WebKit-https.git
4 years agoRemoving IndexedDB WebsiteData can fail for some users.
beidson@apple.com [Tue, 2 Aug 2016 21:28:12 +0000 (21:28 +0000)]
Removing IndexedDB WebsiteData can fail for some users.
https://bugs.webkit.org/show_bug.cgi?id=160463

Reviewed by Alex Christensen.

Source/WebCore:

Covered by changes to API test IndexedDB.StoreBlobThenDelete.

For a few months in trunk WebKit, deleting an IndexedDB using WebsiteDataStore API would only
delete the "IndexedDB.sqlite3" file, but would leave the "-wal" and "-shm" files behind.

We'd then fail to delete the directory itself, tricking API clients to think there are still
databases in the given origin.

Furthermore, the code to delete a database first checks for the existence of "IndexedDB.sqlite3"
then refuses to do any further cleanup if it doesn't exist.

We should just always run the entire delete procedure even if IndexedDB.sqlite does not exist.

* Modules/indexeddb/server/IDBServer.cpp:
(WebCore::IDBServer::removeAllDatabasesForOriginPath):

* platform/sql/SQLiteFileSystem.cpp:
(WebCore::SQLiteFileSystem::deleteDatabaseFile):

Tools:

* TestWebKitAPI/Tests/WebKit2Cocoa/StoreBlobThenDelete.mm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204041 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFix Mac CMake build.
achristensen@apple.com [Tue, 2 Aug 2016 21:17:37 +0000 (21:17 +0000)]
Fix Mac CMake build.

* PlatformMac.cmake:
Some ObjC bindings have been frozen.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204040 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoMarking two media tests as flaky crashes on ios-simulator-wk2
ryanhaddad@apple.com [Tue, 2 Aug 2016 21:15:04 +0000 (21:15 +0000)]
Marking two media tests as flaky crashes on ios-simulator-wk2
https://bugs.webkit.org/show_bug.cgi?id=160367

Unreviewed test gardening.

* platform/ios-simulator-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204039 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoMore cleanup of MathML operator parsing
fred.wang@free.fr [Tue, 2 Aug 2016 20:26:08 +0000 (20:26 +0000)]
More cleanup of MathML operator parsing
https://bugs.webkit.org/show_bug.cgi?id=160336

Patch by Frederic Wang <fwang@igalia.com> on 2016-08-02
Reviewed by Darin Adler.

No new tests, behavior is unchanged.

* rendering/mathml/RenderMathMLBlock.cpp:
(WebCore::parseMathMLLength): Deleted. This function is no longer used.
* rendering/mathml/RenderMathMLBlock.h: Ditto.
* rendering/mathml/RenderMathMLFencedOperator.h: Move some length members from
RenderMathMLOperator and override the accessors. Also override useMathOperator() since it is
always true for anonymous mfenced operators.
* rendering/mathml/RenderMathMLOperator.cpp:
(WebCore::RenderMathMLOperator::textContent): No need to ASSERT !isAnonymous since this
function is overriden by RenderMathMLFencedOperator.
(WebCore::RenderMathMLOperator::isInvisibleOperator): Use a temporary variable to call
textContent only once.
(WebCore::RenderMathMLOperator::leadingSpace): Helper function to determine the leading space.
(WebCore::RenderMathMLOperator::trailingSpace): Helper function to determine the trailing space.
(WebCore::RenderMathMLOperator::minSize): Helper function to determine the min space.
(WebCore::RenderMathMLOperator::maxSize): Helper function to determine the max space.
(WebCore::RenderMathMLOperator::setOperatorProperties): This is now only used to set m_isVertical.
(WebCore::RenderMathMLOperator::stretchTo): Call maxSize and minSize when necessary.
(WebCore::RenderMathMLOperator::computePreferredLogicalWidths): Call leadingSpace and trailingSpace.
(WebCore::RenderMathMLOperator::layoutBlock): Ditto.
(WebCore::RenderMathMLOperator::useMathOperator): Remove the case for anonymous operators as
this is now handled in the derived class.
(WebCore::RenderMathMLOperator::paint): Call leadingSpace.
* rendering/mathml/RenderMathMLOperator.h: Remove length members and declare helper functions
to get these lengths from MathMLOperatorElement.
(WebCore::RenderMathMLOperator::setLeadingSpace): Deleted. This is no longer used.
(WebCore::RenderMathMLOperator::setTrailingSpace): Deleted. This is no longer used.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204038 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoMove parsing of operator length attributes to MathMLOperatorElement
fred.wang@free.fr [Tue, 2 Aug 2016 19:53:57 +0000 (19:53 +0000)]
Move parsing of operator length attributes to MathMLOperatorElement
https://bugs.webkit.org/show_bug.cgi?id=160301

Patch by Frederic Wang <fwang@igalia.com> on 2016-08-02
Reviewed by Darin Adler.

Source/WebCore:

We move parsing of operator attributes lspace, rspace minsize, maxsize into the class
MathMLOperatorElement. We rely on the existing MathMLElement::Length structure and expand
it to handle the "infinity" case of maxsize that we always resolve as intMaxForLayoutUnit.
We also improve the resolution of a negative value for minsize and maxsize since the MathML
recommendation says it should be rounded up to the nearest valid value (which is zero)
instead of ignoring the attribute and using the default value. This is similar to r203285
for mfrac@linethickness. We also update the existing test for minsize/maxsize to take into
account that change.

No new tests, already covered by existing tests.

* mathml/MathMLElement.h: Add an "Infinity" type for MathML Length.
* mathml/MathMLOperatorElement.cpp:
(WebCore::MathMLOperatorElement::leadingSpace): Helper function to retrieve the cached value
for the lspace attribute, parse it if necessary.
(WebCore::MathMLOperatorElement::trailingSpace): Ditto for rspace.
(WebCore::MathMLOperatorElement::minSize): Ditto for minsize.
(WebCore::MathMLOperatorElement::maxSize): Ditto for maxsize. This attribute also accepts the
"infinity" value.
(WebCore::MathMLOperatorElement::parseAttribute): Mark attributes dirty.
* mathml/MathMLOperatorElement.h: Declare Length members and accessors.
* rendering/mathml/RenderMathMLBlock.cpp:
(WebCore::toUserUnits): Resolve Infinity as intMaxForLayoutUnit.
* rendering/mathml/RenderMathMLOperator.cpp:
(WebCore::RenderMathMLOperator::setOperatorProperties): We use toUserUnits to resolve the
lengths. Negative maxsize and minsize values now fallback to 0 instead of their default
values. We also remove the !isAnonymous() check since setOperatorProperties is overriden
in the RenderMathMLFencedOperator class.

LayoutTests:

* mathml/presentation/stretchy-minsize-maxsize-expected.html: Update the test for negative
minsize and maxsize to align on the behavior suggested in the MathML recommendation.
* mathml/presentation/stretchy-minsize-maxsize.html: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204037 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed build error fix.
fred.wang@free.fr [Tue, 2 Aug 2016 19:23:26 +0000 (19:23 +0000)]
Unreviewed build error fix.

* mathml/MathMLOperatorElement.cpp:
(WebCore::propertyFlagToAttributeName): Return a nullQName to avoid warnings on some platforms.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204036 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoStop building Objective-C bindings that aren't exposed anywhere
andersca@apple.com [Tue, 2 Aug 2016 19:16:08 +0000 (19:16 +0000)]
Stop building Objective-C bindings that aren't exposed anywhere
https://bugs.webkit.org/show_bug.cgi?id=160458

Reviewed by Darin Adler.

* DerivedSources.make:
* WebCore.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204035 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoHTMLCollection's named getter should only do 'name' attribute matching for HTMLElements
cdumez@apple.com [Tue, 2 Aug 2016 19:13:42 +0000 (19:13 +0000)]
HTMLCollection's named getter should only do 'name' attribute matching for HTMLElements
https://bugs.webkit.org/show_bug.cgi?id=160456

Reviewed by Darin Adler.

LayoutTests/imported/w3c:

* web-platform-tests/dom/nodes/Document-getElementsByTagName-expected.txt:
Rebaseline. We are still failing but on the next check.

* web-platform-tests/dom/nodes/Element-children-expected.txt:
Rebaseline now that one more check is passing.

Source/WebCore:

HTMLCollection's named getter should only do 'name' attribute matching for HTMLElements:
- https://dom.spec.whatwg.org/#dom-htmlcollection-nameditem

Our slow path which relies on HTMLCollection::updateNamedElementCache() does the right
thing. However, we have a fast path in
CachedHTMLCollection<HTMLCollectionClass, traversalType>::namedItem() that was missing a
check.

Firefox and Chrome both behave correctly here.

No new tests, rebaselined existing tests.

* html/CachedHTMLCollection.h:
(WebCore::traversalType>::namedItem):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204034 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoKill [StrictTypeChecking] IDL extended attribute
cdumez@apple.com [Tue, 2 Aug 2016 19:11:27 +0000 (19:11 +0000)]
Kill [StrictTypeChecking] IDL extended attribute
https://bugs.webkit.org/show_bug.cgi?id=160382

Reviewed by Darin Adler.

Kill [StrictTypeChecking] IDL extended attribute. Its only remaining
use was on DOMString parameters for a few CanvasRenderingContext2D
operations in order to work around a bug in our overload resolution.

However, as of r204028, we implement the Web IDL overload resolution
algorithm and [StrictTypeChecking] is now a no-op and is no longer
needed.

* bindings/scripts/CodeGeneratorJS.pm:
(GenerateParametersCheckExpression):
* bindings/scripts/IDLAttributes.txt:
* bindings/scripts/test/TestObj.idl:
* bindings/scripts/test/TestTypedefs.idl:
* html/canvas/CanvasRenderingContext2D.idl:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204033 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoMove parsing of boolean operator properties into MathMLOperatorElement
fred.wang@free.fr [Tue, 2 Aug 2016 18:47:14 +0000 (18:47 +0000)]
Move parsing of boolean operator properties into MathMLOperatorElement
https://bugs.webkit.org/show_bug.cgi?id=160190

Patch by Frederic Wang <fwang@igalia.com> on 2016-08-02
Reviewed by Darin Adler.

No new tests, already covered by existing tests.

* mathml/MathMLOperatorDictionary.h: Add a bit mask with all the property flags set.
* mathml/MathMLOperatorElement.cpp:
(WebCore::attributeNameToPropertyFlag): helper function to map from attribute name to flag.
(WebCore::MathMLOperatorElement::computeOperatorFlag): Helper function to update one
bit of m_properties.flags from the corresponding boolean attribute. The default value is
taken from the operator dictionary data stored in m_dictionaryProperty.flags.
(WebCore::MathMLOperatorElement::hasProperty): Returns whether the operator has a property,
parsing it again if the corresponding attribute is dirty.
(WebCore::MathMLOperatorElement::childrenChanged): Make all properties dirty.
(WebCore::propertyFlagToAttributeName): helper function to map from flag to attribute name.
(WebCore::MathMLOperatorElement::parseAttribute): Make all properties dirty if the form
attribute changed. Make each property dirty when the corresponding attribute changed.
(WebCore::MathMLOperatorElement::flags): Deleted. Replaced with the finer hasProperty function.
* mathml/MathMLOperatorElement.h: Define new structure, member and functions to handle
operator properties.
* rendering/mathml/RenderMathMLFencedOperator.cpp:
(WebCore::RenderMathMLFencedOperator::RenderMathMLFencedOperator): Move m_operatorFlags from
the base class to the derived class.
* rendering/mathml/RenderMathMLFencedOperator.h: Ditto.
* rendering/mathml/RenderMathMLOperator.cpp:
(WebCore::RenderMathMLOperator::RenderMathMLOperator): Ditto.
(WebCore::RenderMathMLOperator::hasOperatorFlag): Just call hasOperatorFlag from the
MathMLOperatorElement class.
(WebCore::RenderMathMLOperator::setOperatorProperties): We do not initialize m_operatorFlags
since it has been removed from the base class. We also do not parse attributes since this
has been moved to the MathMLOperatorElement class.
(WebCore::RenderMathMLOperator::setOperatorFlagFromAttribute): Deleted.
(WebCore::RenderMathMLOperator::setOperatorFlagFromAttributeValue): Deleted.
* rendering/mathml/RenderMathMLOperator.h: Move m_operatorFlags from the base class to the
derived class, remove some parsing helper functions and update hasOperatorFlag to make it
overridable.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204032 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFreeze another bunch of Objective-C binding files
andersca@apple.com [Tue, 2 Aug 2016 17:42:27 +0000 (17:42 +0000)]
Freeze another bunch of Objective-C binding files
https://bugs.webkit.org/show_bug.cgi?id=160435

Reviewed by Alex Christensen.

* DerivedSources.make:
* WebCore.xcodeproj/project.pbxproj:
* bindings/objc/DOMAbstractView.h: Added.
* bindings/objc/DOMAbstractView.mm:
* bindings/objc/DOMAbstractViewInternal.h: Added.
* bindings/objc/DOMMediaList.h: Added.
* bindings/objc/DOMMediaList.mm: Added.
* bindings/objc/DOMMediaListInternal.h: Added.
* bindings/objc/DOMNodeFilter.h: Added.
* bindings/objc/DOMNodeIterator.h: Added.
* bindings/objc/DOMNodeIterator.mm: Added.
* bindings/objc/DOMNodeIteratorInternal.h: Added.
* bindings/objc/DOMRange.h: Added.
* bindings/objc/DOMRange.mm: Added.
* bindings/objc/DOMRangeInternal.h: Added.
* bindings/objc/DOMStyleSheet.h: Added.
* bindings/objc/DOMStyleSheet.mm: Added.
* bindings/objc/DOMStyleSheetInternal.h: Added.
* bindings/objc/DOMStyleSheetList.h: Added.
* bindings/objc/DOMStyleSheetList.mm: Added.
* bindings/objc/DOMStyleSheetListInternal.h: Added.
* bindings/objc/DOMTreeWalker.h: Added.
* bindings/objc/DOMTreeWalker.mm: Added.
* bindings/objc/DOMTreeWalkerInternal.h: Added.
* bindings/objc/DOMXPathExpression.h: Added.
* bindings/objc/DOMXPathExpression.mm: Added.
* bindings/objc/DOMXPathExpressionInternal.h: Added.
* bindings/objc/DOMXPathNSResolver.h: Added.
* bindings/objc/DOMXPathResult.h: Added.
* bindings/objc/DOMXPathResult.mm: Added.
* bindings/objc/DOMXPathResultInternal.h: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204031 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[MathML] Use more auto for local variables
commit-queue@webkit.org [Tue, 2 Aug 2016 17:27:37 +0000 (17:27 +0000)]
[MathML] Use more auto for local variables
https://bugs.webkit.org/show_bug.cgi?id=160453

Patch by Frederic Wang <fwang@igalia.com> on 2016-08-02
Reviewed by Darin Adler.

No new tests, behavior is unchanged.

* mathml/MathMLElement.cpp:
(WebCore::MathMLElement::colSpan):
(WebCore::MathMLElement::rowSpan):
(WebCore::MathMLElement::childShouldCreateRenderer):
(WebCore::MathMLElement::defaultEventHandler):
* mathml/MathMLSelectElement.cpp:
(WebCore::MathMLSelectElement::getSelectedActionChild):
(WebCore::MathMLSelectElement::getSelectedSemanticsChild):
* rendering/mathml/RenderMathMLFenced.cpp:
(WebCore::RenderMathMLFenced::updateFromElement):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204030 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoAttempt to fix build after r204014.
ryanhaddad@apple.com [Tue, 2 Aug 2016 17:03:40 +0000 (17:03 +0000)]
Attempt to fix build after r204014.

Unreviewed build fix.

* loader/mac/ResourceLoaderMac.mm:
(WebCore::ResourceLoader::willCacheResponse):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204029 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[WebIDL] Implement overload resolution algorithm
cdumez@apple.com [Tue, 2 Aug 2016 17:02:23 +0000 (17:02 +0000)]
[WebIDL] Implement overload resolution algorithm
https://bugs.webkit.org/show_bug.cgi?id=160394

Reviewed by Darin Adler.

Source/WebCore:

Implement overload resolution algorithm:
- http://heycam.github.io/webidl/#es-overloads

This means that our support for operation overloading in our IDL is
now a lot more extensive than it used to be and is now compliant with
the Web IDL specification.

In particular, overloading should now work for a lot more parameter
types which means that:
- We should be able to drop some custom bindings code in a follow-up patch.
- We will be able to drop the [StrictTypeChecking] attribute which was
  a hack used to make overloading work for parameters of type DOMString.

Also, the order of the overloads in the IDL no longer impacts the
generated bindings code. It used to be that you needed to put the
overloads with the more specific parameter types first in order for the
generated bindings code to be somewhat correct.

No new tests, rebaselined bindings tests.

* bindings/scripts/CodeGenerator.pm:
(IsStringOrEnumType):
* bindings/scripts/CodeGeneratorJS.pm:
(IsNullableType):
(StripNullable):
(ComputeEffectiveOverloadSet):
(AreTypesDistinguishableForOverloadResolution):
(GetDistinguishingArgumentIndex):
(GetOverloadThatMatches):
(GenerateOverloadedFunction):
* bindings/scripts/test/JS/JSTestGlobalObject.cpp:
(WebCore::jsTestGlobalObjectInstanceFunctionEnabledAtRuntimeOperation):
* bindings/scripts/test/JS/JSTestObj.cpp:
(WebCore::jsTestObjPrototypeFunctionEnabledAtRuntimeOperation):
(WebCore::jsTestObjPrototypeFunctionOverloadedMethod):
(WebCore::jsTestObjPrototypeFunctionOverloadedMethodWithOptionalParameter1):
(WebCore::jsTestObjPrototypeFunctionOverloadedMethodWithOptionalParameter):
(WebCore::jsTestObjConstructorFunctionOverloadedMethod1):
(WebCore::jsTestObjPrototypeFunctionTestPromiseOverloadedFunction):
(WebCore::jsTestObjConstructorFunctionOverloadedMethod12): Deleted.
(WebCore::jsTestObjPrototypeFunctionTestPromiseOverloadedFunction2Promise): Deleted.
* bindings/scripts/test/TestObj.idl:
* dom/EventTarget.h:
* dom/EventTarget.idl:
* html/HTMLOptionsCollection.h:
* html/HTMLOptionsCollection.idl:
* html/HTMLSelectElement.h:
(WebCore::HTMLSelectElement::add):
* html/HTMLSelectElement.idl:

LayoutTests:

Update / Rebaseline existing test. There is no major behavior change.
We sometimes get a different exception message than we used to.

* fast/canvas/canvas-clip-path-expected.txt:
* fast/canvas/canvas-fill-path-expected.txt:
* fast/canvas/canvas-path-addPath-expected.txt:
* fast/canvas/canvas-path-isPointInPath-expected.txt:
* fast/canvas/canvas-path-isPointInStroke-expected.txt:
* fast/canvas/canvas-putImageData-expected.txt:
* fast/canvas/canvas-putImageData.js:
* fast/canvas/canvas-stroke-path-expected.txt:
* fast/canvas/webgl/script-tests/texImageTest.js:
* fast/canvas/webgl/texImageTest-expected.txt:
* fast/dom/HTMLSelectElement/add-expected.txt:
* fast/dom/HTMLSelectElement/add.html:
* fast/dom/HTMLSelectElement/options-collection-add-expected.txt:
* fast/dom/HTMLSelectElement/options-collection-add.html:
* fast/dom/incompatible-operations-expected.txt:
* js/dom/select-options-add-expected.txt:
* webaudio/audiobuffer-expected.txt:
* webaudio/audiobuffer.html:
* webgl/1.0.2/resources/webgl_test_files/conformance/textures/tex-image-with-invalid-data.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204028 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[macOS] Guard workaround in r203314 for only OSes which need it
mmaxfield@apple.com [Tue, 2 Aug 2016 16:57:20 +0000 (16:57 +0000)]
[macOS] Guard workaround in r203314 for only OSes which need it
https://bugs.webkit.org/show_bug.cgi?id=160440

Reviewed by Simon Fraser.

The workarond introduced in r203314 only needs to be applied on
certain OSes.

No new tests because there is no behavior change.

* platform/text/mac/TextBoundaries.mm:
(WebCore::findNextWordFromIndex):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204027 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoRemove CachedResourceRequest initiatingDocumentLoader
commit-queue@webkit.org [Tue, 2 Aug 2016 15:46:10 +0000 (15:46 +0000)]
Remove CachedResourceRequest initiatingDocumentLoader
https://bugs.webkit.org/show_bug.cgi?id=160449

Patch by Youenn Fablet <youenn@apple.com> on 2016-08-02
Reviewed by Alex Christensen.

No change of behavior.

* loader/DocumentLoader.cpp:
(WebCore::DocumentLoader::startLoadingMainResource): Removing call to setInitiator.
* loader/cache/CachedResourceRequest.cpp:
(WebCore::CachedResourceRequest::setInitiator): Deleted.
* loader/cache/CachedResourceRequest.h:
(WebCore::CachedResourceRequest::initiatingDocumentLoader): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204026 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[ARM] Disable Inline Caching on ARMv7 traditional until proper fix
ossy@webkit.org [Tue, 2 Aug 2016 15:37:23 +0000 (15:37 +0000)]
[ARM] Disable Inline Caching on ARMv7 traditional until proper fix
https://bugs.webkit.org/show_bug.cgi?id=159759

Reviewed by Saam Barati.

* jit/JITMathIC.h:
(JSC::JITMathIC::generateInline):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204025 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed test gardening.
pvollan@apple.com [Tue, 2 Aug 2016 13:22:33 +0000 (13:22 +0000)]
Unreviewed test gardening.

Update test expectations on Windows for many failing tests. These failures are tracked in
https://bugs.webkit.org/show_bug.cgi?id=160447.

* platform/win/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204024 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoSkip fetch API worker tests in Debug mode
commit-queue@webkit.org [Tue, 2 Aug 2016 12:45:04 +0000 (12:45 +0000)]
Skip fetch API worker tests in Debug mode
https://bugs.webkit.org/show_bug.cgi?id=160448

Unreviewed.

Patch by Youenn Fablet <youenn@apple.com> on 2016-08-02

LayoutTests/imported/w3c:

* web-platform-tests/fetch/api/cors/cors-cookies-worker-expected.txt: Rebasing

LayoutTests:

* TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204023 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoMove parsing of the form attribute to MathMLOperatorElement
fred.wang@free.fr [Tue, 2 Aug 2016 11:27:19 +0000 (11:27 +0000)]
Move parsing of the form attribute to MathMLOperatorElement
https://bugs.webkit.org/show_bug.cgi?id=160239

Patch by Frederic Wang <fwang@igalia.com> on 2016-07-28
Reviewed by Darin Adler.

We move the parsing of the mo@form attribute to the MathMLOperatorElement class. Note that
when the attribute is not explicit, the form may also be guessed by searching into the
operator dictionary. Hence we also start moving the determination of the default dictionary
properties. Moving the actual parsing of the corresponding attributes will be done in
follow-up commits.

No new tests, already covered by existing tests.

* mathml/MathMLInlineContainerElement.cpp:
(WebCore::MathMLInlineContainerElement::childrenChanged): Make the form of operators dirty
if its siblings have changed.
* mathml/MathMLOperatorDictionary.cpp:
(WebCore::MathMLOperatorDictionary::search): Merge old getEntry functions into one helper
function that returns an Optional<Entry>.
(WebCore::MathMLOperatorDictionary::getEntry): Deleted
* mathml/MathMLOperatorDictionary.h: Declare MathMLOperatorDictionary::find.
* mathml/MathMLOperatorElement.cpp:
(WebCore::MathMLOperatorElement::dictionaryProperty): We determine the operator form by
parsing the corresponding attribute or by using heuristics if that attribute is not
specified. We also read dictionary the corresponding dictionary properties.
(WebCore::MathMLOperatorElement::flags): Return the flags read from the dictionary.
(WebCore::MathMLOperatorElement::defaultLeadingSpace): Return the space read from the dictionary.
(WebCore::MathMLOperatorElement::defaultTrailingSpace): Return the space read from the dictionary.
(WebCore::MathMLOperatorElement::childrenChanged): Make the dictionary properties dirty.
(WebCore::MathMLOperatorElement::parseAttribute): Make the dictionary properties dirty when
the form changes.
* mathml/MathMLOperatorElement.h: New member to store dictionary properties and expose them.
* rendering/mathml/RenderMathMLFencedOperator.cpp:
(WebCore::RenderMathMLFencedOperator::setOperatorProperties): We implement the case specific
to anonymous mfenced operators here.
* rendering/mathml/RenderMathMLFencedOperator.h: Move the m_operatorForm member here and
declare the overriden function setOperatorProperties.
* rendering/mathml/RenderMathMLOperator.cpp:
(WebCore::RenderMathMLOperator::setOperatorProperties): Simplify this code since the logic
has been moved to the element classes.
(WebCore::RenderMathMLOperator::setOperatorPropertiesFromOpDictEntry): Deleted.
* rendering/mathml/RenderMathMLOperator.h: Remove setOperatorPropertiesFromOpDictEntry, make
members accessible to RenderMathMLFencedOperator and remove m_operatorForm.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204022 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUse Optional members to store parsed MathML attributes.
fred.wang@free.fr [Tue, 2 Aug 2016 09:43:01 +0000 (09:43 +0000)]
Use Optional members to store parsed MathML attributes.
https://bugs.webkit.org/show_bug.cgi?id=160400

Patch by Frederic Wang <fwang.igalia.com> on 2016-08-02
Reviewed by Darin Adler.

Parsed MathML attributes are stored on the element classes using the memoize pattern to
minimize the number of parsing updates. Currently, a dirty flag for each member
indicate when it must be parsed again. We change this to wrap these members into an
Optional<T> container instead, where a null value indicates that the member is dirty.

No new tests, behavior is unchanged.

* mathml/MathMLElement.cpp:
(WebCore::MathMLElement::cachedMathMLLength):
(WebCore::MathMLElement::cachedBooleanAttribute):
(WebCore::MathMLElement::parseMathVariantAttribute):
(WebCore::MathMLElement::specifiedDisplayStyle):
(WebCore::MathMLElement::specifiedMathVariant):
* mathml/MathMLElement.h:
* mathml/MathMLFractionElement.cpp:
(WebCore::MathMLFractionElement::lineThickness):
(WebCore::MathMLFractionElement::cachedFractionAlignment):
(WebCore::MathMLFractionElement::parseAttribute):
* mathml/MathMLFractionElement.h:
* mathml/MathMLInlineContainerElement.cpp:
(WebCore::MathMLInlineContainerElement::parseAttribute):
* mathml/MathMLMathElement.cpp:
(WebCore::MathMLMathElement::specifiedDisplayStyle):
(WebCore::MathMLMathElement::parseAttribute):
* mathml/MathMLOperatorElement.cpp:
(WebCore::MathMLOperatorElement::operatorText):
* mathml/MathMLPaddedElement.cpp:
(WebCore::MathMLPaddedElement::parseAttribute):
* mathml/MathMLPaddedElement.h:
* mathml/MathMLScriptsElement.cpp:
(WebCore::MathMLScriptsElement::parseAttribute):
* mathml/MathMLScriptsElement.h:
* mathml/MathMLSpaceElement.cpp:
(WebCore::MathMLSpaceElement::parseAttribute):
* mathml/MathMLSpaceElement.h:
* mathml/MathMLTextElement.cpp:
(WebCore::MathMLTextElement::parseAttribute):
* mathml/MathMLUnderOverElement.cpp:
(WebCore::MathMLUnderOverElement::parseAttribute):
* mathml/MathMLUnderOverElement.h:
* rendering/mathml/MathMLStyle.cpp:
(WebCore::MathMLStyle::resolveMathMLStyle):
* rendering/mathml/MathMLStyle.h:
* rendering/mathml/RenderMathMLToken.cpp:
(WebCore::mathVariant):
(WebCore::RenderMathMLToken::updateMathVariantGlyph):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204021 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Fetch API] Fetch promises should not reject or resolve when ActiveDOMObjects are...
commit-queue@webkit.org [Tue, 2 Aug 2016 08:07:28 +0000 (08:07 +0000)]
[Fetch API] Fetch promises should not reject or resolve when ActiveDOMObjects are being stopped
https://bugs.webkit.org/show_bug.cgi?id=160420

Patch by Youenn Fablet <youenn@apple.com> on 2016-08-02
Reviewed by Alex Christensen.

LayoutTests/imported/w3c:

* web-platform-tests/fetch/api/cors/cors-preflight-redirect-worker-expected.txt: Added.
* web-platform-tests/fetch/nosniff/worker-expected.txt: Added.

Source/WebCore:

Test: http/tests/fetch/fetch-in-worker-crash.html

Promises should not be rejected when FetchResponse/FetchRequest are stopped as ActiveDOMObject.

* Modules/fetch/FetchBody.h:
(WebCore::FetchBody::cleanConsumePromise): Allowing to nullify the promise used to consume body.
* Modules/fetch/FetchBodyOwner.cpp:
(WebCore::FetchBodyOwner::stop): Nullifying the body consume promise if any.
* Modules/fetch/FetchResponse.cpp:
(WebCore::FetchResponse::BodyLoader::stop): Nullifying the fetch promise if any.

LayoutTests:

* TestExpectations:
* http/tests/fetch/fetch-in-worker-crash-expected.txt: Added.
* http/tests/fetch/fetch-in-worker-crash.html: Added.
* http/tests/fetch/fetch-in-worker.js: Added.
(onmessage):
* platform/ios-simulator-wk2/imported/w3c/web-platform-tests/fetch/api/cors/cors-basic-worker-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204020 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Fetch API] Add support to ReferrerPolicy
commit-queue@webkit.org [Tue, 2 Aug 2016 07:20:23 +0000 (07:20 +0000)]
[Fetch API] Add support to ReferrerPolicy
https://bugs.webkit.org/show_bug.cgi?id=160404

Patch by Youenn Fablet <youenn@apple.com> on 2016-08-02
Reviewed by Alex Christensen.

LayoutTests/imported/w3c:

* web-platform-tests/fetch/api/basic/referrer-expected.txt: Added.
* web-platform-tests/fetch/api/basic/referrer-worker-expected.txt: Added.
* web-platform-tests/fetch/api/basic/referrer-worker.html: Added.
* web-platform-tests/fetch/api/basic/referrer.html: Added.
* web-platform-tests/fetch/api/basic/referrer.js: Added.
(runTest):
* web-platform-tests/fetch/api/cors/cors-preflight-referrer-expected.txt:
* web-platform-tests/fetch/api/cors/cors-preflight-referrer-worker-expected.txt:
* web-platform-tests/fetch/api/cors/cors-preflight-referrer.js:
(corsPreflightReferrer): fix infrastructure bug in test and fix expected referrer policy in default case.
* web-platform-tests/fetch/api/cors/cors-preflight-worker-expected.txt:

Source/WebCore:

Tests: imported/w3c/web-platform-tests/fetch/api/basic/referrer-worker.html
       imported/w3c/web-platform-tests/fetch/api/basic/referrer.html

Refactored ThreadableLoader API to take more references.

Updated loading code in CachedResource to take into account referrer policy, that can be set by fetch API.
To enable correct handling of OriginWhenCrossOrigin policy, the computation of the cross-origin status of a request is
moved from ResourceLoader to CachedResource.

Referrer is passed to the loading code through HTTP headers.
This was triggering preflighting for Workers request as WorkerThreadableLoader was setting the referrer.
It is now done in DocumentThreadableLoader::loadRequest so that preflighting is done before setting the referrer.
Note that this referrer setting is only a way to pass a specific referrer value to the loading code.
CachedResource code will compute the actual referer value based on the referrer policy.
It might be good in the future to have a different way of conveying this information to CachedResource, maybe
through CachedResourceRequest.

* Modules/fetch/FetchLoader.cpp:
(WebCore::FetchLoader::start): Refactoring of ThreadableLoader API.
* Modules/fetch/FetchRequest.cpp:
(WebCore::FetchRequest::internalRequest): Setting referrer if it has a specific value.
* fileapi/FileReaderLoader.cpp:
(WebCore::FileReaderLoader::start): Ractoring of ThreadableLoader API.
* loader/CrossOriginPreflightChecker.cpp:
(WebCore::CrossOriginPreflightChecker::startPreflight): Set referrer in the preflight request if any.
(WebCore::CrossOriginPreflightChecker::doPreflight): Ditto.
* loader/DocumentThreadableLoader.cpp:
(WebCore::DocumentThreadableLoader::loadResourceSynchronously): Refactoring of ThreadableLoader API.
(WebCore::DocumentThreadableLoader::create): Ditto.
(WebCore::DocumentThreadableLoader::DocumentThreadableLoader): Ditto.
(WebCore::DocumentThreadableLoader::loadRequest): Ditto.
* loader/DocumentThreadableLoader.h:
(WebCore::DocumentThreadableLoader::referrer): Storing referrer in case one is defined.
* loader/SubresourceLoader.cpp:
(WebCore::SubresourceLoader::init): Removing of cross origin computation since it is done earlier in CachedResource.
* loader/ThreadableLoader.cpp:
(WebCore::ThreadableLoader::create): Refactoring of ThreadableLoader API.
(WebCore::ThreadableLoader::loadResourceSynchronously): Ditto.
* loader/ThreadableLoader.h:
* loader/ThreadableLoaderClientWrapper.h:
(WebCore::ThreadableLoaderClientWrapper::create): Ditto.
(WebCore::ThreadableLoaderClientWrapper::ThreadableLoaderClientWrapper): Ditto.
* loader/WorkerThreadableLoader.cpp:
(WebCore::WorkerThreadableLoader::WorkerThreadableLoader): Ditto.
(WebCore::WorkerThreadableLoader::loadResourceSynchronously): Removing setting of referrer in header and passing
it as a parameter.
(WebCore::WorkerThreadableLoader::MainThreadBridge::MainThreadBridge): Ditto.
* loader/WorkerThreadableLoader.h:
(WebCore::WorkerThreadableLoader::create): Refactoring of ThreadableLoader API.
* loader/cache/CachedResource.cpp:
(WebCore::addAdditionalRequestHeadersToRequest): Computation of referrer based on referrer policy.
(WebCore::CachedResource::addAdditionalRequestHeaders): Ditto.
(WebCore::CachedResource::load): Computation of cross origin status.
(WebCore::CachedResource::varyHeaderValuesMatch):
* loader/cache/CachedResource.h:
(WebCore::CachedResource::options):
* page/EventSource.cpp:
(WebCore::EventSource::connect): Refactoring of ThreadableLoader API.
* workers/WorkerScriptLoader.cpp:
(WebCore::WorkerScriptLoader::loadSynchronously): Ditto.
(WebCore::WorkerScriptLoader::loadAsynchronously): Ditto.
* xml/XMLHttpRequest.cpp:
(WebCore::XMLHttpRequest::createRequest): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204019 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoRefactor data passed along for a "GetRecord" request.
beidson@apple.com [Tue, 2 Aug 2016 07:19:58 +0000 (07:19 +0000)]
Refactor data passed along for a "GetRecord" request.
https://bugs.webkit.org/show_bug.cgi?id=160352

Reviewed by Tim Horton.

Source/WebCore:

No new tests (No behavior change).

Adding this new object lets us easily add new data passed along for GetRecord
requests going forward.

* WebCore.xcodeproj/project.pbxproj:
* CMakeLists.txt:

* Modules/indexeddb/IDBObjectStore.cpp:
(WebCore::IDBObjectStore::get):

* Modules/indexeddb/IDBTransaction.cpp:
(WebCore::IDBTransaction::requestGetRecord):
(WebCore::IDBTransaction::requestIndexRecord):
(WebCore::IDBTransaction::getRecordOnServer):
* Modules/indexeddb/IDBTransaction.h:

* Modules/indexeddb/client/IDBConnectionProxy.cpp:
(WebCore::IDBClient::IDBConnectionProxy::getRecord):
* Modules/indexeddb/client/IDBConnectionProxy.h:

* Modules/indexeddb/client/IDBConnectionToServer.cpp:
(WebCore::IDBClient::IDBConnectionToServer::getRecord):
* Modules/indexeddb/client/IDBConnectionToServer.h:
* Modules/indexeddb/client/IDBConnectionToServerDelegate.h:

* Modules/indexeddb/server/IDBServer.cpp:
(WebCore::IDBServer::IDBServer::getRecord):
* Modules/indexeddb/server/IDBServer.h:

* Modules/indexeddb/server/UniqueIDBDatabase.cpp:
(WebCore::IDBServer::UniqueIDBDatabase::getRecord):
* Modules/indexeddb/server/UniqueIDBDatabase.h:

* Modules/indexeddb/server/UniqueIDBDatabaseTransaction.cpp:
(WebCore::IDBServer::UniqueIDBDatabaseTransaction::getRecord):
* Modules/indexeddb/server/UniqueIDBDatabaseTransaction.h:

* Modules/indexeddb/shared/IDBGetRecordData.cpp: Added.
(WebCore::IDBGetRecordData::isolatedCopy):
* Modules/indexeddb/shared/IDBGetRecordData.h: Added.
(WebCore::IDBGetRecordData::encode):
(WebCore::IDBGetRecordData::decode):

* Modules/indexeddb/shared/InProcessIDBServer.cpp:
(WebCore::InProcessIDBServer::getRecord):
* Modules/indexeddb/shared/InProcessIDBServer.h:

Source/WebKit2:

* DatabaseProcess/IndexedDB/WebIDBConnectionToClient.cpp:
(WebKit::WebIDBConnectionToClient::getRecord):
* DatabaseProcess/IndexedDB/WebIDBConnectionToClient.h:
* DatabaseProcess/IndexedDB/WebIDBConnectionToClient.messages.in:

* WebProcess/Databases/IndexedDB/WebIDBConnectionToServer.cpp:
(WebKit::WebIDBConnectionToServer::getRecord):
* WebProcess/Databases/IndexedDB/WebIDBConnectionToServer.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204018 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Fetch API] Pass directly FetchRequest fetch options to ThreadableLoader
commit-queue@webkit.org [Tue, 2 Aug 2016 06:46:57 +0000 (06:46 +0000)]
[Fetch API] Pass directly FetchRequest fetch options to ThreadableLoader
https://bugs.webkit.org/show_bug.cgi?id=160407

Patch by Youenn Fablet <youenn@apple.com> on 2016-08-01
Reviewed by Sam Weinig.

No change of behavior.

Added a ResourceLoaderOptions constructor that takes a FetchOptions as input.
Removed setters and getters of ResourceLoaderOptions as it is a struct.
Updated member fields accordingly.

Updated code accordingly.
Removed explicit option setting if it is the same as the default value.

* Modules/fetch/FetchLoader.cpp:
(WebCore::FetchLoader::start):
* css/CSSFontFaceSrcValue.cpp:
(WebCore::CSSFontFaceSrcValue::cachedFont):
* css/CSSImageSetValue.cpp:
(WebCore::CSSImageSetValue::cachedImageSet):
* css/CSSImageValue.cpp:
(WebCore::CSSImageValue::cachedImage):
* dom/ScriptElement.cpp:
(WebCore::ScriptElement::requestScript):
* fileapi/FileReaderLoader.cpp:
(WebCore::FileReaderLoader::start):
* html/HTMLLinkElement.cpp:
(WebCore::HTMLLinkElement::process):
* inspector/InspectorNetworkAgent.cpp:
(WebCore::InspectorNetworkAgent::loadResource):
* loader/CrossOriginPreflightChecker.cpp:
(WebCore::CrossOriginPreflightChecker::validatePreflightResponse):
(WebCore::CrossOriginPreflightChecker::startPreflight):
* loader/DocumentThreadableLoader.cpp:
(WebCore::DocumentThreadableLoader::DocumentThreadableLoader):
(WebCore::DocumentThreadableLoader::makeCrossOriginAccessRequest):
(WebCore::DocumentThreadableLoader::makeSimpleCrossOriginAccessRequest):
(WebCore::DocumentThreadableLoader::redirectReceived):
(WebCore::DocumentThreadableLoader::didReceiveResponse):
(WebCore::DocumentThreadableLoader::preflightSuccess):
(WebCore::DocumentThreadableLoader::loadRequest):
* loader/ImageLoader.cpp:
(WebCore::ImageLoader::updateFromElement):
* loader/ResourceLoader.cpp:
(WebCore::ResourceLoader::ResourceLoader):
(WebCore::ResourceLoader::init):
(WebCore::ResourceLoader::start):
(WebCore::ResourceLoader::setDefersLoading):
(WebCore::ResourceLoader::setDataBufferingPolicy):
(WebCore::ResourceLoader::addDataOrBuffer):
(WebCore::ResourceLoader::willSendRequestInternal):
(WebCore::ResourceLoader::didReceiveResponse):
(WebCore::ResourceLoader::didReceiveDataOrBuffer):
(WebCore::ResourceLoader::didFinishLoadingOnePart):
(WebCore::ResourceLoader::cleanupForError):
(WebCore::ResourceLoader::shouldUseCredentialStorage):
(WebCore::ResourceLoader::didReceiveAuthenticationChallenge):
* loader/ResourceLoader.h:
(WebCore::ResourceLoader::shouldSendResourceLoadCallbacks):
(WebCore::ResourceLoader::setSendCallbackPolicy):
(WebCore::ResourceLoader::shouldSniffContent):
(WebCore::ResourceLoader::shouldIncludeCertificateInfo):
* loader/ResourceLoaderOptions.h:
(WebCore::ResourceLoaderOptions::ResourceLoaderOptions):
(WebCore::ResourceLoaderOptions::sendLoadCallbacks): Deleted.
(WebCore::ResourceLoaderOptions::setSendLoadCallbacks): Deleted.
(WebCore::ResourceLoaderOptions::sniffContent): Deleted.
(WebCore::ResourceLoaderOptions::setSniffContent): Deleted.
(WebCore::ResourceLoaderOptions::dataBufferingPolicy): Deleted.
(WebCore::ResourceLoaderOptions::setDataBufferingPolicy): Deleted.
(WebCore::ResourceLoaderOptions::allowCredentials): Deleted.
(WebCore::ResourceLoaderOptions::setAllowCredentials): Deleted.
(WebCore::ResourceLoaderOptions::securityCheck): Deleted.
(WebCore::ResourceLoaderOptions::setSecurityCheck): Deleted.
(WebCore::ResourceLoaderOptions::certificateInfoPolicy): Deleted.
(WebCore::ResourceLoaderOptions::setCertificateInfoPolicy): Deleted.
(WebCore::ResourceLoaderOptions::contentSecurityPolicyImposition): Deleted.
(WebCore::ResourceLoaderOptions::setContentSecurityPolicyImposition): Deleted.
(WebCore::ResourceLoaderOptions::defersLoadingPolicy): Deleted.
(WebCore::ResourceLoaderOptions::setDefersLoadingPolicy): Deleted.
(WebCore::ResourceLoaderOptions::cachingPolicy): Deleted.
(WebCore::ResourceLoaderOptions::setCachingPolicy): Deleted.
* loader/SubresourceLoader.cpp:
(WebCore::SubresourceLoader::checkRedirectionCrossOriginAccessControl):
* loader/TextTrackLoader.cpp:
(WebCore::TextTrackLoader::load):
* loader/cache/CachedRawResource.cpp:
(WebCore::CachedRawResource::setDataBufferingPolicy):
* loader/cache/CachedResource.cpp:
(WebCore::CachedResource::load):
* loader/cache/CachedResource.h:
(WebCore::CachedResource::shouldSendResourceLoadCallbacks):
(WebCore::CachedResource::dataBufferingPolicy):
(WebCore::CachedResource::allowsCaching):
* loader/cache/CachedResourceLoader.cpp:
(WebCore::CachedResourceLoader::canRequest):
* loader/cache/CachedResourceRequest.cpp:
(WebCore::CachedResourceRequest::setAsPotentiallyCrossOrigin):
* loader/cache/CachedResourceRequest.h:
(WebCore::CachedResourceRequest::allowsCaching):
* loader/cache/CachedResourceRequestInitiators.cpp:
(WebCore::CachedResourceRequestInitiators::CachedResourceRequestInitiators):
* loader/cache/CachedResourceRequestInitiators.h:
* loader/mac/ResourceLoaderMac.mm:
(WebCore::ResourceLoader::willCacheResponse):
(WebCore::ResourceLoader::didReceiveDataArray):
* page/EventSource.cpp:
(WebCore::EventSource::connect):
* style/StylePendingResources.cpp:
(WebCore::Style::loadPendingImage):
(WebCore::Style::loadPendingSVGFilters):
* svg/SVGFEImageElement.cpp:
(WebCore::SVGFEImageElement::requestImageResource):
* svg/SVGFontFaceUriElement.cpp:
(WebCore::SVGFontFaceUriElement::loadFont):
* svg/SVGUseElement.cpp:
(WebCore::SVGUseElement::updateExternalDocument):
* workers/WorkerScriptLoader.cpp:
(WebCore::WorkerScriptLoader::loadSynchronously):
(WebCore::WorkerScriptLoader::loadAsynchronously):
* xml/XMLHttpRequest.cpp:
(WebCore::XMLHttpRequest::createRequest):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204014 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[GTK] Move the redirected XComposite window to the web process
carlosgc@webkit.org [Tue, 2 Aug 2016 06:42:18 +0000 (06:42 +0000)]
[GTK] Move the redirected XComposite window to the web process
https://bugs.webkit.org/show_bug.cgi?id=160389

Reviewed by Žan Doberšek.

Source/WebCore:

Add helper methods to PlatformDisplayX11 to query X extensions supported by the display.

* platform/graphics/x11/PlatformDisplayX11.cpp:
(WebCore::PlatformDisplayX11::supportsXComposite):
(WebCore::PlatformDisplayX11::supportsXDamage):
* platform/graphics/x11/PlatformDisplayX11.h:

Source/WebKit2:

In the current code, the UI process creates the redirected window that the web process uses to render
accelerated contents. The redirected window is sent to the web process as native surface handle, and using
XDamage extension the UI process takes a pixmap of the redirected window to render into the widget when there
are updates. This requires several points of synchronization between UI and web processes. When the web view is
resized, the UI process first resizes the redirected window and then sends a new backing store ID to the web
process. The time between the redirected window is resized and the web process renders the new contents the UI
process keeps rendering the previous contents with the previous size in the new window with the new size. This
makes the resize process slow, and it produces rendering artifacts quite often. The redirected window is created
when the web view is realized, to be able to inherit the XVisual from the parent window, and the native window
handle is sent to the web process. The time until the window is realized, the web process doesn't have a context
to render into, so the UI process simply renders an empty page. When the web view is unrealized, for example if
the web view is reparented, the redirected window is destroyed, and a sync message is sent to the web process to
destroy the current gl context and stop drawing. This needs to happen synchronously, because the UI process
can't remove the redirected window until the web process has stopped rendering into it. This makes also the
reparenting process quite unstable and risky.
To all those synchronization points we now have to add the synchronization with the compositing thread when
using the threaded compositor. The threaded compositor made resizing, reparenting, etc. even worse. We can't
avoid the synchronization with the threaded compositor, but we can reduce the synchronization points and improve
the current ones by moving the redirected window to the web process. In this case is web process who creates the
redirected window, so we can be sure that it always has a valid native surface handle to render into. This means
we no longer need the IPC message to send the native surface handle from the UI process to the web process, nor
the sync message to destroy it either. This also means we no longer need to wait until the view is realized to
start rendering accelerated contents, and we don't need to stop when it's unrealized either. We don't really
need to inherit the XVisual from the parent window if the redirected window always uses always a RGBA visual
when available. That way we always render into a transparent window that is composed into the web view
widget. And when the web view is resized, we no longer need to destroy the GL context either, because we use the
same redirected window as the native handle, but create a new pixmap that is what we send to the UI process as
layer tree context ID. The layer tree context ID is already sent to the UI process by the drawing area as part
of the backing store update process, so we don't need any new IPC message for this. When the web view is
resized, the UI process sends a backing store state update message to the web process that updates its size,
relayouts and then renders the new contents, so that when the update backing store state reply gets to the UI
process, we already have a new pixmap with the new contents updated. This makes resizing smooth again, and
avoids flickering and rendering artifacts. And finally all this also prevents several race conditions that were
causing X errors and web process crashes.

* PlatformGTK.cmake:
* Shared/CoordinatedGraphics/threadedcompositor/ThreadedCompositor.cpp:
(WebKit::ThreadedCompositor::create): Make it possible to optionally pass a native surface handle for
compositing to the ThreadedCompositor contructor.
(WebKit::ThreadedCompositor::ThreadedCompositor): Initialize the native surface handle and make the scene active
after the thread initialization if we already have a valid handle.
* Shared/CoordinatedGraphics/threadedcompositor/ThreadedCompositor.h:
* UIProcess/API/gtk/PageClientImpl.cpp:
(WebKit::PageClientImpl::enterAcceleratedCompositingMode): Pass the layer tree context ID to the web view.
(WebKit::PageClientImpl::updateAcceleratedCompositingMode): Notify the web view about the update.
(WebKit::PageClientImpl::willEnterAcceleratedCompositingMode): Deleted.
* UIProcess/API/gtk/PageClientImpl.h:
* UIProcess/API/gtk/WebKitWebViewBase.cpp:
(webkitWebViewBaseSetAcceleratedCompositingPixmap): Helper function to update the pixmap used to render the
accelerated contents.
(webkitWebViewBaseRealize): Remove the code to create the redirected window.
(webkitWebViewBaseUnrealize): Remove the code to destroy the redirected window.
(webkitWebViewBaseDispose): Reset the accelerated compositing pixmap.
(webkitWebViewRenderAcceleratedCompositingResults): Use the accelerated compositing pixmap surface.
(webkitWebViewBaseSizeAllocate): Remove the code to resize the redirected window.
(webkitWebViewBaseEnterAcceleratedCompositingMode): Update the accelerated compositing pixmap.
(webkitWebViewBaseUpdateAcceleratedCompositingMode): Ditto.
(webkitWebViewBaseExitAcceleratedCompositingMode): Reset the accelerated compositing pixmap.
(webkitWebViewBaseDidRelaunchWebProcess): Remove the code to send native surface handle to the web process.
(webkitWebViewBasePageClosed): Reset the accelerated compositing pixmap.
(_WebKitWebViewBasePrivate::updateViewStateTimerFired): Deleted.
* UIProcess/API/gtk/WebKitWebViewBasePrivate.h:
* UIProcess/DrawingAreaProxy.h:
* UIProcess/DrawingAreaProxy.messages.in:
* UIProcess/DrawingAreaProxyImpl.cpp:
(WebKit::DrawingAreaProxyImpl::didUpdateBackingStoreState): Handle the case when the layer tree context ID
changes without leaving the accelerated compositing mode, calling updateAcceleratedCompositingMode().
(WebKit::DrawingAreaProxyImpl::exitAcceleratedCompositingMode): Let the web view know we are leaving accelerated
compositing mode even when it's forced, since in case of web process crash we need to leave the accelerated
compositing mode in the UI process.
(WebKit::DrawingAreaProxyImpl::willEnterAcceleratedCompositingMode): This message was added only to prepare the
redirected window when the web process was about to enter accelerated compositing mode, so it's no longer needed.
(WebKit::DrawingAreaProxyImpl::setNativeSurfaceHandleForCompositing): This is now only used when building
without redirected XComposite window support.
* UIProcess/DrawingAreaProxyImpl.h:
* UIProcess/PageClient.h:
* UIProcess/WebPageProxy.cpp:
(WebKit::WebPageProxy::willEnterAcceleratedCompositingMode): Deleted.
* UIProcess/WebPageProxy.h:
* UIProcess/efl/WebView.h:
* UIProcess/gtk/RedirectedXCompositeWindow.cpp: Removed.
* UIProcess/gtk/XDamageNotifier.cpp: Helper class to be notified about XDamage events.
(WebKit::XDamageNotifier::singleton):
(WebKit::XDamageNotifier::XDamageNotifier):
(WebKit::XDamageNotifier::add):
(WebKit::XDamageNotifier::remove):
(WebKit::XDamageNotifier::filterXDamageEvent):
(WebKit::XDamageNotifier::notify):
* UIProcess/gtk/XDamageNotifier.h: Added.
* UIProcess/ios/PageClientImplIOS.h:
* UIProcess/ios/PageClientImplIOS.mm:
(WebKit::PageClientImpl::willEnterAcceleratedCompositingMode): Deleted.
* UIProcess/mac/PageClientImpl.h:
* UIProcess/mac/PageClientImpl.mm:
(WebKit::PageClientImpl::willEnterAcceleratedCompositingMode): Deleted.
* WebProcess/WebPage/AcceleratedDrawingArea.cpp:
(WebKit::AcceleratedDrawingArea::enterAcceleratedCompositingMode): Remove code to send
WillEnterAcceleratedCompositingMode message.
(WebKit::AcceleratedDrawingArea::setNativeSurfaceHandleForCompositing): This is now only used when building
without redirected XComposite window support.
* WebProcess/WebPage/AcceleratedDrawingArea.h:
* WebProcess/WebPage/CoordinatedGraphics/ThreadedCoordinatedLayerTreeHost.cpp:
(WebKit::ThreadedCoordinatedLayerTreeHost::ThreadedCoordinatedLayerTreeHost): Create a redirected window and
pass the window ID as native surface handle to the threaded compositor. Use the redirected window pixmap as
layer tree context.
(WebKit::ThreadedCoordinatedLayerTreeHost::invalidate): Destroy the redirected window.
(WebKit::ThreadedCoordinatedLayerTreeHost::deviceOrPageScaleFactorChanged): Resize the redirected window and
update the layer tree context.
(WebKit::ThreadedCoordinatedLayerTreeHost::sizeDidChange): Ditto.
* WebProcess/WebPage/CoordinatedGraphics/ThreadedCoordinatedLayerTreeHost.h:
* WebProcess/WebPage/DrawingArea.h:
* WebProcess/WebPage/DrawingArea.messages.in:
* WebProcess/WebPage/gtk/LayerTreeHostGtk.cpp:
(WebKit::LayerTreeHostGtk::LayerTreeHostGtk): Create a redirected window and texture mapper that uses the window
ID as native surface handle. Use the redirected window pixmap as layer tree context.
(WebKit::LayerTreeHostGtk::makeContextCurrent): Use the redirected window as native handle.
(WebKit::LayerTreeHostGtk::invalidate): Destroy the redirected window.
(WebKit::LayerTreeHostGtk::sizeDidChange): Resize the redirected window and update the layer tree context.
(WebKit::LayerTreeHostGtk::deviceOrPageScaleFactorChanged): Ditto.
(WebKit::LayerTreeHostGtk::createTextureMapper): Helper function to create the texture mapper.
(WebKit::LayerTreeHostGtk::setNativeSurfaceHandleForCompositing): Use createTextureMapper().
* WebProcess/WebPage/gtk/LayerTreeHostGtk.h:
* WebProcess/WebPage/gtk/RedirectedXCompositeWindow.cpp: Added.
(WebKit::RedirectedXCompositeWindow::create):
(WebKit::RedirectedXCompositeWindow::RedirectedXCompositeWindow):
(WebKit::RedirectedXCompositeWindow::~RedirectedXCompositeWindow):
(WebKit::RedirectedXCompositeWindow::resize):
* WebProcess/WebPage/gtk/RedirectedXCompositeWindow.h: Renamed from Source/WebKit2/UIProcess/gtk/RedirectedXCompositeWindow.h.
(WebKit::RedirectedXCompositeWindow::window):
(WebKit::RedirectedXCompositeWindow::pixmap):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204013 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoDatabase Process: ASSERTION FAILED: filePaths.size() == result.handles().size() with...
carlosgc@webkit.org [Tue, 2 Aug 2016 05:40:01 +0000 (05:40 +0000)]
Database Process: ASSERTION FAILED: filePaths.size() == result.handles().size() with SANDBOX_EXTENSIONS disabled
https://bugs.webkit.org/show_bug.cgi?id=160398

Reviewed by Brady Eidson.

Same assert as in bug #160188, but this time in preregisterSandboxExtensionsIfNecessary.

* DatabaseProcess/DatabaseProcess.cpp:
* DatabaseProcess/DatabaseProcess.h:
* DatabaseProcess/DatabaseProcess.messages.in:
* DatabaseProcess/IndexedDB/WebIDBConnectionToClient.cpp:
(WebKit::WebIDBConnectionToClient::handleGetResult):
* NetworkProcess/NetworkConnectionToWebProcess.cpp:
(WebKit::NetworkConnectionToWebProcess::preregisterSandboxExtensionsForOptionallyFileBackedBlob):
* UIProcess/Databases/DatabaseProcessProxy.cpp:
* UIProcess/Databases/DatabaseProcessProxy.h:
* UIProcess/Databases/DatabaseProcessProxy.messages.in:
* WebProcess/Databases/IndexedDB/WebIDBConnectionToServer.cpp:
(WebKit::preregisterSandboxExtensionsIfNecessary):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204011 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoREGRESSION (r203990): JSC Debug test stress/arity-check-ftl-throw.js failing
fpizlo@apple.com [Tue, 2 Aug 2016 05:38:58 +0000 (05:38 +0000)]
REGRESSION (r203990): JSC Debug test stress/arity-check-ftl-throw.js failing
https://bugs.webkit.org/show_bug.cgi?id=160438

Reviewed by Mark Lam.

In r203990 I fixed a bug where CommonSlowPaths.h/arityCheckFor() was basically failing at
catching stack overflow due to large parameter count. It would only catch regular old stack
overflow, like if the frame pointer was already past the limit.

This had a secondary problem: unfortunately all of our tests for what happens when you overflow
the stack due to large parameter count were not going down that path at all, so we haven't had
test coverage for this in ages.  There were bugs in all tiers of the engine when handling this
case.

We need to be able to roll back the topCallFrame on paths that are meant to throw an exception
from the caller. Otherwise, we'd crash in StackVisitor because it would see a busted stack
frame. Rolling back like this "just works" except when the caller is the VM entry frame. I had
some choices here. I could have forced anyone who is rolling back to always skip VM entry
frames. They can't do it in a way that changes the value of VM::topVMEntryFrame, which is what
a stack frame roll back normally does, since exception unwinding needs to see the current value
of topVMEntryFrame. So, we have a choice to either try to magically avoid all of the paths that
look at topCallFrame, or give topCallFrame a state that unambiguously signals that we are
sitting right on top of a VM entry frame without having succeeded at making a JS call. The only
place that really needs to know is StackVisitor, which wants to start scanning at topCallFrame.
To signal this, I could have either made topCallFrame point to the real top JS call frame
without also rolling back topVMEntryFrame, or I could make topCallFrame == topVMEntryFrame. The
latter felt somehow cleaner. I filed a bug (https://bugs.webkit.org/show_bug.cgi?id=160441) for
converting topCallFrame to a void*, which would give us a chance to harden the rest of the
engine against this case.

* interpreter/StackVisitor.cpp:
(JSC::StackVisitor::StackVisitor):
We may do ShadowChicken processing, which invokes StackVisitor, when we have topCallFrame
pointing at topVMEntryFrame. This teaches StackVisitor how to handle this case. I believe that
StackVisitor is the only place that needs to be taught about this at this time, because it's
one of the few things that access topCallFrame along this special path.

* jit/JITOperations.cpp: Roll back the top call frame.
* runtime/CommonSlowPaths.cpp:
(JSC::SLOW_PATH_DECL): Roll back the top call frame.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204010 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[JSC][ARM64] Fix branchTest32/64 taking an immediate as mask
commit-queue@webkit.org [Tue, 2 Aug 2016 05:02:27 +0000 (05:02 +0000)]
[JSC][ARM64] Fix branchTest32/64 taking an immediate as mask
https://bugs.webkit.org/show_bug.cgi?id=160439

Patch by Benjamin Poulain <bpoulain@apple.com> on 2016-08-01
Reviewed by Filip Pizlo.

Source/JavaScriptCore:

* assembler/MacroAssemblerARM64.h:
(JSC::MacroAssemblerARM64::branchTest64):
* b3/air/AirOpcode.opcodes:
Fix the ARM64 codegen to lower BitImm64 without using a scratch register.

Source/WTF:

* wtf/MathExtras.h:
(getLSBSet):
This was not working at all for MacroAssembler.
Since TrustedImm32/64 are signed integers, the arithmetic shift would
never get rid of the top bit and we get an infinite loop.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204009 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago_WKDownloadDelegate.didReceiveResponse should be called before decideDestinationWithS...
commit-queue@webkit.org [Tue, 2 Aug 2016 03:59:18 +0000 (03:59 +0000)]
_WKDownloadDelegate.didReceiveResponse should be called before decideDestinationWithSuggestedFilename
https://bugs.webkit.org/show_bug.cgi?id=160437
<rdar://problem/27578272>

Patch by Alex Christensen <achristensen@webkit.org> on 2016-08-01
Reviewed by Brady Eidson.

Source/WebKit2:

This preserves the behavior of the NSURLConnection/NSURLDownload code path.
decideDestinationWithSuggestedFilename might need some information from the didReceiveResponse callback
in order to decide what the suggested filename should be.

* NetworkProcess/Downloads/Download.cpp:
(WebKit::Download::didReceiveAuthenticationChallenge):
(WebKit::Download::didReceiveResponse):
(WebKit::Download::didReceiveData):
* NetworkProcess/Downloads/Download.h:
* NetworkProcess/NetworkConnectionToWebProcess.cpp:
(WebKit::NetworkConnectionToWebProcess::convertMainResourceLoadToDownload):
* NetworkProcess/NetworkLoad.cpp:
(WebKit::NetworkLoad::convertTaskToDownload):
(WebKit::NetworkLoad::setPendingDownloadID):
(WebKit::NetworkLoad::didReceiveResponseNetworkSession):
* NetworkProcess/NetworkLoad.h:
(WebKit::NetworkLoad::pendingDownloadID):
* NetworkProcess/NetworkProcess.cpp:
(WebKit::NetworkProcess::pendingDownloadCanceled):
(WebKit::NetworkProcess::findPendingDownloadLocation):
* NetworkProcess/NetworkProcess.h:
* NetworkProcess/cocoa/NetworkSessionCocoa.mm:
(-[WKNetworkSessionDelegate URLSession:dataTask:didBecomeDownloadTask:]):

Tools:

* TestWebKitAPI/Tests/WebKit2Cocoa/Download.mm:
(-[DownloadDelegate _download:didReceiveResponse:]):
(-[DownloadDelegate _download:decideDestinationWithSuggestedFilename:allowOverwrite:]):
(runTest):
Add a boolean to make sure that didReceiveResponse is called before decideDestinationWithSuggestedFilename.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204008 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Tools] The GDB backtrace report tool don't checks the gdb return code.
clopez@igalia.com [Tue, 2 Aug 2016 01:50:16 +0000 (01:50 +0000)]
[Tools] The GDB backtrace report tool don't checks the gdb return code.
https://bugs.webkit.org/show_bug.cgi?id=160414

Reviewed by Michael Catanzaro.

* Scripts/webkitpy/port/linux_get_crash_log.py:
(GDBCrashLogGenerator._wait_for_corefile):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204007 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoShrink MediaQuerySets to fit after parsing.
akling@apple.com [Tue, 2 Aug 2016 01:44:43 +0000 (01:44 +0000)]
Shrink MediaQuerySets to fit after parsing.
<https://webkit.org/b/160419>

Reviewed by Antti Koivisto.

Add shrinkToFit() helpers to MediaQuerySet and MediaQuery
and call it after parsing to remove extra capacity from their
internal vectors.

This reduces memory usage by 1.24MB on theverge.com.

* css/CSSParser.cpp:
(WebCore::CSSParser::createMediaRule):
* css/MediaList.cpp:
(WebCore::MediaQuerySet::parse):
(WebCore::MediaQuerySet::shrinkToFit):
* css/MediaList.h:
* css/MediaQuery.h:
(WebCore::MediaQuery::shrinkToFit):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204006 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago<attachment> action label doesn't wrap, but needs to for some localizations
timothy_horton@apple.com [Tue, 2 Aug 2016 01:39:26 +0000 (01:39 +0000)]
<attachment> action label doesn't wrap, but needs to for some localizations
https://bugs.webkit.org/show_bug.cgi?id=160431
<rdar://problem/27526464>

Reviewed by Simon Fraser.

Test: fast/attachment/attachment-wrapping-action.html

* rendering/RenderThemeIOS.mm:
(WebCore::AttachmentInfo::buildWrappedLines):
(WebCore::AttachmentInfo::AttachmentInfo):
(WebCore::AttachmentInfo::buildTitleLines): Deleted.
Rework buildTitleLines into buildWrappedLines so that we have a
generic way to add lines that wrap up to N lines and then center-truncate.

Make use of it for the action text, and follow the same rules/width as
the title text already does.

* fast/attachment/attachment-wrapping-action-expected-mismatch.html: Added.
* fast/attachment/attachment-wrapping-action.html: Added.
Add a test that ensures that the action label wraps by filling one with
non-breaking and breaking spaces, which will wrap onto two lines, and one
with just a single space, which will always fit in one line, and ensuring
that they do not match.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204005 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[iOS] Fix fast/events/touch/document-create-touch-list-ios.html
cdumez@apple.com [Tue, 2 Aug 2016 00:31:26 +0000 (00:31 +0000)]
[iOS] Fix fast/events/touch/document-create-touch-list-ios.html
https://bugs.webkit.org/show_bug.cgi?id=160429

Reviewed by Ryosuke Niwa.

Fix several bugs in fast/events/touch/document-create-touch-list-ios.html:
1. Use proper parameters to TouchEvent.initTouchEvent() so that it no longer throws.
2. For the check testing passing bad parameters to createTouchList(), print PASS
   if it throws instead of FAIL since this is the expected behavior.

* fast/events/touch/document-create-touch-list-ios-expected.txt:
* fast/events/touch/script-tests/document-create-touch-list-ios.js:
(catch): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204004 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoSierra test expectations gardening for editing/input/password-echo-textnode.html.
ap@apple.com [Tue, 2 Aug 2016 00:19:04 +0000 (00:19 +0000)]
Sierra test expectations gardening for editing/input/password-echo-textnode.html.

* platform/mac/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204002 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoSource/WebCore:
andersca@apple.com [Tue, 2 Aug 2016 00:11:55 +0000 (00:11 +0000)]
Source/WebCore:
Build fix.

* WebCore.xcodeproj/project.pbxproj:

Source/WebKit/mac:
Bring back a file - it needs to be copied by WebCore.

* MigrateHeaders.make:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204001 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoRemove another file.
andersca@apple.com [Mon, 1 Aug 2016 23:51:52 +0000 (23:51 +0000)]
Remove another file.

* MigrateHeaders.make:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@204000 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoRemove a file.
andersca@apple.com [Mon, 1 Aug 2016 23:37:16 +0000 (23:37 +0000)]
Remove a file.

* MigrateHeaders.make:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203997 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[B3] Fusing immediates into test instructions should work again
fpizlo@apple.com [Mon, 1 Aug 2016 23:22:15 +0000 (23:22 +0000)]
[B3] Fusing immediates into test instructions should work again
https://bugs.webkit.org/show_bug.cgi?id=160073

Reviewed by Sam Weinig.

When we introduced BitImm, we forgot to change the Branch(BitAnd(value, constant))
fusion.  This emits test instructions, so it should use BitImm for the constant.  But it
was still using Imm!  This meant that isValidForm() always returned false.

This fixes the code path to use BitImm, and turns off our use of BitImm64 on x86 since
it provides no benefit on x86 and has some risk (the code appears to play fast and loose
with the scratch register).

This is not an obvious progression on anything, so I added comprehensive tests to
testb3, which check that we selected the optimal instruction in a variety of situations.
We should add more tests like this!

Rolling this back in after fixing ARM64. The bug was that branchTest32|64 on ARM64 doesn't
actually support BitImm or BitImm64, at least not yet. Disabling that in AirOpcodes makes
this patch not a regression on ARM64. That change was reviewed by Benjamin Poulain.

* b3/B3BasicBlock.h:
(JSC::B3::BasicBlock::successorBlock):
* b3/B3LowerToAir.cpp:
(JSC::B3::Air::LowerToAir::createGenericCompare):
* b3/B3LowerToAir.h:
* b3/air/AirArg.cpp:
(JSC::B3::Air::Arg::isRepresentableAs):
(JSC::B3::Air::Arg::usesTmp):
* b3/air/AirArg.h:
(JSC::B3::Air::Arg::isRepresentableAs):
(JSC::B3::Air::Arg::castToType):
(JSC::B3::Air::Arg::asNumber):
* b3/air/AirCode.h:
(JSC::B3::Air::Code::size):
(JSC::B3::Air::Code::at):
* b3/air/AirOpcode.opcodes:
* b3/air/AirValidate.h:
* b3/air/opcode_generator.rb:
* b3/testb3.cpp:
(JSC::B3::compile):
(JSC::B3::compileAndRun):
(JSC::B3::lowerToAirForTesting):
(JSC::B3::testSomeEarlyRegister):
(JSC::B3::testBranchBitAndImmFusion):
(JSC::B3::zero):
(JSC::B3::run):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203996 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoAnother build fix.
andersca@apple.com [Mon, 1 Aug 2016 23:21:48 +0000 (23:21 +0000)]
Another build fix.

* WebCore.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203995 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWeb Inspector: Don't set `width:calc(100% - 0px); left:0px` on console messages
nvasilyev@apple.com [Mon, 1 Aug 2016 23:04:17 +0000 (23:04 +0000)]
Web Inspector: Don't set `width:calc(100% - 0px); left:0px` on console messages
https://bugs.webkit.org/show_bug.cgi?id=160428
<rdar://problem/27646033>

Reviewed by Timothy Hatcher.

Console messages inside console.group() are indented. Console messages outside of
console.group() aren't and there is no reason to set inline styles that do nothing
on each one of them.

* UserInterface/Views/LogContentView.js:
(WebInspector.LogContentView.prototype.didAppendConsoleMessageView):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203994 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFreeze the last of the HTML Objective-C bindings
andersca@apple.com [Mon, 1 Aug 2016 22:51:04 +0000 (22:51 +0000)]
Freeze the last of the HTML Objective-C bindings
https://bugs.webkit.org/show_bug.cgi?id=160430

Reviewed by Tim Horton.

Also, remove a bunch of form validation related methods that were never exposed/used.

* DerivedSources.make:
* WebCore.xcodeproj/project.pbxproj:
* bindings/objc/DOMBlob.h: Added.
* bindings/objc/DOMBlob.mm: Added.
* bindings/objc/DOMBlobInternal.h: Added.
* bindings/objc/DOMDOMTokenList.h: Added.
* bindings/objc/DOMDOMTokenList.mm: Added.
* bindings/objc/DOMDOMTokenListInternal.h: Added.
* bindings/objc/DOMFile.h: Added.
* bindings/objc/DOMFile.mm: Added.
* bindings/objc/DOMFileInternal.h: Added.
* bindings/objc/DOMFileList.h: Added.
* bindings/objc/DOMFileList.mm: Added.
* bindings/objc/DOMFileListInternal.h: Added.
* bindings/objc/DOMHTMLButtonElement.mm:
* bindings/objc/DOMHTMLFieldSetElement.mm:
* bindings/objc/DOMHTMLInputElement.mm:
* bindings/objc/DOMHTMLInputElementPrivate.h:
* bindings/objc/DOMHTMLObjectElement.mm:
* bindings/objc/DOMHTMLSelectElement.mm:
* bindings/objc/DOMHTMLTextAreaElement.mm:
* bindings/objc/DOMHTMLTextAreaElementPrivate.h:
* bindings/objc/DOMMediaError.h: Added.
* bindings/objc/DOMMediaError.mm: Added.
* bindings/objc/DOMMediaErrorInternal.h: Added.
* bindings/objc/DOMTimeRanges.h: Added.
* bindings/objc/DOMTimeRanges.mm: Added.
* bindings/objc/DOMTimeRangesInternal.h: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203993 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoLand updated baseline for rdar://problem/27611932.
ryanhaddad@apple.com [Mon, 1 Aug 2016 22:44:00 +0000 (22:44 +0000)]
Land updated baseline for rdar://problem/27611932.

* editing/mac/attributed-string/font-style-variant-effect-expected.txt:
* platform/mac-elcapitan/editing/mac/attributed-string/font-style-variant-effect-expected.txt: Copied from LayoutTests/editing/mac/attributed-string/font-style-variant-effect-expected.txt.
* platform/mac-wk1/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203992 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWeb Inspector: Grid column resizer elements positioned incorrectly
mattbaker@apple.com [Mon, 1 Aug 2016 22:31:02 +0000 (22:31 +0000)]
Web Inspector: Grid column resizer elements positioned incorrectly
https://bugs.webkit.org/show_bug.cgi?id=160311
<rdar://problem/27595586>

Reviewed by Brian Burg.

Perform a resize layout on a newly shown tab content view if the tab
browser was resized before showing the tab.

* UserInterface/Views/TabBrowser.js:
(WebInspector.TabBrowser):
(WebInspector.TabBrowser.prototype.layout):
Mark all recent tab content views (except for the current view)
as needing a resize layout.

(WebInspector.TabBrowser.prototype._tabBarItemSelected):
If the tab browser was resized while the view was hidden, update the
view's layout for the new content view container size.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203991 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoRationalize varargs stack overflow checks
fpizlo@apple.com [Mon, 1 Aug 2016 22:20:49 +0000 (22:20 +0000)]
Rationalize varargs stack overflow checks
https://bugs.webkit.org/show_bug.cgi?id=160425

Reviewed by Michael Saboff.

* ftl/FTLLink.cpp:
(JSC::FTL::link): AboveOrEqual 0 is a tautology. The code meant GreaterThanOrEqual, since the error code is -1.
* runtime/CommonSlowPaths.h:
(JSC::CommonSlowPaths::arityCheckFor): Use roundUpToMultipleOf(), which is almost certainly what we meant when we said %.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203990 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFix build.
andersca@apple.com [Mon, 1 Aug 2016 21:32:50 +0000 (21:32 +0000)]
Fix build.

* WebCore.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203989 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFreeze all the DOMHTML* Objective-C bindings
andersca@apple.com [Mon, 1 Aug 2016 21:16:39 +0000 (21:16 +0000)]
Freeze all the DOMHTML* Objective-C bindings
https://bugs.webkit.org/show_bug.cgi?id=160424

Reviewed by Tim Horton.

Source/WebCore:

* DerivedSources.make:
* WebCore.xcodeproj/project.pbxproj:
* bindings/objc/DOMHTMLAnchorElement.h: Added.
* bindings/objc/DOMHTMLAnchorElement.mm: Added.
* bindings/objc/DOMHTMLAnchorElementInternal.h: Added.
* bindings/objc/DOMHTMLAppletElement.h: Added.
* bindings/objc/DOMHTMLAppletElement.mm: Added.
* bindings/objc/DOMHTMLAppletElementInternal.h: Added.
* bindings/objc/DOMHTMLAreaElement.h: Added.
* bindings/objc/DOMHTMLAreaElement.mm: Added.
* bindings/objc/DOMHTMLAreaElementInternal.h: Added.
* bindings/objc/DOMHTMLBRElement.h: Added.
* bindings/objc/DOMHTMLBRElement.mm: Added.
* bindings/objc/DOMHTMLBRElementInternal.h: Added.
* bindings/objc/DOMHTMLBaseElement.h: Added.
* bindings/objc/DOMHTMLBaseElement.mm: Added.
* bindings/objc/DOMHTMLBaseElementInternal.h: Added.
* bindings/objc/DOMHTMLBaseFontElement.h: Added.
* bindings/objc/DOMHTMLBaseFontElement.mm: Added.
* bindings/objc/DOMHTMLBaseFontElementInternal.h: Added.
* bindings/objc/DOMHTMLBodyElement.h: Added.
* bindings/objc/DOMHTMLBodyElement.mm: Added.
* bindings/objc/DOMHTMLBodyElementInternal.h: Added.
* bindings/objc/DOMHTMLButtonElement.h: Added.
* bindings/objc/DOMHTMLButtonElement.mm: Added.
* bindings/objc/DOMHTMLButtonElementInternal.h: Added.
* bindings/objc/DOMHTMLCanvasElement.h: Added.
* bindings/objc/DOMHTMLCanvasElement.mm: Added.
* bindings/objc/DOMHTMLCanvasElementInternal.h: Added.
* bindings/objc/DOMHTMLCollection.h: Added.
* bindings/objc/DOMHTMLCollection.mm: Added.
* bindings/objc/DOMHTMLCollectionInternal.h: Added.
* bindings/objc/DOMHTMLDListElement.h: Added.
* bindings/objc/DOMHTMLDListElement.mm: Added.
* bindings/objc/DOMHTMLDListElementInternal.h: Added.
* bindings/objc/DOMHTMLDirectoryElement.h: Added.
* bindings/objc/DOMHTMLDirectoryElement.mm: Added.
* bindings/objc/DOMHTMLDirectoryElementInternal.h: Added.
* bindings/objc/DOMHTMLDivElement.h: Added.
* bindings/objc/DOMHTMLDivElement.mm: Added.
* bindings/objc/DOMHTMLDivElementInternal.h: Added.
* bindings/objc/DOMHTMLDocument.h: Added.
* bindings/objc/DOMHTMLDocument.mm: Added.
* bindings/objc/DOMHTMLDocumentInternal.h: Added.
* bindings/objc/DOMHTMLElement.h: Added.
* bindings/objc/DOMHTMLElement.mm: Added.
* bindings/objc/DOMHTMLElementInternal.h: Added.
* bindings/objc/DOMHTMLEmbedElement.h: Added.
* bindings/objc/DOMHTMLEmbedElement.mm: Added.
* bindings/objc/DOMHTMLEmbedElementInternal.h: Added.
* bindings/objc/DOMHTMLFieldSetElement.h: Added.
* bindings/objc/DOMHTMLFieldSetElement.mm: Added.
* bindings/objc/DOMHTMLFieldSetElementInternal.h: Added.
* bindings/objc/DOMHTMLFontElement.h: Added.
* bindings/objc/DOMHTMLFontElement.mm: Added.
* bindings/objc/DOMHTMLFontElementInternal.h: Added.
* bindings/objc/DOMHTMLFormElement.h: Added.
* bindings/objc/DOMHTMLFormElement.mm: Added.
* bindings/objc/DOMHTMLFormElementInternal.h: Added.
* bindings/objc/DOMHTMLFrameElement.h: Added.
* bindings/objc/DOMHTMLFrameElement.mm: Added.
* bindings/objc/DOMHTMLFrameElementInternal.h: Added.
* bindings/objc/DOMHTMLFrameSetElement.h: Added.
* bindings/objc/DOMHTMLFrameSetElement.mm: Added.
* bindings/objc/DOMHTMLFrameSetElementInternal.h: Added.
* bindings/objc/DOMHTMLHRElement.h: Added.
* bindings/objc/DOMHTMLHRElement.mm: Added.
* bindings/objc/DOMHTMLHRElementInternal.h: Added.
* bindings/objc/DOMHTMLHeadElement.h: Added.
* bindings/objc/DOMHTMLHeadElement.mm: Added.
* bindings/objc/DOMHTMLHeadElementInternal.h: Added.
* bindings/objc/DOMHTMLHeadingElement.h: Added.
* bindings/objc/DOMHTMLHeadingElement.mm: Added.
* bindings/objc/DOMHTMLHeadingElementInternal.h: Added.
* bindings/objc/DOMHTMLHtmlElement.h: Added.
* bindings/objc/DOMHTMLHtmlElement.mm: Added.
* bindings/objc/DOMHTMLHtmlElementInternal.h: Added.
* bindings/objc/DOMHTMLIFrameElement.h: Added.
* bindings/objc/DOMHTMLIFrameElement.mm: Added.
* bindings/objc/DOMHTMLIFrameElementInternal.h: Added.
* bindings/objc/DOMHTMLImageElement.h: Added.
* bindings/objc/DOMHTMLImageElement.mm: Added.
* bindings/objc/DOMHTMLImageElementInternal.h: Added.
* bindings/objc/DOMHTMLInputElement.h: Added.
* bindings/objc/DOMHTMLInputElement.mm: Added.
* bindings/objc/DOMHTMLInputElementInternal.h: Added.
* bindings/objc/DOMHTMLInputElementPrivate.h: Added.
* bindings/objc/DOMHTMLLIElement.h: Added.
* bindings/objc/DOMHTMLLIElement.mm: Added.
* bindings/objc/DOMHTMLLIElementInternal.h: Added.
* bindings/objc/DOMHTMLLabelElement.h: Added.
* bindings/objc/DOMHTMLLabelElement.mm: Added.
* bindings/objc/DOMHTMLLabelElementInternal.h: Added.
* bindings/objc/DOMHTMLLegendElement.h: Added.
* bindings/objc/DOMHTMLLegendElement.mm: Added.
* bindings/objc/DOMHTMLLegendElementInternal.h: Added.
* bindings/objc/DOMHTMLLinkElement.h: Added.
* bindings/objc/DOMHTMLLinkElement.mm: Added.
* bindings/objc/DOMHTMLLinkElementInternal.h: Added.
* bindings/objc/DOMHTMLMapElement.h: Added.
* bindings/objc/DOMHTMLMapElement.mm: Added.
* bindings/objc/DOMHTMLMapElementInternal.h: Added.
* bindings/objc/DOMHTMLMarqueeElement.h: Added.
* bindings/objc/DOMHTMLMarqueeElement.mm: Added.
* bindings/objc/DOMHTMLMarqueeElementInternal.h: Added.
* bindings/objc/DOMHTMLMediaElement.h: Added.
* bindings/objc/DOMHTMLMediaElement.mm: Added.
* bindings/objc/DOMHTMLMediaElementInternal.h: Added.
* bindings/objc/DOMHTMLMenuElement.h: Added.
* bindings/objc/DOMHTMLMenuElement.mm: Added.
* bindings/objc/DOMHTMLMenuElementInternal.h: Added.
* bindings/objc/DOMHTMLMetaElement.h: Added.
* bindings/objc/DOMHTMLMetaElement.mm: Added.
* bindings/objc/DOMHTMLMetaElementInternal.h: Added.
* bindings/objc/DOMHTMLModElement.h: Added.
* bindings/objc/DOMHTMLModElement.mm: Added.
* bindings/objc/DOMHTMLModElementInternal.h: Added.
* bindings/objc/DOMHTMLOListElement.h: Added.
* bindings/objc/DOMHTMLOListElement.mm: Added.
* bindings/objc/DOMHTMLOListElementInternal.h: Added.
* bindings/objc/DOMHTMLObjectElement.h: Added.
* bindings/objc/DOMHTMLObjectElement.mm: Added.
* bindings/objc/DOMHTMLObjectElementInternal.h: Added.
* bindings/objc/DOMHTMLObjectElementPrivate.h: Added.
* bindings/objc/DOMHTMLOptGroupElement.h: Added.
* bindings/objc/DOMHTMLOptGroupElement.mm: Added.
* bindings/objc/DOMHTMLOptGroupElementInternal.h: Added.
* bindings/objc/DOMHTMLOptionElement.h: Added.
* bindings/objc/DOMHTMLOptionElement.mm: Added.
* bindings/objc/DOMHTMLOptionElementInternal.h: Added.
* bindings/objc/DOMHTMLOptionsCollection.h: Added.
* bindings/objc/DOMHTMLOptionsCollection.mm: Added.
* bindings/objc/DOMHTMLOptionsCollectionInternal.h: Added.
* bindings/objc/DOMHTMLParagraphElement.h: Added.
* bindings/objc/DOMHTMLParagraphElement.mm: Added.
* bindings/objc/DOMHTMLParagraphElementInternal.h: Added.
* bindings/objc/DOMHTMLParamElement.h: Added.
* bindings/objc/DOMHTMLParamElement.mm: Added.
* bindings/objc/DOMHTMLParamElementInternal.h: Added.
* bindings/objc/DOMHTMLPreElement.h: Added.
* bindings/objc/DOMHTMLPreElement.mm: Added.
* bindings/objc/DOMHTMLPreElementInternal.h: Added.
* bindings/objc/DOMHTMLQuoteElement.h: Added.
* bindings/objc/DOMHTMLQuoteElement.mm: Added.
* bindings/objc/DOMHTMLQuoteElementInternal.h: Added.
* bindings/objc/DOMHTMLScriptElement.h: Added.
* bindings/objc/DOMHTMLScriptElement.mm: Added.
* bindings/objc/DOMHTMLScriptElementInternal.h: Added.
* bindings/objc/DOMHTMLSelectElement.h: Added.
* bindings/objc/DOMHTMLSelectElement.mm: Added.
* bindings/objc/DOMHTMLSelectElementInternal.h: Added.
* bindings/objc/DOMHTMLStyleElement.h: Added.
* bindings/objc/DOMHTMLStyleElement.mm: Added.
* bindings/objc/DOMHTMLStyleElementInternal.h: Added.
* bindings/objc/DOMHTMLTableCaptionElement.h: Added.
* bindings/objc/DOMHTMLTableCaptionElement.mm: Added.
* bindings/objc/DOMHTMLTableCaptionElementInternal.h: Added.
* bindings/objc/DOMHTMLTableCellElement.h: Added.
* bindings/objc/DOMHTMLTableCellElement.mm: Added.
* bindings/objc/DOMHTMLTableCellElementInternal.h: Added.
* bindings/objc/DOMHTMLTableColElement.h: Added.
* bindings/objc/DOMHTMLTableColElement.mm: Added.
* bindings/objc/DOMHTMLTableColElementInternal.h: Added.
* bindings/objc/DOMHTMLTableElement.h: Added.
* bindings/objc/DOMHTMLTableElement.mm: Added.
* bindings/objc/DOMHTMLTableElementInternal.h: Added.
* bindings/objc/DOMHTMLTableRowElement.h: Added.
* bindings/objc/DOMHTMLTableRowElement.mm: Added.
* bindings/objc/DOMHTMLTableRowElementInternal.h: Added.
* bindings/objc/DOMHTMLTableSectionElement.h: Added.
* bindings/objc/DOMHTMLTableSectionElement.mm: Added.
* bindings/objc/DOMHTMLTableSectionElementInternal.h: Added.
* bindings/objc/DOMHTMLVideoElement.h: Added.
* bindings/objc/DOMHTMLVideoElement.mm: Added.
* bindings/objc/DOMHTMLVideoElementInternal.h: Added.

Source/WebKit/mac:

Don't migrate DOMHTMLEmbedElementPrivate.h - it was just an empty header.

* MigrateHeaders.make:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203988 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoMake wpt server generate its certificates in layout test results folder
cdumez@apple.com [Mon, 1 Aug 2016 21:06:36 +0000 (21:06 +0000)]
Make wpt server generate its certificates in layout test results folder
https://bugs.webkit.org/show_bug.cgi?id=160422

Patch by Youenn Fablet <youenn@apple.com> on 2016-08-01
Reviewed by Chris Dumez.

LayoutTests/imported/w3c:

* resources/_wpt_certs/.gitignore: Removed.
* resources/_wpt_certs/localhost.key: Removed.
* resources/_wpt_certs/localhost.pem: Removed.
* resources/config.json:

Tools:

* Scripts/webkitpy/layout_tests/servers/web_platform_test_server.py:
(WebPlatformTestServer._copy_webkit_test_files): Setting certificate folder to layout tests results directory.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203987 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWeb Inspector: Waterfall graph in Network tab shows too much whitespace/dead time
mattbaker@apple.com [Mon, 1 Aug 2016 21:03:44 +0000 (21:03 +0000)]
Web Inspector: Waterfall graph in Network tab shows too much whitespace/dead time
https://bugs.webkit.org/show_bug.cgi?id=160373

Reviewed by Timothy Hatcher.

Track the start and end time separate from the ruler to remove
leading/trailing whitespace in the graph. Also prevent network
record bars from being clipped.

* UserInterface/Views/NetworkGridContentView.js:
(WebInspector.NetworkGridContentView):
(WebInspector.NetworkGridContentView.prototype.get zeroTime):
Set equal to the ruler start time.
(WebInspector.NetworkGridContentView.prototype.reset):
Reset start and end time.
(WebInspector.NetworkGridContentView.prototype.layout):
Set the initial values for ruler start/zero time if needed.
(WebInspector.NetworkGridContentView.prototype._networkTimelineRecordAdded):
Set the start time if needed.
Reduce debounce from 250ms to 150ms. This prevents the current time from
exceeding the end time by as wide a margin, while keeping the current
time "timer" from being restarted too frequently.

(WebInspector.NetworkGridContentView.prototype._stopUpdatingCurrentTime):
Update the ruler end time and schedule a layout.
Add padding equal to the time needed to draw a network bar at minimum width
to prevent bars from being clipped.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203986 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoREGRESSION(r198943): drop-down menu navigation on fiddlevideo.com doesn't appear...
antti@apple.com [Mon, 1 Aug 2016 20:49:53 +0000 (20:49 +0000)]
REGRESSION(r198943): drop-down menu navigation on fiddlevideo.com doesn't appear on iOS, works on OS X
https://bugs.webkit.org/show_bug.cgi?id=160406
Source/WebCore:

rdar://problem/26310261

Reviewed by Simon Fraser.

On iOS we generate synthetic mouse events from taps. Click event is generated on tap only if the move event
doesn't produce visible changes to the document. This is important to make certain types of drop down menus
work.

The information on mutations is passed via WKContentObservation side channel which is updated from varous parts
of the code. Newly visible elements are detected CheckForVisibilityChangeOnRecalcStyle during style resolution.
This got broken by the style refactoring because it assumes that renderer is mutated along with style computation.
However mutation is now a separate step performed by RenderTreeUpdater.

Fix by moving CheckForVisibilityChange to RenderTreeUpdater.

Test: fast/content-observation/click-event-suppression-on-content-change.html

* style/RenderTreeUpdater.cpp:
(WebCore::RenderTreeUpdater::Parent::Parent):
(WebCore::RenderTreeUpdater::updateElementRenderer):
(WebCore::RenderTreeUpdater::tearDownRenderer):
(WebCore::elementImplicitVisibility):
(WebCore::CheckForVisibilityChange::CheckForVisibilityChange):
(WebCore::CheckForVisibilityChange::~CheckForVisibilityChange):
* style/StyleTreeResolver.cpp:
(WebCore::Style::TreeResolver::createAnimatedElementUpdate):
(WebCore::Style::TreeResolver::pushParent):
(WebCore::Style::TreeResolver::resolveComposedTree):
(WebCore::Style::elementImplicitVisibility): Deleted.
(WebCore::Style::CheckForVisibilityChangeOnRecalcStyle::CheckForVisibilityChangeOnRecalcStyle): Deleted.
(WebCore::Style::CheckForVisibilityChangeOnRecalcStyle::~CheckForVisibilityChangeOnRecalcStyle): Deleted.

LayoutTests:

Reviewed by Simon Fraser.

This stuff has had zero test coverage. Adding a basic UIScript based test.

* TestExpectations:
* fast/content-observation/click-event-suppression-on-content-change-expected.txt: Added.
* fast/content-observation/click-event-suppression-on-content-change.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203985 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[iOS] A video element that does not pause after exiting from fullscreen should be...
eric.carlson@apple.com [Mon, 1 Aug 2016 20:48:35 +0000 (20:48 +0000)]
[iOS] A video element that does not pause after exiting from fullscreen should be allowed to continue playing inline
https://bugs.webkit.org/show_bug.cgi?id=160416
<rdar://problem/27409854>

Reviewed by Alex Christensen.

* html/HTMLMediaElement.cpp:
(WebCore::HTMLMediaElement::exitFullscreen): If playback normally requires fullscreen but the
element was not paused when exiting from fullscreen, set the 'playsinline' attribute so we won't
force fullscreen if playback is paused and resumes, and set the 'controls' attribute so the
user can control playback.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203984 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoBuild fix.
andersca@apple.com [Mon, 1 Aug 2016 20:34:02 +0000 (20:34 +0000)]
Build fix.

* WebCore.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203983 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Mac][iOS] Adopt MediaRemote "seek to playback position"
eric.carlson@apple.com [Mon, 1 Aug 2016 19:49:02 +0000 (19:49 +0000)]
[Mac][iOS] Adopt MediaRemote "seek to playback position"
https://bugs.webkit.org/show_bug.cgi?id=160405
<rdar://problem/27547583>

Reviewed by Dean Jackson.

Source/WebCore:

Test: media/remote-control-command-seek.html

* Modules/webaudio/AudioContext.h: Update for didReceiveRemoteControlCommand argument change.

* html/HTMLMediaElement.cpp:
(WebCore::HTMLMediaElement::didReceiveRemoteControlCommand): Support SeekToPlaybackPositionCommand.
Drive by fix, support Stop command.
(WebCore::HTMLMediaElement::supportsSeeking): New.
* html/HTMLMediaElement.h:

* platform/RemoteCommandListener.h:
(WebCore::RemoteCommandListenerClient::didReceiveRemoteControlCommand): Add command argument.
(WebCore::RemoteCommandListenerClient::supportsSeeking): New.
(WebCore::RemoteCommandListener::updateSupportedCommands): Ditto.
(WebCore::RemoteCommandListener::client): Ditto.

* platform/audio/PlatformMediaSession.cpp:
(WebCore::PlatformMediaSession::didReceiveRemoteControlCommand): Add command argument.
(WebCore::PlatformMediaSession::supportsSeeking): New, pass through to client.
* platform/audio/PlatformMediaSession.h:

* platform/audio/PlatformMediaSessionManager.cpp:
(WebCore::PlatformMediaSessionManager::setCurrentSession): Tell remote command listener to
update supported commands.
(WebCore::PlatformMediaSessionManager::currentSession): Make const.
(WebCore::PlatformMediaSessionManager::didReceiveRemoteControlCommand): Add command argument.
(WebCore::PlatformMediaSessionManager::supportsSeeking): New, pass through to session.
* platform/audio/PlatformMediaSessionManager.h:

* platform/ios/RemoteCommandListenerIOS.h:
(WebCore::RemoteCommandListenerIOS::createWeakPtr):
* platform/ios/RemoteCommandListenerIOS.mm:
(WebCore::RemoteCommandListenerIOS::RemoteCommandListenerIOS): Support changePlaybackPositionCommand.
(WebCore::RemoteCommandListenerIOS::~RemoteCommandListenerIOS): Remove seekToTime target.
(WebCore::RemoteCommandListenerIOS::updateSupportedCommands): Update changePlaybackPositionCommand.

* platform/mac/MediaRemoteSoftLink.cpp:
* platform/mac/MediaRemoteSoftLink.h:

* platform/mac/RemoteCommandListenerMac.h:
* platform/mac/RemoteCommandListenerMac.mm:
(WebCore::RemoteCommandListenerMac::updateSupportedCommands): New, split out of constructor.
(WebCore::RemoteCommandListenerMac::RemoteCommandListenerMac): Split setup logic out into
updateSupportedCommands. Support MRMediaRemoteCommandSeekToPlaybackPosition. Don't assert when
receiving an unsupported command, it happens. Return error when a command isn't supported or
fails.

* testing/Internals.cpp:
(WebCore::Internals::postRemoteControlCommand): Add command argument parameter. Support
seektoplaybackposition.
* testing/Internals.h:
* testing/Internals.idl:

LayoutTests:

* media/remote-control-command-seek-expected.txt: Added.
* media/remote-control-command-seek.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203982 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFreeze DOMHTMLTitleElement and DOMHTMLUListElement bindings
andersca@apple.com [Mon, 1 Aug 2016 19:35:11 +0000 (19:35 +0000)]
Freeze DOMHTMLTitleElement and DOMHTMLUListElement bindings
https://bugs.webkit.org/show_bug.cgi?id=160415

Reviewed by Sam Weinig.

* DerivedSources.make:
* WebCore.xcodeproj/project.pbxproj:
* bindings/objc/DOMHTMLTitleElement.h: Added.
* bindings/objc/DOMHTMLTitleElement.mm: Added.
* bindings/objc/DOMHTMLTitleElementInternal.h: Added.
* bindings/objc/DOMHTMLUListElement.h: Added.
* bindings/objc/DOMHTMLUListElement.mm: Added.
* bindings/objc/DOMHTMLUListElementInternal.h: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203981 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoSub should be a Math IC
sbarati@apple.com [Mon, 1 Aug 2016 18:48:14 +0000 (18:48 +0000)]
Sub should be a Math IC
https://bugs.webkit.org/show_bug.cgi?id=160270

Reviewed by Mark Lam.

This makes Sub an IC like Mul and Add. I'm seeing the following
improvements of average Sub size on Unity and JetStream:

           |   JetStream  |  Unity 3D  |
     ------| -------------|--------------
      Old  |   202 bytes  |  205 bytes |
     ------| -------------|--------------
      New  |   134  bytes |  134 bytes |
     ------------------------------------

* bytecode/CodeBlock.cpp:
(JSC::CodeBlock::addJITMulIC):
(JSC::CodeBlock::addJITSubIC):
(JSC::CodeBlock::findStubInfo):
(JSC::CodeBlock::dumpMathICStats):
* bytecode/CodeBlock.h:
(JSC::CodeBlock::stubInfoBegin):
(JSC::CodeBlock::stubInfoEnd):
* dfg/DFGSpeculativeJIT.cpp:
(JSC::DFG::SpeculativeJIT::compileArithSub):
* ftl/FTLLowerDFGToB3.cpp:
(JSC::FTL::DFG::LowerDFGToB3::compileArithAddOrSub):
* jit/JITArithmetic.cpp:
(JSC::JIT::emit_op_sub):
(JSC::JIT::emitSlow_op_sub):
(JSC::JIT::emit_op_pow):
* jit/JITMathIC.h:
* jit/JITMathICForwards.h:
* jit/JITOperations.cpp:
* jit/JITOperations.h:
* jit/JITSubGenerator.cpp:
(JSC::JITSubGenerator::generateInline):
(JSC::JITSubGenerator::generateFastPath):
* jit/JITSubGenerator.h:
(JSC::JITSubGenerator::JITSubGenerator):
(JSC::JITSubGenerator::isLeftOperandValidConstant):
(JSC::JITSubGenerator::isRightOperandValidConstant):
(JSC::JITSubGenerator::arithProfile):
(JSC::JITSubGenerator::didEmitFastPath): Deleted.
(JSC::JITSubGenerator::endJumpList): Deleted.
(JSC::JITSubGenerator::slowPathJumpList): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203979 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoAdd an option to run the MotionMark tests with classic tile size (512x512)
commit-queue@webkit.org [Mon, 1 Aug 2016 18:45:17 +0000 (18:45 +0000)]
Add an option to run the MotionMark tests with classic tile size (512x512)
https://bugs.webkit.org/show_bug.cgi?id=160371

Patch by Said Abou-Hallawa <sabouhallawa@apple.com> on 2016-08-01
Reviewed by Darin Adler.

Setting the body to large size (3000x3000) and overflow to scroll forces
the classic tile size (512x512). This option does not affect the canvas
tests because the canvas is displayed on one tile regardless of its size.

* Animometer/developer.html:
* Animometer/resources/debug-runner/animometer.css:
(body.showing-test-container.tiles-big):
(body.showing-test-container.tiles-classic):
* Animometer/resources/debug-runner/animometer.js:
* Animometer/resources/runner/animometer.js:
(window.benchmarkController.startBenchmark):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203978 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUpdate Apache configuration for Arch Linux
commit-queue@webkit.org [Mon, 1 Aug 2016 18:30:38 +0000 (18:30 +0000)]
Update Apache configuration for Arch Linux
https://bugs.webkit.org/show_bug.cgi?id=160408

Patch by Adrian Perez de Castro <aperez@igalia.com> on 2016-08-01
Reviewed by Ryosuke Niwa.

Arch Linux now ships Apache 2.4 and PHP 7, so the configuration file
used for the layout tests is updated to be compatible with the newer
versions.

* http/conf/archlinux-httpd.conf: Updated.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203977 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoREGRESSION (r196383): Drop down CSS menus not working on cnet.com, apmex.com
antti@apple.com [Mon, 1 Aug 2016 18:00:48 +0000 (18:00 +0000)]
REGRESSION (r196383): Drop down CSS menus not working on cnet.com, apmex.com
https://bugs.webkit.org/show_bug.cgi?id=160390

Reviewed by Simon Fraser.

Source/WebCore:

The case here is that we have a rule like

    .enableHover:hover .child { ... }

and the "enableHover" class is added dynamically. The class change invalidation optimization code would figure out
that nothing needs to be invalidated as the class change doesn't make the rule match (since :hover doesn't match).

However for event driven hover to actually work the hover element needs to have its childrenAffectedByHover bit set.
This bits is set when the selector match is attempted, whether it actually matches or not. Since we optimized away
the style invalidation we never set the bit either.

Fix by treating :hover as always matching (==ignored) when collecting rules for invalidation optimization purposes.
Dynamic pseudo elements are already treated this way for similar reasons.

Test: fast/selectors/hover-invalidation-descendant-dynamic.html

* css/SelectorChecker.cpp:
(WebCore::SelectorChecker::checkOne):

    Match always in CollectingRulesIgnoringVirtualPseudoElements mode (now slightly misnamed).

    This mode is used for optimization purposes in StyleInvalidationAnalysis (which we care about here) and
    StyleSharingResolver. The change is fine for both.

* cssjit/SelectorCompiler.cpp:
(WebCore::SelectorCompiler::SelectorCodeGenerator::generateElementIsHovered):

    Same change for the slow path selector checker.

LayoutTests:

* fast/selectors/hover-invalidation-descendant-dynamic-expected.txt: Added.
* fast/selectors/hover-invalidation-descendant-dynamic.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203976 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFunction Call on NULL Object
commit-queue@webkit.org [Mon, 1 Aug 2016 17:49:57 +0000 (17:49 +0000)]
Function Call on NULL Object
https://bugs.webkit.org/show_bug.cgi?id=160325

Patch by Jonathan Bedard <jbedard@apple.com> on 2016-08-01
Reviewed by Darin Adler.

* WebKitTestRunner/InjectedBundle/InjectedBundle.cpp:
(WTR::InjectedBundle::didReceiveMessageToPage): Changed removeAllWebNotificationPermissions() call to static function call.
* WebKitTestRunner/InjectedBundle/TestRunner.h: Made all web-notification functions static.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203975 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Cocoa] Freeze Objective-C bindings and stop autogenerating them: Step 1 - Convert...
darin@apple.com [Mon, 1 Aug 2016 17:49:50 +0000 (17:49 +0000)]
[Cocoa] Freeze Objective-C bindings and stop autogenerating them: Step 1 - Convert a single file
https://bugs.webkit.org/show_bug.cgi?id=160386

Reviewed by Anders Carlsson.

No behavior change, no tests changed. I tested this locally by putting "#error" into the
files in the DerivedSources directory to make sure the build doesn't use those files even
if there are stale copies still sitting there.

Moving the bindings from WebCore to the legacy part of WebKit is something we also want to
do, but that is quite challenging to do one file at a time, so we will do that later.

For this first file, I did some hand edits to make the code in a the file a little less
peculiar. It's still awkward to have all those JSMainThreadNullState in there; should double
check that we really need them.

An important follow up step will be to cut down on the bindings. We have lots of bindings
that are only in the Private.h header, and it would be good to find a way to cut down on those.
And on iOS we would like eliminate these bindings entirely since they are not API there.

* DerivedSources.make: Moved HTMLTextAreaElement from a list we use for both JavaScript and
Objective-C DOM bindings to a list we use only for JavaScript bindings.
* WebCore.xcodeproj/project.pbxproj: Removed references to generated DOMHTMLTextAreaElement
files and instead reference the checked-in copies.
* bindings/objc/DOMHTMLTextAreaElement.h: Added.
* bindings/objc/DOMHTMLTextAreaElement.mm: Added.
* bindings/objc/DOMHTMLTextAreaElementInternal.h: Added.
* bindings/objc/DOMHTMLTextAreaElementPrivate.h: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203974 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoIntroduce a RenderMathMLFencedOperator class
fred.wang@free.fr [Mon, 1 Aug 2016 07:44:20 +0000 (07:44 +0000)]
Introduce a RenderMathMLFencedOperator class
https://bugs.webkit.org/show_bug.cgi?id=160245

Patch by Frederic Wang <fwang@igalia.com> on 2016-08-01
Reviewed by Darin Adler.

We introduce a class to handle all the code specific to anonymous mfenced operators. In
particular many members of the RenderMathMLOperator class are going to be removed while
fixing bug 156537 but they may still need to stored and exposed in some way by mfenced
operators.

No new tests, behavior is unchanged.

* CMakeLists.txt: Add RenderMathMLFencedOperator.
* WebCore.xcodeproj/project.pbxproj: Ditto.
* rendering/RenderObject.h:
(WebCore::RenderObject::isRenderMathMLFencedOperator): Function to allow casting to
RenderMathMLFencedOperator.
* rendering/mathml/RenderMathMLFenced.cpp: Use RenderMathMLFencedOperator instead of
RenderMathMLOperator and use the new name updateOperatorContent.
(WebCore::RenderMathMLFenced::updateFromElement):
(WebCore::RenderMathMLFenced::createMathMLOperator):
(WebCore::RenderMathMLFenced::makeFences):
(WebCore::RenderMathMLFenced::addChild):
* rendering/mathml/RenderMathMLFenced.h: Use RenderMathMLFencedOperator instead of
RenderMathMLOperator.
* rendering/mathml/RenderMathMLFencedOperator.cpp: New class derived from RenderMathMLOperator
(WebCore::RenderMathMLFencedOperator::RenderMathMLFencedOperator): Init form and operator
flags with the specified parameters and init the text content.
(WebCore::RenderMathMLFencedOperator::updateOperatorContent): We move the
RenderMathMLOperator::updateTokenContent function used by anonymous operators here. It is
renamed to avoid build errors due to conflicts with the other
RenderMathMLOperator::updateTokenContent function.
* rendering/mathml/RenderMathMLFencedOperator.h: New class for anonymous mfenced operators
with getter and setter for the m_textContent. Also override RenderMathMLFencedOperator to
allow casting to RenderMathMLFencedOperator.
* rendering/mathml/RenderMathMLOperator.cpp:
(WebCore::RenderMathMLOperator::RenderMathMLOperator): Simplify the constructors. Some
initializations is moved in RenderMathMLOperator.h while other code is moved to
RenderMathMLFencedOperator.
(WebCore::RenderMathMLOperator::textContent): No need to handle the anonymous case here since
it is handled in RenderMathMLFencedOperator.
(WebCore::RenderMathMLOperator::updateTokenContent): We delete the version used for anonymous
operators and move it into RenderMathMLFencedOperator.
* rendering/mathml/RenderMathMLOperator.h: Make the class and some members overridable and
exposed to the derived RenderMathMLFencedOperator class. We also move initialization of some
members here.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203973 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWe should not keep the JavaScript tests inside the Source/JavaScriptCore/ directory.
keith_miller@apple.com [Mon, 1 Aug 2016 07:28:55 +0000 (07:28 +0000)]
We should not keep the JavaScript tests inside the Source/JavaScriptCore/ directory.
https://bugs.webkit.org/show_bug.cgi?id=160372

Rubber stamped by Geoffrey Garen.

This patch moves all the JavaScript tests from Source/JavaScriptCore/tests to
a new top level directory, JSTests. Having the tests in the Source directory
was both confusing and inconvenient for people that just want to checkout the
source code of WebKit. Since there is no other obvious place to put all the
JavaScript tests a new top level directory seemed the most sensible.

* JSTests/: Copied from Source/JavaScriptCore/tests.
* Source/JavaScriptCore/tests/: Deleted.
* Scripts/import-test262-tests:
* Scripts/run-javascriptcore-tests:
* Scripts/update-javascriptcore-test-res:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203972 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoReduce the number of ResourceRequest copies in DocumentThreadableLoader
commit-queue@webkit.org [Mon, 1 Aug 2016 07:02:35 +0000 (07:02 +0000)]
Reduce the number of ResourceRequest copies in DocumentThreadableLoader
https://bugs.webkit.org/show_bug.cgi?id=160338

Patch by Youenn Fablet <youennf@gmail.com> on 2016-08-01
Reviewed by Darin Adler.

No observable change of behavior.

Updating one of CachedResourceReques constructor to take a ResourceRequest&&.
This allows reducing a ResourceRequest copy in DocumentThreadableLoader.
Updating DocumentLoader and MediaResourceLoader accordingly (future optimization steps may look at making these
two loaders more efficient with regards to Resource Request).

Updating ThreadableLoader to take a ResourceRequest&&.
This allows reducing the number of copies in DocumentThreadableLoader.
Updating clients of ThreadableLoader accordingly.

* Modules/fetch/FetchLoader.cpp:
(WebCore::FetchLoader::start):
* dom/Document.cpp:
(WebCore::Document::didLoadResourceSynchronously):
* dom/Document.h:
* dom/ScriptExecutionContext.cpp:
(WebCore::ScriptExecutionContext::didLoadResourceSynchronously):
* dom/ScriptExecutionContext.h:
* fileapi/FileReaderLoader.cpp:
(WebCore::FileReaderLoader::start):
* inspector/InspectorNetworkAgent.cpp:
(WebCore::InspectorNetworkAgent::loadResource):
* loader/DocumentLoader.cpp:
(WebCore::DocumentLoader::startLoadingMainResource):
* loader/DocumentThreadableLoader.cpp:
(WebCore::DocumentThreadableLoader::loadResourceSynchronously):
(WebCore::DocumentThreadableLoader::create):
(WebCore::DocumentThreadableLoader::DocumentThreadableLoader):
(WebCore::DocumentThreadableLoader::makeCrossOriginAccessRequest):
(WebCore::DocumentThreadableLoader::makeSimpleCrossOriginAccessRequest):
(WebCore::DocumentThreadableLoader::redirectReceived):
(WebCore::DocumentThreadableLoader::preflightSuccess):
(WebCore::DocumentThreadableLoader::loadRequest):
* loader/DocumentThreadableLoader.h:
* loader/MediaResourceLoader.cpp:
(WebCore::MediaResourceLoader::requestResource):
* loader/ThreadableLoader.cpp:
(WebCore::ThreadableLoader::create):
(WebCore::ThreadableLoader::loadResourceSynchronously):
* loader/ThreadableLoader.h:
* loader/WorkerThreadableLoader.cpp:
(WebCore::WorkerThreadableLoader::WorkerThreadableLoader):
(WebCore::WorkerThreadableLoader::loadResourceSynchronously):
(WebCore::WorkerThreadableLoader::MainThreadBridge::MainThreadBridge):
* loader/WorkerThreadableLoader.h:
(WebCore::WorkerThreadableLoader::create):
* loader/cache/CachedResourceRequest.cpp:
(WebCore::CachedResourceRequest::CachedResourceRequest):
* loader/cache/CachedResourceRequest.h:
* page/EventSource.cpp:
(WebCore::EventSource::connect):
* workers/WorkerScriptLoader.cpp:
(WebCore::WorkerScriptLoader::loadSynchronously):
(WebCore::WorkerScriptLoader::loadAsynchronously):
* xml/XMLHttpRequest.cpp:
(WebCore::XMLHttpRequest::createRequest):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203971 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoFetch Response built-ins should use @makeThisTypeError
commit-queue@webkit.org [Mon, 1 Aug 2016 06:00:56 +0000 (06:00 +0000)]
Fetch Response built-ins should use @makeThisTypeError
https://bugs.webkit.org/show_bug.cgi?id=160290

Patch by Youenn Fablet <youenn@apple.com> on 2016-07-31
Reviewed by Darin Adler.

Source/WebCore:

Covered by updated test.

Fixed type error checks.
Making use of @makeThisTypeError to have the correct error message.
Updating arrayBuffer, blob, json and text to return rejected promises in lieu of throwing
in case the 'this' value is not q Response object.

* Modules/fetch/FetchResponse.js:
(clone): Updated instanceof check and making use of @makeThisTypeError.
(arrayBuffer): Ditto.
(blob): Ditto.
(formData): Ditto.
(json): Ditto.
(text): Ditto.

LayoutTests:

* fetch/fetch-error-messages-expected.txt:
* fetch/fetch-error-messages.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203961 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[GTK] Implement missing WebCore::moveFile() using GLib functions
commit-queue@webkit.org [Mon, 1 Aug 2016 05:43:57 +0000 (05:43 +0000)]
[GTK] Implement missing WebCore::moveFile() using GLib functions
https://bugs.webkit.org/show_bug.cgi?id=160363

Patch by Adrian Perez de Castro <aperez@igalia.com> on 2016-07-31
Reviewed by Carlos Garcia Campos.

* platform/glib/FileSystemGlib.cpp:
(WebCore::moveFile): Function added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203960 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[JSC] Should check Test262Error correctly
utatane.tea@gmail.com [Mon, 1 Aug 2016 00:37:14 +0000 (00:37 +0000)]
[JSC] Should check Test262Error correctly
https://bugs.webkit.org/show_bug.cgi?id=159862

Reviewed by Saam Barati.

Test262Error in the harness does not have "name" property.
Rather than checking "name" property, peforming `instanceof` is better to check the class of the exception.

* jsc.cpp:
(checkUncaughtException):
* runtime/JSObject.h:
* tests/test262.yaml:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203957 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoDrop [StrictTypeChecking] in cases where it is a no-op
cdumez@apple.com [Sun, 31 Jul 2016 23:36:20 +0000 (23:36 +0000)]
Drop [StrictTypeChecking] in cases where it is a no-op
https://bugs.webkit.org/show_bug.cgi?id=160387

Reviewed by Darin Adler.

Drop [StrictTypeChecking] in cases where it is a no-op. After r203949 & r203950,
[StrictTypeChecking] only has an impact when used on operation parameters of
string types and is currently used as a workaround for bugs in our overload
resolution algorithm.

* Modules/mediastream/RTCPeerConnection.idl:
* Modules/mediastream/RTCRtpSender.idl:
* bindings/scripts/test/TestObj.idl:
* bindings/scripts/test/TestTypedefs.idl:
* html/canvas/ANGLEInstancedArrays.idl:
* html/canvas/OESVertexArrayObject.idl:
* html/canvas/WebGL2RenderingContext.idl:
* html/canvas/WebGLDebugShaders.idl:
* html/canvas/WebGLLoseContext.idl:
* html/canvas/WebGLRenderingContextBase.idl:
* svg/SVGAngle.idl:
* svg/SVGColor.idl:
* svg/SVGLength.idl:
* svg/SVGLengthList.idl:
* svg/SVGMatrix.idl:
* svg/SVGNumberList.idl:
* svg/SVGPaint.idl:
* svg/SVGPathSegList.idl:
* svg/SVGPoint.idl:
* svg/SVGPointList.idl:
* svg/SVGStringList.idl:
* svg/SVGTransform.idl:
* svg/SVGTransformList.idl:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203956 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoAX: Add a check for touch event listener on iOS accessibility object
n_wang@apple.com [Sun, 31 Jul 2016 20:02:54 +0000 (20:02 +0000)]
AX: Add a check for touch event listener on iOS accessibility object
https://bugs.webkit.org/show_bug.cgi?id=160388

Reviewed by Chris Fleizach.

Source/WebCore:

dispatchTouchEvent() is not working correctly within AXPress() sometimes. Need to
investigate it more in the future. Now, adding a check for the touch event listener
on the object's node so that iOS can handle dispatching the touch event instead.

Test: accessibility/ios-simulator/has-touch-event-listener.html

* accessibility/AccessibilityObject.h:
* accessibility/ios/AccessibilityObjectIOS.mm:
(WebCore::AccessibilityObject::accessibilityPlatformIncludesObject):
(WebCore::AccessibilityObject::hasTouchEventListener):
* accessibility/ios/WebAccessibilityObjectWrapperIOS.mm:
(appendStringToResult):
(-[WebAccessibilityObjectWrapper _accessibilityHasTouchEventListener]):
(-[WebAccessibilityObjectWrapper _accessibilityValueIsAutofilled]):

Tools:

* DumpRenderTree/ios/AccessibilityUIElementIOS.mm:
(AccessibilityUIElement::boolAttributeValue):
* WebKitTestRunner/InjectedBundle/ios/AccessibilityUIElementIOS.mm:
(WTR::AccessibilityUIElement::boolAttributeValue):

LayoutTests:

* accessibility/ios-simulator/has-touch-event-listener-expected.txt: Added.
* accessibility/ios-simulator/has-touch-event-listener.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203955 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[GTK][Unix] Implement missing WebKit::SharedMemory::create() function
commit-queue@webkit.org [Sun, 31 Jul 2016 08:31:14 +0000 (08:31 +0000)]
[GTK][Unix] Implement missing WebKit::SharedMemory::create() function
https://bugs.webkit.org/show_bug.cgi?id=160364

Patch by Adrian Perez de Castro <aperez@igalia.com> on 2016-07-31
Reviewed by Carlos Garcia Campos.

The WebKit::SharedMemory::create() function is missing for the Unix
platform, which is also used by the GTK+ port. The latter is going
to need this in place to use the common content filtering code.

* Platform/unix/SharedMemoryUnix.cpp:
(WebKit::accessModeMMap): Added helper function to convert a
SharedMemory::Protection value into flags useable with mmap().
(WebKit::SharedMemory::create): Added. Implementation reuses code
existing in the SharedMemory::allocate() function.
(WebKit::SharedMemory::allocate): Reimplemented in terms of
SharedMemory::create().

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203954 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[ES6] Module binding can be exported by multiple names
utatane.tea@gmail.com [Sun, 31 Jul 2016 07:04:57 +0000 (07:04 +0000)]
[ES6] Module binding can be exported by multiple names
https://bugs.webkit.org/show_bug.cgi?id=160343

Reviewed by Saam Barati.

ES6 Module can export the same local binding by using multiple names.
For example,

    ```
    var value = 42;

    export { value };
    export { value as value2 };
    ```

Currently, we only allowed one local binding to be exported with one name. So, in the above case,
the local binding "value" is exported as "value2" and "value" name is not exported. This is wrong.

To fix this issue, we collect the correspondence (local name => exported name) to the local bindings
in the parser. Previously, we only maintained the exported local bindings in the parser. And utilize
this information when creating the export entries in ModuleAnalyzer.

And this patch also moves ModuleScopeData from the Scope object to the Parser class since exported
names should be managed per-module, not per-scope.

This change fixes several test262 failures.

* JavaScriptCore.xcodeproj/project.pbxproj:
* parser/ModuleAnalyzer.cpp:
(JSC::ModuleAnalyzer::exportVariable):
(JSC::ModuleAnalyzer::analyze):
(JSC::ModuleAnalyzer::exportedBinding): Deleted.
(JSC::ModuleAnalyzer::declareExportAlias): Deleted.
* parser/ModuleAnalyzer.h:
* parser/ModuleScopeData.h: Copied from Source/JavaScriptCore/parser/ModuleAnalyzer.h.
(JSC::ModuleScopeData::create):
(JSC::ModuleScopeData::exportedBindings):
(JSC::ModuleScopeData::exportName):
(JSC::ModuleScopeData::exportBinding):
* parser/Nodes.cpp:
(JSC::ProgramNode::ProgramNode):
(JSC::ModuleProgramNode::ModuleProgramNode):
(JSC::EvalNode::EvalNode):
(JSC::FunctionNode::FunctionNode):
* parser/Nodes.h:
(JSC::ModuleProgramNode::moduleScopeData):
* parser/NodesAnalyzeModule.cpp:
(JSC::ExportDefaultDeclarationNode::analyzeModule):
(JSC::ExportNamedDeclarationNode::analyzeModule): Deleted.
* parser/Parser.cpp:
(JSC::Parser<LexerType>::Parser):
(JSC::Parser<LexerType>::parseModuleSourceElements):
(JSC::Parser<LexerType>::parseVariableDeclarationList):
(JSC::Parser<LexerType>::createBindingPattern):
(JSC::Parser<LexerType>::parseFunctionDeclaration):
(JSC::Parser<LexerType>::parseClassDeclaration):
(JSC::Parser<LexerType>::parseExportSpecifier):
(JSC::Parser<LexerType>::parseExportDeclaration):
* parser/Parser.h:
(JSC::Parser::exportName):
(JSC::Parser<LexerType>::parse):
(JSC::ModuleScopeData::create): Deleted.
(JSC::ModuleScopeData::exportedBindings): Deleted.
(JSC::ModuleScopeData::exportName): Deleted.
(JSC::ModuleScopeData::exportBinding): Deleted.
(JSC::Scope::Scope): Deleted.
(JSC::Scope::setSourceParseMode): Deleted.
(JSC::Scope::moduleScopeData): Deleted.
(JSC::Scope::setIsModule): Deleted.
* tests/modules/aliased-names.js: Added.
* tests/modules/aliased-names/main.js: Added.
(change):
* tests/stress/modules-syntax-error-with-names.js:
(export.Cocoa):
(SyntaxError.Cannot.export.a.duplicate.name):
* tests/test262.yaml:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203953 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoAssertion failure while setting the length of an ArrayClass array.
mark.lam@apple.com [Sun, 31 Jul 2016 01:08:37 +0000 (01:08 +0000)]
Assertion failure while setting the length of an ArrayClass array.
https://bugs.webkit.org/show_bug.cgi?id=160381
<rdar://problem/27328703>

Reviewed by Filip Pizlo.

Source/JavaScriptCore:

When setting large length values, we're currently treating ArrayClass as a
ContiguousIndexingType array.  This results in an assertion failure.  This is
now fixed.

There are currently only 2 places where we create arrays with indexing type
ArrayClass: ArrayPrototype and RuntimeArray.  The fix in JSArray:;setLength()
takes care of ArrayPrototype.

RuntimeArray already checks for the setting of its length property, and will
throw a RangeError.  Hence, there's no change is needed for the RuntimeArray.
Instead, I added some test cases ensure that the check and throw behavior does
not change without notice.

* runtime/JSArray.cpp:
(JSC::JSArray::setLength):
* tests/stress/array-setLength-on-ArrayClass-with-large-length.js: Added.
(toString):
(assertEqual):
* tests/stress/array-setLength-on-ArrayClass-with-small-length.js: Added.
(toString):
(assertEqual):

LayoutTests:

Test that RuntimeArrays will throw an error if we try to set its length.

* platform/mac/fast/dom/wrapper-classes-objc.html:
* platform/mac/fast/dom/wrapper-classes-objc-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203952 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed, rebaseline bindings tests.
cdumez@apple.com [Sun, 31 Jul 2016 00:46:03 +0000 (00:46 +0000)]
Unreviewed, rebaseline bindings tests.

* bindings/scripts/test/GObject/WebKitDOMTestObj.cpp:
(webkit_dom_test_obj_set_property): Deleted.
(webkit_dom_test_obj_get_property): Deleted.
(webkit_dom_test_obj_class_init): Deleted.
* bindings/scripts/test/GObject/WebKitDOMTestObj.h:
* bindings/scripts/test/JS/JSTestObj.cpp:
(WebCore::jsTestObjStrictTypeCheckingAttribute): Deleted.
(WebCore::jsTestObjStrictFloat): Deleted.
(WebCore::setJSTestObjStrictTypeCheckingAttribute): Deleted.
(WebCore::setJSTestObjStrictFloat): Deleted.
* bindings/scripts/test/ObjC/DOMTestObj.h:
* bindings/scripts/test/ObjC/DOMTestObj.mm:
(-[DOMTestObj strictTypeCheckingAttribute]): Deleted.
(-[DOMTestObj setStrictTypeCheckingAttribute:]): Deleted.
(-[DOMTestObj strictFloat]): Deleted.
(-[DOMTestObj setStrictFloat:]): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203951 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoEnable strict type checking for Window dictionary members
cdumez@apple.com [Sun, 31 Jul 2016 00:29:19 +0000 (00:29 +0000)]
Enable strict type checking for Window dictionary members
https://bugs.webkit.org/show_bug.cgi?id=160356

Reviewed by Darin Adler.

LayoutTests/imported/w3c:

Rebaseline W3C test now that one more check is passing.

* web-platform-tests/dom/events/Event-subclasses-constructors-expected.txt:

Source/WebCore:

Enable strict type checking for Window dictionary members. Technically,
we should do strict type checking of all wrapper types but this patch
focuses on Window because it is common to pass a Window dictionary
member to Event constructors.

By strict type checking, I mean that we should throw a TypeError is
the value is not null/undefined and does not implement the Window
interface:
- http://heycam.github.io/webidl/#es-interface

Firefox and Chrome comply with the specification already.

No new tests, updated / rebaselined existing tests.

* bindings/js/JSDictionary.cpp:
(WebCore::JSDictionary::convertValue):

LayoutTests:

Update existing tests to reflect behavior change.

* fast/events/constructors/composition-event-constructor-expected.txt:
* fast/events/constructors/composition-event-constructor.html:
* fast/events/constructors/focus-event-constructor-expected.txt:
* fast/events/constructors/focus-event-constructor.html:
* fast/events/constructors/keyboard-event-constructor-expected.txt:
* fast/events/constructors/keyboard-event-constructor.html:
* fast/events/constructors/mouse-event-constructor.html:
* fast/events/constructors/ui-event-constructor-expected.txt:
* fast/events/constructors/ui-event-constructor.html:
* fast/events/constructors/wheel-event-constructor.html:
* platform/mac/fast/events/constructors/mouse-event-constructor-expected.txt:
* platform/mac/fast/events/constructors/wheel-event-constructor-expected.txt:

* resources/js-test-pre.js:
Add a shouldThrowErrorName() utility function that is similar to shouldThrow()
but only checks the error name instead of the full error message. Checking
only the error name has the benefit of working across browsers and facilitating
refactoring of error messages.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203950 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[WebIDL] Enable strict type checking for nullable attribute setters of wrapper types
cdumez@apple.com [Sun, 31 Jul 2016 00:23:21 +0000 (00:23 +0000)]
[WebIDL] Enable strict type checking for nullable attribute setters of wrapper types
https://bugs.webkit.org/show_bug.cgi?id=160375

Reviewed by Darin Adler.

Source/WebCore:

Enable strict type checking for nullable attribute setters of wrapper types:
- http://heycam.github.io/webidl/#es-nullable-type
- http://heycam.github.io/webidl/#es-interface

For such attributes, if the JS tries to assign a value that is not null /
undefined and does not have the expected wrapper type, then we now throw a
TypeError instead of silently converting the value to null.

This behavior is consistent with Chrome and Firefox. It also helps identify
bugs in JavaScript code.

No new tests, updated existing tests.

* Modules/webaudio/AudioBufferSourceNode.idl:
* bindings/scripts/CodeGeneratorJS.pm:
(GenerateImplementation):
* bindings/scripts/test/JS/JSTestObj.cpp:
(WebCore::setJSTestObjTestNullableObjAttr):
(WebCore::setJSTestObjMutablePoint):
(WebCore::setJSTestObjImmutablePoint):
* bindings/scripts/test/TestObj.idl:
* dom/Document.idl:
* html/HTMLTableElement.idl:
* svg/SVGAngle.idl:
* svg/SVGAnimatedBoolean.idl:
* svg/SVGAnimatedEnumeration.idl:
* svg/SVGAnimatedInteger.idl:
* svg/SVGAnimatedNumber.idl:
* svg/SVGLength.idl:
* svg/SVGMatrix.idl:
* svg/SVGNumber.idl:
* svg/SVGPathSegArcAbs.idl:
* svg/SVGPathSegArcRel.idl:
* svg/SVGPathSegCurvetoCubicAbs.idl:
* svg/SVGPathSegCurvetoCubicRel.idl:
* svg/SVGPathSegCurvetoCubicSmoothAbs.idl:
* svg/SVGPathSegCurvetoCubicSmoothRel.idl:
* svg/SVGPathSegCurvetoQuadraticAbs.idl:
* svg/SVGPathSegCurvetoQuadraticRel.idl:
* svg/SVGPathSegCurvetoQuadraticSmoothAbs.idl:
* svg/SVGPathSegCurvetoQuadraticSmoothRel.idl:
* svg/SVGPathSegLinetoAbs.idl:
* svg/SVGPathSegLinetoHorizontalAbs.idl:
* svg/SVGPathSegLinetoHorizontalRel.idl:
* svg/SVGPathSegLinetoRel.idl:
* svg/SVGPathSegLinetoVerticalAbs.idl:
* svg/SVGPathSegLinetoVerticalRel.idl:
* svg/SVGPathSegMovetoAbs.idl:
* svg/SVGPathSegMovetoRel.idl:
* svg/SVGPoint.idl:
* svg/SVGPreserveAspectRatio.idl:
* svg/SVGRect.idl:

LayoutTests:

Rebaseline / update existing tests to reflect behavior change.

* fast/speechsynthesis/speech-synthesis-crash-on-bad-utterance-expected.txt:
* fast/speechsynthesis/speech-synthesis-crash-on-bad-utterance.html:
* webaudio/convolver-setBuffer-null-expected.txt:
* webaudio/convolver-setBuffer-null.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203949 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agomacOS Sierra test result gardening for media/video-loop.html.
ap@apple.com [Sat, 30 Jul 2016 21:08:57 +0000 (21:08 +0000)]
macOS Sierra test result gardening for media/video-loop.html.

* platform/mac/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203947 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agomacOS Sierra test result gardening.
ap@apple.com [Sat, 30 Jul 2016 20:30:01 +0000 (20:30 +0000)]
macOS Sierra test result gardening.

* platform/mac-wk1/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203945 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[Xcode] WebKit.framework is touched on incremental build even if nothing’s changed
mitz@apple.com [Sat, 30 Jul 2016 19:27:23 +0000 (19:27 +0000)]
[Xcode] WebKit.framework is touched on incremental build even if nothing’s changed
https://bugs.webkit.org/show_bug.cgi?id=160383

Reviewed by Tim Horton.

* WebKit2.xcodeproj/project.pbxproj: Removed the output files listed under the
  Copy iOS Sandbox Profiles for Manual Sandboxing script build phase, because it doesn’t
  produce those files on macOS. With no outputs listed, the script will still run every
  time, but the fact that it has run won’t trigger touching the framework.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203944 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoSynchronous preflight should check for successful responses
commit-queue@webkit.org [Sat, 30 Jul 2016 19:27:09 +0000 (19:27 +0000)]
Synchronous preflight should check for successful responses
https://bugs.webkit.org/show_bug.cgi?id=159350

Patch by Youenn Fablet <youenn@apple.com> on 2016-07-30
Reviewed by Darin Adler.

Source/WebCore:

Test: http/tests/xmlhttprequest/access-control-preflight-not-successful.html

* loader/CrossOriginPreflightChecker.cpp:
(WebCore::CrossOriginPreflightChecker::doPreflight): Adding successful response check for synchronous
preflighting.

LayoutTests:

* http/tests/xmlhttprequest/access-control-preflight-not-successful-expected.txt: Added.
* http/tests/xmlhttprequest/access-control-preflight-not-successful.html: Added, not all tests are passing as CORS checks.
for redirections are not well supported for synchronous loading.
* http/tests/xmlhttprequest/resources/status-404-without-body.php:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203943 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUnreviewed: add myself to the reviewers list.
rego@igalia.com [Sat, 30 Jul 2016 14:16:37 +0000 (14:16 +0000)]
Unreviewed: add myself to the reviewers list.

* Scripts/webkitpy/common/config/contributors.json:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203942 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[WebIDL] Enable strict type checking for operations' nullable parameters of wrapper...
cdumez@apple.com [Sat, 30 Jul 2016 08:35:06 +0000 (08:35 +0000)]
[WebIDL] Enable strict type checking for operations' nullable parameters of wrapper types
https://bugs.webkit.org/show_bug.cgi?id=160374

Reviewed by Youenn Fablet.

LayoutTests/imported/w3c:

Rebaseline W3C test now that one more check is passing.

* web-platform-tests/dom/nodes/DOMImplementation-createDocument-expected.txt:

Source/WebCore:

Enable strict type checking for operations' nullable parameters of
wrapper types to comply with the Web IDL specification:
- http://heycam.github.io/webidl/#es-nullable-type
- http://heycam.github.io/webidl/#es-interface

For such parameters, if the JS passes a value that is not null /
undefined and does not have the expected wrapper type, then we now
throw a TypeError instead of silently converting the value to null.

This behavior is consistent with Chrome and Firefox. It also helps
identify bugs in JavaScript code.

No new tests, rebaselined existing tests.

* bindings/scripts/CodeGeneratorJS.pm:
(GenerateParametersCheck):
* bindings/scripts/test/JS/JSTestObj.cpp:
(WebCore::jsTestObjPrototypeFunctionMethodWithOptionalNullableWrapper):
(WebCore::jsTestObjPrototypeFunctionMethodWithOptionalNullableWrapperIsNull):
(WebCore::jsTestObjPrototypeFunctionMethodWithOptionalXPathNSResolver):
(WebCore::jsTestObjPrototypeFunctionOverloadedMethod1):
(WebCore::jsTestObjPrototypeFunctionOverloadedMethod2):
(WebCore::jsTestObjPrototypeFunctionOverloadedMethod6):
(WebCore::jsTestObjPrototypeFunctionOverloadedMethodWithOptionalParameter1):
(WebCore::jsTestObjPrototypeFunctionOverloadedMethodWithOptionalParameter2):
(WebCore::jsTestObjPrototypeFunctionConvert2):
(WebCore::jsTestObjPrototypeFunctionStrictFunctionWithSequence):
(WebCore::jsTestObjPrototypeFunctionConditionalMethod3): Deleted.
(WebCore::jsTestObjPrototypeFunctionOverloadedMethod5): Deleted.
(WebCore::jsTestObjPrototypeFunctionOverloadedMethod): Deleted.
(WebCore::jsTestObjPrototypeFunctionOrange): Deleted.

LayoutTests:

Update / rebaseline existing tests to reflect behavior change. Also fix
several bugs in tests that became obvious now that we throw.

* editing/execCommand/indent-pre-expected.txt:
* editing/execCommand/indent-pre.html:
* editing/selection/4895428-2.html:
* editing/selection/move-by-word-visually-null-box.html:
* fast/dom/DOMImplementation/detached-doctype.html:
* fast/dom/MutationObserver/observe-attributes.html:
* fast/dom/Node/contains-method-expected.txt:
* fast/dom/Node/script-tests/contains-method.js:
* fast/dom/Window/webkitConvertPoint.html:
* fast/dom/adopt-node-crash-2.html:
* fast/dom/incompatible-operations-expected.txt:
* fast/dom/incompatible-operations.html:
* fast/events/selectstart-by-arrow-keys.html:
* fast/images/image-copy-memory-usage.html:
* fast/xsl/xslt-processor-expected.txt:
* http/tests/misc/webtiming-cross-origin-and-back1.html:
* platform/mac/fast/dom/Window/webkitConvertPoint-expected.txt:
* svg/custom/unicode-in-tspan-multi-svg-crash.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203941 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUndo r203939, and copy over a rule that clips the logos.
ap@apple.com [Sat, 30 Jul 2016 06:39:37 +0000 (06:39 +0000)]
Undo r203939, and copy over a rule that clips the logos.

Sierra wasn't the only logo without transparency.

* BuildSlaveSupport/build.webkit.org-config/public_html/dashboard/Images/Sierra.png:
* BuildSlaveSupport/build.webkit.org-config/public_html/dashboard/Images/Sierra@2x.png:
* BuildSlaveSupport/build.webkit.org-config/public_html/dashboard/Styles/Main.css:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203940 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoThe Sierra asset at build.webkit.org/dashboard/images/Sierra*.png has a white background
commit-queue@webkit.org [Sat, 30 Jul 2016 04:42:09 +0000 (04:42 +0000)]
The Sierra asset at build.webkit.org/dashboard/images/Sierra*.png has a white background
https://bugs.webkit.org/show_bug.cgi?id=160377

Patch by Alan Sien Wei Hshieh <hshieh@apple.com> on 2016-07-29
Reviewed by Brian Weinstein.

Update assets to remove the white background and use a transparent background instead.

* BuildSlaveSupport/build.webkit.org-config/public_html/dashboard/Images/Sierra.png:
* BuildSlaveSupport/build.webkit.org-config/public_html/dashboard/Images/Sierra@2x.png:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203939 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoANGLE: Fix global constructors and exit-time destructors
ddkilzer@apple.com [Sat, 30 Jul 2016 03:22:19 +0000 (03:22 +0000)]
ANGLE: Fix global constructors and exit-time destructors
<https://webkit.org/b/160332>

Reviewed by Darin Adler.

* Configurations/Base.xcconfig:
- Add warning flags.

* src/common/angleutils.cpp:
(FormatString):
* src/common/angleutils.h:
(MakeStaticString):
* src/common/debug.cpp:
* src/common/mathutil.cpp:
(gl::g_sharedexp_max):
(gl::convertRGBFloatsTo999E5):
* src/compiler/translator/ShaderLang.cpp:
(ShGetUniformRegisterMap):
* src/libANGLE/Caps.cpp:
(gl::TextureCapsMap::get):
* src/libANGLE/Device.cpp:
(egl::GetDeviceSet):
* src/libANGLE/Display.cpp:
(egl::Display::getClientExtensionString):
* src/libANGLE/formatutils.cpp:
(gl::BuildFormatMap):
(gl::BuildInternalFormatInfoMap):
(gl::GetInternalFormatMap):
(gl::BuildAllSizedInternalFormatSet):
(gl::GetSizedInternalFormat):
(gl::GetAllSizedInternalFormats):
* src/libANGLE/validationES3.cpp:
(gl::BuildES3FormatSet):
(gl::ValidateTexImageFormatCombination):
(gl::BuildSizedEffectiveInternalFormatList):
(gl::BuildUnsizedEffectiveInternalFormatList):
(gl::GetEffectiveInternalFormat):
(gl::BuildValidES3CopyTexImageCombinations):
(gl::IsValidES3CopyTexImageCombination):
- Fix global constructors and exit-time destructors.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203938 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoTypedArray super constructor has some incompatabilities
keith_miller@apple.com [Sat, 30 Jul 2016 02:02:48 +0000 (02:02 +0000)]
TypedArray super constructor has some incompatabilities
https://bugs.webkit.org/show_bug.cgi?id=160369

Reviewed by Filip Pizlo.

This patch fixes the length proprety of the TypedArray super constructor.
Additionally, the TypedArray super constructor should no longer be callable.

Also, this patch fixes the expected result of some test262 tests.

* runtime/JSTypedArrayViewConstructor.cpp:
(JSC::JSTypedArrayViewConstructor::finishCreation):
(JSC::constructTypedArrayView):
(JSC::JSTypedArrayViewConstructor::getCallData):
* tests/test262.yaml:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203937 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoIgnore most check-webkit-style checks for ANGLE project
ddkilzer@apple.com [Sat, 30 Jul 2016 01:25:31 +0000 (01:25 +0000)]
Ignore most check-webkit-style checks for ANGLE project
<https://webkit.org/b/160333>

Reviewed by Daniel Bates.

* Scripts/webkitpy/style/checker.py: Ignore all checkers except:
- No tabs.
- No spaces at the end of lines.
- No carriage returns (DOS line endings).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203936 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWindow's named properties should be exposed on a WindowProperties object in its prototype
cdumez@apple.com [Sat, 30 Jul 2016 01:08:41 +0000 (01:08 +0000)]
Window's named properties should be exposed on a WindowProperties object in its prototype
https://bugs.webkit.org/show_bug.cgi?id=160354

Reviewed by Gavin Barraclough.

LayoutTests/imported/w3c:

Rebaseline W3C test now that one more check is passing.

* web-platform-tests/html/dom/interfaces-expected.txt:

Source/WebCore:

Window's named properties should be exposed on a WindowProperties object
in its prototype:
- http://heycam.github.io/webidl/#named-properties-object

Firefox and Chrome both comply with the specification. However, WebKit
had no "WindowProperties" object in the Window prototype chain and the
named properties are exposed on the Window object itself.

No new tests, rebaselined existing tests.

* CMakeLists.txt:
* WebCore.xcodeproj/project.pbxproj:
* bindings/js/JSBindingsAllInOne.cpp:
* bindings/js/JSDOMWindowCustom.cpp:
(WebCore::JSDOMWindow::getOwnPropertySlot):
(WebCore::JSDOMWindow::getOwnPropertySlotByIndex):
(WebCore::jsDOMWindowGetOwnPropertySlotRestrictedAccess): Deleted.
(WebCore::JSDOMWindow::put): Deleted.
(WebCore::JSDOMWindow::putByIndex): Deleted.
(WebCore::JSDOMWindow::getEnumerableLength): Deleted.
* bindings/js/JSDOMWindowProperties.cpp: Added.
(WebCore::jsDOMWindowPropertiesGetOwnPropertySlotNamedItemGetter):
(WebCore::JSDOMWindowProperties::getOwnPropertySlot):
(WebCore::JSDOMWindowProperties::getOwnPropertySlotByIndex):
* bindings/js/JSDOMWindowProperties.h: Added.
(WebCore::JSDOMWindowProperties::create):
(WebCore::JSDOMWindowProperties::createStructure):
(WebCore::JSDOMWindowProperties::JSDOMWindowProperties):
* bindings/js/JSDOMWindowShell.cpp:
(WebCore::JSDOMWindowShell::setWindow):

LayoutTests:

* fast/dom/Window/es52-globals-expected.txt:
Update / Rebaseline test now that named properties are no longer reported as "own"
properties on the Window object. I have verified that the test gives the
same result in Firefox and Chrome.

* fast/loader/window-clearing-expected.txt:
Rebaseline test that prints one more line because there is one more
object in Window's prototype chain.

* http/tests/security/window-named-proto-expected.txt:
* http/tests/security/window-named-valueOf-expected.txt:
Rebaseline 2 security tests that give slightly different output. The new
output is identical to the one in Firefox and Chrome. The tests are not
failing since they are not alert'ing content from the other frame.
The reason those tests were logging a security error is because we would
previously prevent named property access if the frame name conflicts with
a property name in the Window prototype, and we now no longer
differentiate this case.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203935 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoMarking http/tests/xmlhttprequest/workers/methods-async.html as flaky on ios-simulato...
ryanhaddad@apple.com [Sat, 30 Jul 2016 00:54:10 +0000 (00:54 +0000)]
Marking http/tests/xmlhttprequest/workers/methods-async.html as flaky on ios-simulator release WK2
https://bugs.webkit.org/show_bug.cgi?id=160368

Unreviewed test gardening.

* platform/ios-simulator-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203934 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoRebaseline fast/block/float/overhanging-tall-block.html for Sierra.
ryanhaddad@apple.com [Sat, 30 Jul 2016 00:39:43 +0000 (00:39 +0000)]
Rebaseline fast/block/float/overhanging-tall-block.html for Sierra.

Unreviewed test gardening.

* platform/mac-elcapitan/fast/block/float/overhanging-tall-block-expected.png: Renamed from LayoutTests/platform/mac/fast/block/float/overhanging-tall-block-expected.png.
* platform/mac-elcapitan/fast/block/float/overhanging-tall-block-expected.txt: Copied from LayoutTests/platform/mac/fast/block/float/overhanging-tall-block-expected.txt.
* platform/mac/TestExpectations:
* platform/mac/fast/block/float/overhanging-tall-block-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203932 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoCrash under HTMLMediaElement::{resolve, reject}PendingPlayPromises() when playback...
dbates@webkit.org [Sat, 30 Jul 2016 00:30:13 +0000 (00:30 +0000)]
Crash under HTMLMediaElement::{resolve, reject}PendingPlayPromises() when playback is interrupted
https://bugs.webkit.org/show_bug.cgi?id=160366
<rdar://problem/27317407>

Reviewed by Eric Carlson.

Source/WebCore:

Fixes a crash/assertion failure in DeferredWrapper::{resolve, rejectWithValue}() caused by a Promise
being settled twice. In particular, if a system interruption occurs when media.play() is invoked
the returned Promise may ultimately be settled twice upon cessation of the interruption.

A Promise can be settled (resolved) exactly once. When a system interruption occurs media
playback is paused and resumes on cessation of the interruption. Currently we also immediately
reject the Promise p retuned by media.play() if the interruption occurs during its invocation.
So, when we resume playback on cessation of an interruption we try to resolve p again. But a
Promise can only be resolved once and hence we violate the assertions that p has both a valid
reference to a JSPromiseDeferred object and a reference to the global object of the page.

Tests: media/non-existent-video-playback-interrupted.html
       media/video-playback-interrupted.html

* html/HTMLMediaElement.cpp:
(WebCore::HTMLMediaElement::play): Modified to reject the Promise and return immediately if
playInternal() returns false.
(WebCore::HTMLMediaElement::playInternal): Treat an interruption as success and return true
so that HTMLMediaElement::play() adds the Promise to the end of the list of pending promises.
We treat an interruption as a success because we will resume playback (assuming the media
can be loaded and is well-formed) upon cessation of the interruption and therefore can either
fulfill or reject the Promise object returned by media.play().

LayoutTests:

* media/non-existent-video-playback-interrupted-expected.txt: Added.
* media/non-existent-video-playback-interrupted.html: Added.
* media/video-playback-interrupted-expected.txt: Added.
* media/video-playback-interrupted.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203931 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[iOS] HTMLMediaElement::updateVolume() calls MediaPlayer::volume() on null media...
dbates@webkit.org [Sat, 30 Jul 2016 00:13:49 +0000 (00:13 +0000)]
[iOS] HTMLMediaElement::updateVolume() calls MediaPlayer::volume() on null media player
https://bugs.webkit.org/show_bug.cgi?id=160353

Reviewed by Eric Carlson.

Using testing functions HTMLMediaElement::updateVolume() can be called on a <video> element
without an associated video. We should support this usage on iOS just as we do for non-iOS
ports.

The use of testing functions to call HTMLMediaElement::updateVolume() on a <video> element
without an associated video is demonstrated by the test LayoutTests/media/video-muted-after-setting-page-muted-state.html.

* html/HTMLMediaElement.cpp:
(WebCore::HTMLMediaElement::updateVolume):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203930 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoLand test expectations for rdar://problem/27611932.
ryanhaddad@apple.com [Sat, 30 Jul 2016 00:11:32 +0000 (00:11 +0000)]
Land test expectations for rdar://problem/27611932.

* platform/mac-wk1/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203929 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoMedia controls are not displayed for some autoplaying videos at certain browser dimen...
wenson_hsieh@apple.com [Sat, 30 Jul 2016 00:11:27 +0000 (00:11 +0000)]
Media controls are not displayed for some autoplaying videos at certain browser dimensions
https://bugs.webkit.org/show_bug.cgi?id=160360
<rdar://problem/27179484>

Reviewed by Myles C. Maxfield.

Source/WebCore:

Previously, if a video's aspect ratio fell outside of the range [0.5, 1.8], we would
not consider it main content and would subsequently not show media controls for it.
This meant that on many websites that scale video dimensions to match the mainframe,
if the mainframe is too wide (e.g. full bounds on a widescreen display) we would not
consider the video to be main content. To fix this, we only consider aspect ratio to
be a requirement if the video does not already take up most of the space in the
mainframe.

Covered by two new TestWebKitAPI unit tests.

* html/MediaElementSession.cpp:
(WebCore::isElementLargeRelativeToMainFrame):
(WebCore::isElementLargeEnoughForMainContent):

Tools:

Adds two tests verifying that videos may be considered main content as long as they
are large enough and cover a majority of the mainframe's viewport.

* TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
* TestWebKitAPI/Tests/WebKit2Cocoa/VideoControlsManager.mm:
(TestWebKitAPI::TEST):
* TestWebKitAPI/Tests/WebKit2Cocoa/full-size-autoplaying-video-with-audio.html: Added.
* TestWebKitAPI/Tests/WebKit2Cocoa/skinny-autoplaying-video-with-audio.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203928 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoTest gardening on ios-simualtor after some media tests were unskipped in r203906.
ryanhaddad@apple.com [Sat, 30 Jul 2016 00:07:37 +0000 (00:07 +0000)]
Test gardening on ios-simualtor after some media tests were unskipped in r203906.

Unreviewed test gardening.

* platform/ios-simulator/TestExpectations:
* platform/ios-simulator/media/audio-controls-rendering-expected.txt: Added.
* platform/ios-simulator/media/audio-repaint-expected.txt: Added.
* platform/ios-simulator/media/controls-strict-expected.txt: Added.
* platform/ios-simulator/media/controls-styling-strict-expected.txt: Added.
* platform/ios-simulator/media/controls-without-preload-expected.txt: Added.
* platform/ios-simulator/media/media-controls-clone-expected.txt: Added.
* platform/ios-simulator/media/track/track-cue-rendering-horizontal-expected.txt: Added.
* platform/ios-simulator/media/track/track-cue-rendering-vertical-expected.txt: Added.
* platform/ios-simulator/media/video-aspect-ratio-expected.txt: Added.
* platform/ios-simulator/media/video-controls-captions-trackmenu-expected.txt: Added.
* platform/ios-simulator/media/video-controls-captions-trackmenu-hide-on-click-expected.txt: Added.
* platform/ios-simulator/media/video-controls-captions-trackmenu-localized-expected.txt: Added.
* platform/ios-simulator/media/video-controls-captions-trackmenu-sorted-expected.txt: Added.
* platform/ios-simulator/media/video-controls-rendering-expected.txt: Added.
* platform/ios-simulator/media/video-display-toggle-expected.txt: Added.
* platform/ios-simulator/media/video-empty-source-expected.txt: Added.
* platform/ios-simulator/media/video-layer-crash-expected.txt: Added.
* platform/ios-simulator/media/video-no-audio-expected.txt: Added.
* platform/ios-simulator/media/video-transformed-expected.txt: Added.
* platform/ios-simulator/media/video-volume-slider-expected.txt: Added.
* platform/ios-simulator/media/video-zoom-controls-expected.txt: Added.
* platform/ios-simulator/media/video-zoom-expected.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203927 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoUndefined Behavior in JSValue cast from NaN
commit-queue@webkit.org [Fri, 29 Jul 2016 23:42:08 +0000 (23:42 +0000)]
Undefined Behavior in JSValue cast from NaN
https://bugs.webkit.org/show_bug.cgi?id=160322

Patch by Jonathan Bedard <jbedard@apple.com> on 2016-07-29
Reviewed by Mark Lam.

JSValues can be constructed from doubles, and in some cases, are deliberately constructed with NaN values.

In circumstances where NaN is bound through the default JSValue constructor, however, an undefined conversion
to int32_t occurs.  While the subsequent if statement should fail and construct the JSValue through the explicit
double constructor, given that the deliberate use of NaN is fairly common, it seems that the jsNaN() function
should immediately call the explicit double constructor both for efficiency and to prevent inadvertent
suppressing of any other bugs which may be instantiating a JSValue with a NaN double.

* runtime/JSCJSValueInlines.h:
(JSC::jsNaN): Explicit double construction for NaN JSValues to avoid undefined behavior.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203925 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoDo not set negative rate on AVSampleBufferRenderSynchronizer.
zalan@apple.com [Fri, 29 Jul 2016 23:41:54 +0000 (23:41 +0000)]
Do not set negative rate on AVSampleBufferRenderSynchronizer.
https://bugs.webkit.org/show_bug.cgi?id=160326
<rdar://problem/27355214>

Reviewed by Eric Carlson.

Source/WebCore:

Test: http/tests/media/media-source/mediasource-play-then-seek-back-with-remote-control.html

* platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.mm:
(WebCore::MediaPlayerPrivateMediaSourceAVFObjC::setRateDouble):

LayoutTests:

Negative rate is not supported at the moment.

* http/tests/media/media-source/mediasource-play-then-seek-back-with-remote-control-expected.txt: Added.
* http/tests/media/media-source/mediasource-play-then-seek-back-with-remote-control.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203924 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoRefactor DFG::Node::hasLocal() to accessesStack()
msaboff@apple.com [Fri, 29 Jul 2016 22:04:02 +0000 (22:04 +0000)]
Refactor DFG::Node::hasLocal() to accessesStack()
https://bugs.webkit.org/show_bug.cgi?id=160357

Reviewed by Filip Pizlo.

Refactoring in preparation for using register arguments for JavaScript calls.

Renamed Node::hasLocal() to Node::accessesStack() and changed all uses accordingly.
Also changed uses of Node::hasVariableAccessData() to accessesStack() where that
use guards stack operation logic associated with the Node's VariableAccessData.

The hasVariableAccessData() check now implies no more than the node has a
VariableAccessData and nothing about its use of that data to coordinate stack
accesses.

* dfg/DFGGraph.cpp:
(JSC::DFG::Graph::dump):
(JSC::DFG::Graph::methodOfGettingAValueProfileFor):
* dfg/DFGLiveCatchVariablePreservationPhase.cpp:
(JSC::DFG::LiveCatchVariablePreservationPhase::handleBlock):
* dfg/DFGMaximalFlushInsertionPhase.cpp:
(JSC::DFG::MaximalFlushInsertionPhase::treatRegularBlock):
(JSC::DFG::MaximalFlushInsertionPhase::treatRootBlock):
* dfg/DFGNode.h:
(JSC::DFG::Node::containsMovHint):
(JSC::DFG::Node::accessesStack):
(JSC::DFG::Node::hasLocal): Deleted.
* dfg/DFGPredictionInjectionPhase.cpp:
(JSC::DFG::PredictionInjectionPhase::run):
* dfg/DFGValidate.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203923 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[macOS Sierra] Test gardening after r198859
mmaxfield@apple.com [Fri, 29 Jul 2016 22:01:03 +0000 (22:01 +0000)]
[macOS Sierra] Test gardening after r198859
https://bugs.webkit.org/show_bug.cgi?id=160358
<rdar://problem/26422274>

Reviewed by Zalan Bujtas.

RTL Scrollbars are not enabled on WK1, so this test should be marked in the
same way as all the other RTL scrollbars tests.

* TestExpectations:
* platform/mac-wk1/TestExpectations:
* platform/mac-wk2/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203922 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years ago[JSC] Use the same data structures for DFG and Air Liveness Analysis
benjamin@webkit.org [Fri, 29 Jul 2016 20:58:35 +0000 (20:58 +0000)]
[JSC] Use the same data structures for DFG and Air Liveness Analysis
https://bugs.webkit.org/show_bug.cgi?id=160346

Reviewed by Geoffrey Garen.

In Air, we minimized memory accesses during liveness analysis
with a couple of tricks:
-Use a single Sparse Set ADT for the live value of each block.
-Manipulate compact positive indices instead of hashing values.

This patch brings the same ideas to DFG.

This patch still uses the same fixpoint algorithms.
The reason is Edge's KillStatus used by other phases. We cannot
use a block-boundary liveness algorithm and update KillStatus
simultaneously. It's something I'll probably revisit at some point.

* dfg/DFGAbstractInterpreterInlines.h:
(JSC::DFG::AbstractInterpreter<AbstractStateType>::forAllValues):
(JSC::DFG::AbstractInterpreter<AbstractStateType>::dump):
* dfg/DFGBasicBlock.h:
* dfg/DFGGraph.h:
(JSC::DFG::Graph::maxNodeCount):
(JSC::DFG::Graph::nodeAt):
* dfg/DFGInPlaceAbstractState.cpp:
(JSC::DFG::setLiveValues):
(JSC::DFG::InPlaceAbstractState::endBasicBlock):
* dfg/DFGLivenessAnalysisPhase.cpp:
(JSC::DFG::LivenessAnalysisPhase::LivenessAnalysisPhase):
(JSC::DFG::LivenessAnalysisPhase::run):
(JSC::DFG::LivenessAnalysisPhase::processBlock):
(JSC::DFG::LivenessAnalysisPhase::addChildUse):
(JSC::DFG::LivenessAnalysisPhase::process): Deleted.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203921 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoWeb Inspector: Inactive/active network bar segments overlap when latency is zero
mattbaker@apple.com [Fri, 29 Jul 2016 20:43:42 +0000 (20:43 +0000)]
Web Inspector: Inactive/active network bar segments overlap when latency is zero
https://bugs.webkit.org/show_bug.cgi?id=160147
<rdar://problem/27516007>

Reviewed by Joseph Pecoraro.

TimelineRecordBar should hide the inactive segment when its duration
is less than the minimum displayable size. The active segment can
assume the whole width of the bar.

* UserInterface/Views/TimelineOverviewGraph.js:
(WebInspector.TimelineOverviewGraph.prototype.get secondsPerPixel):
Make secondsPerPixel available during TimelineRecordBars.refresh.

* UserInterface/Views/TimelineRecordBar.js:
(WebInspector.TimelineRecordBar.prototype.refresh):
Check inactive duration against the minimum display size,
and add/remove DOM nodes and bar styles as needed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203920 268f45cc-cd09-0410-ab3c-d52691b4dbfc

4 years agoRemove a Sierra specific expectation for http/tests/navigation/forward-and-cancel...
ap@apple.com [Fri, 29 Jul 2016 20:29:20 +0000 (20:29 +0000)]
Remove a Sierra specific expectation for http/tests/navigation/forward-and-cancel.html,
we already have an identical one in platform/mac.

* platform/mac-wk1/TestExpectations:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@203919 268f45cc-cd09-0410-ab3c-d52691b4dbfc