Cap lifetime of persistent cookies created client-side through document.cookie

[WebKit-https.git] / Source / WebCore / ChangeLog

2018-09-24  John Wilander  <wilander@apple.com>

        Cap lifetime of persistent cookies created client-side through document.cookie
        https://bugs.webkit.org/show_bug.cgi?id=189933
        <rdar://problem/44741888>

        Reviewed by Chris Dumez.

        Test: http/tests/cookies/capped-lifetime-for-cookie-set-in-js.html

        As pointed out in https://github.com/mikewest/http-state-tokens:

        1) Cookies are available to JavaScript by default via document.cookie, which
        enables a smooth upgrade from one-time XSS to theft of persistent credentials
        and also makes cookies available to Spectre-like attacks on memory.

        2) Though the HttpOnly attribute was introduced well over a decade ago, only
        ~8.31% of Set-Cookie operations use it today (stats from Chrome). We need
        developer incentives to put proper protections in place.

        3) The median (uncompressed) Cookie request header is 409 bytes, while the 90th
        percentile is 1,589 bytes, the 95th 2,549 bytes, the 99th 4,601 bytes, and
        ~0.1% of Cookie headers are over 10kB (stats from Chrome). This is bad for load
        performance.

        In addition to this, third-party scripts running in first-party contexts can
        read user data through document.cookie and even store cross-site tracking data
        in them.

        Authentication cookies should be HttpOnly and thus not be affected by
        restrictions to document.cookie. Cookies that persist for a long time should
        be Secure, HttpOnly, and SameSite to provide good security and privacy.

        By capping the lifetime of persistent cookies set through document.cookie we
        embark on a journey towards better cookie management on the web.

        * platform/network/cocoa/NetworkStorageSessionCocoa.mm:
        (WebCore::filterCookies):
            Now caps the life time of persistent cookies to one week (seven days).
        * testing/Internals.cpp:
        (WebCore::Internals::getCookies const):
            New test function to get to cookie meta data such as expiry.
        * testing/Internals.h:
        * testing/Internals.idl:

2018-09-24  Simon Fraser  <simon.fraser@apple.com>

        Remove filterRes parameter from SVG filters
        https://bugs.webkit.org/show_bug.cgi?id=129565
        <rdar://problem/44714340>

        Reviewed by Dean Jackson.
        
        Remove support for the "filterRes" attribute on SVG filters. It's marked as
        deprecated in https://drafts.fxtf.org/filter-effects/#element-attrdef-filter-filterres
        and no longer supported by Chrome or Firefox.
        
        Removed existing filterRes tests, added new test checking that it has no effect.

        Tests: svg/filters/filterRes-is-noop.svg

        * rendering/svg/RenderSVGResourceFilter.cpp:
        (WebCore::RenderSVGResourceFilter::applyResource):
        * svg/SVGElement.cpp:
        (WebCore::SVGElement::animatableAttributeForName):
        * svg/SVGFilterElement.cpp:
        (WebCore::SVGFilterElement::registerAttributes):
        (WebCore::SVGFilterElement::parseAttribute):
        (WebCore::SVGFilterElement::filterResXIdentifier): Deleted.
        (WebCore::SVGFilterElement::filterResYIdentifier): Deleted.
        (WebCore::SVGFilterElement::setFilterRes): Deleted.
        * svg/SVGFilterElement.h:
        * svg/SVGFilterElement.idl:
        * svg/svgattrs.in:

2018-09-24  Ryosuke Niwa  <rniwa@webkit.org>

        Don't cause a crash even when some IDL attribute is missing CEReactions
        https://bugs.webkit.org/show_bug.cgi?id=189937

        Reviewed by Simon Fraser.

        Replaced release assertions in ElementQueue::add and ElementQueue::invokeAll by debug assertions
        since a missing CEReactions resulting in a crash is a terrible user experience.

        Also made the iteration in invokeAll safe when more elements were added to m_elements.

        No new tests since we would still hit debug assertions, and this behavior should only come up
        when some IDL attribute is erroneously missing CEReactions.

        * dom/CustomElementReactionQueue.cpp:
        (WebCore::CustomElementReactionQueue::ElementQueue::add):
        (WebCore::CustomElementReactionQueue::ElementQueue::invokeAll):

2018-09-24  Wenson Hsieh  <wenson_hsieh@apple.com>

        Refactor Editor::fontAttributesForSelectionStart to be platform-agnostic
        https://bugs.webkit.org/show_bug.cgi?id=189918
        Work towards <rdar://problem/44648705>

        Reviewed by Tim Horton.

        Refactors the functionality in Editor::fontAttributesForSelectionStart to not be Cocoa-only. Rename this to
        fontAttributesAtSelectionStart (to be consistent with `EditingStyle::styleAtSelectionStart`) and move it from
        EditorCocoa.mm to Editor.cpp; instead of creating and populating an NSDictionary with font attribute
        information, create and populate a new `FontAttributes` struct that contains the same information. Cocoa clients
        in WebKitLegacy may then create an `NSDictionary` as needed from the `FontAttributes`.

        * SourcesCocoa.txt:
        * WebCore.xcodeproj/project.pbxproj:
        * editing/Editor.cpp:
        (WebCore::Editor::platformFontAttributesAtSelectionStart const):

        Add a hook to allow platforms to supply additional information in FontAttributes. On Cocoa, this adds a UIFont
        or NSFont to FontAttributes; otherwise, this is a no-op.

        (WebCore::Editor::fontAttributesAtSelectionStart const):
        * editing/Editor.h:
        * editing/FontAttributeChanges.cpp:
        (WebCore::cssValueListForShadow):
        * editing/FontAttributeChanges.h:
        (): Deleted.
        (WebCore::FontShadow::encode const): Deleted.
        (WebCore::FontShadow::decode): Deleted.
        * editing/FontAttributes.h: Added.

        Introduce a new struct that contains font attribute information. May be converted into an NSDictionary for use
        by Cocoa clients in WebKitLegacy and WebKit. In a future patch, this will become serializable over IPC for use
        in WebKit2.

        * editing/FontShadow.h: Added.

        Move FontShadow out into a separate header file, included in `FontAttributeChanges.h` and `FontAttributes.h`.

        (WebCore::FontShadow::encode const):
        (WebCore::FontShadow::decode):
        * editing/cocoa/EditorCocoa.mm:

        Add a helper function to convert a WebCore::Color to either `UIColor` on iOS or `NSColor` when AppKit is being
        used.

        (WebCore::Editor::platformFontAttributesAtSelectionStart const):
        (WebCore::Editor::getTextDecorationAttributesRespectingTypingStyle const): Deleted.

        Remove a helper function that was only used to compute text decoration attributes in
        fontAttributesForSelectionStart.

        (WebCore::Editor::fontAttributesForSelectionStart const): Deleted.
        * editing/cocoa/FontAttributesCocoa.mm: Added.
        (WebCore::FontAttributes::createDictionary const):
        * editing/cocoa/FontShadowCocoa.mm: Added.
        (WebCore::FontShadow::createShadow const):
        * editing/cocoa/HTMLConverter.mm:
        (_webKitBundle):
        (HTMLConverter::_colorForElement):
        (_platformColor): Deleted.

        Adopt platformColor().

        * platform/graphics/cocoa/ColorCocoa.h: Added.
        * platform/graphics/cocoa/ColorCocoa.mm: Added.
        (WebCore::platformColor):
        * platform/graphics/metal/GPURenderPassDescriptorMetal.mm:

        Build fix due to changes in unified sources.

        * platform/mac/WebCoreNSFontManagerExtras.mm:
        (WebCore::computedFontAttributeChanges):
        * platform/mediastream/mac/RealtimeIncomingVideoSourceCocoa.mm:
        * platform/mediastream/mac/RealtimeIncomingVideoSourceCocoa.mm:

        More build fixes due to changes in unified sources.

2018-09-24  Jer Noble  <jer.noble@apple.com>

        SharedBuffer should have an equality test
        https://bugs.webkit.org/show_bug.cgi?id=189919

        Reviewed by Alex Christensen.

        Test: TestWebKitAPI SharedBuffer.isEqualTo.

        * platform/SharedBuffer.cpp:
        * platform/SharedBuffer.h:
        (WebCore::operator==):
        (WebCore::operator!=):

2018-09-24  Ryosuke Niwa  <rniwa@webkit.org>

        imported/w3c/web-platform-tests/shadow-dom/slotchange.html is a flaky failure
        https://bugs.webkit.org/show_bug.cgi?id=167652

        Reviewed by Saam Barati.

        The bug appears to be caused by the JS wrappers of slot elements getting prematurely collected.
        Deployed GCReachableRef introduced in r236376 to fix the bug.

        Test: fast/shadow-dom/signal-slot-list-retains-js-wrappers.html

        * dom/MutationObserver.cpp:
        (WebCore::signalSlotList):
        (WebCore::MutationObserver::enqueueSlotChangeEvent):
        (WebCore::MutationObserver::notifyMutationObservers):

2018-09-24  Ryosuke Niwa  <rniwa@webkit.org>

        Release assert when using paper-textarea due to autocorrect IDL attribute missing CEReactions
        https://bugs.webkit.org/show_bug.cgi?id=174629
        <rdar://problem/33407620>

        Reviewed by Simon Fraser.

        The bug was caused by autocorrect and autocapitalize IDL attributes missing CEReactions.

        Test: fast/custom-elements/autocorrect-autocapitalize-idl-attributes-crash.html

        * html/HTMLElement.idl:

2018-09-24  Chris Dumez  <cdumez@apple.com>

        No-op document.open() calls should not have any side effects
        https://bugs.webkit.org/show_bug.cgi?id=189373
        <rdar://problem/44282702>

        Reviewed by Geoffrey Garen.

        Update document.open() implementation to match the specification [1] more closely.
        In particular, URLs updates should happen much later, at step 11. They were happening
        too early and would cause side effects when returning early.

        [1] https://html.spec.whatwg.org/multipage/dynamic-markup-insertion.html#document-open-steps

        No new tests, rebaselined existing test.

        * dom/Document.cpp:
        (WebCore::Document::open):

2018-09-24  Daniel Bates  <dabates@apple.com>

        Include more headers in IOSurface.mm, PixelBufferResizer.{h, mm}
        https://bugs.webkit.org/show_bug.cgi?id=189928

        Reviewed by Andy Estes.

        Make building IOSurface.mm and PixelBufferResizer.mm deterministic regardless of
        which bundle of unified sources they are built in.

        * platform/graphics/cocoa/IOSurface.mm: Include headers HostWindow.h and PlatformScreen.h
        since we make use of functionality from these headers.
        * platform/graphics/cv/PixelBufferResizer.h: Include header IntSize.h since
        we need the size of an IntSize for m_size.
        * platform/graphics/cv/PixelBufferResizer.mm: Include header Logging.h since
        this file makes use of logging facilities. Also substitute #import for #include
        while I am here since this file is an Objective-C++ file.

2018-09-24  Andy Estes  <aestes@apple.com>

        [Payment Request] Events cleanup
        https://bugs.webkit.org/show_bug.cgi?id=189925

        Reviewed by Simon Fraser.

        1. Constructed MerchantValidationEvents (and ApplePayValidateMerchantEvents) with rvalue
        references to validationURLs.
        2. Instead of MerchantValidationEvent and PaymentRequestUpdateEvent having a
        RefPtr<PaymentRequest>, downcasted their target to a PaymentRequest. Trusted versions of
        these events are always dispatched to a PaymentRequest object.
        3. Defined MerchantValidationEventInit in MerchantValidationEvent.idl instead of having a
        separate .idl and .h for this dictionary.

        No new tests. No change in behavior.

        * CMakeLists.txt:
        * DerivedSources.make:
        * Modules/applepay/ApplePaySession.cpp:
        (WebCore::ApplePaySession::validateMerchant):
        * Modules/applepay/ApplePaySession.h:
        * Modules/applepay/ApplePayValidateMerchantEvent.cpp:
        (WebCore::ApplePayValidateMerchantEvent::ApplePayValidateMerchantEvent):
        * Modules/applepay/ApplePayValidateMerchantEvent.h:
        (WebCore::ApplePayValidateMerchantEvent::create):
        * Modules/applepay/PaymentCoordinator.cpp:
        (WebCore::PaymentCoordinator::validateMerchant):
        * Modules/applepay/PaymentCoordinator.h:
        * Modules/applepay/PaymentSession.h:
        * Modules/applepay/paymentrequest/ApplePayPaymentHandler.cpp:
        (WebCore::ApplePayPaymentHandler::validateMerchant):
        * Modules/applepay/paymentrequest/ApplePayPaymentHandler.h:
        * Modules/paymentrequest/MerchantValidationEvent.cpp:
        (WebCore::MerchantValidationEvent::create):
        (WebCore::MerchantValidationEvent::MerchantValidationEvent):
        (WebCore::MerchantValidationEvent::complete):
        * Modules/paymentrequest/MerchantValidationEvent.h:
        * Modules/paymentrequest/MerchantValidationEvent.idl:
        * Modules/paymentrequest/MerchantValidationEventInit.h: Removed.
        * Modules/paymentrequest/MerchantValidationEventInit.idl: Removed.
        * Modules/paymentrequest/PaymentMethodChangeEvent.cpp:
        (WebCore::PaymentMethodChangeEvent::PaymentMethodChangeEvent):
        * Modules/paymentrequest/PaymentMethodChangeEvent.h:
        * Modules/paymentrequest/PaymentRequest.cpp:
        (WebCore::PaymentRequest::shippingAddressChanged):
        (WebCore::PaymentRequest::shippingOptionChanged):
        (WebCore::PaymentRequest::paymentMethodChanged):
        * Modules/paymentrequest/PaymentRequest.h:
        (isType):
        * Modules/paymentrequest/PaymentRequestUpdateEvent.cpp:
        (WebCore::PaymentRequestUpdateEvent::PaymentRequestUpdateEvent):
        (WebCore::PaymentRequestUpdateEvent::updateWith):
        * Modules/paymentrequest/PaymentRequestUpdateEvent.h:
        * Sources.txt:
        * WebCore.xcodeproj/project.pbxproj:
        * dom/EventTarget.cpp:
        (WebCore::EventTarget::isPaymentRequest const):
        * dom/EventTarget.h:

2018-09-24  Daniel Bates  <dabates@apple.com>

        Separate Mac and iOS implementation of windowsKeyCodeForCharCode()
        https://bugs.webkit.org/show_bug.cgi?id=189898

        Reviewed by Tim Horton.

        Extract the Mac and iOS implementations into a windowsKeyCodeForCharCode() defined in file
        KeyEventMac.mm and file KeyEventIOS, respectively.

        Mac and iOS have significant differences in how they represent function keys. It is not worthwhile
        to share windowsKeyCodeForCharCode() between them given these differences. On Mac function keys
        are represented by a char code in the range 0xF700-0xF8FF. On iOS these keys may not have a unique
        char code (e.g. F1 = 0x10 = F10) and must be identified either by special string (e.g. UIKeyInputUpArrow)
        or key code.

        * platform/cocoa/KeyEventCocoa.mm:
        (WebCore::windowsKeyCodeForCharCode): Deleted.
        * platform/ios/KeyEventIOS.mm:
        (WebCore::windowsKeyCodeForCharCode): Added. Remove the handling of NS*FunctionKey char codes
        as function keys are not represented using them as of iOS 12.
        * platform/mac/KeyEventMac.mm:
        (WebCore::windowsKeyCodeForCharCode): Added.

2018-09-21  Simon Fraser  <simon.fraser@apple.com>

        Remove the old "AcceleratedCompositingForOverflowScroll" code
        https://bugs.webkit.org/show_bug.cgi?id=189870

        Reviewed by Zalan Bujtas.

        The "AcceleratedCompositingForOverflowScroll" code was added to allow overflow:scroll to use
        composited scrolling if an overflow:scroll could be made a stacking context without affecting
        z-order. We need overflow:scroll to be accelerated always, so a different approach is needed.
        Remove this old code (unused by any platform?) to make working on new code easier.

        * page/Settings.yaml:
        * rendering/RenderLayer.cpp:
        (WebCore::RenderLayer::RenderLayer):
        (WebCore::RenderLayer::setHasVisibleContent):
        (WebCore::RenderLayer::updateDescendantDependentFlags):
        (WebCore::RenderLayer::dirty3DTransformedDescendantStatus):
        (WebCore::RenderLayer::stackingContext const):
        (WebCore::compositingContainer):
        (WebCore::RenderLayer::addChild):
        (WebCore::RenderLayer::removeChild):
        (WebCore::RenderLayer::hasAcceleratedTouchScrolling const):
        (WebCore::RenderLayer::usesAcceleratedScrolling const):
        (WebCore::adjustedScrollDelta):
        (WebCore::RenderLayer::updateCompositingLayersAfterScroll):
        (WebCore::RenderLayer::updateScrollInfoAfterLayout):
        (WebCore::RenderLayer::enclosingFragmentedFlowAncestor const):
        (WebCore::RenderLayer::calculateClipRects const):
        (WebCore::RenderLayer::acceleratedCompositingForOverflowScrollEnabled const): Deleted.
        (WebCore::RenderLayer::updateDescendantsAreContiguousInStackingOrder): Deleted.
        (WebCore::RenderLayer::updateDescendantsAreContiguousInStackingOrderRecursive): Deleted.
        (WebCore::RenderLayer::positionNewlyCreatedOverflowControls): Deleted.
        (WebCore::RenderLayer::canBeStackingContainer const): Deleted.
        (WebCore::RenderLayer::stackingContainer const): Deleted.
        (WebCore::RenderLayer::needsCompositedScrolling const): Deleted.
        (WebCore::RenderLayer::updateNeedsCompositedScrolling): Deleted.
        * rendering/RenderLayer.h:
        (WebCore::RenderLayer::clearZOrderLists):
        (WebCore::RenderLayer::updateZOrderLists):
        * rendering/RenderLayerBacking.cpp:
        (WebCore::RenderLayerBacking::updateAfterLayout):
        (WebCore::RenderLayerBacking::computeParentGraphicsLayerRect const):
        (WebCore::RenderLayerBacking::updateGeometry):
        (WebCore::RenderLayerBacking::requiresHorizontalScrollbarLayer const):
        (WebCore::RenderLayerBacking::requiresVerticalScrollbarLayer const):
        (WebCore::RenderLayerBacking::requiresScrollCornerLayer const):
        (WebCore::RenderLayerBacking::compositingOpacity const):
        (WebCore::traverseVisibleNonCompositedDescendantLayers):
        (WebCore::RenderLayerBacking::hasUnpositionedOverflowControlsLayers const): Deleted.
        * rendering/RenderLayerBacking.h:
        * rendering/RenderLayerCompositor.cpp:
        (WebCore::RenderLayerCompositor::updateScrollCoordinatedLayersAfterFlush):
        (WebCore::RenderLayerCompositor::updateCustomLayersAfterFlush):
        (WebCore::styleChangeRequiresLayerRebuild):
        (WebCore::RenderLayerCompositor::enclosingNonStackingClippingLayer const):
        (WebCore::RenderLayerCompositor::computeCompositingRequirements):
        (WebCore::RenderLayerCompositor::rebuildCompositingLayerTree):
        (WebCore::RenderLayerCompositor::requiresCompositingLayer const):
        (WebCore::RenderLayerCompositor::requiresOwnBackingStore const):
        (WebCore::RenderLayerCompositor::reasonsForCompositing const):
        (WebCore::RenderLayerCompositor::requiresCompositingForBackfaceVisibility const):
        (WebCore::RenderLayerCompositor::isViewportConstrainedFixedOrStickyLayer const):
        (WebCore::RenderLayerCompositor::requiresCompositingForPosition const):
        (WebCore::RenderLayerCompositor::requiresCompositingForOverflowScrolling const):
        (WebCore::RenderLayerCompositor::requiresCompositingForScrolling const): Deleted.
        * rendering/RenderLayerCompositor.h:

2018-09-24  Youenn Fablet  <youenn@apple.com>

        Enable conversion of libwebrtc internal frames as CVPixelBuffer
        https://bugs.webkit.org/show_bug.cgi?id=189892

        Reviewed by Eric Carlson.

        Make sure to handle the case of libwebrtc frames that are not backed by CVPixelBuffer.
        No observable change of behavior.

        * platform/mediastream/libwebrtc/LibWebRTCProviderCocoa.cpp:
        (WebCore::LibWebRTCProviderCocoa::createDecoderFactory):
        (WebCore::LibWebRTCProviderCocoa::createEncoderFactory):
        Update according renamed methods.
        * platform/mediastream/mac/RealtimeIncomingVideoSourceCocoa.h:
        * platform/mediastream/mac/RealtimeIncomingVideoSourceCocoa.mm:
        (WebCore::RealtimeIncomingVideoSourceCocoa::pixelBufferFromVideoFrame):
        In case of libwebrtc frame that are not backed by CVPixelBuffer, we create
        a CVPixelBuffer from a pixel buffer pool.
        This CVPixelBuffer is then filled as part of webrtc::pixelBufferFromFrame.
        * platform/mediastream/mac/RealtimeOutgoingVideoSourceCocoa.mm:
        (WebCore::RealtimeOutgoingVideoSourceCocoa::convertToYUV):
        Make sure to use preferred pixel buffer format.

2018-09-24  Eric Carlson  <eric.carlson@apple.com>

        [MediaStream] Add mock window capture source
        https://bugs.webkit.org/show_bug.cgi?id=189843
        <rdar://problem/44687445>

        Reviewed by Youenn Fablet.

        No new tests, the API is disabled and it isn't possible to test yet.

        * platform/mediastream/mac/AVVideoCaptureSource.mm:
        (WebCore::AVVideoCaptureSource::processNewFrame): Remove an extra blank line.

        * platform/mock/MockMediaDevice.h:
        (WebCore::MockDisplayProperties::encode const): Get rid of defaultFrameRate, add type.
        (WebCore::MockDisplayProperties::decode): Ditto.
        (WebCore::MockMediaDevice::type const):

        * platform/mock/MockRealtimeMediaSourceCenter.cpp:
        (WebCore::defaultDevices): Add mock window devices.
        (WebCore::MockRealtimeMediaSourceCenter::audioDevices): Cleanup.
        (WebCore::MockRealtimeMediaSourceCenter::videoDevices): Cleanup.
        (WebCore::MockRealtimeMediaSourceCenter::displayDevices): New.

        * platform/mock/MockRealtimeVideoSource.cpp:
        (WebCore::MockRealtimeVideoSource::MockRealtimeVideoSource): Set default size on displays.
        (WebCore::MockRealtimeVideoSource::supportsSizeAndFrameRate): Call RealtimeVideoSource for 
        mock camera, base class for device.
        (WebCore::MockRealtimeVideoSource::setSizeAndFrameRate): Ditto.
        (WebCore::MockRealtimeVideoSource::generatePresets): ASSERT if called as a camera.
        (WebCore::MockRealtimeVideoSource::capabilities): updateCapabilities is only appropriate for cameras.
        (WebCore::MockRealtimeVideoSource::settings): Camera and Device are different surface types.
        (WebCore::MockRealtimeVideoSource::drawText): Render name, not ID.
        (WebCore::MockRealtimeVideoSource::mockDisplayType const):

        * platform/mock/MockRealtimeVideoSource.h:
        (WebCore::MockRealtimeVideoSource::mockDisplay const):
        (WebCore::MockRealtimeVideoSource::mockScreen const):
        (WebCore::MockRealtimeVideoSource::mockWindow const):

2018-09-24  Daniel Bates  <dabates@apple.com>

        [iOS] Key code is 0 for many hardware keyboard keys
        https://bugs.webkit.org/show_bug.cgi?id=189604

        Reviewed by Wenson Hsieh.

        Based off a patch by Jeremy Jones.

        Add iOS-specific implementation of windowsKeyCodeForKeyCode() to map an iOS virtual key code to
        the corresponding Windows virtual key code. Only hardware keyboard-generated events have a
        virtual key code. For software-generated keyboard events we do what we do now and compute the
        Windows virtual key code from the character string associated with the event.

        When a WebEvent is instantiated with a non-zero iOS virtual key code (keyCode) we now always
        convert it to its corresponding Windows virtual key code without considering the specified
        charactersIgnoringModifiers character string. Currently we prefer computing the key code from
        charactersIgnoringModifiers regardless of whether a non-zero iOS virtual key code was given.
        However this causes special keys, including function keys (e.g. F10) to be misidentified because
        keyboard layouts in iOS (at least iOS 12) map such special keys to ASCII control characters (e.g.
        F10 maps to ASCII control character "data link escape" = 0x10) as opposed to special 16-bit
        integral constants as we do on Mac (e.g. F10 maps to NSF10FunctionKey = 0xF70D on Mac). I will
        look to fix up the computation of a Windows virtual key code from a char code on iOS in a
        subsequent commit(s). For now, computing the Windows virtual key code directly from the iOS
        virtual key code specified to the WebEvent constructor avoids the misidentification using
        an ANSI US keyboard layout.

        * platform/cocoa/KeyEventCocoa.mm:
        (WebCore::windowsKeyCodeForKeyCode): Deleted; moved to KeyEventMac.mm as this mapping is specific to Mac.
        * platform/ios/KeyEventIOS.mm:
        (WebCore::windowsKeyCodeForKeyCode): Added.
        * platform/ios/WebEvent.mm:
        (-[WebEvent initWithKeyEventType:timeStamp:characters:charactersIgnoringModifiers:modifiers:isRepeating:withFlags:keyCode:isTabKey:characterSet:]): Address the NOTE comment and compute the Windows virtual key code from
        the iOS virtual key code when we have one. Also inline the value of an unncessary local variable.
        (-[WebEvent initWithKeyEventType:timeStamp:characters:charactersIgnoringModifiers:modifiers:isRepeating:withFlags:withInputManagerHint:keyCode:isTabKey:]): Ditto.
        * platform/mac/KeyEventMac.mm:
        (WebCore::windowsKeyCodeForKeyCode): Moved from KeyEventCocoa.mm. Updated code to make use of WTF_ARRAY_LENGTH() instead
        of hardcoding the upper bound of the lookup table.

2018-09-24  Simon Fraser  <simon.fraser@apple.com>

        feMorphology filter in CSS doesn't update when element moves
        https://bugs.webkit.org/show_bug.cgi?id=189895

        Reviewed by Dean Jackson.
        
        SourceAlpha needs to be invalidated from clearIntermediateResults(),
        so get it from the SVGFilterBuilder (which always creates one) and store in
        a member variable.

        Test: css3/filters/invalidate-sourceAlpha.html

        * rendering/CSSFilter.cpp:
        (WebCore::CSSFilter::buildReferenceFilter):
        (WebCore::CSSFilter::clearIntermediateResults):
        * rendering/CSSFilter.h:
        * svg/graphics/filters/SVGFilterBuilder.h:

2018-09-24  Simon Fraser  <simon.fraser@apple.com>

        CSS reference filter with feDisplacementMap shows buffer corruption on Retina displays
        https://bugs.webkit.org/show_bug.cgi?id=188486
        <rdar://problem/43189750>

        Reviewed by Dean Jackson.
        
        The paintSize needs to be scaled by filterScale on Retina displays.

        Test: css3/filters/hidpi-feDisplacementMap.html

        * platform/graphics/filters/FEDisplacementMap.cpp:
        (WebCore::FEDisplacementMap::platformApplySoftware):
        * platform/graphics/filters/FEDisplacementMap.h:

2018-09-24  Simon Fraser  <simon.fraser@apple.com>

        ReferenceFilterOperation doesn't need to store the FilterEffect
        https://bugs.webkit.org/show_bug.cgi?id=189904

        Reviewed by Dean Jackson.

        ReferenceFilterOperation doesn't do anything with m_filterEffect so don't store it.

        * platform/graphics/filters/FilterOperation.cpp:
        (WebCore::ReferenceFilterOperation::setFilterEffect): Deleted.
        * platform/graphics/filters/FilterOperation.h:
        (WebCore::ReferenceFilterOperation::filterEffect const): Deleted.
        * rendering/CSSFilter.cpp:
        (WebCore::CSSFilter::build):

2018-09-24  Simon Fraser  <simon.fraser@apple.com>

        Garbled rendering of image when applied feConvolveMatrix to it, on Retina display
        https://bugs.webkit.org/show_bug.cgi?id=189748
        <rdar://problem/44621494>

        Reviewed by Jon Lee.
        
        feConvolveMatrix needs to scale the paintSize by the filter scale (2x on Retina displays),
        otherwise parts of the output buffer are uninitialized and the result is incorrect.

        Test: css3/filters/hidpi-feConvolveMatrix.html

        * platform/graphics/filters/FEConvolveMatrix.cpp:
        (WebCore::FEConvolveMatrix::platformApplySoftware):

2018-09-22  Dean Jackson  <dino@apple.com>

        Ensure PointerEvent is not visible when disabled
        https://bugs.webkit.org/show_bug.cgi?id=189889
        <rdar://problem/44708253>

        Reviewed by Eric Carlson.

        Test: pointerevents/disabled.html

        * bindings/js/WebCoreBuiltinNames.h: Now that it is enabled at runtime, it needs
        a built-in name.
        * dom/PointerEvent.idl: Add EnabledAtRuntime.

2018-09-24  Alicia Boya García  <aboya@igalia.com>

        [MSE][GStreamer] Pull demuxed samples in batches
        https://bugs.webkit.org/show_bug.cgi?id=189871

        Reviewed by Xabier Rodriguez-Calvar.

        After this patch, only the notifications of "new samples available"
        (appsink-new-sample bus messages) travel from the streaming thread to
        the main thread through the bus and the main thread is the responsible
        of pulling as many samples as it can from appsink. Before, the samples
        were pulled from appsink in the non-main thread and traveled to the
        main thread through the bus one by one.

        This reduces drastically the amount of context switches and waiting
        time in the streaming thread, resulting in a noticeable performance
        improvement.

        This fixes stutter while loading YouTube videos.

        * platform/graphics/gstreamer/MediaSampleGStreamer.cpp:
        (WebCore::MediaSampleGStreamer::MediaSampleGStreamer):
        * platform/graphics/gstreamer/mse/AppendPipeline.cpp:
        (WebCore::AppendPipeline::AppendPipeline):
        (WebCore::AppendPipeline::~AppendPipeline):
        (WebCore::AppendPipeline::clearPlayerPrivate):
        (WebCore::AppendPipeline::handleApplicationMessage):
        (WebCore::AppendPipeline::appsinkNewSample):
        (WebCore::AppendPipeline::consumeAppSinkAvailableSamples):
        (WebCore::AppendPipeline::resetPipeline):
        (WebCore::AppendPipeline::handleNewAppsinkSample):
        * platform/graphics/gstreamer/mse/AppendPipeline.h:

2018-09-24  Per Arne Vollan  <pvollan@apple.com>

        [WebVTT] Change name of WebVTT region attribute 'height' to 'lines'.
        https://bugs.webkit.org/show_bug.cgi?id=189862

        Reviewed by Eric Carlson.

        The WebVTT region attribute 'height' has been renamed to 'lines', see https://w3c.github.io/webvtt/#region-settings.

        No new tests, covered by existing tests.

        * html/track/VTTRegion.cpp:
        (WebCore::VTTRegion::setLines):
        (WebCore::VTTRegion::updateParametersFromRegion):
        (WebCore::VTTRegion::scanSettingName):
        (WebCore::VTTRegion::parseSettingValue):
        (WebCore::VTTRegion::prepareRegionDisplayTree):
        (WebCore::VTTRegion::setHeight): Deleted.
        * html/track/VTTRegion.h:
        * html/track/VTTRegion.idl:

2018-09-24  Alicia Boya García  <aboya@igalia.com>

        [MSE][GStreamer] Use no-more-pads event for noticing initialization segments
        https://bugs.webkit.org/show_bug.cgi?id=189868

        Reviewed by Xabier Rodriguez-Calvar.

        Fixes the following YTTV 2018 tests:
        62.VideoDimensionVP9
        63.PlaybackStateVP9

        This removes the hack that was making supporting multiple tracks in
        the same file in MSE impossible.

        For WebM, this GStreamer patch is required:
        https://bugzilla.gnome.org/show_bug.cgi?id=797187
        "matroskademux: Emit no-more-pads after parsing Tracks"

        * platform/graphics/gstreamer/mse/AppendPipeline.cpp:
        (WebCore::AppendPipeline::AppendPipeline):
        (WebCore::AppendPipeline::handleApplicationMessage):
        (WebCore::AppendPipeline::demuxerNoMorePads):
        (WebCore::AppendPipeline::appsinkCapsChanged):
        (WebCore::AppendPipeline::connectDemuxerSrcPadToAppsink):
        (WebCore::AppendPipeline::appendPipelineDemuxerNoMorePadsFromAnyThread):
        (WebCore::appendPipelineDemuxerNoMorePads):
        * platform/graphics/gstreamer/mse/AppendPipeline.h:

2018-09-24  Thibault Saunier  <tsaunier@igalia.com>

        [WPE][GTK][WebRTC] Fix leaks in the libwebrtc Decoder and Encoder
        https://bugs.webkit.org/show_bug.cgi?id=189835

        Reviewed by Philippe Normand.

        - Rework memory management to avoid leaking encoded frames (basically use the same
          strategy as other libwebrtc encoder implementation).
        - Plug a GstCaps leak.

        * platform/mediastream/gstreamer/GStreamerVideoCapturer.cpp:
        * platform/mediastream/libwebrtc/GStreamerVideoDecoderFactory.cpp:
        * platform/mediastream/libwebrtc/GStreamerVideoEncoderFactory.cpp:
        (WebCore::GStreamerVideoEncoder::InitEncode):
        (WebCore::GStreamerVideoEncoder::newSampleCallback):
        (WebCore::GStreamerVideoEncoder::Fragmentize):
        (WebCore::GStreamerVideoEncoder::SetRestrictionCaps):

2018-09-24  Philippe Normand  <pnormand@igalia.com>

        [GStreamer] Utilities cleanups
        https://bugs.webkit.org/show_bug.cgi?id=189699
        <rdar://problem/44634143>

        Reviewed by Xabier Rodriguez-Calvar.

        The GstMappedBuffer now has a move constructor so that it can be easily
        reused in the webaudiosrc element. The now-unused corresponding
        buffer-mapping utilities are removed from the code-base.

        The HTTP source element used to handle a GstBuffer in its private
        structure but this is no longer required since data is now pushed
        in chunks, see bug #182829.

        * platform/audio/gstreamer/WebKitWebAudioSourceGStreamer.cpp:
        (webKitWebAudioSrcLoop):
        * platform/graphics/gstreamer/GStreamerCommon.cpp:
        (WebCore::createGstBuffer): Deleted.
        (WebCore::createGstBufferForData): Deleted.
        (WebCore::getGstBufferDataPointer): Deleted.
        (WebCore::mapGstBuffer): Deleted.
        (WebCore::unmapGstBuffer): Deleted.
        * platform/graphics/gstreamer/GStreamerCommon.h:
        (WebCore::GstMappedBuffer::create): New method returning a
        reference to a newly created GstMappedBuffer instance.
        * platform/graphics/gstreamer/WebKitWebSourceGStreamer.cpp:
        (webKitWebSrcStop): Remove reference to unused GstBuffer.
        (CachedResourceStreamingClient::dataReceived): Ditto.

2018-09-24  Enrique Ocaña González  <eocanha@igalia.com>

        [MSE][GStreamer] Don't update duration when it was not previously NaN
        https://bugs.webkit.org/show_bug.cgi?id=189869

        Reviewed by Xabier Rodriguez-Calvar.

        This is what the spec mandates. The spec doesn't say anything about
        updating duration when it had been previously set, even if the new
        init segment says that the duration is growing.

        This fixes MSE YTTV 2018 69.MediaSourceDurationVP9.

        * platform/graphics/gstreamer/mse/AppendPipeline.cpp:
        (WebCore::AppendPipeline::connectDemuxerSrcPadToAppsink):

2018-09-23  Frederic Wang  <fwang@igalia.com>

        Add missing includes in TextCodecReplacement.cpp
        https://bugs.webkit.org/show_bug.cgi?id=189894

        Reviewed by Darin Adler.

        No new tests, behavior unchanged.

        * platform/text/TextCodecReplacement.cpp: Add missing headers.

2018-09-22  Adrian Perez de Castro  <aperez@igalia.com>

        [ARM] Building FELightingNEON.cpp fails due to missing lightVector member
        https://bugs.webkit.org/show_bug.cgi?id=189890

        Reviewed by Darin Adler.

        No new tests needed.

        * platform/graphics/cpu/arm/filters/FELightingNEON.h:
        (WebCore::FELighting::platformApplyNeon): Adapt to new layout of "struct PaintingData" after r225122.

2018-09-22  Zan Dobersek  <zdobersek@igalia.com>

        [Cairo] Null-check cairo_pattern_t gradient objects
        https://bugs.webkit.org/show_bug.cgi?id=189820

        Reviewed by Alex Christensen.

        Cairo-specific implementation of Gradient::createPlatformGradient() can
        now return a nullptr value when a conic gradient is described by the
        Gradient object. Cairo doesn't have a way to create cairo_pattern_t
        objects for such gradients.

        Null-checks are now done on return values of createPlatformGradient(),
        in order to avoid proceeding to paint a null cairo_pattern_t object.

        * platform/graphics/cairo/GradientCairo.cpp:
        (WebCore::Gradient::fill):
        * platform/graphics/cairo/GraphicsContextImplCairo.cpp:
        (WebCore::GraphicsContextImplCairo::fillRect):

2018-09-21  Ryosuke Niwa  <rniwa@webkit.org>

        Cannot start a drag inside a shadow tree when an inclusive-ancestor of its shadow host is a draggable element
        https://bugs.webkit.org/show_bug.cgi?id=136836

        Reviewed by Wenson Hsieh.

        Fixed the bug by simply generalizing the existing code path existed for video / input type=color.

        Tests: fast/shadow-dom/dragging-element-inside-shadow-tree.html
               fast/shadow-dom/dragging-element-with-shadow-tree.html

        * page/DragController.cpp:
        (WebCore::DragController::startDrag):

2018-09-22  Chris Dumez  <cdumez@apple.com>

        FontDataCache should use Ref<Font> instead of a RefPtr<Font>
        https://bugs.webkit.org/show_bug.cgi?id=189861

        Reviewed by Antti Koivisto.

        * platform/graphics/FontCache.cpp:
        (WebCore::FontCache::fontForPlatformData):
        (WebCore::FontCache::purgeInactiveFontData):

2018-09-21  Justin Michaud  <justin_michaud@apple.com>

        Implement initialValue support for CSS Custom Properties and Values API
        https://bugs.webkit.org/show_bug.cgi?id=189819

        Reviewed by Simon Fraser.

        * css/CSSComputedStyleDeclaration.cpp:
        (WebCore::ComputedStyleExtractor::customPropertyValue):
        * css/CSSCustomPropertyValue.cpp:
        (WebCore::CSSCustomPropertyValue::resolveVariableReferences const):
        * css/CSSCustomPropertyValue.h:
        * css/CSSRegisteredCustomProperty.h:
        * css/CSSVariableData.cpp:
        (WebCore::CSSVariableData::resolveVariableFallback const):
        (WebCore::CSSVariableData::resolveVariableReference const):
        (WebCore::CSSVariableData::resolveVariableReferences const):
        (WebCore::CSSVariableData::resolveTokenRange const):
        * css/CSSVariableData.h:
        * css/DOMCSSRegisterCustomProperty.cpp:
        (WebCore::DOMCSSRegisterCustomProperty::registerProperty):
        * css/DOMCSSRegisterCustomProperty.h:
        * css/DOMCSSRegisterCustomProperty.idl:
        * css/StyleResolver.cpp:
        (WebCore::StyleResolver::resolvedVariableValue):
        (WebCore::StyleResolver::applyCascadedProperties):
        * css/parser/CSSParser.cpp:
        (WebCore::CSSParser::parseValueWithVariableReferences):
        * css/parser/CSSParser.h:
        * dom/Document.h:
        (WebCore::Document::getCSSRegisteredCustomPropertySet const):
        * rendering/style/RenderStyle.cpp:
        (WebCore::RenderStyle::checkVariablesInCustomProperties):
        * rendering/style/RenderStyle.h:

2018-09-21  Dean Jackson  <dino@apple.com>

        Add PointerEvent, plus feature flag, plus Web Platform Tests
        https://bugs.webkit.org/show_bug.cgi?id=189867
        <rdar://problem/44697384>

        Reviewed by Simon Fraser.

        Add the PointerEvent interface.

        Tests: imported/w3c/web-platform-tests/pointerevents/extension/idlharness.window.html
               imported/w3c/web-platform-tests/pointerevents/extension/pointerevent_constructor.html
               imported/w3c/web-platform-tests/pointerevents/extension/pointerevent_touch-action-verification.html
               imported/w3c/web-platform-tests/pointerevents/idlharness.window.html
               imported/w3c/web-platform-tests/pointerevents/pointerevent_constructor.html
               imported/w3c/web-platform-tests/pointerevents/pointerevent_on_event_handlers.html
               imported/w3c/web-platform-tests/pointerevents/pointerevent_touch-action-illegal.html
               imported/w3c/web-platform-tests/pointerevents/pointerevent_touch-action-verification.html

        * DerivedSources.make:
        * Sources.txt:
        * WebCore.xcodeproj/project.pbxproj:
                Add the new files.

        * dom/Event.h:
        (WebCore::Event::isPointerEvent const): New virtual identification method.

        * dom/EventNames.in: Add PointerEvent so the InterfaceType code is generated.

        * dom/PointerEvent.cpp: The PointerEvent interface, as specified by W3C.
        (WebCore::PointerEvent::PointerEvent):
        (WebCore::PointerEvent::eventInterface const):
        * dom/PointerEvent.h: Added.
        * dom/PointerEvent.idl: Added.

        * page/RuntimeEnabledFeatures.h: Clean up the ordering of the features, so there aren't
        confusing blank lines interspersed with #if USE macros.
        (WebCore::RuntimeEnabledFeatures::setPointerEventsEnabled): Add a new flag for Pointer Events.
        (WebCore::RuntimeEnabledFeatures::pointerEventsEnabled const):

2018-09-21  Ryosuke Niwa  <rniwa@webkit.org>

        Custom elements in a reaction queue can lose its JS wrapper and become HTMLUnknownElement
        https://bugs.webkit.org/show_bug.cgi?id=184307

        Reviewed by Keith Miller.

        The bug was caused by the custom elements reaction queue not reporting its content to GC during marking.

        When there is no JS reference to the JS wrappers of those custom element, and if those custom elements
        are disconnected, GC would happily collect those the wrappers. Unfortunately, the same bug exists for
        any asynchronous events and other WebCore code which keeps elements alive for a later use but doesn't
        report them to GC (e.g. during visitChildren).

        This patch, therefore, introduces a generic mechanism to keep these elements' wrappers alive. Namely,
        we introduce GCReachableRef, a new smart pointer type for Node's subclasses, which keeps element as well
        as its wrappers alive. GCReachableRef works by adding its Node to a global hash counted set when it's
        created and making JSNodeOwner::isReachableFromOpaqueRoots return true when the node is in the set.

        Test: fast/custom-elements/custom-elements-reaction-queue-retains-js-wrapper.html

        * Sources.txt:
        * WebCore.xcodeproj/project.pbxproj:
        * bindings/js/JSNodeCustom.cpp:
        (WebCore::isReachableFromDOM):
        * dom/CustomElementReactionQueue.cpp:
        (WebCore::CustomElementReactionQueue::ElementQueue::invokeAll): Don't swap the vector of elements in
        in the queue. Assuming each DOM API has an appropriate CustomElementsReactionStack, we should never
        append a new element to this queue while invoking custom element reactions.
        * dom/CustomElementReactionQueue.h:
        * dom/GCReachableRef.cpp: Added.
        * dom/GCReachableRef.h: Added.
        (WebCore::GCReachableRefMap::contains): Added.
        (WebCore::GCReachableRefMap::add): Added.
        (WebCore::GCReachableRefMap::remove): Added.
        (WebCore::GCReachableRef::GCReachableRef): Added. We need isNull() check since WTFMove may have been
        called on the source GCReachableRef.
        (WebCore::GCReachableRef::~GCReachableRef): Ditto.
        (WebCore::GCReachableRef::operator-> const): Added.
        (WebCore::GCReachableRef::get const): Added.
        (WebCore::GCReachableRef::operator T& const): Added.
        (WebCore::GCReachableRef::operator! const): Added.
        (WebCore::GCReachableRef::isNull const): Added. Returns true if WTFMove had been called on Ref.

2018-09-21  Alex Christensen  <achristensen@webkit.org>

        Use a Variant for FormDataElement
        https://bugs.webkit.org/show_bug.cgi?id=189777

        Reviewed by Chris Dumez.

        * platform/network/FormData.cpp:
        (WebCore::FormData::FormData):
        (WebCore::FormDataElement::lengthInBytes const):
        (WebCore::FormDataElement::isolatedCopy const):
        (WebCore::FormData::appendData):
        (WebCore::FormData::flatten const):
        (WebCore::FormData::resolveBlobReferences):
        (WebCore::FormData::generateFiles):
        (WebCore::FormData::hasGeneratedFiles const):
        (WebCore::FormData::hasOwnedGeneratedFiles const):
        (WebCore::FormData::removeGeneratedFilesIfNeeded):
        (WebCore::FormData::asSharedBuffer const):
        (WebCore::FormData::asBlobURL const):
        (WebCore::FormData::expandDataStore): Deleted.
        * platform/network/FormData.h:
        (WebCore::FormDataElement::FormDataElement):
        (WebCore::FormDataElement::encode const):
        (WebCore::FormDataElement::decode):
        (WebCore::FormDataElement::EncodedFileData::isolatedCopy const):
        (WebCore::FormDataElement::EncodedFileData::operator== const):
        (WebCore::FormDataElement::EncodedFileData::encode const):
        (WebCore::FormDataElement::EncodedFileData::decode):
        (WebCore::FormDataElement::EncodedBlobData::operator== const):
        (WebCore::FormDataElement::EncodedBlobData::encode const):
        (WebCore::FormDataElement::EncodedBlobData::decode):
        (WebCore::FormDataElement::operator== const):
        (WebCore::FormDataElement::operator!= const):
        * platform/network/cf/FormDataStreamCFNet.cpp:
        (WebCore::advanceCurrentStream):
        (WebCore::createHTTPBodyCFReadStream):
        (WebCore::setHTTPBody):
        * platform/network/curl/CurlFormDataStream.cpp:
        (WebCore::CurlFormDataStream::computeContentLength):
        (WebCore::CurlFormDataStream::read):
        (WebCore::CurlFormDataStream::readFromFile):
        (WebCore::CurlFormDataStream::readFromData):
        * platform/network/curl/CurlFormDataStream.h:

2018-09-20  Simon Fraser  <simon.fraser@apple.com>

        Simplify the logic around has*ScrollbarWithAutoBehavior
        https://bugs.webkit.org/show_bug.cgi?id=189813

        Reviewed by Zalan Bujtas.

        The boolean logic in scrollsOverflowX() and hasHorizontalScrollbarWithAutoBehavior() (and the vertical
        equivalents) reduces simply to hasOverflowClip() && (style().overflowX() == Overflow::Scroll || style().overflowX() == Overflow::Auto);
        
        Similarly, RenderBox::intrinsicScrollbarLogicalWidth() just needs the part of the logic
        that asks whether the theme uses overlay scrollbars which are not customized (and thus
        turned into non-overlay scrollbars).

        * rendering/RenderBox.cpp:
        (WebCore::RenderBox::intrinsicScrollbarLogicalWidth const):
        (WebCore::RenderBox::canUseOverlayScrollbars const):
        (WebCore::RenderBox::hasVerticalScrollbarWithAutoBehavior const):
        (WebCore::RenderBox::hasHorizontalScrollbarWithAutoBehavior const):
        * rendering/RenderBox.h:
        (WebCore::RenderBox::scrollsOverflowX const):
        (WebCore::RenderBox::scrollsOverflowY const):
        * rendering/RenderLayer.cpp:
        (WebCore::RenderLayer::updateScrollbarsAfterLayout):

2018-09-21  Michael Catanzaro  <mcatanzaro@igalia.com>

        Unreviewed, rolling out r236255.

        Many WebAudio crashes

        Reverted changeset:

        "[GStreamer] Utilities cleanups"
        https://bugs.webkit.org/show_bug.cgi?id=189699
        https://trac.webkit.org/changeset/236255

2018-09-21  Jer Noble  <jer.noble@apple.com>

        Move AVVideoPerformanceMetrics into AVFoundationSPI.h
        https://bugs.webkit.org/show_bug.cgi?id=189842

        Reviewed by Jon Lee.

        * platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.mm:

2018-09-21  Chris Dumez  <cdumez@apple.com>

        WebSQL: User cannot grant quota increase if the JS provides an expected usage value that is too low
        https://bugs.webkit.org/show_bug.cgi?id=189801
        <rdar://problem/43592498>

        Reviewed by Youenn Fablet.

        User was unable to grant a quota increase for WebSQL if the JS provided an expected usage value that
        is too low. This is because WebKit was passing this provided expectedUsage value to the client for
        the purpose of quota increase, even when this expectedUsage value does not make any sense (i.e. it
        is lower than the current database size). As a result, the client would grant a quota that is equal
        to the previous quota and the JS would not be able to insert any data.

        In order to address the issue, when the current quota is exceeded and Database::didExceedQuota()
        is called, we now make sure that the expectedUsage value is greater than the current quota. If it
        is not, we provide `current quota + 5MB` as expected usage to the client. This way, the client will
        grant a quota that is actually increased (provided that the user accepts).

        Test: storage/websql/transaction-database-expand-quota.html

        * Modules/webdatabase/Database.cpp:
        (WebCore::Database::setEstimatedSize):
        (WebCore::Database::didExceedQuota):
        * Modules/webdatabase/Database.h:

2018-09-21  Youenn Fablet  <youenn@apple.com>

        Use biplanar CVPixelBuffer for black frames sent to libwebrtc
        https://bugs.webkit.org/show_bug.cgi?id=189837

        Reviewed by Eric Carlson.

        Covered by webrtc/video-mute.html.

        Add support to call CVPixelBufferGetBytesPerRowOfPlane.
        Make createBlackPixelBuffer use a biplanar CVPixelBuffer as this is better supported in libwebrtc.
        It is also what is being used in iOS for capture.

        * platform/cocoa/CoreVideoSoftLink.cpp:
        * platform/cocoa/CoreVideoSoftLink.h:
        * platform/mediastream/mac/RealtimeIncomingVideoSourceCocoa.mm:
        (WebCore::createBlackPixelBuffer):

2018-09-21  Youenn Fablet  <youenn@apple.com>

        Add RTCCodecStats support
        https://bugs.webkit.org/show_bug.cgi?id=189792
        <rdar://problem/32370668>

        Reviewed by Eric Carlson.

        Covered by updated and rebased tests.

        * Modules/mediastream/RTCStatsReport.h:
        Removed fields that are already defined in the base class.
        (WebCore::RTCStatsReport::CodecStats::CodecStats):
        Add support for RTCCodecStats.
        * Modules/mediastream/RTCStatsReport.idl:
        * Modules/mediastream/libwebrtc/LibWebRTCMediaEndpoint.cpp:
        (WebCore::LibWebRTCMediaEndpoint::OnStatsDelivered):
        * Modules/mediastream/libwebrtc/LibWebRTCStatsCollector.cpp:
        (WebCore::fillRTCRTPStreamStats):
        (WebCore::fillRTCCodecStats):
        (WebCore::LibWebRTCStatsCollector::OnStatsDelivered):
        Add routines to fill RTCCodecStats from libwebrtc stats.

2018-09-20  Simon Fraser  <simon.fraser@apple.com>

        Make "overflow: overlay" a synonym for "overflow: auto"
        https://bugs.webkit.org/show_bug.cgi?id=189811

        Reviewed by Zalan Bujtas.
        
        The "overlay" value for overflow was added for an internal Safari feature, and only has
        an effect (allow the scrollbar to overlap the content) with legacy scrollbars on macOS.
        It's little used on the web.

        To simplify code in rendering, just make "overflow: overlay" behave like "overflow: auto".
        It's still parsed, but turns into an "auto" value internally, and will be returned from getComputedStyle
        as "auto".

        Test: fast/css/getComputedStyle/getComputedStyle-overflow.html

        * css/CSSPrimitiveValueMappings.h:
        (WebCore::CSSPrimitiveValue::CSSPrimitiveValue):
        (WebCore::CSSPrimitiveValue::operator Overflow const):
        * css/CSSProperties.json:
        * css/CSSValueKeywords.in:
        * css/StyleResolver.cpp:
        (WebCore::isScrollableOverflow):
        * css/parser/CSSParserFastPaths.cpp:
        (WebCore::CSSParserFastPaths::isValidKeywordPropertyAndValue):
        * page/ios/FrameIOS.mm:
        (WebCore::Frame::nodeRespondingToScrollWheelEvents):
        * rendering/RenderBox.cpp:
        (WebCore::RenderBox::hasVerticalScrollbarWithAutoBehavior const):
        (WebCore::RenderBox::hasHorizontalScrollbarWithAutoBehavior const):
        * rendering/RenderLayer.cpp:
        (WebCore::styleDefinesAutomaticScrollbar):
        * rendering/RenderLayerCompositor.cpp:
        (WebCore::isScrollableOverflow):
        * rendering/style/RenderStyleConstants.h:

2018-09-21  Youenn Fablet  <youenn@apple.com>

        Centralize which CVPixelBuffer format is being used
        https://bugs.webkit.org/show_bug.cgi?id=189772

        Reviewed by Eric Carlson.

        Get the format type from a single point.
        This changes the video capture and mock realtime video sources on Mac to use a biplanar format. 
        No observable change of behavior.

        * WebCore.xcodeproj/project.pbxproj:
        * platform/mediastream/mac/AVVideoCaptureSource.mm:
        (WebCore::AVVideoCaptureSource::setSizeAndFrameRateWithPreset):
        (WebCore::AVVideoCaptureSource::setupCaptureSession):
        (WebCore::AVVideoCaptureSource::captureOutputDidOutputSampleBufferFromConnection):
        * platform/mediastream/mac/MockRealtimeVideoSourceMac.mm:
        (WebCore::MockRealtimeVideoSourceMac::updateSampleBuffer):
        (WebCore::MockRealtimeVideoSourceMac::setSizeAndFrameRateWithPreset):
        * platform/mediastream/mac/RealtimeIncomingVideoSourceCocoa.mm:
        (WebCore::createBlackPixelBuffer):
        * platform/mediastream/mac/RealtimeOutgoingVideoSourceCocoa.cpp:
        (WebCore::RealtimeOutgoingVideoSourceCocoa::sampleBufferUpdated):
        * platform/mediastream/mac/RealtimeVideoUtilities.h: Added.

2018-09-21  Antoine Quint  <graouts@apple.com>

        REGRESSION (r235962-r235963): Layout Test animations/suspend-resume-animation-events.html is a flaky failure
        https://bugs.webkit.org/show_bug.cgi?id=189607
        <rdar://problem/44652315>

        Reviewed by Dean Jackson.

        There is no reason we shouldn't return the document timeline's time when suspended as otherwise animations may
        report an unresolved current time when suspended which would wreak havoc when invalidating what DOM events to
        dispatch for CSS Animations and Transitions. We also shouldn't be invalidation DOM events when suspended.

        * animation/DocumentTimeline.cpp:
        (WebCore::DocumentTimeline::currentTime):
        (WebCore::DocumentTimeline::DocumentTimeline::performInvalidationTask):

2018-09-21  Mike Gorse  <mgorse@suse.com>

        Build tools should work when the /usr/bin/python is python3
        https://bugs.webkit.org/show_bug.cgi?id=156674

        Reviewed by Michael Catanzaro.

        No new tests (no behavior change).

        * platform/network/create-http-header-name-table: remove xreadlines.

2018-09-21  Jer Noble  <jer.noble@apple.com>

        [EME] Introduce the concept of CDMInstanceSession.
        https://bugs.webkit.org/show_bug.cgi?id=189725

        Reviewed by Eric Carlson.

        Currently, the same CDMInstance owned by a MediaKeys object is passed to every MediaKeySession created by that
        MediaKeys, and since the CDMInstance has only a single CDMInstanceClient, subsequent MediaKeySessions prevent
        previous ones from getting updates.

        Add a new virtual interface, CDMInstanceSession, to be passed to MediaKeySession upon creation. Refactor
        CDMInstanceClearKey and CDMInstanceFairPlayStreamingAVFObjC to adopt this new interface.

        Drive-by fixes: Made a number of virtual overrides in final classes final themselves.

        * Modules/encryptedmedia/MediaKeySession.cpp:
        (WebCore::MediaKeySession::create):
        (WebCore::MediaKeySession::MediaKeySession):
        (WebCore::MediaKeySession::generateRequest):
        (WebCore::MediaKeySession::load):
        (WebCore::MediaKeySession::update):
        (WebCore::MediaKeySession::remove):
        (WebCore::MediaKeySession::updateKeyStatuses):
        * Modules/encryptedmedia/MediaKeySession.h:
        * Modules/encryptedmedia/MediaKeys.cpp:
        (WebCore::MediaKeys::createSession):
        * WebCore.xcodeproj/project.pbxproj:
        * platform/encryptedmedia/CDMInstance.h:
        (WebCore::CDMInstance::setHDCPStatus):
        (WebCore::CDMInstance::setClient): Deleted.
        (WebCore::CDMInstance::clearClient): Deleted.
        * platform/encryptedmedia/CDMInstanceSession.h: Copied from Source/WebCore/platform/encryptedmedia/CDMInstance.h.
        (WebCore::CDMInstanceSession::setClient):
        (WebCore::CDMInstanceSession::clearClient):
        * platform/encryptedmedia/clearkey/CDMClearKey.cpp:
        (WebCore::parseLicenseFormat):
        (WebCore::CDMInstanceClearKey::keySystem const):
        (WebCore::CDMInstanceClearKey::createSession):
        (WebCore::CDMInstanceSessionClearKey::requestLicense):
        (WebCore::CDMInstanceSessionClearKey::keys const):
        (WebCore::CDMInstanceSessionClearKey::updateLicense):
        (WebCore::CDMInstanceSessionClearKey::loadSession):
        (WebCore::CDMInstanceSessionClearKey::closeSession):
        (WebCore::CDMInstanceSessionClearKey::removeSessionData):
        (WebCore::CDMInstanceSessionClearKey::storeRecordOfKeyUsage):
        (WebCore::CDMInstanceClearKey::requestLicense): Deleted.
        (WebCore::CDMInstanceClearKey::keys const): Deleted.
        (WebCore::CDMInstanceClearKey::updateLicense): Deleted.
        (WebCore::CDMInstanceClearKey::loadSession): Deleted.
        (WebCore::CDMInstanceClearKey::closeSession): Deleted.
        (WebCore::CDMInstanceClearKey::removeSessionData): Deleted.
        (WebCore::CDMInstanceClearKey::storeRecordOfKeyUsage): Deleted.
        * platform/encryptedmedia/clearkey/CDMClearKey.h:
        * platform/graphics/avfoundation/objc/CDMInstanceFairPlayStreamingAVFObjC.h:
        * platform/graphics/avfoundation/objc/CDMInstanceFairPlayStreamingAVFObjC.mm:
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::createSession):
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::processContentKeyRequestForSession):
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::processNextContentKeyRequest):
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::keySystem const):
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::didProvideRequest):
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::didProvideRenewingRequest):
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::didProvidePersistableRequest):
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::didFailToProvideRequest):
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::requestDidSucceed):
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::shouldRetryRequestForReason):
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::sessionIdentifierChanged):
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::outputObscuredDueToInsufficientExternalProtectionChanged):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::CDMInstanceSessionFairPlayStreamingAVFObjC):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::keyIDs):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::requestLicense):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::updateLicense):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::loadSession):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::closeSession):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::removeSessionData):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::storeRecordOfKeyUsage):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::setClient):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::clearClient):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::didProvideRequest):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::didProvideRenewingRequest):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::didProvidePersistableRequest):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::didFailToProvideRequest):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::requestDidSucceed):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::shouldRetryRequestForReason):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::sessionIdentifierChanged):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::outputObscuredDueToInsufficientExternalProtectionChanged):
        (WebCore::CDMInstanceSessionFairPlayStreamingAVFObjC::isLicenseTypeSupported const):
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::isLicenseTypeSupported const): Deleted.
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::keyIDs): Deleted.
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::requestLicense): Deleted.
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::updateLicense): Deleted.
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::loadSession): Deleted.
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::closeSession): Deleted.
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::removeSessionData): Deleted.
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::storeRecordOfKeyUsage): Deleted.
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::setClient): Deleted.
        (WebCore::CDMInstanceFairPlayStreamingAVFObjC::clearClient): Deleted.
        * testing/MockCDMFactory.cpp:
        (WebCore::MockCDMInstance::keySystem const):
        (WebCore::MockCDMInstance::createSession):
        (WebCore::MockCDMInstanceSession::MockCDMInstanceSession):
        (WebCore::MockCDMInstanceSession::requestLicense):
        (WebCore::MockCDMInstanceSession::updateLicense):
        (WebCore::MockCDMInstanceSession::loadSession):
        (WebCore::MockCDMInstanceSession::closeSession):
        (WebCore::MockCDMInstanceSession::removeSessionData):
        (WebCore::MockCDMInstanceSession::storeRecordOfKeyUsage):
        (WebCore::MockCDMInstance::requestLicense): Deleted.
        (WebCore::MockCDMInstance::updateLicense): Deleted.
        (WebCore::MockCDMInstance::loadSession): Deleted.
        (WebCore::MockCDMInstance::closeSession): Deleted.
        (WebCore::MockCDMInstance::removeSessionData): Deleted.
        (WebCore::MockCDMInstance::storeRecordOfKeyUsage): Deleted.
        * testing/MockCDMFactory.h:
        (WebCore::MockCDMInstance::factory const):
        (WebCore::MockCDMInstance::distinctiveIdentifiersAllowed const):
        (WebCore::MockCDMInstance::persistentStateAllowed const):

2018-09-21  Alicia Boya García  <aboya@igalia.com>

        [MSE] Fix comparsion with uninitialized greatestDecodeDuration
        https://bugs.webkit.org/show_bug.cgi?id=189805

        Reviewed by Michael Catanzaro.

        This bug was causing greatestDecodeDuration to never be initialized,
        which in turned caused unintended frame erase as distant appends where
        not being recognized as distinct coded frame groups.

        A test reproducing the sequence of appends that caused unintended
        frame deletion has also been added (media-source-append-out-of-order.html).

        * Modules/mediasource/SourceBuffer.cpp:
        (WebCore::SourceBuffer::sourceBufferPrivateDidReceiveSample):

2018-09-21  Antoine Quint  <graouts@apple.com>

        [Web Animations] Accelerated animations don't get suspended
        https://bugs.webkit.org/show_bug.cgi?id=189783
        <rdar://problem/44652315>

        Unreviewed, correct a merge error in the previous commit.

        * animation/DocumentTimeline.cpp:
        (WebCore::DocumentTimeline::performInvalidationTask):

2018-09-21  Antoine Quint  <graouts@apple.com>

        [Web Animations] Accelerated animations don't get suspended
        https://bugs.webkit.org/show_bug.cgi?id=189783
        <rdar://problem/43033568>

        Reviewed by Dean Jackson.

        Test: webanimations/accelerated-animation-suspension.html

        We used to set the flag that marked the timeline as suspended prior to notifying animations that they need to be suspended.
        However, since the timeline was marked as suspended, querying the running state of the animations would indicate that the
        animations weren't running since a suspended timeline would identify its animations as not running. As such we would fail
        to pause the accelerated animations because they were already not marked as running. We now set the suspended flag on the
        timeline _after_ suspending its animations.

        We also fix a bug in the new internals.acceleratedAnimationsForElement() test function so that we read from the actual
        CA animations and not from a stale list of animations which would not indicate the correct animation speeds.

        * animation/DocumentTimeline.cpp:
        (WebCore::DocumentTimeline::suspendAnimations):
        * platform/graphics/ca/GraphicsLayerCA.cpp:
        (WebCore::GraphicsLayerCA::acceleratedAnimationsForTesting):

2018-09-21  Zan Dobersek  <zdobersek@igalia.com>

        TransformationMatrix::toColumnMajorFloatArray() should return a std::array<> object
        https://bugs.webkit.org/show_bug.cgi?id=189823

        Reviewed by Michael Catanzaro.

        Alias the TransformationMatrix::FloatMatrix4 type to
        std::array<float, 16>. Instead of filling out the array object that's
        passed in through a reference parameter, return the std::array<>
        object from the function.

        * Modules/webvr/VRFrameData.cpp:
        (WebCore::matrixToArray):
        * Modules/webvr/VRStageParameters.cpp:
        (WebCore::VRStageParameters::sittingToStandingTransform const):
        * platform/graphics/texmap/TextureMapperShaderProgram.cpp:
        (WebCore::TextureMapperShaderProgram::setMatrix):
        * platform/graphics/transforms/TransformationMatrix.cpp:
        (WebCore::TransformationMatrix::toColumnMajorFloatArray const):
        * platform/graphics/transforms/TransformationMatrix.h:

2018-09-21  Zan Dobersek  <zdobersek@igalia.com>

        FloatQuad point getters should return const references
        https://bugs.webkit.org/show_bug.cgi?id=189821

        Reviewed by Yusuke Suzuki.

        Be pedantic and have the FloatQuad point getters return const references
        to the FloatPoint member variables, instead of technically creating
        copies of them (though much of this copying is eliminated when the
        getters are inlined).

        * platform/graphics/FloatQuad.h:
        (WebCore::FloatQuad::p1 const):
        (WebCore::FloatQuad::p2 const):
        (WebCore::FloatQuad::p3 const):
        (WebCore::FloatQuad::p4 const):

2018-09-20  Antoine Quint  <graouts@apple.com>

        [Web Animations] DocumentTimeline::updateAnimations() is called endlessly
        https://bugs.webkit.org/show_bug.cgi?id=189784
        <rdar://problem/41705679>

        Reviewed by Dean Jackson.

        Test: webanimations/accelerated-animation-interruption-display-none.html

        We have code that keeps queueing pending accelerated actions for an animation that does not have a renderer until it has one
        so that we can deal with situations where animations are ready to commited before its composited renderer is available. This
        code ended up running continuously when an element with an accelerated animation had its renderer removed without the animation
        being removed itself, such as setting "display: none" on an element with an acceelerated CSS Animation targeting it.

        We fix this by queueing up a "Stop" accelerated action when updating the accelerated state if there is no renderer for the current
        animation target. Then, we no longer re-queue pending accelerated actions if the last queued operation is "Stop". This ensures that
        we no longer queue actions endlessly when there is no longer a visible animation.

        To test this, we add a new internals.numberOfAnimationTimelineInvalidations() method that indicates the number of times the current
        document's animation timeline was invalidated.

        * animation/DocumentTimeline.cpp:
        (WebCore::DocumentTimeline::updateAnimations):
        (WebCore::DocumentTimeline::numberOfAnimationTimelineInvalidationsForTesting const):
        * animation/DocumentTimeline.h:
        * animation/KeyframeEffectReadOnly.cpp:
        (WebCore::KeyframeEffectReadOnly::updateAcceleratedAnimationState): If the animation target does not have a renderer and it's still
        marked as running, enqueue a "Stop" accelerated action.
        (WebCore::KeyframeEffectReadOnly::addPendingAcceleratedAction): If we enqueue a "Stop" accelerated action, remove any other queued
        action so that we only process the "Stop" action, which would have superseded all previously queued actions anyway.
        (WebCore::KeyframeEffectReadOnly::applyPendingAcceleratedActions): Only re-queue pending accelerated actions when a composited renderer
        is not yet available if we don't have a "Stop" action queued.
        * testing/Internals.cpp:
        (WebCore::Internals::numberOfAnimationTimelineInvalidations const):
        * testing/Internals.h:
        * testing/Internals.idl:

2018-09-21  Yacine Bandou  <yacine.bandou@softathome.com>

        [EME] Fix typo in WebM sanitization variable
        https://bugs.webkit.org/show_bug.cgi?id=189789

        Reviewed by Xabier Rodriguez-Calvar.

        This commit corrects a typo in the name of a local variable, sanitizedBuffer
        instead of sanitazedBuffer.

        * Modules/encryptedmedia/InitDataRegistry.cpp:
        (WebCore::extractKeyIDsWebM):

2018-09-20  Dean Jackson  <dino@apple.com>

        Restrict the total combined size of backdrop filters
        https://bugs.webkit.org/show_bug.cgi?id=189812
        <rdar://problem/44532782>

        Reviewed by Simon Fraser.

        If the total area of all backdrop filters on the page gets
        too large, the universe collapses in on itself and we enter
        the Quantum Realm (i.e. crash horribly).

        Put a hard limit on the total coverage, and ignore any backdrop
        filters after the limit. This might break some content, but
        such content is likely not doing things in the most optimal manner.
        There isn't any reason to have a backdrop larger than the size of
        the screen, because you'd be better off applying a foreground
        filter to the main content and showing something above it.

        Tests: css3/filters/backdrop/resource-use-add-more-layers.html
               css3/filters/backdrop/resource-use-excessive.html
               css3/filters/backdrop/resource-use-ok.html
               css3/filters/backdrop/resource-use-remove-some-layers.html

        * platform/graphics/ca/GraphicsLayerCA.cpp: Pick a fairly small maximum size. We
        can consider increasing this if necessary, and as devices with less RAM are
        upgraded.
        (WebCore::GraphicsLayerCA::recursiveCommitChanges): Gather the accumulated size
        of backdrop filters into the commit state as we are recursing through the tree.
        (WebCore::GraphicsLayerCA::commitLayerChangesBeforeSublayers): Force any layer
        with backdrop filters, or any that is removing backdrop filters, into an update.
        (WebCore::GraphicsLayerCA::updateBackdropFilters): Update the logic to first
        check if this backdrop layer causes us to exceed the total allowed size, and if
        it does, forbid it from getting the GraphicsLayer that composits the backdrop.

        * platform/graphics/ca/GraphicsLayerCA.h: Remove const from some parameters so
        that we can use the CommitState to hold the accumulated size.

2018-09-20  Benjamin Poulain  <benjamin@webkit.org>

        Adopt safe-area-insets on ImageDocument
        https://bugs.webkit.org/show_bug.cgi?id=189774

        Reviewed by Tim Horton.
        rdar://problem/44624432

        By having the safe-area insets on the image, we ensure that they only
        grow the document if there is not enough space. This also ensures the image
        does not have parts under UI elements.

        * html/ImageDocument.cpp:
        (WebCore::ImageDocument::createDocumentStructure):
        (WebCore::ImageDocument::imageUpdated):

2018-09-20  Zalan Bujtas  <zalan@apple.com>

        Release assert under RenderView::pageOrViewLogicalHeight
        https://bugs.webkit.org/show_bug.cgi?id=189798
        <rdar://problem/43659749>

        Reviewed by Simon Fraser.

        Only the mainframe's render view is sized to the page while printing.
        Use the matching check (see RenderView::layout) when accessing m_pageLogicalSize.

        Test: printing/crash-while-formatting-subframe-for-printing.html

        * rendering/RenderView.cpp:
        (WebCore::RenderView::pageOrViewLogicalHeight const):

2018-09-20  Sihui Liu  <sihui_liu@apple.com>

        REGRESSION(r196265): WKWebView fires mouseover, mouseenter, and mouseleave events even when it's in a background window
        https://bugs.webkit.org/show_bug.cgi?id=187545
        <rdar://problem/42401575>

        Reviewed by Ryosuke Niwa.

        When the window is not active, we should only update the scrollbar for mouse events. GTK
        apps have different expectation on this behavior.

        Test: fast/events/inactive-window-no-mouse-event.html

        * page/EventHandler.cpp:
        (WebCore::EventHandler::handleMouseMoveEvent):
        (WebCore::EventHandler::shouldSendMouseEventsToInactiveWindows const):
        * page/EventHandler.h:

2018-09-20  Alex Christensen  <achristensen@webkit.org>

        Unreviewed, rolling out r235976.

        Broke ARM

        Reverted changeset:

        "Use a Variant instead of a union in CSSSelector"
        https://bugs.webkit.org/show_bug.cgi?id=188559
        https://trac.webkit.org/changeset/235976

2018-09-20  Oriol Brufau  <obrufau@igalia.com>

        Fix 'border' serialization with both common and uncommon values
        https://bugs.webkit.org/show_bug.cgi?id=189597

        Reviewed by Simon Fraser.

        Remove CommonValueMode enum and make borderPropertyValue always return null
        when there are uncommon values (the previous ReturnNullOnUncommonValues mode).

        Test: fast/css/getPropertyValue-border.html
        Test: fast/dom/css-shorthand-common-value.html

        * css/StyleProperties.cpp:
        (WebCore::StyleProperties::getPropertyValue const):
        (WebCore::StyleProperties::borderPropertyValue const):
        (WebCore::StyleProperties::asText const):
        * css/StyleProperties.h:

2018-09-20  Justin Michaud  <justin_michaud@apple.com>

        Implement CSS Custom Properties and Values Skeleton
        https://bugs.webkit.org/show_bug.cgi?id=189694

        Reviewed by Simon Fraser.

        Add feature flag, CSS.registerProperty binding and registered property set in Document.h
        for the css custom properties and values api.

        Test: css-properties-values-api/registerProperty.html

        * CMakeLists.txt:
        * DerivedSources.make:
        * Sources.txt:
        * WebCore.xcodeproj/project.pbxproj:
        * css/CSSRegisteredCustomProperty.h: Added.
        * css/DOMCSSCustomPropertyDescriptor.h: Added.
        * css/DOMCSSCustomPropertyDescriptor.idl: Added.
        * css/DOMCSSNamespace.h:
        * css/DOMCSSRegisterCustomProperty.cpp: Added.
        (WebCore::DOMCSSRegisterCustomProperty::registerProperty):
        (WebCore::DOMCSSRegisterCustomProperty::from):
        (WebCore::DOMCSSRegisterCustomProperty::supplementName):
        * css/DOMCSSRegisterCustomProperty.h: Added.
        * css/DOMCSSRegisterCustomProperty.idl: Added.
        * css/parser/CSSParserContext.cpp:
        * dom/Document.cpp:
        (WebCore::Document::registerCSSProperty):
        * dom/Document.h:
        * features.json:
        * page/RuntimeEnabledFeatures.h:
        (WebCore::RuntimeEnabledFeatures::setCSSCustomPropertiesAndValuesEnabled):
        (WebCore::RuntimeEnabledFeatures::cssCustomPropertiesAndValuesEnabled const):

2018-09-20  Justin Michaud  <justin_michaud@apple.com>

        JS bindings generator should support EnabledAtRuntime for static methods
        https://bugs.webkit.org/show_bug.cgi?id=189729

        Reviewed by Chris Dumez.

        Add support for EnabledAtRuntime to static methods in the JS bindings
        code generator.

        * bindings/scripts/CodeGeneratorJS.pm:
        (GenerateRuntimeEnableConditionalStringForExposed):
        (GenerateRuntimeEnableConditionalString):
        (GetRuntimeEnabledStaticProperties):
        (GenerateConstructorHelperMethods):
        * bindings/scripts/test/JS/JSTestGlobalObject.cpp:
        (WebCore::JSTestGlobalObjectConstructor::initializeProperties):
        (WebCore::JSTestGlobalObject::finishCreation):
        (WebCore::jsTestGlobalObjectConstructorFunctionEnabledAtRuntimeOperationStaticBody):
        (WebCore::jsTestGlobalObjectConstructorFunctionEnabledAtRuntimeOperationStatic):
        * bindings/scripts/test/JS/JSTestObj.cpp:
        (WebCore::JSTestObjConstructor::initializeProperties):
        (WebCore::jsTestObjConstructorEnabledAtRuntimeAttributeStaticGetter):
        (WebCore::jsTestObjConstructorEnabledAtRuntimeAttributeStatic):
        (WebCore::setJSTestObjConstructorEnabledAtRuntimeAttributeStaticSetter):
        (WebCore::setJSTestObjConstructorEnabledAtRuntimeAttributeStatic):
        (WebCore::jsTestObjConstructorFunctionEnabledAtRuntimeOperationStaticBody):
        (WebCore::jsTestObjConstructorFunctionEnabledAtRuntimeOperationStatic):
        * bindings/scripts/test/TestGlobalObject.idl:
        * bindings/scripts/test/TestObj.idl:

2018-09-20  Per Arne Vollan  <pvollan@apple.com>

        [WebVTT] Update the parser according to the new region syntax.
        https://bugs.webkit.org/show_bug.cgi?id=189767

        Reviewed by Eric Carlson.

        The majority of the code added in this patch is adopted from the Chromium project, which has added
        support for the new region syntax. The complete parser specification can be found at
        https://w3c.github.io/webvtt/#file-parsing. One small difference in behavior is that the new parser
        will not add regions with empty id.

        No new tests, covered by existing tests.

        * html/track/WebVTTParser.cpp:
        (WebCore::WebVTTParser::getNewRegions):
        (WebCore::WebVTTParser::parse):
        (WebCore::WebVTTParser::collectRegionSettings):
        (WebCore::WebVTTParser::collectWebVTTBlock):
        (WebCore::WebVTTParser::checkAndRecoverCue):
        (WebCore::WebVTTParser::checkAndCreateRegion):
        (WebCore::WebVTTParser::checkAndStoreRegion):
        (WebCore::WebVTTParser::collectMetadataHeader): Deleted.
        (WebCore::WebVTTParser::createNewRegion): Deleted.
        * html/track/WebVTTParser.h:

2018-09-20  Alicia Boya García  <aboya@igalia.com>

        [GStreamer][MSE] Add a default sample duration
        https://bugs.webkit.org/show_bug.cgi?id=189788

        Some WebM files don't provide sample durations, so we need to provide
        a safe default in order for them to be playable.

        Reviewed by Michael Catanzaro.

        * platform/graphics/gstreamer/MediaSampleGStreamer.cpp:
        (WebCore::MediaSampleGStreamer::MediaSampleGStreamer):

2018-09-20  Alicia Boya García  <aboya@igalia.com>

        [MSE] Use some tolerance when deciding whether a frame should be appended to the decode queue
        https://bugs.webkit.org/show_bug.cgi?id=189782

        Reviewed by Xabier Rodriguez-Calvar.

        Ideally, container formats should use exact timestamps and frames
        should not overlap. Unfortunately, there are lots of files out there
        where this is not always the case.

        This is particularly a problem in WebM, where timestamps are expressed
        in a power of 10 timescale, which forces some rounding.

        This patch makes SourceBuffer allow frames with a small overlaps
        (<=1ms) as those usually found in WebM. 1 ms is chosen because it's
        the default time scale of WebM files.

        * Modules/mediasource/SourceBuffer.cpp:
        (WebCore::SourceBuffer::sourceBufferPrivateDidReceiveSample):

2018-09-20  Yacine Bandou  <yacine.bandou@softathome.com>

        [EME] Add WebM sanitization
        https://bugs.webkit.org/show_bug.cgi?id=189740

        Reviewed by Xabier Rodriguez-Calvar.

        This patch adds support for sanitizing the WebM initialization data,
        ensures there are no bogus values.
        See https://www.w3.org/TR/encrypted-media/#dom-mediakeysession-generaterequest.

        Tests: imported/w3c/web-platform-tests/encrypted-media/clearkey-generate-request-disallowed-input.https.html

        * Modules/encryptedmedia/InitDataRegistry.cpp:
        (WebCore::sanitizeWebM): Added implementation, check if the initialization data doesn't empty and its size
        should be less than 64KB, return the buffer copy if it is ok, otherwise a nullptr.
        (WebCore::extractKeyIDsWebM): Added implementation.

2018-09-20  Philippe Normand  <pnormand@igalia.com>

        [GStreamer] Utilities cleanups
        https://bugs.webkit.org/show_bug.cgi?id=189699

        Reviewed by Xabier Rodriguez-Calvar.

        The GstMappedBuffer now has a move constructor so that it can be easily
        reused in the webaudiosrc element. The now-unused corresponding
        buffer-mapping utilities are removed from the code-base.

        The HTTP source element used to handle a GstBuffer in its private
        structure but this is no longer required since data is now pushed
        in chunks, see bug #182829.

        * platform/audio/gstreamer/WebKitWebAudioSourceGStreamer.cpp:
        (webKitWebAudioSrcLoop):
        * platform/graphics/gstreamer/GStreamerCommon.cpp:
        (WebCore::createGstBuffer): Deleted.
        (WebCore::createGstBufferForData): Deleted.
        (WebCore::getGstBufferDataPointer): Deleted.
        (WebCore::mapGstBuffer): Deleted.
        (WebCore::unmapGstBuffer): Deleted.
        * platform/graphics/gstreamer/GStreamerCommon.h:
        (WebCore::GstMappedBuffer::create): New method returning a
        reference to a newly created GstMappedBuffer instance.
        * platform/graphics/gstreamer/WebKitWebSourceGStreamer.cpp:
        (webKitWebSrcStop): Remove reference to unused GstBuffer.
        (CachedResourceStreamingClient::dataReceived): Ditto.

2018-09-20  Simon Fraser  <simon.fraser@apple.com>

        Fix crash under FontCache::purgeInactiveFontData() when a memory warning fires
        https://bugs.webkit.org/show_bug.cgi?id=189722
        rdar://problem/44182860

        Reviewed by Myles C. Maxfield.
        
        Hashing of FontPlatformData for cachedFonts() is somewhat broken because CFEqual() on CTFont
        can return false when the fonts are actually the same, and have the same CFHash(). This 
        can result in multiple entries in cachedFonts() with the same Font.
        
        Then in FontCache::purgeInactiveFontData(), the loop that appends fonts to fontsToDelete
        gets the value by reference, and WTFMoves it into fontsToDelete. This nulls out all
        the entries sharing the same value, leaving null entries in the hash table.
        We later crash at font->hasOneRef() when using one of those null entries.
        
        Fix by making a copy of the RefPtr<Font> in the loop, so the WTFMove doesn't nuke
        the hash table entries. The entries will get removed at cachedFonts().remove() lower down.

        * platform/graphics/FontCache.cpp:
        (WebCore::FontCache::purgeInactiveFontData):

2018-09-20  Antoine Quint  <graouts@apple.com>

        [Web Animations] Provide a way to query accelerated animations for internal testing
        https://bugs.webkit.org/show_bug.cgi?id=189762

        Reviewed by Dean Jackson.

        Expose a new internals.acceleratedAnimationsForElement(element) method to allow layout tests to query the current list
        of accelerated animations for a given element. Currently only the animated property and animation speed are exposed, which
        will allow us to identify missing, running and paused accelerated animations.

        * animation/DocumentTimeline.cpp:
        (WebCore::DocumentTimeline::acceleratedAnimationsForElement const):
        * animation/DocumentTimeline.h:
        * platform/graphics/GraphicsLayer.h:
        (WebCore::GraphicsLayer::acceleratedAnimationsForTesting const):
        * platform/graphics/ca/GraphicsLayerCA.cpp:
        (WebCore::GraphicsLayerCA::createTransformAnimationsFromKeyframes):
        * platform/graphics/ca/GraphicsLayerCA.h:
        * testing/Internals.cpp:
        (WebCore::Internals::acceleratedAnimationsForElement):
        * testing/Internals.h:
        * testing/Internals.idl:

2018-09-19  Ryosuke Niwa  <rniwa@webkit.org>

        Improve node statistics for rare data
        https://bugs.webkit.org/show_bug.cgi?id=189775

        Reviewed by Simon Fraser.

        Report reasons we created NodeRareData and ElementRareData in node statistics.

        Added NodeRareData::useTypes and ElementRareData::useTypes which returns OptionSet<NodeRareData::UseType>
        indicating which instance member of the rare data is currently in use.

        * dom/Element.cpp:
        * dom/Element.h:
        * dom/ElementRareData.h:
        (WebCore::defaultMinimumSizeForResizing):
        (WebCore::ElementRareData::useTypes const):
        * dom/Node.cpp:
        (WebCore::stringForRareDataUseType):
        (WebCore::Node::dumpStatistics):
        * dom/NodeRareData.cpp:
        * dom/NodeRareData.h:
        (WebCore::NodeRareData::useTypes const):

2018-09-19  Ryosuke Niwa  <rniwa@webkit.org>

        REGRESSION(r235917): 2% regression in Dromaeo CSS selector on MacBookPro11,4
        https://bugs.webkit.org/show_bug.cgi?id=189738

        Reviewed by Yusuke Suzuki.

        The regression was caused by the regundant walk to the parent element. Removed it to fix the regression.

        * cssjit/SelectorCompiler.cpp:
        (WebCore::SelectorCompiler::SelectorCodeGenerator::generateNthChildParentCheckAndRelationUpdate):
        (WebCore::SelectorCompiler::SelectorCodeGenerator::generateNthLastChildParentCheckAndRelationUpdate):

2018-09-19  John Wilander  <wilander@apple.com>

        Resource Load Statistics: Add optional cap on partitioned cache max age
        https://bugs.webkit.org/show_bug.cgi?id=189711
        <rdar://problem/39246837>

        Reviewed by Antti Koivisto and Chris Dumez.

        Test: http/tests/resourceLoadStatistics/cap-cache-max-age-for-prevalent-resource.html

        * platform/network/NetworkStorageSession.h:
        * platform/network/cf/NetworkStorageSessionCFNet.cpp:
        (WebCore::NetworkStorageSession::maxAgeCacheCap):
            Checks if a max age cap is set and returns it if the request
            represents a prevalent resource.
        (WebCore::NetworkStorageSession::setCacheMaxAgeCapForPrevalentResources):
        (WebCore::NetworkStorageSession::resetCacheMaxAgeCapForPrevalentResources):
            New functionality to receive a max age cap setting in the session.

2018-09-19  Youenn Fablet  <youenn@apple.com>

        Layout Test webrtc/video-mute.html is flaky.
        https://bugs.webkit.org/show_bug.cgi?id=177501

        Reviewed by Eric Carlson.

        Covered by updated test expectation.

        * platform/mediastream/mac/RealtimeIncomingVideoSourceCocoa.mm:
        (WebCore::createBlackPixelBuffer): Make sure the whole buffer is properly initialized.

2018-09-19  David Kilzer  <ddkilzer@apple.com>

        WebCoreNSURLSession leaks an NSString (_sessionDescription) in -dealloc
        <https://webkit.org/b/189742>
        <rdar://problem/44589774>

        Reviewed by Joseph Pecoraro.

        * platform/network/cocoa/WebCoreNSURLSession.h:
        (WebCoreNSURLSession._sessionDescription): Change type from
        NSString * to RetainPtr<NSString>.
        * platform/network/cocoa/WebCoreNSURLSession.mm: Remove
        @synthesized statement for sessionDescription so that custom
        methods can be implemented to handle RetainPtr<NSString>.
        (-[WebCoreNSURLSession sessionDescription]): Add.
        (-[WebCoreNSURLSession setSessionDescription:]): Add.

2018-09-19  Youenn Fablet  <youenn@apple.com>

        Implement sender/receiver getStats
        https://bugs.webkit.org/show_bug.cgi?id=189707

        Reviewed by Eric Carlson.

        Add support for sender and receiver getStats.
        Also add support for peer connection selector parameter.

        Add the plumbing of the selector to LibWebRTCMediaEndpoint.
        Then make use of libwebrtc overloaded methods to retrieve the right stats.

        Covered by updated/rebased tests.

        * Modules/mediastream/PeerConnectionBackend.h:
        * Modules/mediastream/RTCPeerConnection.cpp:
        (WebCore::RTCPeerConnection::getStats):
        * Modules/mediastream/RTCPeerConnection.h:
        * Modules/mediastream/RTCPeerConnection.idl:
        * Modules/mediastream/RTCRtpReceiver.cpp:
        (WebCore::RTCRtpReceiver::RTCRtpReceiver):
        (WebCore::RTCRtpReceiver::getStats):
        * Modules/mediastream/RTCRtpReceiver.h:
        (WebCore::RTCRtpReceiver::create):
        (WebCore::RTCRtpReceiver::backend):
        * Modules/mediastream/RTCRtpReceiver.idl:
        * Modules/mediastream/RTCRtpSender.cpp:
        (WebCore::RTCRtpSender::create):
        (WebCore::RTCRtpSender::RTCRtpSender):
        (WebCore::RTCRtpSender::getStats):
        * Modules/mediastream/RTCRtpSender.h:
        * Modules/mediastream/RTCRtpSender.idl:
        * Modules/mediastream/libwebrtc/LibWebRTCMediaEndpoint.cpp:
        (WebCore::LibWebRTCMediaEndpoint::getStats):
        * Modules/mediastream/libwebrtc/LibWebRTCMediaEndpoint.h:
        * Modules/mediastream/libwebrtc/LibWebRTCPeerConnectionBackend.cpp:
        (WebCore::LibWebRTCPeerConnectionBackend::getStats):
        (WebCore::backendFromRTPSender):
        (WebCore::createReceiverForSource):
        (WebCore::LibWebRTCPeerConnectionBackend::createReceiver):
        (WebCore::LibWebRTCPeerConnectionBackend::videoReceiver):
        (WebCore::LibWebRTCPeerConnectionBackend::audioReceiver):
        (WebCore::LibWebRTCPeerConnectionBackend::addTrack):
        (WebCore::LibWebRTCPeerConnectionBackend::addUnifiedPlanTransceiver):
        (WebCore::LibWebRTCPeerConnectionBackend::addTransceiver):
        (WebCore::LibWebRTCPeerConnectionBackend::newRemoteTransceiver):
        * Modules/mediastream/libwebrtc/LibWebRTCPeerConnectionBackend.h:
        * Modules/mediastream/libwebrtc/LibWebRTCRtpReceiverBackend.h:

2018-09-19  Jer Noble  <jer.noble@apple.com>

        REGRESSION (r236006): New waitingForKey() requirement breaks Modern EME tests.
        https://bugs.webkit.org/show_bug.cgi?id=189720
        <rdar://problem/44572140>

        Reviewed by Xabier Rodriguez-Calvar.

        Always call waitingForKey() after calling initializationDataEncountered().

        * platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.mm:
        (WebCore::MediaPlayerPrivateAVFoundationObjC::shouldWaitForLoadingOfResource):
        * platform/graphics/avfoundation/objc/MediaPlayerPrivateMediaSourceAVFObjC.mm:
        (WebCore::MediaPlayerPrivateMediaSourceAVFObjC::initializationDataEncountered):

2018-09-19  Philippe Normand <pnormand@igalia.com>

        [GStreamer] Add support for AV1 decoding
        https://bugs.webkit.org/show_bug.cgi?id=189647

        Tweaked by Xabier Rodriguez Calvar <calvaris@igalia.com>.
        Reviewed by Žan Doberšek.

        AV1 can be muxed in MP4 and WebM containers. The test is an adaptation from Chromium's unittest:
        https://chromium.googlesource.com/chromium/src/+/master/content/browser/media/media_canplaytype_browsertest.cc

        Test: media/media-can-play-av1.html

        * platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.cpp:
        (WebCore::MediaPlayerPrivateGStreamer::supportsType): Add AV1
        support. The av01 codec has to be explicitely checked, along with
        the presence of a compatible AV1 decoder.

2018-09-18  Basuke Suzuki  <Basuke.Suzuki@sony.com>

        [Curl] Limit capturing extra metrics for Web Inspector when not required.
        https://bugs.webkit.org/show_bug.cgi?id=189520

        Reviewed by Alex Christensen.

        Respect the value of NetworkDataTask::shouldCaptureExtraNetworkLoadMetrics() to reduce the process
        time when they are not needed.

        No new tests because there's no behavior change.

        * platform/network/curl/CurlContext.cpp:
        (WebCore::CurlHandle::getNetworkLoadMetrics):
        (WebCore::CurlHandle::addExtraNetworkLoadMetrics):
        * platform/network/curl/CurlContext.h:
        * platform/network/curl/CurlRequest.cpp:
        (WebCore::CurlRequest::CurlRequest):
        (WebCore::CurlRequest::updateNetworkLoadMetrics):
        * platform/network/curl/CurlRequest.h:
        (WebCore::CurlRequest::create):
        * platform/network/curl/ResourceHandleCurl.cpp:
        (WebCore::ResourceHandle::createCurlRequest):

2018-09-18  Megan Gardner  <megan_gardner@apple.com>

        Support Images Module Level 4's double-position gradient color stop syntax
        https://bugs.webkit.org/show_bug.cgi?id=186154
        <rdar://problem/44158152>

        Reviewed by Simon Fraser.

        The CSS spec for all gradients allows for each color stop to have two angles to be used for hints.
        This makes pie chart and checkerboard conic gradients much simpler to write.
        Any time you want to have a hard line in a gradient, this syntax simplifies the gradient specification.

        Test: fast/gradients/conic-two-hints.html
        Test: fast/gradients/linear-two-hints-angle.html
        Test: fast/gradients/linear-two-hints.html
        Test: fast/gradients/radial-two-hints.html

        * css/parser/CSSPropertyParserHelpers.cpp:
        (WebCore::CSSPropertyParserHelpers::consumeAngularGradientColorStops): Removed.
        (WebCore::CSSPropertyParserHelpers::consumeGradientColorStops):

2018-09-18  Simon Fraser  <simon.fraser@apple.com>

        Remove the unused RenderLayerCompositor::enclosingCompositorFlushingLayers()
        https://bugs.webkit.org/show_bug.cgi?id=189689

        Reviewed by Alex Christensen.

        enclosingCompositorFlushingLayers() was added in r76196 but never used. Also use
        a SetForScope<>.

        * rendering/RenderLayerCompositor.cpp:
        (WebCore::RenderLayerCompositor::flushPendingLayerChanges):
        (WebCore::RenderLayerCompositor::enclosingCompositorFlushingLayers const): Deleted.
        * rendering/RenderLayerCompositor.h:

2018-09-18  Joseph Pecoraro  <pecoraro@apple.com>

        [macOS] Frequent leaks seen under WebCore::gpuIDForDisplayMask
        https://bugs.webkit.org/show_bug.cgi?id=189685
        <rdar://problem/44541974>

        Reviewed by Per Arne Vollan.

        * platform/mac/PlatformScreenMac.mm:
        (WebCore::gpuIDForDisplayMask):

2018-09-18  Youenn Fablet  <youenn@apple.com>

        Implement RTCRtpReceiver getContributingSources/getSynchronizationSources
        https://bugs.webkit.org/show_bug.cgi?id=189671

        Reviewed by Eric Carlson.

        Introduce implementation of these two methods by calling the libwebrtc corresponding method.
        Add corresponding IDL as per spec.
        Covered by rebased WPT tests.

        * CMakeLists.txt:
        * DerivedSources.make:
        * Modules/mediastream/RTCRtpContributingSource.h: Added.
        * Modules/mediastream/RTCRtpContributingSource.idl: Added.
        * Modules/mediastream/RTCRtpReceiver.h:
        (WebCore::RTCRtpReceiver::getContributingSources const):
        (WebCore::RTCRtpReceiver::getSynchronizationSources const):
        * Modules/mediastream/RTCRtpReceiver.idl:
        * Modules/mediastream/RTCRtpReceiverBackend.h:
        (WebCore::RTCRtpReceiverBackend::getContributingSources const):
        (WebCore::RTCRtpReceiverBackend::getSynchronizationSources const):
        * Modules/mediastream/RTCRtpSynchronizationSource.h: Added.
        * Modules/mediastream/RTCRtpSynchronizationSource.idl: Added.
        * Modules/mediastream/libwebrtc/LibWebRTCRtpReceiverBackend.cpp:
        (WebCore::fillRTCRtpContributingSource):
        (WebCore::toRTCRtpContributingSource):
        (WebCore::toRTCRtpSynchronizationSource):
        (WebCore::LibWebRTCRtpReceiverBackend::getContributingSources const):
        (WebCore::LibWebRTCRtpReceiverBackend::getSynchronizationSources const):
        * Modules/mediastream/libwebrtc/LibWebRTCRtpReceiverBackend.h:
        * Sources.txt:
        * WebCore.xcodeproj/project.pbxproj:

2018-09-18  Youenn Fablet  <youenn@apple.com>

        Enable Unified Plan by default
        https://bugs.webkit.org/show_bug.cgi?id=189675

        Reviewed by Eric Carlson.

        RTCRtpTransceiver.currentDirection is now gated by the unified plan runtime flag.
        This will allow easy feature detection of unified plan support.
        Covered by updated test.

        * Modules/mediastream/RTCRtpTransceiver.idl:

2018-09-18  Per Arne Vollan  <pvollan@apple.com>

        [WebVTT] change "middle" to "center" for consistency with CSS
        https://bugs.webkit.org/show_bug.cgi?id=158478

        Reviewed by Eric Carlson.

        No new tests, covered by existing tests.

        * html/track/TextTrackCueGeneric.cpp:
        (WebCore::TextTrackCueGenericBoxElement::applyCSSProperties):
        * html/track/VTTCue.cpp:
        (WebCore::centerKeyword):
        (WebCore::VTTCue::initialize):
        (WebCore::VTTCue::align const):
        (WebCore::VTTCue::setAlign):
        (WebCore::VTTCue::calculateDisplayParameters):
        (WebCore::VTTCue::setCueSettings):
        (WebCore::middleKeyword): Deleted.
        * html/track/VTTCue.h:
        * html/track/WebVTTParser.cpp:
        (WebCore::WebVTTTreeBuilder::buildFromString):

2018-09-18  Chris Dumez  <cdumez@apple.com>

        "DidFirstVisuallyNonEmptyLayout" callback does not get called when restoring a page from PageCache
        https://bugs.webkit.org/show_bug.cgi?id=189681
        <rdar://problem/44526171>

        Reviewed by Alex Christensen and Zalan Bujtas.

        The "DidFirstVisuallyNonEmptyLayout" callback was not getting called when restoring a page from PageCache
        because the FrameView is restored from PageCache and we would fail to restore its flags (such as
        m_firstVisuallyNonEmptyLayoutCallbackPending) when entering Page Cache. We now call reset those flags that
        are related to layout miletones when entering PageCache so that layout milestone events properly get sent
        again when restoring from Page Cache.

        * history/CachedFrame.cpp:
        (WebCore::CachedFrame::CachedFrame):

2018-09-18  Manuel Rego Casasnovas  <rego@igalia.com>

        [css-grid] Static position should use content-box, not padding-box
        https://bugs.webkit.org/show_bug.cgi?id=189698

        Reviewed by Javier Fernandez.

        This is a recent change by the CSSWG:
        https://github.com/w3c/csswg-drafts/issues/3020

        The spec text (https://drafts.csswg.org/css-grid/#static-position):
          "The static position of an absolutely-positioned child
           of a grid container is determined as if it were the sole grid item
           in a grid area whose edges coincide with the content edges
           of the grid container."

        Test: imported/w3c/web-platform-tests/css/css-grid/abspos/absolute-positioning-grid-container-parent-001.html

        * rendering/RenderGrid.cpp:
        (WebCore::RenderGrid::prepareChildForPositionedLayout):
        Simple change to use border and padding.

2018-09-18  Xabier Rodriguez Calvar  <calvaris@igalia.com>

        [EME][GStreamer] The current EME implementation doesn't support the waitingforkey event
        https://bugs.webkit.org/show_bug.cgi?id=185590

        Reviewed by Philippe Normand.

        When decryptors are blocked waiting for the key, instruct the
        player to run the Wait for key algorithm. As per spec, if we run
        out of blocks pending to decrypt because we don't have the key, we
        request running the algorithm again.

        Test: imported/w3c/web-platform-tests/encrypted-media/clearkey-mp4-playback-temporary-waitingforkey.https.html.

        * platform/graphics/gstreamer/MediaPlayerPrivateGStreamer.cpp:
        (WebCore::MediaPlayerPrivateGStreamer::handleMessage):
        * platform/graphics/gstreamer/MediaPlayerPrivateGStreamerBase.cpp:
        (WebCore::MediaPlayerPrivateGStreamerBase::initializationDataEncountered):
        (WebCore::MediaPlayerPrivateGStreamerBase::reportWaitingForKey):
        * platform/graphics/gstreamer/MediaPlayerPrivateGStreamerBase.h:
        * platform/graphics/gstreamer/eme/WebKitCommonEncryptionDecryptorGStreamer.cpp:
        (webkitMediaCommonEncryptionDecryptTransformInPlace):
        (webkitMediaCommonEncryptionDecryptSinkEventHandler):

2018-09-12  Ryosuke Niwa  <rniwa@webkit.org>

        Update composedPath to match the latest spec
        https://bugs.webkit.org/show_bug.cgi?id=180378
        <rdar://problem/42843004>

        Reviewed by Darin Adler.

        This patch makes the check for whether a given node in the event path be included in composedPath
        pre-determined at the time of the event dispatching per https://github.com/whatwg/dom/issues/525.
        This was a fix for the issue that if an event listener in a closed shadow tree removes a node in the
        same tree in the event path, then composedPath called on its shadow host, for example, will include
        the removed node since it's no longer in the closed shadow tree.

        Naively, implementing this behavior would require remembering the original document or shadow root
        of each node in the event path as well as its parent shadow root, or worse which node is disclosed
        to every other node at the time of computing the event path.

        This patch takes a more novel and efficient approach to implement the new behavior by adding a single
        integer indicating the number of closed-mode shadow root ancestors of each node in the event path.
        In computePathUnclosedToTarget, any node whose *depth* is greater than the context object is excluded.

        Consider the following example:
        div ------- ShadowRoot (closed)
          +- span     +- slot
        If an event is dispatched on span, then the event path would be [span, slot, ShadowRoot, div]. Then
        the values of integers assigned to each node would be: [0, 1, 1, 0] respectively. When composedPath
        is called on span or div, slot and ShadowRoot are excluded because they have a greater *depth* value.

        Unfortunately, this simplistic solution doesn't work when there are multiple shadow roots of the same
        depth through which an event is dispatched as in:
        section -- ShadowRoot (closed, SR2)
          |          +- slot (s2)
          +div ------ ShadowRoot (closed, SR1)
            +- span     +- slot (s1)
        If an event is dispatched on span, the event path would be [span, s1, SR1, div, s2, SR2, section].
        The values of integers assigned are: [0, 1, 1, 0, 1, 1, 0] respectively. When composedPath is called
        on SR1, the simplistic approach would include s2 and SR2, which would be wrong.

        To account for this case, in computePathUnclosedToTarget, we traverse the event path upwards (i.e.
        ancestors) and downwards (i.e. descendants) from the context object and decrease the *allowed depth*
        of shadow trees when we traverse out of a shadow tree in either direction. When traversing upwards,
        therefore, moving out of a shadow root to its host would would decrease the allowed depth. When
        traversing dowards, moving from a slot element to its assigned node would decrease the allowed depth.

        Note that the depths can be negative when a composed event is dispatched inside a closed shadow tree,
        and it gets out of its shadow host.

        Unfortunately, the latest DOM specification has a bug and doesn't match the behavior of Chrome. This
        patch proposes a new algorithm which can be adopted in https://github.com/whatwg/dom/issues/684.

        Test: imported/w3c/web-platform-tests/shadow-dom/event-composed-path-after-dom-mutation.html

        * dom/EventContext.cpp:
        (WebCore::EventContext::EventContext):
        (WebCore::MouseOrFocusEventContext::MouseOrFocusEventContext):
        (WebCore::TouchEventContext::TouchEventContext):
        * dom/EventContext.h:
        (WebCore::EventContext::closedShadowDepth const): Added.
        * dom/EventPath.cpp:
        (WebCore::WindowEventContext::WindowEventContext):
        (WebCore::EventPath::buildPath): Compute the closed shadow tree's depths for each node in the path.
        (WebCore::computePathUnclosedToTarget const): Implemented the aforementioned algorithm.
        (WebCore::EventPath::EventPath):

2018-09-17  Yusuke Suzuki  <utatane.tea@gmail.com>

        [WTF] Use Semaphore and BinarySemaphore instead of dispatch_semaphore_t
        https://bugs.webkit.org/show_bug.cgi?id=185339

        Reviewed by Mark Lam.

        * fileapi/ThreadableBlobRegistry.cpp:
        (WebCore::ThreadableBlobRegistry::blobSize):
        * platform/cocoa/NetworkExtensionContentFilter.h:
        * platform/cocoa/NetworkExtensionContentFilter.mm:
        (WebCore::NetworkExtensionContentFilter::initialize):
        (WebCore::NetworkExtensionContentFilter::willSendRequest):
        (WebCore::NetworkExtensionContentFilter::responseReceived):
        (WebCore::NetworkExtensionContentFilter::addData):
        (WebCore::NetworkExtensionContentFilter::finishedAddingData):
        (WebCore::NetworkExtensionContentFilter::handleDecision):
        Use per-function BinarySemaphore instead of holding it in NetworkExtensionContentFilter's field.

        * platform/glib/FileMonitorGLib.cpp:
        (WebCore::FileMonitor::FileMonitor):
        (WebCore::FileMonitor::~FileMonitor):
        * platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.h:
        * platform/graphics/avfoundation/objc/MediaPlayerPrivateAVFoundationObjC.mm:
        (WebCore::MediaPlayerPrivateAVFoundationObjC::MediaPlayerPrivateAVFoundationObjC):
        (WebCore::MediaPlayerPrivateAVFoundationObjC::~MediaPlayerPrivateAVFoundationObjC):
        (WebCore::MediaPlayerPrivateAVFoundationObjC::waitForVideoOutputMediaDataWillChange):
        (WebCore::MediaPlayerPrivateAVFoundationObjC::outputMediaDataWillChange):
        Use BinarySemaphore. And remove unused dispatch_semaphore_t.

        * platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.h:
        * platform/graphics/avfoundation/objc/SourceBufferPrivateAVFObjC.mm:
        (-[WebAVStreamDataParserListener streamDataParserWillProvideContentKeyRequestInitializationData:forTrackID:]):
        (-[WebAVStreamDataParserListener streamDataParser:didProvideContentKeyRequestInitializationData:forTrackID:]):
        (WebCore::SourceBufferPrivateAVFObjC::SourceBufferPrivateAVFObjC):
        (WebCore::SourceBufferPrivateAVFObjC::~SourceBufferPrivateAVFObjC):
        (WebCore::SourceBufferPrivateAVFObjC::didProvideContentKeyRequestInitializationDataForTrackID):
        (WebCore::SourceBufferPrivateAVFObjC::abort):
        (WebCore::SourceBufferPrivateAVFObjC::setCDMSession):
        (WebCore::SourceBufferPrivateAVFObjC::setCDMInstance):
        Use Box<Semaphore> and Box<BinarySemaphore>.

        * platform/graphics/cocoa/WebCoreDecompressionSession.h:
        * platform/graphics/cocoa/WebCoreDecompressionSession.mm:
        (WebCore::WebCoreDecompressionSession::WebCoreDecompressionSession):
        Remove unused dispatch_semaphore_t.

        * platform/network/cf/ResourceHandleCFNet.cpp:
        (WebCore::ResourceHandle::platformLoadResourceSynchronously):
        * platform/network/cf/ResourceHandleCFURLConnectionDelegateWithOperationQueue.cpp:
        (WebCore::getRunLoop):
        (WebCore::ResourceHandleCFURLConnectionDelegateWithOperationQueue::willSendRequest):
        (WebCore::ResourceHandleCFURLConnectionDelegateWithOperationQueue::didReceiveResponse):
        (WebCore::ResourceHandleCFURLConnectionDelegateWithOperationQueue::willCacheResponse):
        (WebCore::ResourceHandleCFURLConnectionDelegateWithOperationQueue::canRespondToProtectionSpace):
        * platform/network/mac/WebCoreResourceHandleAsOperationQueueDelegate.h:
        * platform/network/mac/WebCoreResourceHandleAsOperationQueueDelegate.mm:
        (-[WebCoreResourceHandleAsOperationQueueDelegate initWithHandle:messageQueue:]):
        (-[WebCoreResourceHandleAsOperationQueueDelegate detachHandle]):
        (-[WebCoreResourceHandleAsOperationQueueDelegate dealloc]):
        (-[WebCoreResourceHandleAsOperationQueueDelegate connection:willSendRequest:redirectResponse:]):
        (-[WebCoreResourceHandleAsOperationQueueDelegate connection:canAuthenticateAgainstProtectionSpace:]):
        (-[WebCoreResourceHandleAsOperationQueueDelegate connection:didReceiveResponse:]):
        (-[WebCoreResourceHandleAsOperationQueueDelegate connection:willCacheResponse:]):
        Use BinarySemaphore instead.

2018-09-17  Simon Fraser  <simon.fraser@apple.com>

        Many modern media control tests leak documents in testing
        https://bugs.webkit.org/show_bug.cgi?id=189437

        Reviewed by Darin Adler.

        In order to accurately detect leaks in media controls tests which use lots of
        SVGImages, we have to:
        - Fire a zero-delay timer after the postTask, in order for ImagesLoader's m_derefElementTimer
          to clear references to elements.
        - Have releaseCriticalMemory() call CachedResourceLoader's garbageCollectDocumentResources()
          to drop the last handle to the CachedResource for an SVGImage.
        - Call WKBundleReleaseMemory() after the GC and timer, since we need garbageCollectDocumentResources()
          to run again after that timer has fired.
        
        This should fix most of the spurious leak reports involving SVGImage documents.

        * page/MemoryRelease.cpp:
        (WebCore::releaseCriticalMemory):

2018-09-17  Jer Noble  <jer.noble@apple.com>

        Add support for HEVC codec types in Media Capabilities
        https://bugs.webkit.org/show_bug.cgi?id=189565

        Reviewed by Eric Carlson.

        Test: media/hevc-codec-parameters.html

        Add some utility methods for parsing HEVC codec strings, and using those parsed
        values to query the platform for detailed support for HEVC decoding.

        Drive-by fix: Modify MediaEngineConfigurationFactory to allow for null function
        pointers in the encode/decode factory pair.

        * Sources.txt:
        * SourcesCocoa.txt:
        * WebCore.xcodeproj/project.pbxproj:
        * platform/cocoa/VideoToolboxSoftLink.cpp:
        * platform/cocoa/VideoToolboxSoftLink.h:
        * platform/graphics/HEVCUtilities.cpp: Added.
        (WebCore::parseHEVCCodecParameters):
        * platform/graphics/HEVCUtilities.h: Added.
        * platform/graphics/cocoa/HEVCUtilitiesCocoa.cpp: Added.
        (WebCore::validateHEVCParameters):
        * platform/graphics/cocoa/HEVCUtilitiesCocoa.h: Added.
        * platform/graphics/cocoa/MediaEngineConfigurationFactoryCocoa.cpp: Added.
        (WebCore::videoCodecTypeFromRFC4281Type):
        (WebCore::createMediaPlayerDecodingConfigurationCocoa):
        * platform/graphics/cocoa/MediaEngineConfigurationFactoryCocoa.h: Added.
        * platform/mediacapabilities/MediaEngineConfigurationFactory.cpp:
        (WebCore::factories):
        (WebCore::MediaEngineConfigurationFactory::createDecodingConfiguration):
        (WebCore::MediaEngineConfigurationFactory::createEncodingConfiguration):
        * testing/Internals.cpp:
        (WebCore::Internals::parseHEVCCodecParameters):
        * testing/Internals.h:
        * testing/Internals.idl:

2018-09-17  Devin Rousso  <drousso@apple.com>

        Web Inspector: generate CSSKeywordCompletions from backend values
        https://bugs.webkit.org/show_bug.cgi?id=189041

        Reviewed by Joseph Pecoraro.

        Modified existing test inspector/css/getSupportedCSSProperties.html.

        * inspector/agents/InspectorCSSAgent.cpp:
        (WebCore::InspectorCSSAgent::getSupportedCSSProperties):
        Send alias and longhand information for all properties, and any known keyword values for
        those applicable. This makes use of `CSSParserFastPaths::isValidKeywordPropertyAndValue` to
        determine if a given keyword is a valid value for each property. This only generates a list
        for properties who have no non-keyword values.

        * css/makeprop.pl:
        * css/makevalues.pl:
        Create additional helper functions/constants for retrieving strings of each CSS keyword.

        * css/CSSProperty.h:
        (WebCore::CSSProperty::aliasesForProperty):

        * css/CSSPrimitiveValue.cpp:
        (WebCore::valueName):

        * css/CSSValuePool.cpp:
        (WebCore::CSSValuePool::CSSValuePool):
        (WebCore::CSSValuePool::createIdentifierValue):

2018-09-17  Youenn Fablet  <youenn@apple.com>

        track.onmute isn't called for a remote MediaStreamTrack when its counter part track is removed from the peer connection
        https://bugs.webkit.org/show_bug.cgi?id=176281
        <rdar://problem/44525674>

        Reviewed by Eric Carlson.

        Listen to libwebrtc remove track callbacks.
        Implement handling as per https://w3c.github.io/webrtc-pc/#process-remote-track-removal.
        This triggers a mute event on the track.

        Test: webrtc/remove-track.html

        * Modules/mediastream/libwebrtc/LibWebRTCMediaEndpoint.cpp:
        (WebCore::LibWebRTCMediaEndpoint::removeRemoteTrack):
        (WebCore::LibWebRTCMediaEndpoint::OnRemoveTrack):
        * Modules/mediastream/libwebrtc/LibWebRTCMediaEndpoint.h:
        * Modules/mediastream/libwebrtc/LibWebRTCPeerConnectionBackend.cpp:
        (WebCore::LibWebRTCPeerConnectionBackend::shouldOfferAllowToReceive const):
        Drive by fix: Plan B code path does not mandate having an rtc backend for each sender.

2018-09-17  Simon Fraser  <simon.fraser@apple.com>

        Add more Fullscreen logging
        https://bugs.webkit.org/show_bug.cgi?id=189656

        Reviewed by Jer Noble.

        Add some fullscreen logging so I can tell whether WebFullScreenManager ever releases
        the fullscreen element.

        * platform/cocoa/VideoFullscreenModelVideoElement.mm:
        (VideoFullscreenModelVideoElement::VideoFullscreenModelVideoElement):
        (VideoFullscreenModelVideoElement::~VideoFullscreenModelVideoElement):
        (VideoFullscreenModelVideoElement::setVideoElement):

2018-09-14  Simon Fraser  <simon.fraser@apple.com>

        Add support for dumping the GraphicsLayer tree via notifyutil
        https://bugs.webkit.org/show_bug.cgi?id=189639

        Reviewed by Zalan Bujtas.

        Make "notifyutil -p com.apple.WebKit.showGraphicsLayerTree" work. It dumps the GraphicsLayer tree
        for each top-level document (GraphicsLayers are connected across frame boundaries, so this prints
        the entire tree for each main frame).
        
        It uses WTFLogAlways rather than fprintf() so output shows on all platforms (other tree dumps should
        be converted in the same way).

        * page/mac/PageMac.mm:
        (WebCore::Page::platformInitialize):
        * platform/graphics/GraphicsLayer.cpp:
        (showGraphicsLayerTree):
        * rendering/RenderLayerCompositor.cpp:
        (showGraphicsLayerTreeForCompositor):
        * rendering/RenderLayerCompositor.h:
        * rendering/RenderObject.cpp:
        (WebCore::printGraphicsLayerTreeForLiveDocuments):
        * rendering/RenderObject.h:

2018-09-17  Christopher Reid  <chris.reid@sony.com>

        [Curl] Add schema version and enable auto vacuum for cookie database.
        https://bugs.webkit.org/show_bug.cgi?id=189669

        Reviewed by Alex Christensen.

        Turning on auto incremental vacuuming and adding versioning to the database. Right now we
        reset tables if there's an unknown schema or if the none is set. There is placeholder logic
        in place to upgrade databases as the schema changes in the future.

        Tested by checking the database manually after launching MiniBrowser.

        * platform/network/curl/CookieJarDB.cpp:
        (WebCore::CookieJarDB::openDatabase):
        (WebCore::CookieJarDB::verifySchemaVersion):
        (WebCore::CookieJarDB::deleteAllTables):
        (WebCore::CookieJarDB::createPrepareStatement):
        (WebCore::CookieJarDB::getPrepareStatement):
        (WebCore::CookieJarDB::executeSimpleSql):
        * platform/network/curl/CookieJarDB.h:
        * platform/network/curl/NetworkStorageSessionCurl.cpp:
        (WebCore::defaultSession):
        * platform/sql/SQLiteDatabase.cpp:
        (WebCore::SQLiteDatabase::turnOnIncrementalAutoVacuum): fixed a bug.

2018-09-14  Ryosuke Niwa  <rniwa@webkit.org>

        Re-order Node flags based on semantics
        https://bugs.webkit.org/show_bug.cgi?id=189643

        Reviewed by Simon Fraser.

        Re-ordered Node flags based on their category and subclasses of Node which use them.

        * dom/Node.h:

2018-09-17  Simon Fraser  <simon.fraser@apple.com>

        Add more ResourceLoading logging, particularly in MemoryCache code
        https://bugs.webkit.org/show_bug.cgi?id=189651

        Reviewed by Tim Horton.

        Adding more logging to the ResourceLoading log channel, which I found useful
        when trying to understand cached SVGImage lifetimes (bug 189437).

        * loader/cache/CachedResource.cpp:
        (WebCore::CachedResource::deleteIfPossible):
        * loader/cache/CachedResourceLoader.cpp:
        (WebCore::CachedResourceLoader::requestResource):
        (WebCore::CachedResourceLoader::garbageCollectDocumentResources):
        * loader/cache/MemoryCache.cpp:
        (WebCore::MemoryCache::add):
        (WebCore::MemoryCache::pruneLiveResourcesToSize):
        (WebCore::MemoryCache::pruneDeadResources):
        (WebCore::MemoryCache::pruneDeadResourcesToSize):
        (WebCore::MemoryCache::remove):
        (WebCore::MemoryCache::dumpLRULists const):

2018-09-17  Jer Noble  <jer.noble@apple.com>

        Enable USE_MEDIAREMOTE on iOS
        https://bugs.webkit.org/show_bug.cgi?id=189096

        Reviewed by Eric Carlson.

        Migrate to using MediaRemote.framework on iOS from MediaPlayer.framework. This unifies the
        Now Playing implementation on iOS and Mac.

        * SourcesCocoa.txt:
        * WebCore.xcodeproj/project.pbxproj:
        * platform/audio/cocoa/MediaSessionManagerCocoa.cpp: Removed.
        * platform/audio/cocoa/MediaSessionManagerCocoa.h:
        * platform/audio/cocoa/MediaSessionManagerCocoa.mm: Renamed from Source/WebCore/platform/audio/mac/MediaSessionManagerMac.mm.
        (PlatformMediaSessionManager::sharedManager):
        (PlatformMediaSessionManager::sharedManagerIfExists):
        (MediaSessionManagerCocoa::updateSessionState):
        (MediaSessionManagerCocoa::beginInterruption):
        (MediaSessionManagerCocoa::scheduleUpdateNowPlayingInfo):
        (MediaSessionManagerCocoa::sessionWillBeginPlayback):
        (MediaSessionManagerCocoa::sessionDidEndRemoteScrubbing):
        (MediaSessionManagerCocoa::removeSession):
        (MediaSessionManagerCocoa::sessionWillEndPlayback):
        (MediaSessionManagerCocoa::clientCharacteristicsChanged):
        (MediaSessionManagerCocoa::sessionCanProduceAudioChanged):
        (MediaSessionManagerCocoa::nowPlayingEligibleSession):
        (MediaSessionManagerCocoa::updateNowPlayingInfo):
        * platform/audio/ios/MediaSessionManagerIOS.h:
        (): Deleted.
        * platform/audio/ios/MediaSessionManagerIOS.mm:
        (WebCore::MediaSessionManageriOS::nowPlayingEligibleSession): Deleted.
        (WebCore::MediaSessionManageriOS::updateNowPlayingInfo): Deleted.
        (WebCore::MediaSessionManageriOS::sessionWillBeginPlayback): Deleted.
        (WebCore::MediaSessionManageriOS::removeSession): Deleted.
        (WebCore::MediaSessionManageriOS::sessionWillEndPlayback): Deleted.
        (WebCore::MediaSessionManageriOS::clientCharacteristicsChanged): Deleted.
        * platform/audio/mac/MediaSessionManagerMac.h: Removed.

2018-09-17  Frederic Wang  <fwang@igalia.com>

        Build error in ImageBufferCG when compiled with IOSurfacePool
        https://bugs.webkit.org/show_bug.cgi?id=189579

        Reviewed by Tim Horton.

        IOSurface.h might be included with different value of IOSURFACE_CANVAS_BACKING_STORE, causing
        compilation errors when files in the same unified source do not agree on the definition.
        This patch moves the definition of IOSURFACE_CANVAS_BACKING_STORE from ImageBufferDataCG.h
        to Platform.h so that IOSURFACE_CANVAS_BACKING_STORE is set to the same value in all files.
        Finally some minors changes to explicitly declare/define ImageBuffer are performed in order
        to prevent future issues with Unified build rotating.

        No new tests, behavior unchanged.

        * html/HTMLCanvasElement.cpp: Explicitly include ImageBuffer.h since it's used in this file.
        * platform/graphics/cg/ImageBufferDataCG.h: Move definition into Platform.h.
        * platform/graphics/cocoa/IOSurface.h: Forward-declare ImageBuffer since it's used in this
        header.

2018-09-17  Basuke Suzuki  <Basuke.Suzuki@sony.com>

        [Curl] Respond with requested authentication scheme for authentication challenge.
        https://bugs.webkit.org/show_bug.cgi?id=189318

        Reviewed by Alex Christensen.

        Curl port depends on libcurl's authentication handling by enabling CURLAUTH_ANY. With this
        mode, the round-trip communication between the client and the server is handled by libcurl
        internally. That's okay for many cases. But when initial request has a credentials
        (i.e. XMLHttpRequest), there's no valid chance to store credential to the storage because
        the returned response is not 401.

        Passes following tests:
        - http/tests/websocket/tests/hybi/httponly-cookie.pl
        - http/tests/websocket/tests/hybi/secure-cookie-insecure-connection.pl
        - http/tests/websocket/tests/hybi/secure-cookie-secure-connection.pl
        - http/tests/xmlhttprequest/basic-auth-default.html
        - http/tests/xmlhttprequest/cross-origin-authorization.html
        - http/tests/xmlhttprequest/logout.html
        - http/tests/xmlhttprequest/null-auth.php
        - http/tests/xmlhttprequest/re-login-async.html
        - http/tests/xmlhttprequest/re-login.html
        - http/tests/xmlhttprequest/redirect-credentials-responseURL.html
        - http/tests/xmlhttprequest/remember-bad-password.html

        * platform/network/ResourceHandle.h:
        * platform/network/curl/CurlContext.cpp:
        (WebCore::CurlHandle::setHttpAuthUserPass):
        (WebCore::CurlHandle::enableHttpAuthentication): Deleted.
        * platform/network/curl/CurlContext.h:
        * platform/network/curl/CurlRequest.cpp:
        (WebCore::CurlRequest::setAuthenticationScheme):
        (WebCore::CurlRequest::setupTransfer):
        * platform/network/curl/CurlRequest.h:
        * platform/network/curl/ResourceHandleCurl.cpp:
        (WebCore::ResourceHandle::start):
        (WebCore::ResourceHandle::didReceiveAuthenticationChallenge):
        (WebCore::ResourceHandle::receivedCredential):
        (WebCore::ResourceHandle::getCredential):
        (WebCore::ResourceHandle::restartRequestWithCredential):
        (WebCore::ResourceHandle::platformLoadResourceSynchronously):
        (WebCore::ResourceHandle::continueAfterWillSendRequest):

2018-09-17  Youenn Fablet  <youenn@apple.com>

        Enable VCP for iOS and reenable it for MacOS
        https://bugs.webkit.org/show_bug.cgi?id=189635
        <rdar://problem/43621029>

        Reviewed by Eric Carlson.

        Covered by exsiting and modified tests.
        Instead of using libwebrtc YUV frames for black frames, use CVPixelBuffer to make it efficient.
        Add internal API to know whether VCP is enabled so as to make capture-webrtc test pass on all platforms.

        * platform/mediastream/RealtimeOutgoingVideoSource.cpp:
        (WebCore::RealtimeOutgoingVideoSource::sendBlackFramesIfNeeded):
        * platform/mediastream/RealtimeOutgoingVideoSource.h:
        * platform/mediastream/gstreamer/RealtimeOutgoingVideoSourceLibWebRTC.h:
        * platform/mediastream/mac/RealtimeIncomingVideoSourceCocoa.h:
        * platform/mediastream/mac/RealtimeIncomingVideoSourceCocoa.mm:
        (WebCore::createBlackPixelBuffer):
        (WebCore::RealtimeIncomingVideoSourceCocoa::pixelBufferFromVideoFrame):
        * platform/mediastream/mac/RealtimeOutgoingVideoSourceCocoa.cpp:
        (WebCore::RealtimeOutgoingVideoSourceCocoa::createBlackFrame):
        * platform/mediastream/mac/RealtimeOutgoingVideoSourceCocoa.h:
        * testing/Internals.cpp:
        (WebCore::Internals::supportsVCPEncoder):
        * testing/Internals.h:
        * testing/Internals.idl:

2018-09-17  Chris Dumez  <cdumez@apple.com>

        PSON: window.open() with 'noopener' should only process-swap cross-site, not cross-origin
        https://bugs.webkit.org/show_bug.cgi?id=189602
        <rdar://problem/44430549>

        Reviewed by Geoff Garen.

        * loader/DocumentLoader.cpp:
        (WebCore::DocumentLoader::setTriggeringAction):
        * loader/DocumentLoader.h:
        * loader/FrameLoadRequest.h:
        * loader/FrameLoader.cpp:
        (WebCore::FrameLoader::loadURL):
        (WebCore::FrameLoader::loadWithNavigationAction):
        (WebCore::FrameLoader::loadWithDocumentLoader):
        (WebCore::FrameLoader::loadPostRequest):
        (WebCore::FrameLoader::continueLoadAfterNewWindowPolicy):
        (WebCore::FrameLoader::loadDifferentDocumentItem):
        Move NavigationAction's opener setting to loadWithNavigationAction() as this is a better bottleneck.
        Otherwise, we'd have to set it at several call sites. Also move the NavigationAction around instead
        of copying it.

        * loader/FrameLoader.h:
        (WebCore::FrameLoader::loadWithNavigationAction):
        * loader/NavigationAction.h:
        (WebCore::NavigationAction::setShouldOpenExternalURLsPolicy):
        * loader/PolicyChecker.cpp:
        (WebCore::PolicyChecker::checkNavigationPolicy):
        * page/DOMWindow.cpp:
        (WebCore::DOMWindow::createWindow):

2018-09-17  Darin Adler  <darin@apple.com>

        Use OpaqueJSString rather than JSRetainPtr inside WebKit
        https://bugs.webkit.org/show_bug.cgi?id=189652

        Reviewed by Saam Barati.

        * Modules/plugins/QuickTimePluginReplacement.mm:
        (WebCore::jsValueWithDictionaryInContext): Use OpaqueJSString::create.
        (WebCore::jsValueWithAVMetadataItemInContext): Use adoptCF.

        * platform/mac/SerializedPlatformRepresentationMac.mm:
        (WebCore::jsValueWithDictionaryInContext): Use OpaqueJSString::create.

2018-09-08  Darin Adler  <darin@apple.com>

        Streamline JSRetainPtr, fix leaks of JSString and JSGlobalContext
        https://bugs.webkit.org/show_bug.cgi?id=189455

        Reviewed by Keith Miller.

        * Modules/plugins/QuickTimePluginReplacement.mm:
        (WebCore::jsValueWithDictionaryInContext): Adding a missing
        JSStringRelease to fix a leak.

2018-09-15  Rob Buis  <rbuis@igalia.com>

        XMLHttpRequest::createResponseBlob() should create a Blob with type for empty response
        https://bugs.webkit.org/show_bug.cgi?id=189627

        Reviewed by Alexey Proskuryakov.

        Right now we return an empty Blob without type when the response is empty, but
        it should always include the type [1].

        Test: web-platform-tests/xhr/overridemimetype-blob.html

        [1] https://xhr.spec.whatwg.org/#blob-response

        * xml/XMLHttpRequest.cpp:
        (WebCore::XMLHttpRequest::createResponseBlob):

2018-09-14  Basuke Suzuki  <Basuke.Suzuki@sony.com>

        [Curl] Bug fix on some inaccurate values in NetworkLoadMetrics.
        https://bugs.webkit.org/show_bug.cgi?id=189530

        Reviewed by Alex Christensen.

        Curl port uses the start time libcurl provided. But there's a lug between main thread and Curl thread.
        Record the start time of request instead of libcurl's start timing and use it to measure the metrics.
        Also respondEnd was not correctly recorded and fixed.

        No new tests because it cannot be measured from DRT.

        * platform/network/ResourceHandleInternal.h:
        * platform/network/curl/CurlContext.cpp:
        (WebCore::CurlHandle::getNetworkLoadMetrics):
        * platform/network/curl/CurlContext.h:
        * platform/network/curl/CurlRequest.cpp:
        (WebCore::CurlRequest::start):
        (WebCore::CurlRequest::setupTransfer):
        (WebCore::CurlRequest::didCompleteTransfer):
        (WebCore::CurlRequest::updateNetworkLoadMetrics):
        * platform/network/curl/CurlRequest.h:
        (WebCore::CurlRequest::setStartTime):
        * platform/network/curl/ResourceHandleCurl.cpp:
        (WebCore::ResourceHandle::start):
        (WebCore::ResourceHandle::restartRequestWithCredential):
        (WebCore::ResourceHandle::platformLoadResourceSynchronously):
        (WebCore::ResourceHandle::willSendRequest):
        (WebCore::ResourceHandle::continueAfterWillSendRequest):

2018-09-14  Justin Fan  <justin_fan@apple.com>

        WebGL 2 conformance: rgb-format-support.html
        https://bugs.webkit.org/show_bug.cgi?id=189610
        <rdar://problem/44403343>

        Reviewed by Dean Jackson.

        Implementing getInternalformatParameter (emulating on macOS) and updating 
        renderbufferStorage{Multisample} for WebGL 2 conformance. 

        Test: webgl/2.0.0/conformance2/rendering/rgb-format-support.html enabled.

        * html/canvas/WebGL2RenderingContext.cpp:
        (WebCore::isRenderableInternalformat):
        (WebCore::WebGL2RenderingContext::getInternalformatParameter):
        (WebCore::WebGL2RenderingContext::renderbufferStorageMultisample):
        (WebCore::WebGL2RenderingContext::renderbufferStorage):
        (WebCore::WebGL2RenderingContext::baseInternalFormatFromInternalFormat):
        (WebCore::WebGL2RenderingContext::isIntegerFormat):
        * platform/graphics/GraphicsContext3D.h:
        * platform/graphics/opengl/GraphicsContext3DOpenGLCommon.cpp:
        (WebCore::GraphicsContext3D::getInternalformativ):
        (WebCore::GraphicsContext3D::renderbufferStorageMultisample):

2018-09-14  Justin Michaud  <justin_michaud@apple.com>

        Add support for spreadMethod=reflect and repeat on SVG gradients (for CoreGraphics platforms)
        https://bugs.webkit.org/show_bug.cgi?id=5968

        Add support for spreadMethod=repeat and reflect. Also, the opacity of a gradient is now
        the result of multiplying stop-opacity with the opacity of the color.

        Reviewed by Simon Fraser.

        Tests: svg/gradients/spreadMethod-expected.svg
               svg/gradients/spreadMethod.svg
               svg/gradients/spreadMethodAlpha-expected.svg
               svg/gradients/spreadMethodAlpha.svg
               svg/gradients/spreadMethodClose0-expected-mismatch.svg
               svg/gradients/spreadMethodClose0.svg
               svg/gradients/spreadMethodClose1-expected-mismatch.svg
               svg/gradients/spreadMethodClose1.svg
               svg/gradients/spreadMethodClose2-expected.svg
               svg/gradients/spreadMethodClose2.svg
               svg/gradients/spreadMethodDiagonal-expected.svg
               svg/gradients/spreadMethodDiagonal.svg
               svg/gradients/spreadMethodDiagonal2-expected.svg
               svg/gradients/spreadMethodDiagonal2.svg
               svg/gradients/spreadMethodDuplicateStop-expected.svg
               svg/gradients/spreadMethodDuplicateStop.svg
               svg/gradients/spreadMethodReversed-expected.svg
               svg/gradients/spreadMethodReversed.svg
               svg/gradients/stopAlpha-expected.svg
               svg/gradients/stopAlpha.svg

        * platform/graphics/cg/GradientCG.cpp:
        (WebCore::Gradient::paint):
        * svg/SVGStopElement.cpp:
        (WebCore::SVGStopElement::stopColorIncludingOpacity const):

2018-09-14  Ryan Haddad  <ryanhaddad@apple.com>

        Unreviewed, attempt to fix the iOSMac build after r236015.

        * platform/graphics/cv/PixelBufferResizer.mm:
        (WebCore::PixelBufferResizer::PixelBufferResizer):

2018-09-11  Simon Fraser  <simon.fraser@apple.com>

        Make GraphicsLayers ref-counted, so their tree can persist when disconnected from RenderLayerBackings
        https://bugs.webkit.org/show_bug.cgi?id=189521

        Reviewed by Tim Horton.
        
        Make GraphicsLayer be RefCounted<GraphicsLayer>. GraphicsLayers own their children, via a Vector<Ref<GraphicsLayer>>.
        
        RenderLayerBacking and other holders of GraphicsLayers use RefPtr<GraphicsLayer>.
        
        All the other changes are just to adapt to the new ownership patterns.
        
        I verified that no GraphicsLayers were leaked or abandoned after this change.

        No behavior change.

        * page/PageOverlayController.cpp:
        (WebCore::PageOverlayController::layerWithDocumentOverlays):
        (WebCore::PageOverlayController::layerWithViewOverlays):
        (WebCore::PageOverlayController::installPageOverlay):
        (WebCore::PageOverlayController::uninstallPageOverlay):
        (WebCore::PageOverlayController::setPageOverlayNeedsDisplay):
        (WebCore::PageOverlayController::didChangeViewSize):
        (WebCore::PageOverlayController::didChangeDocumentSize):
        (WebCore::PageOverlayController::didChangeSettings):
        (WebCore::PageOverlayController::paintContents):
        (WebCore::PageOverlayController::didChangeOverlayFrame):
        (WebCore::PageOverlayController::didChangeOverlayBackgroundColor):
        * page/PageOverlayController.h:
        * page/mac/ServicesOverlayController.h:
        (WebCore::ServicesOverlayController::Highlight::layer const):
        * page/mac/ServicesOverlayController.mm:
        (WebCore::ServicesOverlayController::Highlight::Highlight):
        (WebCore::ServicesOverlayController::Highlight::invalidate):
        (WebCore::ServicesOverlayController::Highlight::fadeIn):
        (WebCore::ServicesOverlayController::Highlight::fadeOut):
        (WebCore::ServicesOverlayController::Highlight::didFinishFadeOutAnimation):
        (WebCore::ServicesOverlayController::determineActiveHighlight):
        * platform/graphics/GraphicsLayer.cpp:
        (WebCore::GraphicsLayer::GraphicsLayer):
        (WebCore::GraphicsLayer::willBeDestroyed):
        (WebCore::GraphicsLayer::setChildren):
        (WebCore::GraphicsLayer::addChild):
        (WebCore::GraphicsLayer::addChildAtIndex):
        (WebCore::GraphicsLayer::addChildBelow):
        (WebCore::GraphicsLayer::addChildAbove):
        (WebCore::GraphicsLayer::replaceChild):
        (WebCore::GraphicsLayer::removeAllChildren):
        (WebCore::GraphicsLayer::removeFromParent):
        (WebCore::GraphicsLayer::setMaskLayer):
        (WebCore::GraphicsLayer::noteDeviceOrPageScaleFactorChangedIncludingDescendants):
        (WebCore::GraphicsLayer::distributeOpacity):
        (WebCore::GraphicsLayer::traverse):
        (WebCore::dumpChildren):
        * platform/graphics/GraphicsLayer.h:
        (WebCore::GraphicsLayer::children const):
        (WebCore::GraphicsLayer::children):
        (WebCore::GraphicsLayer::maskLayer const):
        (WebCore::GraphicsLayer::replicaLayer const):
        (WebCore::GraphicsLayer::beingDestroyed const):
        (WebCore::GraphicsLayer:: const): Deleted.
        * platform/graphics/GraphicsLayerFactory.h:
        * platform/graphics/ca/GraphicsLayerCA.cpp:
        (WebCore::GraphicsLayer::create):
        (WebCore::GraphicsLayerCA::setChildren):
        (WebCore::GraphicsLayerCA::addChild):
        (WebCore::GraphicsLayerCA::addChildAtIndex):
        (WebCore::GraphicsLayerCA::addChildBelow):
        (WebCore::GraphicsLayerCA::addChildAbove):
        (WebCore::GraphicsLayerCA::replaceChild):
        (WebCore::GraphicsLayerCA::setMaskLayer):
        (WebCore::GraphicsLayerCA::recursiveVisibleRectChangeRequiresFlush const):
        (WebCore::GraphicsLayerCA::recursiveCommitChanges):
        (WebCore::GraphicsLayerCA::updateSublayerList):
        (WebCore::GraphicsLayerCA::createTransformAnimationsFromKeyframes):
        * platform/graphics/ca/GraphicsLayerCA.h:
        * platform/graphics/texmap/GraphicsLayerTextureMapper.cpp:
        (WebCore::GraphicsLayer::create):
        (WebCore::GraphicsLayerTextureMapper::setChildren):
        (WebCore::GraphicsLayerTextureMapper::addChild):
        (WebCore::GraphicsLayerTextureMapper::addChildAtIndex):
        (WebCore::GraphicsLayerTextureMapper::addChildAbove):
        (WebCore::GraphicsLayerTextureMapper::addChildBelow):
        (WebCore::GraphicsLayerTextureMapper::replaceChild):
        (WebCore::GraphicsLayerTextureMapper::setMaskLayer):
        (WebCore::GraphicsLayerTextureMapper::updateBackingStoreIncludingSubLayers):
        * platform/graphics/texmap/GraphicsLayerTextureMapper.h:
        * platform/graphics/texmap/coordinated/CoordinatedGraphicsLayer.cpp:
        (WebCore::GraphicsLayer::create):
        (WebCore::CoordinatedGraphicsLayer::setShouldUpdateVisibleRect):
        (WebCore::CoordinatedGraphicsLayer::addChild):
        (WebCore::CoordinatedGraphicsLayer::addChildAtIndex):
        (WebCore::CoordinatedGraphicsLayer::addChildAbove):
        (WebCore::CoordinatedGraphicsLayer::addChildBelow):
        (WebCore::CoordinatedGraphicsLayer::replaceChild):
        (WebCore::CoordinatedGraphicsLayer::setMaskLayer):
        (WebCore::CoordinatedGraphicsLayer::syncPendingStateChangesIncludingSubLayers):
        (WebCore::CoordinatedGraphicsLayer::updateContentBuffersIncludingSubLayers):
        (WebCore::CoordinatedGraphicsLayer::setCoordinatorIncludingSubLayersIfNeeded):
        * platform/graphics/texmap/coordinated/CoordinatedGraphicsLayer.h:
        * platform/graphics/win/GraphicsLayerDirect2D.cpp:
        (WebCore::GraphicsLayer::create):
        (WebCore::GraphicsLayerDirect2D::GraphicsLayerDirect2D): Deleted.
        (WebCore::GraphicsLayerDirect2D::initialize): Deleted.
        (WebCore::GraphicsLayerDirect2D::~GraphicsLayerDirect2D): Deleted.
        (WebCore::GraphicsLayerDirect2D::setNeedsDisplay): Deleted.
        (WebCore::GraphicsLayerDirect2D::setNeedsDisplayInRect): Deleted.
        * rendering/RenderLayerBacking.cpp:
        (WebCore::RenderLayerBacking::createGraphicsLayer):
        (WebCore::RenderLayerBacking::createPrimaryGraphicsLayer):
        (WebCore::RenderLayerBacking::updateConfiguration):
        (WebCore::RenderLayerBacking::updateInternalHierarchy):
        (WebCore::RenderLayerBacking::updateMaskingLayer):
        (WebCore::RenderLayerBacking::updateChildClippingStrategy):
        (WebCore::RenderLayerBacking::updateScrollingLayers):
        * rendering/RenderLayerBacking.h:
        * rendering/RenderLayerCompositor.cpp:
        (WebCore::RenderLayerCompositor::updateCompositingLayers):
        (WebCore::RenderLayerCompositor::appendDocumentOverlayLayers):
        (WebCore::RenderLayerCompositor::setCompositingParent):
        (WebCore::RenderLayerCompositor::rebuildCompositingLayerTree):
        (WebCore::RenderLayerCompositor::parentFrameContentLayers):
        (WebCore::RenderLayerCompositor::updateLayerForTopOverhangArea):
        (WebCore::RenderLayerCompositor::updateLayerForBottomOverhangArea):
        (WebCore::RenderLayerCompositor::updateLayerForHeader):
        (WebCore::RenderLayerCompositor::updateLayerForFooter):
        (WebCore::RenderLayerCompositor::updateOverflowControlsLayers):
        (WebCore::RenderLayerCompositor::ensureRootLayer):
        (WebCore::RenderLayerCompositor::rootLayerAttachmentChanged):
        * rendering/RenderLayerCompositor.h:

2018-09-14  Eric Carlson  <eric.carlson@apple.com>

        Support arbitrary video resolution in getUserMedia API
        https://bugs.webkit.org/show_bug.cgi?id=178109
        <rdar://problem/35083128>

        Reviewed by Youenn Fablet.

        Support arbitrary video resolutions by configuring the camera to capture at the closest
        larger size it supports and scaling/cropping frames as necessary.

        No new tests, existing tests updated.

        * Modules/mediastream/CanvasCaptureMediaStreamTrack.h:
        * Modules/webaudio/MediaStreamAudioSource.cpp:
        (WebCore::MediaStreamAudioSource::capabilities): Make non-const, it wasn't helpful.
        (WebCore::MediaStreamAudioSource::settings): Ditto.
        (WebCore::MediaStreamAudioSource::capabilities const): Deleted.
        (WebCore::MediaStreamAudioSource::settings const): Deleted.
        * Modules/webaudio/MediaStreamAudioSource.h:

        * SourcesCocoa.txt: Add PixelBufferResizer.

        * WebCore.xcodeproj/project.pbxproj: Ditto.

        * platform/cocoa/CoreVideoSoftLink.h:
        * platform/cocoa/VideoToolboxSoftLink.cpp:
        * platform/cocoa/VideoToolboxSoftLink.h:

        * platform/graphics/cv/PixelBufferResizer.h: Added.
        (WebCore::PixelBufferResizer::canResizeTo):
        * platform/graphics/cv/PixelBufferResizer.mm: Added.
        (WebCore::PixelBufferResizer::PixelBufferResizer):
        (WebCore::PixelBufferResizer::resize):

        * platform/mediastream/RealtimeIncomingAudioSource.cpp:
        (WebCore::RealtimeIncomingAudioSource::capabilities): Make non-const, it wasn't helpful.
        (WebCore::RealtimeIncomingAudioSource::settings): Ditto.
        (WebCore::RealtimeIncomingAudioSource::capabilities const): Deleted.
        (WebCore::RealtimeIncomingAudioSource::settings const): Deleted.
        * platform/mediastream/RealtimeIncomingAudioSource.h:

        * platform/mediastream/RealtimeIncomingVideoSource.cpp:
        (WebCore::RealtimeIncomingVideoSource::capabilities): Make non-const, it wasn't helpful.
        (WebCore::RealtimeIncomingVideoSource::settings): Ditto.
        (WebCore::RealtimeIncomingVideoSource::capabilities const): Deleted.
        (WebCore::RealtimeIncomingVideoSource::settings const): Deleted.
        * platform/mediastream/RealtimeIncomingVideoSource.h:

        * platform/mediastream/RealtimeMediaSource.cpp:
        (WebCore::RealtimeMediaSource::supportsConstraint):
        (WebCore::RealtimeMediaSource::supportsConstraint const): Deleted.
        * platform/mediastream/RealtimeMediaSource.h:

        * platform/mediastream/RealtimeVideoSource.cpp:
        (WebCore::RealtimeVideoSource::presets):
        (WebCore::RealtimeVideoSource::setSupportedPresets):
        (WebCore::standardVideoSizes):
        (WebCore::RealtimeVideoSource::updateCapabilities): Make non-const, it wasn't helpful.
        (WebCore::presetSupportsFrameRate):
        (WebCore::RealtimeVideoSource::supportsCaptureSize):
        (WebCore::RealtimeVideoSource::shouldUsePreset):
        (WebCore::RealtimeVideoSource::bestSupportedSizeAndFrameRate):
        (WebCore::RealtimeVideoSource::setSizeAndFrameRate):
        (WebCore::RealtimeVideoSource::addSupportedCapabilities const): Deleted.
        * platform/mediastream/RealtimeVideoSource.h:
        (WebCore::VideoPresetData::encode const):
        (WebCore::VideoPresetData::decode):
        (WebCore::VideoPreset::create):
        (WebCore::VideoPreset::VideoPreset):
        (WebCore::RealtimeVideoSource::prefersPreset):
        (WebCore::RealtimeVideoSource::canResizeVideoFrames const):
        (WebCore::RealtimeVideoSource::setDefaultSize):
        (WebCore::RealtimeVideoSource::observedFrameRate const):
        (WebCore::VideoPreset::encode const): Deleted.
        (WebCore::VideoPreset::decode): Deleted.

        * platform/mediastream/mac/AVVideoCaptureSource.h:
        (WebCore::AVVideoCaptureSource::videoPresets): Deleted.
        * platform/mediastream/mac/AVVideoCaptureSource.mm:
        (WebCore::AVVideoPreset::create):
        (WebCore::AVVideoPreset::AVVideoPreset):
        (WebCore::AVVideoCaptureSource::AVVideoCaptureSource):
        (WebCore::AVVideoCaptureSource::settings):
        (WebCore::AVVideoCaptureSource::capabilities): Make non-const, it wasn't helpful.
        (WebCore::AVVideoCaptureSource::setFrameRate):
        (WebCore::AVVideoCaptureSource::prefersPreset):
        (WebCore::AVVideoCaptureSource::setSizeAndFrameRateWithPreset):
        (WebCore::AVVideoCaptureSource::frameDurationForFrameRate):
        (WebCore::AVVideoCaptureSource::setupCaptureSession):
        (WebCore::AVVideoCaptureSource::processNewFrame):
        (WebCore::AVVideoCaptureSource::captureOutputDidOutputSampleBufferFromConnection):
        (WebCore::AVVideoCaptureSource::isFrameRateSupported):
        (WebCore::AVVideoCaptureSource::generatePresets):
        (WebCore::updateSizeMinMax): Deleted.
        (WebCore::updateAspectRatioMinMax): Deleted.
        (WebCore::AVVideoCaptureSource::settings const): Deleted.
        (WebCore::AVVideoCaptureSource::capabilities const): Deleted.
        (WebCore::AVVideoCaptureSource::sizeForPreset): Deleted.
        (WebCore::AVVideoCaptureSource::setPreset): Deleted.
        (WebCore::AVVideoCaptureSource::setSizeAndFrameRate): Deleted.
        (WebCore::AVVideoCaptureSource::bestSessionPresetForVideoDimensions): Deleted.
        (WebCore::AVVideoCaptureSource::supportsSizeAndFrameRate): Deleted.

        * platform/mediastream/mac/CoreAudioCaptureSource.cpp:
        (WebCore::CoreAudioCaptureSource::capabilities): Make non-const, it wasn't helpful.
        (WebCore::CoreAudioCaptureSource::settings): Ditto.
        (WebCore::CoreAudioCaptureSource::capabilities const): Deleted.
        (WebCore::CoreAudioCaptureSource::settings const): Deleted.
        * platform/mediastream/mac/CoreAudioCaptureSource.h:

        * platform/mediastream/mac/DisplayCaptureSourceCocoa.cpp:
        (WebCore::DisplayCaptureSourceCocoa::capabilities): Make non-const, it wasn't helpful.
        (WebCore::DisplayCaptureSourceCocoa::settings): Ditto.
        (WebCore::DisplayCaptureSourceCocoa::capabilities const): Deleted.
        (WebCore::DisplayCaptureSourceCocoa::settings const): Deleted.
        * platform/mediastream/mac/DisplayCaptureSourceCocoa.h:

        * platform/mediastream/mac/MockRealtimeVideoSourceMac.h:
        * platform/mediastream/mac/MockRealtimeVideoSourceMac.mm:
        (WebCore::MockRealtimeVideoSourceMac::CMSampleBufferFromPixelBuffer):
        (WebCore::MockRealtimeVideoSourceMac::updateSampleBuffer):
        (WebCore::MockRealtimeVideoSourceMac::setSizeAndFrameRateWithPreset):
        * platform/mock/MockMediaDevice.h:
        (WebCore::MockCameraProperties::decode):

        * platform/mock/MockRealtimeAudioSource.cpp:
        (WebCore::MockRealtimeAudioSource::settings): Make non-const, it wasn't helpful.
        (WebCore::MockRealtimeAudioSource::capabilities): Ditto.
        (WebCore::MockRealtimeAudioSource::settings const): Deleted.
        (WebCore::MockRealtimeAudioSource::capabilities const): Deleted.
        * platform/mock/MockRealtimeAudioSource.h:

        * platform/mock/MockRealtimeMediaSourceCenter.cpp:
        (WebCore::defaultDevices): Change video device presets to trigger resize code more often.

        * platform/mock/MockRealtimeVideoSource.cpp:
        (WebCore::MockRealtimeVideoSource::MockRealtimeVideoSource):
        (WebCore::MockRealtimeVideoSource::generatePresets):
        (WebCore::MockRealtimeVideoSource::capabilities): Make non-const, it wasn't helpful.
        (WebCore::MockRealtimeVideoSource::settings): Ditto.
        (WebCore::MockRealtimeVideoSource::capabilities const): Deleted.
        (WebCore::MockRealtimeVideoSource::settings const): Deleted.
        * platform/mock/MockRealtimeVideoSource.h:

2018-09-14  Frederic Wang  <fwang@igalia.com>

        Bug 189541 - Build error in FontDescriptionKey::computeHash when compiling FontTaggedSettings and FontCascadeFonts together
        https://bugs.webkit.org/show_bug.cgi?id=189541

        Reviewed by Alex Christensen.

        FontDescriptionKey::computeHash() from In FontCache.h requires implicit instantiation of the
        FontTaggedSettings::hash() template function. This instantiation may happen before
        FontTaggedSettings::hash() is actually fully specialized in FontTaggedSettings.cpp. To avoid
        compiler errors when FontCache.h and FontTaggedSettings.cpp are in the same translation unit,
        we declare full specialization of the hash() functions in FontTaggedSettings.h.

        No new tests, behavior unchanged.

        * platform/graphics/FontCache.h: Explicitly include FontTaggedSettings to avoid possible future breakage.
        * platform/graphics/FontTaggedSettings.h: Declare full specialization of FontTaggedSettings::hash().

2018-09-14  Ryan Haddad  <ryanhaddad@apple.com>

        Unreviewed, rolling out r235990.

        Introduced TestWebKitAPI.NowPlayingTest timeouts on iOS

        Reverted changeset:

        "Enable USE_MEDIAREMOTE on iOS"
        https://bugs.webkit.org/show_bug.cgi?id=189096
        https://trac.webkit.org/changeset/235990

2018-09-14  Jer Noble  <jer.noble@apple.com>

        Turn SourceBufferChangeTypeEnabled on by default
        https://bugs.webkit.org/show_bug.cgi?id=189527

        Reviewed by Eric Carlson.

        * page/Settings.yaml:

2018-09-14  Devin Rousso  <webkit@devinrousso.com>

        Web Inspector: Record actions performed on ImageBitmapRenderingContext
        https://bugs.webkit.org/show_bug.cgi?id=181341

        Reviewed by Joseph Pecoraro.

        Test: inspector/canvas/recording-bitmaprenderer.html

        * html/canvas/ImageBitmapRenderingContext.idl:

        * inspector/InspectorCanvas.cpp:
        (WebCore::shouldSnapshotBitmapRendererAction):
        (WebCore::InspectorCanvas::recordAction):

        * inspector/agents/InspectorCanvasAgent.cpp:
        (WebCore::InspectorCanvasAgent::didFinishRecordingCanvasFrame):

        * page/PageConsoleClient.cpp:
        (canvasRenderingContext):

2018-09-14  David Kilzer  <ddkilzer@apple.com>

        REGRESSION (r235954): Fix build failure on watchOS
        <https://webkit.org/b/189605>

        Reviewed by Geoffrey Garen.

        Remove `using WebCore::IndexedDB::KeyType;` from
        Source/WebCore/Modules/indexeddb/IDBKey.h and fix all the
        resulting build failures.

        * Modules/indexeddb/IDBKey.cpp:
        (WebCore::IDBKey::IDBKey):
        (WebCore::IDBKey::isValid const):
        (WebCore::IDBKey::compare const):
        * Modules/indexeddb/IDBKey.h:
        (WebCore::IDBKey::createNumber):
        (WebCore::IDBKey::createDate):
        (WebCore::IDBKey::type const):
        (WebCore::IDBKey::array const):
        (WebCore::IDBKey::string const):
        (WebCore::IDBKey::date const):
        (WebCore::IDBKey::number const):
        (WebCore::IDBKey::binary const):
        (WebCore::IDBKey::compareTypes):
        (WebCore::IDBKey::IDBKey):
        * Modules/indexeddb/IDBKeyData.cpp:
        (WebCore::IDBKeyData::IDBKeyData):
        (WebCore::IDBKeyData::maybeCreateIDBKey const):
        (WebCore::IDBKeyData::isolatedCopy):
        (WebCore::IDBKeyData::encode const):
        (WebCore::IDBKeyData::decode):
        (WebCore::IDBKeyData::compare const):
        (WebCore::IDBKeyData::loggingString const):
        (WebCore::IDBKeyData::setArrayValue):
        (WebCore::IDBKeyData::setBinaryValue):
        (WebCore::IDBKeyData::setStringValue):
        (WebCore::IDBKeyData::setDateValue):
        (WebCore::IDBKeyData::setNumberValue):
        (WebCore::IDBKeyData::isValid const):
        (WebCore::IDBKeyData::operator== const):
        * Modules/indexeddb/IDBKeyData.h:
        (WebCore::IDBKeyData::IDBKeyData):
        (WebCore::IDBKeyData::minimum):
        (WebCore::IDBKeyData::maximum):
        (WebCore::IDBKeyData::type const):
        (WebCore::IDBKeyData::hash const):
        (WebCore::IDBKeyData::string const):
        (WebCore::IDBKeyData::date const):
        (WebCore::IDBKeyData::number const):
        (WebCore::IDBKeyData::binary const):
        (WebCore::IDBKeyData::array const):
        (WebCore::IDBKeyData::encode const):
        (WebCore::IDBKeyData::decode):
        * Modules/indexeddb/server/SQLiteIDBBackingStore.cpp:
        (WebCore::IDBServer::SQLiteIDBBackingStore::uncheckedGetIndexRecordForOneKey):
        * bindings/js/IDBBindingUtilities.cpp:
        (WebCore::toJS):

2018-09-14  Xabier Rodriguez Calvar  <calvaris@igalia.com>

        [EME] Add support the waitingforkey event
        https://bugs.webkit.org/show_bug.cgi?id=189616

        Reviewed by Philippe Normand.

        Crossplatform support to fire the waitingforkey event from the
        player to the element. The element implements the W3C specified
        algorithm.

        * html/HTMLMediaElement.cpp:
        (WebCore::HTMLMediaElement::mediaPlayerWaitingForKey):
        (WebCore::HTMLMediaElement::attemptToResumePlaybackIfNecessary):
        * html/HTMLMediaElement.h:
        * platform/graphics/MediaPlayer.cpp:
        (WebCore::MediaPlayer::waitingForKey):
        * platform/graphics/MediaPlayer.h:
        (WebCore::MediaPlayerClient::mediaPlayerWaitingForKey):

2018-09-14  Mike Gorse  <mgorse@suse.com>

        builtins directory causes name conflict on Python 3
        https://bugs.webkit.org/show_bug.cgi?id=189552

        Reviewed by Michael Catanzaro.

        No new tests (No behavior change).

        * CMakeLists.txt: builtins -> wkbuiltins.
        * DerivedSources.make: builtins -> wkbuiltins.

2018-09-13  Ryosuke Niwa  <rniwa@webkit.org>

        Capturing event listeners are called during bubbling phase for shadow hosts
        https://bugs.webkit.org/show_bug.cgi?id=174288
        <rdar://problem/33530455>

        Reviewed by Darin Adler.

        Implemented the new behavior proposed in https://github.com/whatwg/dom/pull/686 [1] to fix the problem
        that capturing event listeners on a shadow host is invoked during bubbling phase when an event is
        dispatched within its shadow tree.

        To see why this is a problem, suppose we fire a composed event at span#target in the following DOM tree:
          section#hostParent
            + div#host -- ShadowRoot
                            - p#parent
                                - span#target
        Then capturing and bubbling event listeners on #target, #parent, #host, and #hostParent are invoked in
        the following order in WebKit & Chrome right now:

        1. #hostParent, capturing, eventPhase: CAPTURING_PHASE
        2. #parent, capturing, eventPhase: CAPTURING_PHASE
        3. #target, capturing, eventPhase: AT_TARGET
        4. #target, non-capturing, eventPhase: AT_TARGET
        5. #parent, non-capturing, eventPhase: BUBBLING_PHASE
        6. #host, capturing, eventPhase: AT_TARGET
        7. #host, non-capturing, eventPhase: AT_TARGET
        8. #hostParent, non-capturing, eventPhase: BUBBLING_PHASE

        This is counter-intuitive because capturing event listeners on #host isn't invoked until bubblign phase
        started. A more natural ordering would be:

        1. #hostParent, capturing, eventPhase: CAPTURING_PHASE
        2. #host, capturing, eventPhase: AT_TARGET
        3. #parent, capturing, eventPhase: CAPTURING_PHASE
        4. #target, capturing, eventPhase: AT_TARGET
        5. #target, non-capturing, eventPhase: AT_TARGET
        6. #parent, non-capturing, eventPhase: BUBBLING_PHASE
        7. #host, non-capturing, eventPhase: AT_TARGET
        8. #hostParent, non-capturing, eventPhase: BUBBLING_PHASE

        This also happens to be the order by which Gecko's current shadow DOM implementation invoke event listners.
        This patch implements this new behavior using the spec-change proposed in [1]. Note that this patch also
        impacts the invocation order of event listeners when there is no shadow tree. Namely, before this patch,
        event listeners on the event's target is invoked in the registration order. After this patch, all capturing
        event listeners are invoked before bubbling event listeners are invoked.

        To implement this behavior, this patch introduces EventTarget::EventInvokePhase indicating whether we're
        in the capturing phase or bubbling phase to EventTarget::fireEventListeners. We can't use Event's eventPhase
        enum because that's set to Event::Phase::AT_TARGET when we're at a shadow host.

        Test: fast/shadow-dom/capturing-and-bubbling-event-listeners-across-shadow-trees.html

        * Modules/modern-media-controls/media/media-controller-support.js:
        (MediaControllerSupport.prototype.enable): Use capturing event listeners so that we can update the states of
        media controls before author scripts recieve the event.
        (MediaControllerSupport.prototype.disable): Ditto.
        * dom/EventContext.cpp:
        (WebCore::EventContext::handleLocalEvents const):
        (WebCore::MouseOrFocusEventContext::handleLocalEvents const):
        (WebCore::TouchEventContext::handleLocalEvents const):
        * dom/EventContext.h:
        * dom/EventDispatcher.cpp:
        (WebCore::dispatchEventInDOM): Invoke capturing event listners even when target and current target are same.
        This happens when the current target is a shadow host and event's target is in its shadow tree. Also merged
        the special code path for the event's target with the code in the bubbling phase.
        * dom/EventPath.cpp:
        (WebCore::WindowEventContext::handleLocalEvents const):
        * dom/EventTarget.cpp:
        (WebCore::EventTarget::dispatchEvent): Invoke capturing and bubbling event listeners in the order.
        (WebCore::EventTarget::fireEventListeners):
        (WebCore::EventTarget::innerInvokeEventListeners): Renamed from fireEventListeners to match the spec. Use
        EventInvokePhase to filter out event listeners so that we can invoke capturing event listners before bubbling
        event listeners even when eventPhase is Event::Phase::AT_TARGET.
        * dom/EventTarget.h:
        * dom/Node.cpp:
        (WebCore::Node::handleLocalEvents):
        * dom/Node.h:
        * html/HTMLFormElement.cpp:
        (WebCore::HTMLFormElement::handleLocalEvents):
        * html/HTMLFormElement.h:
        * page/DOMWindow.cpp:
        (WebCore::DOMWindow::dispatchEvent):

2018-09-13  Megan Gardner  <megan_gardner@apple.com>

        Fix color stop blending in conic gradients for stops past 1
        https://bugs.webkit.org/show_bug.cgi?id=189532
        <rdar://problem/44158221>

        Reviewed by Simon Fraser.

        Calculation was wrong, fixed it, and wrote a test that failed
        without the change, and passes with it.

        Test: fast/gradients/conic-repeating-last-stop.html

        * css/CSSGradientValue.cpp:
        (WebCore::ConicGradientAdapter::normalizeStopsAndEndpointsOutsideRange):

2018-09-13  Chris Dumez  <cdumez@apple.com>

        Regression(PSON): setting window.opener to null allows process swapping in cases that are not web-compatible
        https://bugs.webkit.org/show_bug.cgi?id=189590
        <rdar://problem/44422725>

        Reviewed by Geoffrey Garen.

        Set a flag on the navigation action to indicate if the page was opened via window.open() without 'noopener'.

        Test: http/tests/navigation/window-open-cross-origin-then-navigated-back-same-origin.html

        * loader/FrameLoader.cpp:
        (WebCore::FrameLoader::loadURL):
        * loader/NavigationAction.h:
        (WebCore::NavigationAction::openedViaWindowOpenWithOpener const):
        (WebCore::NavigationAction::setOpenedViaWindowOpenWithOpener):
        * page/DOMWindow.cpp:
        (WebCore::DOMWindow::createWindow):
        * page/Page.h:
        (WebCore::Page::openedViaWindowOpenWithOpener const):
        (WebCore::Page::setOpenedViaWindowOpenWithOpener):

2018-09-13  Jer Noble  <jer.noble@apple.com>

        Enable USE_MEDIAREMOTE on iOS
        https://bugs.webkit.org/show_bug.cgi?id=189096

        Reviewed by Eric Carlson.

        Migrate to using MediaRemote.framework on iOS from MediaPlayer.framework. This unifies the
        Now Playing implementation on iOS and Mac.

        * SourcesCocoa.txt:
        * WebCore.xcodeproj/project.pbxproj:
        * platform/audio/cocoa/MediaSessionManagerCocoa.cpp: Removed.
        * platform/audio/cocoa/MediaSessionManagerCocoa.h:
        * platform/audio/cocoa/MediaSessionManagerCocoa.mm: Renamed from Source/WebCore/platform/audio/mac/MediaSessionManagerMac.mm.
        (PlatformMediaSessionManager::sharedManager):
        (PlatformMediaSessionManager::sharedManagerIfExists):
        (MediaSessionManagerCocoa::updateSessionState):
        (MediaSessionManagerCocoa::beginInterruption):
        (MediaSessionManagerCocoa::scheduleUpdateNowPlayingInfo):
        (MediaSessionManagerCocoa::sessionWillBeginPlayback):
        (MediaSessionManagerCocoa::sessionDidEndRemoteScrubbing):
        (MediaSessionManagerCocoa::removeSession):
        (MediaSessionManagerCocoa::sessionWillEndPlayback):
        (MediaSessionManagerCocoa::clientCharacteristicsChanged):
        (MediaSessionManagerCocoa::sessionCanProduceAudioChanged):
        (MediaSessionManagerCocoa::nowPlayingEligibleSession):
        (MediaSessionManagerCocoa::updateNowPlayingInfo):
        * platform/audio/ios/MediaSessionManagerIOS.h:
        (): Deleted.
        * platform/audio/ios/MediaSessionManagerIOS.mm:
        (WebCore::MediaSessionManageriOS::nowPlayingEligibleSession): Deleted.
        (WebCore::MediaSessionManageriOS::updateNowPlayingInfo): Deleted.
        (WebCore::MediaSessionManageriOS::sessionWillBeginPlayback): Deleted.
        (WebCore::MediaSessionManageriOS::removeSession): Deleted.
        (WebCore::MediaSessionManageriOS::sessionWillEndPlayback): Deleted.
        (WebCore::MediaSessionManageriOS::clientCharacteristicsChanged): Deleted.
        * platform/audio/mac/MediaSessionManagerMac.h: Removed.

2018-09-13  Fujii Hironori  <Hironori.Fujii@sony.com>

        Remove the workaround for friend class LazyNeverDestroyed<X> statements for MSVC
        https://bugs.webkit.org/show_bug.cgi?id=189576

        Reviewed by Alex Christensen.

        Old MSVC can't compile "friend class LazyNeverDestroyed<X>"
        statements, but "friend LazyNeverDestroyed<X>".

        No new tests (No behavior change).

        * css/CSSInheritedValue.h: Removed the code for COMPILER(MSVC).
        Removed 'class' keyword in "friend class LazyNeverDestroyed<X>"
        statement.
        * css/CSSInitialValue.h: Ditto.
        * css/CSSPrimitiveValue.h: Ditto.
        * css/CSSRevertValue.h: Ditto.
        * css/CSSUnsetValue.h: Ditto.