage to be a wrapper of CSSValues
instead of StyleImages in order to follow the same resource caching
logic as typical CSS image values.
This patch also updates EffectInput::convert() to pass the document's
StyleSheetContents to MutableStylePropertySet::setProperty() so that
relative URLs are correctly resolved to absolute URLs.

BUG=353385

Review URL: https://codereview.chromium.org/196573030

git-svn-id: svn://svn.chromium.org/blink/trunk@169424 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This makes it possible to override the default tickmark behavior for a frame.
The default behavior is to display tickmarks for find results. However the PDF
plugin handles finding itself but still wants to set tickmarks. This CL makes
it possible to specify an array of tickmarks to display.

BUG=303491

Review URL: https://codereview.chromium.org/198683002

git-svn-id: svn://svn.chromium.org/blink/trunk@169423 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=353257
TBR=vsevik

Review URL: https://codereview.chromium.org/202893003

git-svn-id: svn://svn.chromium.org/blink/trunk@169422 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=353257
TBR=vsevik

Review URL: https://codereview.chromium.org/203043003

git-svn-id: svn://svn.chromium.org/blink/trunk@169421 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Revert of Reapply 169354 "DevTools: track stylesheets by document, not local frame" (https://codereview.chromium.org/202623003/)

Reason for revert:
Seems like there are more failures.

Original issue's description:
> Reapply 169354 "DevTools: track stylesheets by document, not local frame"
> 
> BUG= 353257
> TBR=vsevik
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=169413

TBR=vsevik@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG= 353257

Review URL: https://codereview.chromium.org/203113002

git-svn-id: svn://svn.chromium.org/blink/trunk@169420 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Remove unused word width hypen code from computePreferredLogicalWidths
as it is no longer used.

TBR=leviw@chromium.org
BUG=

Review URL: https://codereview.chromium.org/195973008

git-svn-id: svn://svn.chromium.org/blink/trunk@169419 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Remove unused constructTextRunInternal function from RenderBlockFlow.

TBR=dglazkov@chromium.org
BUG=

Review URL: https://codereview.chromium.org/197283029

git-svn-id: svn://svn.chromium.org/blink/trunk@169418 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Unskip complex text tests failures caused by harfbuzz upgrade as the
harfbuzz version has been rolled back.

BUG=353015
TBR=wangxianzhu@chromium.org

Review URL: https://codereview.chromium.org/202153004

git-svn-id: svn://svn.chromium.org/blink/trunk@169417 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch sets default 'flex' property for resource-view items so that
the won't shrink and overlap.

BUG=352254

Review URL: https://codereview.chromium.org/196343022

git-svn-id: svn://svn.chromium.org/blink/trunk@169416 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This is the last inconsistent use of |imp| instead of |impl| in Chromium C++.
(There are a few Objective C++ uses that I'm not touching.)

Followup to
Rename |imp| => |impl| in bindings generation
https://codereview.chromium.org/202203009/

TBR=abarth

Review URL: https://codereview.chromium.org/203073002

git-svn-id: svn://svn.chromium.org/blink/trunk@169415 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Blink should not share StyleSheetContents cache between different WebFrames.

If shares, WebFrameImpl::reload uses StyleSheetContents created by another WebFrameImpl.

So when changing allowImage setting (controlled by chrome) and reloading some page, the
reloaded frame might use StyleSheetContents cache.

Since StyleImages are cached in CSSValues and are not managed by FrameLoader, allowImage
change doesn't work.

BUG=346264
TEST=fast/css/background-image-reload.html, fast/css/two-different-iframe-not-share-style-image.html

Review URL: https://codereview.chromium.org/179873014

git-svn-id: svn://svn.chromium.org/blink/trunk@169414 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG= 353257
TBR=vsevik

Review URL: https://codereview.chromium.org/202623003

git-svn-id: svn://svn.chromium.org/blink/trunk@169413 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Replace elementHasLegalLinkAttribute() helper with virtual function on Element:
hasLegalLinkAttribute(). elementHasLegalLinkAttribute() was a bit complex due
to all the branching. Using a virtual function distributes the code to each
class and results in clearer code.

R=tkent, adamk

Review URL: https://codereview.chromium.org/202893002

git-svn-id: svn://svn.chromium.org/blink/trunk@169412 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Now slightly fewer files are recompiled when touching the header.

BUG=none

Review URL: https://codereview.chromium.org/196353003

git-svn-id: svn://svn.chromium.org/blink/trunk@169411 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Blink is calling new API createAudioDevice().
It needs to remove.

TODO(crogers): remove deprecated API as soon as WebKit calls new API.

This patch is related with https://codereview.chromium.org/201303006/

BUG=353110

Review URL: https://codereview.chromium.org/201723003

git-svn-id: svn://svn.chromium.org/blink/trunk@169410 bbb929c8-8fbe-4397-9dbb-9b2b20218538

1.6% of the overall profile, and 2% of the recalcStyle, in the PeaceKeeper
benchmark is spent inside cacheUserAgentBorderAndBackground() because even
though there is an appearance guard on the constructor, we still run
constructors and destructors for all the members which aren't free since lots
of them have a lot of fields to initialize (ex. FillLayers and BorderData).

This patch moves the check from the constructor of CachedUAStyle into
cacheUserAgentBorderAndBackground() so that we can skip all the
initialization for the common case where the element has no appearance.

Review URL: https://codereview.chromium.org/202223002

git-svn-id: svn://svn.chromium.org/blink/trunk@169408 bbb929c8-8fbe-4397-9dbb-9b2b20218538

One type testing mechanism to rule them all.

BUG=none

Review URL: https://codereview.chromium.org/200713004

git-svn-id: svn://svn.chromium.org/blink/trunk@169407 bbb929c8-8fbe-4397-9dbb-9b2b20218538

adjustRenderStyle is 2% of the recalcStyle cost and 1.5% of the time
in the peacekeeper benchmark. This patch adds an early out when adjusting
the element's style based on the tagName since after one match on the
tagName no other cases can match. It also adds a check for <div> and <span>
which are the most common elements and never need adjustment to skip
all the rest of the tagName checks.

Review URL: https://codereview.chromium.org/202343003

git-svn-id: svn://svn.chromium.org/blink/trunk@169406 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Involving the media controller plainly does not make sense.

There was no test coverage for the CC button with a media controller
present, and adding a negative test for it does not seem useful.

Note that what remains is still not great since it doesn't allow you to
pick between multiple tracks, but that will be fixed at a later time.

BUG=341813
TEST=LayoutTests/media/audio-controls-captions.html
     LayoutTests/media/video-controls-captions.html

Review URL: https://codereview.chromium.org/201123004

git-svn-id: svn://svn.chromium.org/blink/trunk@169405 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch parse the sizes attribute into vector of IntSize, also added sizes in WebIconURL, so it could be used by chromium 

BUG=349039

Review URL: https://codereview.chromium.org/196743002

git-svn-id: svn://svn.chromium.org/blink/trunk@169404 bbb929c8-8fbe-4397-9dbb-9b2b20218538