GitLab

This is based on an earlier patch from <fs@opera.com>.

The performance regression mostly came from excessive tree walks
on SVG_SvgCubics (with only a tiny part of it came also from
extra invalidations).

This change re-introduces what
SVGRenderSupport::checkForSVGRepaintingDuringLayout was doing to
prevent invalidation by checking if your parent had a transform
change, except that we use it to avoid entire subtree walks.

On my Z620, this changes the results of SVG_SvgCubics as follows:
- RAL without patch: 16 ms (std: 0.5)
- RAL with patch: 8.8 ms (std: 0.4)
- Pre-RAL: 9.2 ms (std: 0.3)

BUG=381138

Review URL: https://codereview.chromium.org/327093003

git-svn-id: svn://svn.chromium.org/blink/trunk@175983 bbb929c8-8fbe-4397-9dbb-9b2b20218538

The Dock button should be the rightmost one, otherwise it gets overlapped
by a button to its right (Settings).

R=eustas, pfeldman@chromium.org, vsevik
BUG=383313

Review URL: https://codereview.chromium.org/329113002

git-svn-id: svn://svn.chromium.org/blink/trunk@175978 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL is a bit of a parlor trick, but it seems worthwhile to remove
DeprecatedDirtyCompositingDuringCompositingUpdate.

Review URL: https://codereview.chromium.org/334443005

git-svn-id: svn://svn.chromium.org/blink/trunk@175977 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Review URL: https://codereview.chromium.org/320423004

Review URL: https://codereview.chromium.org/328113005

git-svn-id: svn://svn.chromium.org/blink/trunk@175976 bbb929c8-8fbe-4397-9dbb-9b2b20218538

There is no callers for these functions in
Chromium so they can be safely removed.

Review URL: https://codereview.chromium.org/334473002

git-svn-id: svn://svn.chromium.org/blink/trunk@175975 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This is somewhat hot (about 0.8% of samples while painting glyphs), is
really trivial once inlined (a few moves and a few adds).

SkRect.h is a pretty lightweight header, so this shouldn't affect build
times very much.

Review URL: https://codereview.chromium.org/328163003

git-svn-id: svn://svn.chromium.org/blink/trunk@175974 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Actually two refactorings for navigator.serviceWorker.ready which is
sensitive to lifetimes. No change in behavior.

First, ServiceWorkerContainer is a DOMWindowLifecycleObserver to
access a window for postMessage. But it already has an execution
context that can provide the DOMWindow. Use that instead.

Second, adds an unrelated assertion that the container's
ServiceWorkerContainerClient has been cleaned up by the time that the
container is deleted. To do otherwise is to leave the
ServiceWorkerContainerClient with a dangling pointer to the container.

BUG=363967

Review URL: https://codereview.chromium.org/328663010

git-svn-id: svn://svn.chromium.org/blink/trunk@175973 bbb929c8-8fbe-4397-9dbb-9b2b20218538

* MaskImageChanged: No one did anything based on this notification.
* VideoChanged: There's only want caller and it's simpler to just
  setNeedsCompositingUpdate, which does the same thing.
* FullScreenChanged: No one ever sent this notification.

Review URL: https://codereview.chromium.org/334463002

git-svn-id: svn://svn.chromium.org/blink/trunk@175972 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=340522
TBR=darin

Review URL: https://codereview.chromium.org/324483002

git-svn-id: svn://svn.chromium.org/blink/trunk@175970 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This CL starts the renaming of RenderObject methods from repaint to
paint invalidation. In this CL I've converted the following:
 repaint
 repaintRectangle
 repaintUsingContainer
 repaintAfterLayoutIfNeeded
 repaintOverflow
 repaintOverflowIfNeed

BUG=338691

Review URL: https://codereview.chromium.org/318803002

git-svn-id: svn://svn.chromium.org/blink/trunk@175969 bbb929c8-8fbe-4397-9dbb-9b2b20218538

We don't need the templates in ContainerNodeAlgorithms now,
lets start merging them into ContainerNode.

Review URL: https://codereview.chromium.org/329703002

git-svn-id: svn://svn.chromium.org/blink/trunk@175968 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Replace compile time guard with runtime flag in blink and remove ENABLE_FAST_MOBILE_SCROLLING in gyp file.

The effort needs to be done in 3 steps:
1. Create the runtime flag interface in blink;
2. Enable the flag for Android from chromium even it does not do anything in blink;
3. Replace compile time guard with runtime flag in blink and remove ENABLE_FAST_MOBILE_SCROLLING in gyp file.

This is the 3rd set of changes.

BUG=381391

Review URL: https://codereview.chromium.org/323123004

git-svn-id: svn://svn.chromium.org/blink/trunk@175967 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Reason for revert:
Suspected cause of crash when running Speedometer on the perf bots: https://code.google.com/p/chromium/issues/detail?id=383338

Original issue's description:
> Enable Oilpan by default in modules/speech/
> 
> R=dmazzoni@chromium.org,haraken@chromium.org,tommi@chromium.org,tkent@chromium.org
> BUG=340522
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=175872

TBR=oilpan-reviews@chromium.org,haraken@chromium.org,dmazzoni@chromium.org,tommyw@chromium.org,tommi@chromium.org,tkent@chromium.org,sigbjornf@opera.com
NOTREECHECKS=true
NOTRY=true
BUG=340522

Review URL: https://codereview.chromium.org/330463002

git-svn-id: svn://svn.chromium.org/blink/trunk@175965 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This will allow debugger to pause on "load", "error" and etc. event breakpoints only
when they happen on XHR event targets.

BUG=381470
R=yurys, pfeldman@chromium.org

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=175882

Review URL: https://codereview.chromium.org/320933003

git-svn-id: svn://svn.chromium.org/blink/trunk@175964 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=245436,327641

Review URL: https://codereview.chromium.org/315003008

git-svn-id: svn://svn.chromium.org/blink/trunk@175963 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This change started as a removal of stale entries in the file
(mostly related to font-stretch - that was rolled out - along
with some misc values).

Then I looked at the file and found out that we still referred
to KHTML and CSS_PROP* (none of which make any sense) so I
cleaned up the comments too.

Review URL: https://codereview.chromium.org/324223004

git-svn-id: svn://svn.chromium.org/blink/trunk@175962 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Change the compositing/gestures tests to avoid a problem where one pixel
is one shade of green lighter due to the underline bleeding through the
rounded corners of the rect. Also change position/size calculation to
use getBoundingClientRect which is faster and more reliable.

BUG=378610
R=dglazkov@chromium.org

Review URL: https://codereview.chromium.org/322253004

git-svn-id: svn://svn.chromium.org/blink/trunk@175961 bbb929c8-8fbe-4397-9dbb-9b2b20218538

GraphicsContext::scale takes a FloatSize, which is not the way most
callers wish to invoke it. Here we change it to take x and y scale.

GraphicsContext::getCTM is not truly reflective of the tranforms
applied to get a pixel on the screen, as it ignores compositor applied
transforms. There are numerous locations where the CTM is used to test
pixel-level results, such as the width of a line. These tests are not
accurate in the presence of high DPI or other zoom factors. Each
location gets a FIXME indicating the problem.

R=fmalita,esprehn
BUG=383070,382491

Review URL: https://codereview.chromium.org/323013004

git-svn-id: svn://svn.chromium.org/blink/trunk@175960 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=None
R=loislo@chromium.org

Review URL: https://codereview.chromium.org/324403004

git-svn-id: svn://svn.chromium.org/blink/trunk@175958 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Move this Page-owned controller object to the heap, allowing a pair
of Persistent<> references to be removed in the process.

R=ager@chromium.org
BUG=340522

Review URL: https://codereview.chromium.org/328133002

git-svn-id: svn://svn.chromium.org/blink/trunk@175957 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch introduces WebInspector.Throttler and uses it for throttling
calls to PageAgent in OverridesSupport, as well as throttling calls in StylesSourceMapping.

Throttler is a helper object that throttles execution of processes
(possibly asynchronous).
Throttler is created with a single parameter - throttle interval T.
Throttler satisfies to the following contract: for every two consecutive
runs performed by throttler, time between the end of the first run and
start of the second run is always greater or equal to T.

Throttler.schedule has additional second parameter "asSoonAsPossible" which essentially
sets throttler timeout into 0 for the next process run.

Review URL: https://codereview.chromium.org/319143002

git-svn-id: svn://svn.chromium.org/blink/trunk@175956 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Partial revert of r174548: Make several functions in SVGRenderSupport templated
for performance. This is an speculative attempt to fix a crash in
WebCore::SVGRenderSupport::computeContainerBoundingBoxes<WebCore::RenderSVGRoot>().

R=pdr@chromium.org
BUG=382428

Review URL: https://codereview.chromium.org/323043005

git-svn-id: svn://svn.chromium.org/blink/trunk@175955 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Revert of DevTools: notify backend agents about profiler started/stopped events (https://codereview.chromium.org/321173002/)

Reason for revert:
We came to the conclusion that this patch
moves us in a wrong direction. 
With this approach we would have to manage 4 pairs of states. Disabled+stopped, disabled+started, enabled+stopped, etc up to enabled+started :)
in the each agent. So we would like to try the simplest one, just use enable/disable states and fix the problems related to them.

Original issue's description:
> DevTools: notify backend agents about profiler started/stopped events
> 
> This patch provides a way to notify backend agents about starting a profiler.
> Each agent should switch to a mode which doesn't affect the performance.
> As example DebuggerAgent needs to disable breakpoints, allow v8 optimizations, etc.
> DOM and Network agents may simple disable themselves.
> 
> BUG=381115
> R=pfeldman@chromium.org
> 
> Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=175946

TBR=pfeldman@chromium.org,vsevik@chromium.org
NOTREECHECKS=true
NOTRY=true
BUG=381115

Review URL: https://codereview.chromium.org/327203002

git-svn-id: svn://svn.chromium.org/blink/trunk@175954 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Simplify this stack allocated object's handling of heap references.

R=ager@chromium.org
BUG=340522

Review URL: https://codereview.chromium.org/328113004

git-svn-id: svn://svn.chromium.org/blink/trunk@175953 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Introduce a new class MIDIAccessInitializer which initializes MIDIAccess.
Previously the initialization code were on MIDIAccess itself, but this CL
decouples it.

This CL changes the various initialization-related resource ownerships.
In this CL MIDIAccessInitializer owns all initialization related resources
and when the initialization is done (no matter whether successfull or not)
it releases such resources.

Unfortunately MIDIAccess owns MIDIAccessInitializer in this CL because
there is no other natural owner of MIDIAccessInitializer.
Hopefully MIDIAccess will stop owning MIDIAccessInitializer when [1] lands.

1. https://codereview.chromium.org/311733004/

BUG=361041

Review URL: https://codereview.chromium.org/311773003

git-svn-id: svn://svn.chromium.org/blink/trunk@175952 bbb929c8-8fbe-4397-9dbb-9b2b20218538

RawPtr<> can be used for off-heap classes.  e.g. RawPtrWillBeMember<OffHeap>
So, we should check if T is on-heap for ParamStorageTraits<RawPtr<T>>.

This CL also introduces IsGarbageCollectedType<T> struct to simplify on-heap
class check.

BUG=

Review URL: https://codereview.chromium.org/322343002

git-svn-id: svn://svn.chromium.org/blink/trunk@175951 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Avoid inheriting the zoom-factor when setting the CanvasRenderingContext2D
'font' property to the same font as is computed for the <canvas> element.
In this case the font descriptions would be considered the same, and hence
the font size computed using the effective zoom of the page would be used.

BUG=381989

Review URL: https://codereview.chromium.org/326893003

git-svn-id: svn://svn.chromium.org/blink/trunk@175950 bbb929c8-8fbe-4397-9dbb-9b2b20218538

NOTRY=true
BUG=

Review URL: https://codereview.chromium.org/322303003

git-svn-id: svn://svn.chromium.org/blink/trunk@175949 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Review URL: https://codereview.chromium.org/320423004

BUG=381471

Review URL: https://codereview.chromium.org/321053005

git-svn-id: svn://svn.chromium.org/blink/trunk@175948 bbb929c8-8fbe-4397-9dbb-9b2b20218538

BUG=340522

Review URL: https://codereview.chromium.org/298873004

git-svn-id: svn://svn.chromium.org/blink/trunk@175947 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch provides a way to notify backend agents about starting a profiler.
Each agent should switch to a mode which doesn't affect the performance.
As example DebuggerAgent needs to disable breakpoints, allow v8 optimizations, etc.
DOM and Network agents may simple disable themselves.

BUG=381115
R=pfeldman@chromium.org

Review URL: https://codereview.chromium.org/321173002

git-svn-id: svn://svn.chromium.org/blink/trunk@175946 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This patch introduces Blink specific editing events usage with |addEventListener| for
 - textInput
 - webkitBeforeTextInserted
 - webkitEditableContentChanged

We would like to deprecate these Blink specific events and replace them standard one, e.g. beforeinput, or use another way to implement editing functionality if community agreed.

The usage data collected from use counter introduced by this patch are important for decision.

BUG=382814
TEST=n/a

Review URL: https://codereview.chromium.org/328513004

git-svn-id: svn://svn.chromium.org/blink/trunk@175945 bbb929c8-8fbe-4397-9dbb-9b2b20218538

- This CL adds an ASSERT to ScriptValue to verify that ScriptValue::v8Value() is used in the same world that created the ScriptValue.

- To insert the check, we need to get a current world in ScriptValue::v8Value(). For that goal, this CL makes sure that we're in some context when calling ScriptValue::v8Value().

BUG=341032

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=175837

Review URL: https://codereview.chromium.org/314953005

git-svn-id: svn://svn.chromium.org/blink/trunk@175944 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Safeguard for webMediaPlayer() access in createImageBitmap() incase webMediaPlayer() and m_readystate of media element are out of sync

BUG=350571

Review URL: https://codereview.chromium.org/320843002

git-svn-id: svn://svn.chromium.org/blink/trunk@175943 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Since the renderer for the plugin always gets either a NormalLayer
or a ForcedLayer, it will have a RenderLayer whether it's composited
or not, so all we have to do is update it's self painting layer bit.

This is a followup to https://codereview.chromium.org/328553004.

Review URL: https://codereview.chromium.org/322193002

git-svn-id: svn://svn.chromium.org/blink/trunk@175942 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This function doesn't do anything. The repaint it generates is redundant and
marking the RenderLayerCompositor as needing a rebuild is redundant because
RenderLayerStackingNode will already mark the same dirty bit. It's more
sensible for RenderLayerStackingNode to be responsible for setting this dirty
bit because the reason it needs to be set is that the stacking tree has changed.

Review URL: https://codereview.chromium.org/325063003

git-svn-id: svn://svn.chromium.org/blink/trunk@175941 bbb929c8-8fbe-4397-9dbb-9b2b20218538

They generate Element factory function like |PassRefPtrWillBeRawPtr<T>
create(const QualifiedName&, Document&)|.

We're going to change element factories at least twice.
 1. Replace adoptRefWillBeRefCountedGarbageCollected with adoptRefWillBeNoop
 2-1. Replace PassRefPtrWillBeRawPtr<T> with T* and remove adoptRefWillBeNoop if
   we switch to Oilpan successfully.
 2-2. Replace PassRefPtrWillBeRawPtr<T> with PassRefPtr<T>, and
   adoptRefWillBeNoop with adoptRef if we decide not to switch Oilpan.

This CL makes these changes easier.

BUG=357163

Review URL: https://codereview.chromium.org/327133002

git-svn-id: svn://svn.chromium.org/blink/trunk@175940 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This call site was already trigginer a CompositingUpdateAfterCompositingInputChange.
Calling updateLayerCompositingState could upgrade to CompositingUpdateRebuildTree,
but there's no reason to do that eagerly. We compute whether that upgrade is
necessary during the compositing update itself.

BUG=343756

Review URL: https://codereview.chromium.org/327963005

git-svn-id: svn://svn.chromium.org/blink/trunk@175939 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This function ignores both its arguments and is one line. It's clearer if
we just inline it and match it up with the call to
setNeedsCompositingInputsUpdate.

Review URL: https://codereview.chromium.org/326253002

git-svn-id: svn://svn.chromium.org/blink/trunk@175938 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Conceptually, there's no reason for RenderLayerScrollableArea to call
updateLayerCompositingState and no tests fail when the call is removed.

BUG=343756,336251

Review URL: https://codereview.chromium.org/321283003

git-svn-id: svn://svn.chromium.org/blink/trunk@175937 bbb929c8-8fbe-4397-9dbb-9b2b20218538