Context Navigation

← Previous Period
Next Period →

Timeline

Mar 27, 2021:

8:26 PM Changeset in webkit [275146] by Simon Fraser

3 edits in trunk/Source/WebKit

Have DisplayLink compute its displayNominalFramesPerSecond just once
https://bugs.webkit.org/show_bug.cgi?id=223845

Reviewed by Zalan Bujtas.

Rather than calling CVDisplayLinkGetNominalOutputVideoRefreshPeriod() every time,
just store displayNominalFramesPerSecond in a member variable.

UIProcess/mac/DisplayLink.cpp:

(WebKit::DisplayLink::DisplayLink):
(WebKit::DisplayLink::nominalFramesPerSecondFromDisplayLink):
(WebKit::DisplayLink::nominalFramesPerSecond const): Deleted.

UIProcess/mac/DisplayLink.h:

(WebKit::DisplayLink::nominalFramesPerSecond const):

8:22 PM Changeset in webkit [275145] by Lauro Moura

2 edits
1 add in trunk/Tools/buildstream

[Flatpak SDK] Backport GTK4 Clipboard fix for GdkContentProviderUnion
https://bugs.webkit.org/show_bug.cgi?id=223813

Reviewed by Philippe Normand.

From https://gitlab.gnome.org/GNOME/gtk/-/merge_requests/3346

Will fix some clipbaoard crashes.

elements/sdk/gtk.bst:
patches/gtk-clipboard-Fix-a-crash-when-GdkContentProviderUnion-d.patch: Added.

7:52 PM Changeset in webkit [275144] by Simon Fraser

21 edits in trunk/Source

Allow DisplayRefreshMonitor to be more long-lived objects
https://bugs.webkit.org/show_bug.cgi?id=223844

Reviewed by Chris Dumez.

The existing behavior for DisplayRefreshMonitors was that they were created and destroyed
frequently, as their clients are registered and unregistered. In addition, some of
their subclasses had duplicated logic related to how often they fired without clients
before they were removed.

The 1:1 correspondance between DisplayRefreshMonitors and physical displays makes them
a useful place to store per-display information (like refresh rate), which should
persist whether or not they have clients, so this change prepares for that.

The main changes in this patch are:

Provide virtual startNotificationMechanism()/stopNotificationMechanism() functions on DisplayRefreshMonitor that subclasses can use to start their CVDisplayLink or related functionality.

Provide some shared maxUnscheduledFireCount logic that subclasses can tune to provide the hysteresis used to control the start/stop of the underlying notification mechanism.

Provide a shared dispatchDisplayDidRefresh() function that most implementations can use.

Source/WebCore:

DisplayRefreshMonitorManager no longer destroys DisplayRefreshMonitors when they lose
all their clients. This means that DisplayRefreshMonitors are never removed, but there
should only ever be as many as there are physical displays, and they are only
active if being scheduled by clients.

loader/EmptyClients.cpp:
platform/graphics/DisplayRefreshMonitor.cpp:

(WebCore::DisplayRefreshMonitor::stop):
(WebCore::DisplayRefreshMonitor::removeClient):
(WebCore::DisplayRefreshMonitor::requestRefreshCallback):
(WebCore::DisplayRefreshMonitor::firedAndReachedMaxUnscheduledFireCount):
(WebCore::DisplayRefreshMonitor::displayLinkFired):
(WebCore::DisplayRefreshMonitor::dispatchDisplayDidRefresh):
(WebCore::DisplayRefreshMonitor::displayDidRefresh):

platform/graphics/DisplayRefreshMonitor.h:

(WebCore::DisplayRefreshMonitor::setMaxUnscheduledFireCount):
(WebCore::DisplayRefreshMonitor::mutex):
(WebCore::DisplayRefreshMonitor::displayLinkFired): Deleted.
(WebCore::DisplayRefreshMonitor::stop): Deleted.
(WebCore::DisplayRefreshMonitor::shouldBeTerminated const): Deleted.
(WebCore::DisplayRefreshMonitor::isActive const): Deleted.
(WebCore::DisplayRefreshMonitor::setIsActive): Deleted.

platform/graphics/DisplayRefreshMonitorManager.cpp:

(WebCore::DisplayRefreshMonitorManager::unregisterClient):
(WebCore::DisplayRefreshMonitorManager::displayDidRefresh):
(WebCore::DisplayRefreshMonitorManager::displayWasUpdated):

platform/graphics/gtk/DisplayRefreshMonitorGtk.cpp:

(WebCore::DisplayRefreshMonitorGtk::~DisplayRefreshMonitorGtk):
(WebCore::onFrameClockUpdate):
(WebCore::DisplayRefreshMonitorGtk::stop):
(WebCore::DisplayRefreshMonitorGtk::startNotificationMechanism):
(WebCore::DisplayRefreshMonitorGtk::stopNotificationMechanism):
(WebCore::DisplayRefreshMonitorGtk::requestRefreshCallback): Deleted.
(WebCore::DisplayRefreshMonitorGtk::displayLinkFired): Deleted.

platform/graphics/gtk/DisplayRefreshMonitorGtk.h:
platform/graphics/ios/DisplayRefreshMonitorIOS.h:
platform/graphics/ios/DisplayRefreshMonitorIOS.mm:

(-[WebDisplayLinkHandler setPaused:]):
(WebCore::DisplayRefreshMonitorIOS::DisplayRefreshMonitorIOS):
(WebCore::DisplayRefreshMonitorIOS::~DisplayRefreshMonitorIOS):
(WebCore::DisplayRefreshMonitorIOS::stop):
(WebCore::DisplayRefreshMonitorIOS::startNotificationMechanism):
(WebCore::DisplayRefreshMonitorIOS::stopNotificationMechanism):
(WebCore::DisplayRefreshMonitorIOS::requestRefreshCallback): Deleted.
(WebCore::DisplayRefreshMonitorIOS::displayLinkFired): Deleted.

platform/graphics/mac/LegacyDisplayRefreshMonitorMac.cpp:

(WebCore::LegacyDisplayRefreshMonitorMac::LegacyDisplayRefreshMonitorMac):
(WebCore::LegacyDisplayRefreshMonitorMac::stop):
(WebCore::LegacyDisplayRefreshMonitorMac::dispatchDisplayDidRefresh):
(WebCore::LegacyDisplayRefreshMonitorMac::startNotificationMechanism):
(WebCore::LegacyDisplayRefreshMonitorMac::stopNotificationMechanism):
(WebCore::LegacyDisplayRefreshMonitorMac::requestRefreshCallback): Deleted.
(WebCore::LegacyDisplayRefreshMonitorMac::displayLinkFired): Deleted.

platform/graphics/mac/LegacyDisplayRefreshMonitorMac.h:
platform/graphics/win/DisplayRefreshMonitorWin.cpp:

(WebCore::DisplayRefreshMonitorWin::DisplayRefreshMonitorWin):
(WebCore::DisplayRefreshMonitorWin::startNotificationMechanism):
(WebCore::DisplayRefreshMonitorWin::stopNotificationMechanism):
(WebCore::DisplayRefreshMonitorWin::requestRefreshCallback): Deleted.
(WebCore::DisplayRefreshMonitorWin::displayLinkFired): Deleted.

platform/graphics/win/DisplayRefreshMonitorWin.h:

Source/WebKit:

Shared/CoordinatedGraphics/threadedcompositor/ThreadedDisplayRefreshMonitor.cpp:
Shared/CoordinatedGraphics/threadedcompositor/ThreadedDisplayRefreshMonitor.h:
UIProcess/mac/DisplayLink.cpp:

(WebKit::DisplayLink::DisplayLink):
(WebKit::DisplayLink::~DisplayLink):
(WebKit::DisplayLink::addObserver):
(WebKit::DisplayLink::notifyObserversDisplayWasRefreshed):

WebProcess/WebPage/RemoteLayerTree/RemoteLayerTreeDisplayRefreshMonitor.h:
WebProcess/WebPage/RemoteLayerTree/RemoteLayerTreeDisplayRefreshMonitor.mm:

(WebKit::RemoteLayerTreeDisplayRefreshMonitor::requestRefreshCallback):

WebProcess/WebPage/mac/DisplayRefreshMonitorMac.cpp:

(WebKit::DisplayRefreshMonitorMac::DisplayRefreshMonitorMac):
(WebKit::DisplayRefreshMonitorMac::~DisplayRefreshMonitorMac):
(WebKit::DisplayRefreshMonitorMac::dispatchDisplayDidRefresh):
(WebKit::DisplayRefreshMonitorMac::startNotificationMechanism):
(WebKit::DisplayRefreshMonitorMac::stopNotificationMechanism):
(WebKit::DisplayRefreshMonitorMac::requestRefreshCallback): Deleted.
(WebKit::DisplayRefreshMonitorMac::displayLinkFired): Deleted.

WebProcess/WebPage/mac/DisplayRefreshMonitorMac.h:

5:40 PM Changeset in webkit [275143] by rniwa@webkit.org

2 edits in trunk/Source/WebCore

Don't add Frame as an opaque root of DOMWindow
https://bugs.webkit.org/show_bug.cgi?id=223838

Reviewed by Simon Fraser.

Don't add Frame as an opaque root of DOMWindow since we don't use Frame
as an opaque root of any JS wrappers since r248276.

bindings/js/JSDOMWindowCustom.cpp:

(WebCore::JSDOMWindow::visitAdditionalChildren):

4:26 PM Changeset in webkit [275142] by ddkilzer@apple.com

2 edits in trunk/Source/WebCore

UBSan: runtime error: load of value <unknown>, which is not a valid value for type 'const WebCore::RealtimeMediaSourceCapabilities::EchoCancellation'
<https://webkit.org/b/223821>

Reviewed by Chris Dumez.

Tests: fast/mediastream/MediaDevices-addEventListener.html

fast/mediastream/constraint-intrinsic-size.html
http/tests/media/media-stream/enumerate-devices-iframe-allow-attribute.html
http/tests/media/media-stream/get-display-media-iframe-allow-attribute.html
imported/w3c/web-platform-tests/mediacapture-record/MediaRecorder-bitrate.https.html
webrtc/addICECandidate-closed.html
webrtc/ice-candidate-sdpMLineIndex.html
webrtc/libwebrtc/descriptionGetters.html
webrtc/peerconnection-page-cache.html

platform/mediastream/RealtimeMediaSourceCapabilities.h:

(WebCore::CapabilityValueOrRange::union ValueUnion):

Make asDouble the first union member so C++ member initializers set all memory.

(WebCore::CapabilityValueOrRange::m_minOrValue):
(WebCore::CapabilityValueOrRange::m_max):
(WebCore::CapabilityValueOrRange::m_type):
(WebCore::RealtimeMediaSourceCapabilities::m_echoCancellation):

Provide member initializers. Setting a default value for m_echoCancellation fixes the bug.

(WebCore::RealtimeMediaSourceCapabilities::encode const):

Drive-by clean-up of encoder statement.

3:34 PM Changeset in webkit [275141] by graouts@webkit.org

7 edits in trunk/Source/WebCore

Fix build with ENABLE_CSS3_TEXT
https://bugs.webkit.org/show_bug.cgi?id=223841

Reviewed by Darin Adler.

We also remove the -webkit- prefix for the "hanging" and "each-line" keywords since
these are well specified by the CSS Text Module Level 3 spec.

css/CSSComputedStyleDeclaration.cpp:

(WebCore::ComputedStyleExtractor::valueForPropertyInStyle):

css/CSSValueKeywords.in:
css/parser/CSSPropertyParser.cpp:

(WebCore::consumeTextIndent):

rendering/style/RenderStyle.h:

(WebCore::RenderStyle::setTextIndentType):
(WebCore::RenderStyle::setTextAlignLast):
(WebCore::RenderStyle::setTextJustify):

rendering/style/StyleRareInheritedData.cpp:

(WebCore::StyleRareInheritedData::StyleRareInheritedData):

style/StyleBuilderCustom.h:

(WebCore::Style::BuilderCustom::applyValueTextIndent):

2:05 PM Changeset in webkit [275140] by Chris Dumez

2 edits in trunk/Source/WTF

wtf/Vector.h:1276:88: runtime error: reference binding to null pointer of type 'const WebCore::Style::RuleFeature'
https://bugs.webkit.org/show_bug.cgi?id=223828

Reviewed by David Kilzer.

Fix issue found by UBSan in Vector::append(const U*, size_t):
wtf/Vector.h:1276:88: runtime error: reference binding to null pointer of type 'const WebCore::Style::RuleFeature'
wtf/Vector.h:1276:88: runtime error: reference binding to null pointer of type 'const WebCore::LayoutRect'
wtf/Vector.h:1276:88: runtime error: reference binding to null pointer of type 'const WTF::RefPtr<WebCore::CSSStyleSheet, WTF::RawPtrTraits<WebCore::CSSStyleSheet>, WTF::DefaultRefDerefTraits<WebCore::CSSStyleSheet> >'

wtf/Vector.h:

(WTF::Malloc>::append):
Return early if dataSize is 0. For some call sites, the pointer may be null and dataSize is 0. In such cases,
we should just return early instead of proceeding and using the null pointer. I believe this can happen if
you call Vector::appendVector() with an empty Vector for example.

1:57 PM Changeset in webkit [275139] by Russell Epstein

8 edits in branches/safari-612.1.7-branch/Source

Versioning.

WebKit-7612.1.7.11

1:54 PM Changeset in webkit [275138] by Russell Epstein

8 edits in branches/safari-612.1.7-branch/Source

Versioning.

WebKit-7612.1.7.10

11:52 AM Changeset in webkit [275137] by Kate Cheney

21 edits in trunk

PCM: Send report to both click source and attribution destination website
https://bugs.webkit.org/show_bug.cgi?id=223615
<rdar://problem/75849443>

Reviewed by Brent Fulgham.

Source/WebCore:

Introduce 2 new structs for storing the earliest time to send and
seconds until send for the source and destination sites.

loader/PrivateClickMeasurement.cpp:

(WebCore::PrivateClickMeasurement::isValid const):
(WebCore::PrivateClickMeasurement::hasPreviouslyBeenReported):
(WebCore::randomlyBetweenTwentyFourAndFortyEightHours):
(WebCore::PrivateClickMeasurement::attributeAndGetEarliestTimeToSend):

loader/PrivateClickMeasurement.h:

(WebCore::PrivateClickMeasurement::AttributionSecondsUntilSendData::hasValidSecondsUntilSendValues):
(WebCore::PrivateClickMeasurement::AttributionSecondsUntilSendData::minSecondsUntilSend):
(WebCore::PrivateClickMeasurement::AttributionSecondsUntilSendData::encode const):
(WebCore::PrivateClickMeasurement::AttributionSecondsUntilSendData::decode):
(WebCore::PrivateClickMeasurement::AttributionTimeToSendData::earliestTimeToSend):
(WebCore::PrivateClickMeasurement::AttributionTimeToSendData::latestTimeToSend):
(WebCore::PrivateClickMeasurement::AttributionTimeToSendData::attributionReportEndpoint):
(WebCore::PrivateClickMeasurement::AttributionTimeToSendData::encode const):
(WebCore::PrivateClickMeasurement::AttributionTimeToSendData::decode):
(WebCore::PrivateClickMeasurement::timesToSend const):
(WebCore::PrivateClickMeasurement::setTimesToSend):
(WebCore::PrivateClickMeasurement::encode const):
(WebCore::PrivateClickMeasurement::decode):
(WebCore::PrivateClickMeasurement::earliestTimeToSend const): Deleted.
(WebCore::PrivateClickMeasurement::setEarliestTimeToSend): Deleted.

Source/WebKit:

NetworkProcess/Classifier/ResourceLoadStatisticsDatabaseStore.cpp:

Move these queries to the correct INSERT OR REPLACE category. Stop
inserting null for earliestTimeToSendToDestination and starting
binding a parameter to it now that we are supporting reports to both
sites.

Now that earliestTimeToSend* can be null if a report has been sent
to a site, we need queries to set the value to null, and also need
to sort attributions by the minimum of either the two reporting times,
or the non-null time if one is null.

(WebKit::ResourceLoadStatisticsDatabaseStore::destroyStatements):
(WebKit::ResourceLoadStatisticsDatabaseStore::buildPrivateClickMeasurementFromDatabase):
(WebKit::ResourceLoadStatisticsDatabaseStore::insertPrivateClickMeasurement):
(WebKit::ResourceLoadStatisticsDatabaseStore::attributePrivateClickMeasurement):
We should not attribute a PCM value if it has already been reported to
either the source or destination. This is covered by checking
secondsUntilSend.hasValidSecondsUntilSendValues() and
previouslyAttributed.value().hasPreviouslyBeenReported() before
inserting anything into the attributed PCM table.

(WebKit::ResourceLoadStatisticsDatabaseStore::earliestTimesToSend):
(WebKit::ResourceLoadStatisticsDatabaseStore::markReportAsSentToSource):
(WebKit::ResourceLoadStatisticsDatabaseStore::markReportAsSentToDestination):
(WebKit::ResourceLoadStatisticsDatabaseStore::clearSentAttribution):
Clear a value from the attributed table only if it has been sent to
both source and destination site. Otherwise, set the corresponding
attribution endpoint to null so we don't send it here again.

(WebKit::ResourceLoadStatisticsDatabaseStore::markAttributedPrivateClickMeasurementsAsExpiredForTesting):
For the sake of testing we can set the destination earliest time to
send to null. We are only confirming here that the expired attribution
gets sent.

NetworkProcess/Classifier/ResourceLoadStatisticsDatabaseStore.h:
NetworkProcess/Classifier/ResourceLoadStatisticsMemoryStore.h:
NetworkProcess/Classifier/ResourceLoadStatisticsStore.h:
NetworkProcess/Classifier/WebResourceLoadStatisticsStore.cpp:

(WebKit::WebResourceLoadStatisticsStore::attributePrivateClickMeasurement):
(WebKit::WebResourceLoadStatisticsStore::clearSentAttribution):

NetworkProcess/Classifier/WebResourceLoadStatisticsStore.h:
NetworkProcess/PrivateClickMeasurementManager.cpp:

(WebKit::PrivateClickMeasurementManager::storeUnattributed):
(WebKit::PrivateClickMeasurementManager::getTokenPublicKey):
We currently have no way of setting the destination token URL site for
testing. To avoid flakiness, we should not make a ping load for the
token URL if we are reporting to the destination in test mode.

(WebKit::PrivateClickMeasurementManager::attribute):
(WebKit::PrivateClickMeasurementManager::fireConversionRequest):
(WebKit::PrivateClickMeasurementManager::fireConversionRequestImpl):
(WebKit::PrivateClickMeasurementManager::clearSentAttribution):
(WebKit::PrivateClickMeasurementManager::firePendingAttributionRequests):

NetworkProcess/PrivateClickMeasurementManager.h:

Tools:

Update API tests to check for a valid time to send for both the source
and destination site.

TestWebKitAPI/Tests/WebCore/PrivateClickMeasurement.cpp:

(TestWebKitAPI::TEST):

LayoutTests:

Layout test coverage.

http/tests/privateClickMeasurement/resources/conversionFilePath.py:
http/tests/privateClickMeasurement/resources/conversionReport.py:
http/tests/privateClickMeasurement/resources/fraudPreventionTestURL.py:
http/tests/privateClickMeasurement/resources/getConversionData.py:
http/tests/privateClickMeasurement/send-attribution-conversion-request-expected.txt:
http/tests/privateClickMeasurement/send-attribution-conversion-request.html:

11:31 AM Changeset in webkit [275136] by commit-queue@webkit.org

2 edits in trunk/Source/WebKit

Remove DisplayRefreshMonitor::handleDisplayRefreshedNotificationOnMainThread()
https://bugs.webkit.org/show_bug.cgi?id=223837

Patch by Tyler Wilcock <Tyler Wilcock> on 2021-03-27
Reviewed by Simon Fraser.

Fixes build for GTK + WPE after
https://trac.webkit.org/changeset/275134/webkit.

Quoting that revision:

handleDisplayRefreshedNotificationOnMainThread() is an anachronism left over from
the non-main-thread nature of the CVDisplayLink callback. There's no need to burden
all subclasses of DisplayRefreshMonitor with that detail.

Shared/CoordinatedGraphics/threadedcompositor/ThreadedDisplayRefreshMonitor.cpp:

(WebKit::ThreadedDisplayRefreshMonitor::invalidate):
(WebKit::ThreadedDisplayRefreshMonitor::displayRefreshCallback):
Replace call to DisplayRefreshMonitor::handleDisplayRefreshedNotificationOnMainThread(this)
with displayDidRefresh().

10:15 AM Changeset in webkit [275135] by ddkilzer@apple.com

2 edits in trunk/Source/WebKit

Fix leaks in WKInspectorResourceURLSchemeHandler
<https://webkit.org/b/223824>

Reviewed by Joseph Pecoraro.

UIProcess/Inspector/mac/WKInspectorResourceURLSchemeHandler.mm:

(-[WKInspectorResourceURLSchemeHandler webView:startURLSchemeTask:]):

Make use of adoptNS() when assigning a newly allocated, +1 retained Objective-C object to a RetainPtr<>.
Also fix leak of local NSHTTPURLResponse object by using a RetainPtr<>.

9:01 AM Changeset in webkit [275134] by Simon Fraser

10 edits in trunk/Source

Remove DisplayRefreshMonitor::handleDisplayRefreshedNotificationOnMainThread()
https://bugs.webkit.org/show_bug.cgi?id=223837

Reviewed by Tim Horton.
Source/WebCore:

handleDisplayRefreshedNotificationOnMainThread() is an anachronism left over from
the non-main-thread nature of the CVDisplayLink callback. There's no need to burden
all subclasses of DisplayRefreshMonitor with that detail.

platform/graphics/DisplayRefreshMonitor.cpp:

(WebCore::DisplayRefreshMonitor::displayDidRefresh):
(WebCore::DisplayRefreshMonitor::handleDisplayRefreshedNotificationOnMainThread): Deleted.

platform/graphics/DisplayRefreshMonitor.h:
platform/graphics/gtk/DisplayRefreshMonitorGtk.cpp:

(WebCore::DisplayRefreshMonitorGtk::displayLinkFired):

platform/graphics/ios/DisplayRefreshMonitorIOS.mm:

(WebCore::DisplayRefreshMonitorIOS::displayLinkFired):

platform/graphics/mac/LegacyDisplayRefreshMonitorMac.cpp:

(WebCore::LegacyDisplayRefreshMonitorMac::displayLinkFired):

platform/graphics/win/DisplayRefreshMonitorWin.cpp:

(WebCore::DisplayRefreshMonitorWin::displayLinkFired):

Source/WebKit:

WebProcess/WebPage/RemoteLayerTree/RemoteLayerTreeDisplayRefreshMonitor.mm:

(WebKit::RemoteLayerTreeDisplayRefreshMonitor::didUpdateLayers):

WebProcess/WebPage/mac/DisplayRefreshMonitorMac.cpp:

(WebKit::DisplayRefreshMonitorMac::displayLinkFired):

6:51 AM Changeset in webkit [275133] by Alan Bujtas

4 edits
2 adds in trunk

[Multicolumn] Do not try to re-validate a multicol spanner when the renderer is moved internally
https://bugs.webkit.org/show_bug.cgi?id=223836
<rdar://75742694>

Reviewed by Ryosuke Niwa.

Source/WebCore:

When a renderer becomes a multicol spanner

it is moved out of its original tree position and placed as a sibling of the enclosing multicolumn flow and
a spanner placeholder is constructed and inserted at the original position

This patch fixes the case when an already placed multicol spanner is internally moved (e.g. collapsing an anonymous block parent)
and we attempt to re-insert this spanner to the multicolumn flow. No spanner state should change due to internal re-parenting.

Test: fast/multicol/spanner-get-re-added-on-move-crash.html

rendering/updating/RenderTreeBuilder.cpp:

(WebCore::RenderTreeBuilder::attachToRenderElementInternal):

LayoutTests:

TestExpectations:
fast/multicol/spanner-get-re-added-on-move-crash-expected.txt: Added.
fast/multicol/spanner-get-re-added-on-move-crash.html: Added.

3:52 AM Changeset in webkit [275132] by Philippe Normand

3 edits in trunk

REGRESSION(r275111) [GLIB] Fix build with new derived sources and forwarding headers scheme
https://bugs.webkit.org/show_bug.cgi?id=223834

Unreviewed build fix.

Source/cmake/OptionsGTK.cmake:
Source/cmake/OptionsWPE.cmake:

Mar 26, 2021:

9:47 PM Changeset in webkit [275131] by ddkilzer@apple.com

2 edits in trunk/Source/WebKitLegacy/mac

WebFeature class fails to copy/retain its instance variables
<https://webkit.org/b/223823>

Reviewed by Sam Weinig.

WebView/WebFeature.m:

(-[WebFeature initWithKey:preferenceKey:name:details:defaultValue:hidden:]):

Copy (and thereby retain) instance variables, fulfilling the promise of its @property declarations and preventing use-after-release issues.

(-[WebFeature dealloc]): Add.

Release the instance variables copied in the constructor.

9:24 PM Changeset in webkit [275130] by commit-queue@webkit.org

2 edits in trunk/Source/WebCore

Dirty layout for floating children of inline on full layout
https://bugs.webkit.org/show_bug.cgi?id=223660

Patch by Ian Gilbert <iang@apple.com> on 2021-03-26
Reviewed by Zalan Bujtas.

Avoid consuming an object on an empty line when that object is floating or out-of-flow.

rendering/line/BreakingContext.h:

(WebCore::BreakingContext::handleEndOfLine):

8:52 PM Changeset in webkit [275129] by Chris Dumez

2 edits in trunk/Source/WebCore

Avoid heap allocation on the audio thread in BaseAudioContext::scheduleNodeDeletion()
https://bugs.webkit.org/show_bug.cgi?id=223825

Reviewed by Darin Adler.

Avoid heap allocation on the audio thread in BaseAudioContext::scheduleNodeDeletion()
for performance. Stop calling appendVector() on m_nodesToDelete since m_nodesToDelete
is always empty. We can use std::exchange() instead of appendVector() and clear() to
avoid the heap allocation.

The reason m_nodesToDelete is always empty is that BaseAudioContext::scheduleNodeDeletion()
relies on the m_isDeletionScheduled flag to avoid adding to m_nodesToDelete until
m_nodesToDelete is cleared on the main thread and the m_isDeletionScheduled gets reset.

Modules/webaudio/BaseAudioContext.cpp:

(WebCore::BaseAudioContext::clear):
(WebCore::BaseAudioContext::scheduleNodeDeletion):

8:27 PM Changeset in webkit [275128] by Patrick Angle

5 edits in trunk/Source

Web Inspector: Grid layout labels can be drawn outside the viewport
https://bugs.webkit.org/show_bug.cgi?id=221972

Reviewed by BJ Burg.

Source/WebCore:

Added logic for a best-effort attempt to make sure that layout labels are drawn within the document's bounds, or
at least within the grid itself. Labels are measured and adjusted so that if they would be drawn outside the
document's bounds, they will be pulled inside the grid. This does not guarantee that all labels will be visible
all the time. It is still possible an entire side of a grid will be outside the document's bounds, and this does
not attempt to correct for this case, as the desired anchor points for labels will be outside of the visible
area of the document.

inspector/InspectorOverlay.cpp:

(WebCore::InspectorOverlay::fontForLayoutLabel):

Added helper that creates the FontCascade for layout labels, which is done in a few different places.

(WebCore::InspectorOverlay::backgroundPathForLayoutLabel):

Added helper to create the Path for the background of layout labels, which is now used in WebKit::WKInspectorHighlightView.

(WebCore::expectedSizeForLayoutLabel):

Gets the expected size of the label based on the text and arrow direction, and does so without needing to

create the entire label background's path.
(WebCore::InspectorOverlay::drawLayoutLabel):

Support new LabelArrowEdgePosition property

(WebCore::InspectorOverlay::drawGridOverlay):
(WebCore::buildLabel):
(WebCore::InspectorOverlay::buildGridOverlay):

inspector/InspectorOverlay.h:

(WebCore::InspectorOverlay::Highlight::GridHighlightOverlay::Label::encode const):
(WebCore::InspectorOverlay::Highlight::GridHighlightOverlay::Label::decode):

Source/WebKit:

Add support for the new WebCore::InspectorOverlay::LabelArrowEdgePosition property to grid overlays on iOS.

UIProcess/Inspector/ios/WKInspectorHighlightView.mm:

(createLayoutLabelLayer):

Support the new WebCore::InspectorOverlay::LabelArrowEdgePosition.
Use new helpers in WebCore::InspectorOverlay to reduce code duplication.

(-[WKInspectorHighlightView _createGridOverlayLayer:scale:]):

8:09 PM Changeset in webkit [275127] by Lauro Moura

4 edits in trunk

REGRESSION(r275111) [GLIB] Fix build with new derived sources and forwarding headers scheme
https://bugs.webkit.org/show_bug.cgi?id=223834

Unreviewed build fix.

Buildfix for now. Actual removal of older scheme will come in the
future.

Source/cmake/OptionsWPE.cmake: Add some missing variables with the

per framework source and header dirs.

Tools:

TestWebKitAPI/glib/CMakeLists.txt: Add some missing variables with

needed include headers.

6:37 PM Changeset in webkit [275126] by Alan Bujtas

3 edits in trunk/Source/WebCore

[RenderTreeBuilder] No need to update the counters when the renderer is moved internally
https://bugs.webkit.org/show_bug.cgi?id=223801
<rdar://75835208>

Reviewed by Ryosuke Niwa.

When the renderer re-parenting is initiated by an "internal move" (e.g. anonymous block parent is not needed anymore so all the descendants
are moved "one level" up), we don't need to re-compute the counters. Their state should not be any different after this re-parenting.

rendering/updating/RenderTreeBuilder.cpp:

(WebCore::RenderTreeBuilder::attachToRenderElementInternal):
(WebCore::RenderTreeBuilder::move):

rendering/updating/RenderTreeBuilder.h:

6:00 PM Changeset in webkit [275125] by Chris Dumez

2 edits in trunk/Source/WTF

REGRESSION (r275038?): Perf test IndexedDB/stress/large-binary-keys.html failing with logging about memory pressure events
https://bugs.webkit.org/show_bug.cgi?id=223826
<rdar://problem/75906134>

Reviewed by Darin Adler.

Use RELEASE_LOG() instead of WTFLogAlways() to log memory pressure events, so that they don't
show on stdout in the performance tests and make them fail.

wtf/cocoa/MemoryPressureHandlerCocoa.mm:

(WTF::MemoryPressureHandler::install):

5:56 PM Changeset in webkit [275124] by Robert Jenner

2 edits in trunk/LayoutTests

[ macOS ] inspector/debugger/csp-exceptions.html is a flakey text failure
https://bugs.webkit.org/show_bug.cgi?id=223830

Unreviewed test gardening.

platform/mac/TestExpectations: Updated test expectations to Pass Failure Timeout while test is being reviewed. I added Timeout to the expectation because while the failure is a text failure, the text failure indicates that it's a timeout.

5:02 PM Changeset in webkit [275123] by commit-queue@webkit.org

2 edits in trunk/Tools

Fix TypeError: Can't mix strings and bytes in path components in Tools/Scripts/dump-class-layout
https://bugs.webkit.org/show_bug.cgi?id=223819

Patch by Tyler Wilcock <Tyler Wilcock> on 2021-03-26
Reviewed by Jonathan Bedard.

Prior to this patch, I received this error when trying to run the
dump-class-layout script. This happens because webkit_build_dir() returns
bytes rather than a string.

$ dump-class-layout -c Release WebCore Pair
Traceback (most recent call last):

File "/home/twilco/projects/webkit/Tools/Scriptsdump-class-layout", line 88, in <module>

main()

File "/home/twilco/projects/webkit/Tools/Scriptsdump-class-layout", line 80, in main

target_path = os.path.join(build_dir, args.config, args.framework + ".framework", args.framework);

File "/usr/lib/python3.9/posixpath.py", line 90, in join

genericpath._check_arg_types('join', a, *p)

File "/usr/lib/python3.9/genericpath.py", line 155, in _check_arg_types

raise TypeError("Can't mix strings and bytes in path components") from None

TypeError: Can't mix strings and bytes in path components

Scripts/dump-class-layout:

Update webkit_build_dir() to decode its output into a UTF-8 string.

4:47 PM Changeset in webkit [275122] by Jonathan Bedard

7 edits in trunk/Tools

System installing webkitcorepy can break certain scripts
https://bugs.webkit.org/show_bug.cgi?id=223822
<rdar://problem/75905163>

Rubber-stamped by Alan Coon.

Scripts/libraries/resultsdbpy/resultsdbpy/example/environment.py:
Scripts/test-lldb-webkit: Replace sys.path.append with sys.path.insert.
Scripts/webkitpy/init.py: Only put webkitcorepy path into sys.path if it isn't already there.
Scripts/webkitpy/benchmark_runner/http_server_driver/http_server/twisted_http_server.py:

Replace sys.path.append with sys.path.insert.

Scripts/webkitpy/common/system/executive_unittest.py: Ditto.
Scripts/webkitpy/port/config_standalone.py: Ditto.

4:32 PM Changeset in webkit [275121] by Chris Fleizach

2 edits in trunk/Source/WebCore

AX: crash at aom.apple.WebCore: WebCore::AccessibilityProgressIndicator::element const
https://bugs.webkit.org/show_bug.cgi?id=223800
rdar://75861367

Reviewed by Zalan Bujtas.

Prevent unintended renderer access.
No steps to reproduce were available, so a test was not able to be created.

accessibility/AccessibilityProgressIndicator.cpp:

(WebCore::AccessibilityProgressIndicator::element const):

4:12 PM Changeset in webkit [275120] by Wenson Hsieh

10 edits
2 adds in trunk

Use PUICQuickboardController for text input when HAVE(QUICKBOARD_CONTROLLER) is defined
https://bugs.webkit.org/show_bug.cgi?id=223809
<rdar://problem/56730407>

Reviewed by Tim Horton.

Source/WebKit:

Enable PUICQuickboardController by default, and remove a runtime preference that previously disabled this
feature by default when HAVE(QUICKBOARD_CONTROLLER) is defined.

UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView _updateTextInputTraits:]):

Add a couple of missing selector checks before attempting to configure smart quotes and smart dash editing
behaviors.

(-[WKContentView _createQuickboardController:]):

Additionally opt into a quickboard controller API to bypass the interstitial Quickboard view controller when
using PUICQuickboardController.

(-[WKContentView presentViewControllerForCurrentFocusedElement]):

Remove a check for the (now-removed) preference.

Source/WTF:

Remove the QuickboardControllerForTextInputEnabled WebKit preference, now that the feature is enabled by
default behind a build-time flag.

Scripts/Preferences/WebPreferencesInternal.yaml:

LayoutTests:

Remove some extraneous instancees of QuickboardControllerForTextInputEnabled=true from watch-specific layout
tests, now that the feature is enabled by default.

fast/forms/watchos/delete-content-in-text-field.html:
fast/forms/watchos/edit-text-field-calls-injected-bundle.html:
fast/forms/watchos/enter-text-with-spellcheck-disabled-expected.txt: Added.
fast/forms/watchos/enter-text-with-spellcheck-disabled.html: Added.

Add a new test to exercise the missing selector checks in -[WKContentView _updateTextInputTraits:].

fast/forms/watchos/form-control-label-text.html:
fast/forms/watchos/time-picker-value-change.html:
fast/forms/watchos/username-text-content-type.html:

4:10 PM Changeset in webkit [275119] by Jonathan Bedard

5 edits in trunk/Tools

[webkit-patch] Use commit.webkit.org instead of remote SVN to compute identifier
https://bugs.webkit.org/show_bug.cgi?id=223820
<rdar://problem/75902125>

Reviewed by Dewei Zhu.

Scripts/webkitpy/tool/commands/download_unittest.py:

(DownloadCommandsTest.mock_svn_remote): Replace mock Svn with mock requests.

Scripts/webkitpy/tool/commands/upload_unittest.py:
Scripts/webkitpy/tool/comments.py:

(bug_comment_from_svn_revision): Use commits.webkit.org instead of svn.webkit.org.

Scripts/webkitpy/tool/steps/closebugforlanddiff_unittest.py:

(CloseBugForLandDiffTest.test_empty_state): Replace mock Svn with mock requests.

3:41 PM Changeset in webkit [275118] by Chris Dumez

2 edits in trunk/Source/WebCore

Source/WebCore/editing/cocoa/HTMLConverter.mm:1278:44: runtime error: member call on null pointer of type 'WebCore::ArchiveResource'
https://bugs.webkit.org/show_bug.cgi?id=223696

Reviewed by Alex Christensen.

Fix bug found by UBSan.

editing/cocoa/HTMLConverter.mm:

(HTMLConverter::_addAttachmentForElement):

Drop duplicate call to dataSource->subresource(url) as it doesn't seem it can return something else than null if it returned null the first time around.
Make sure resource is null-checked before we dereference it to get the MIME type.

3:18 PM Changeset in webkit [275117] by Alan Coon

1 copy in tags/Safari-612.1.8

Tag Safari-612.1.8.

3:05 PM Changeset in webkit [275116] by beidson@apple.com

25 edits in trunk/Source/WebKit

Refactor NetworkSessionCocoa to prepare for per-WebPageProxy sessions
https://bugs.webkit.org/show_bug.cgi?id=223776

Reviewed by Alex Christensen.

No new tests (No behavior change refactor)

A NetworkSessionCocoa has (n) NSURLSessions associated with it.

When a network load needs to happen - based on the context for that load - the appropriate
NSURLSession is chosen (or created).

In a future patch, there will be an additional bit of context used for choosing the appropriate
NSURLSession - The WKWebView the load was triggered from.

This patch takes those (n) NSURLSessions and wraps them in a SessionSet object.
It also adds the concept of looking up the appropriate SessionSet based on the WebPageProxyIdentifier
triggering the load.

This patch by itself should not be a behavior change, but allows the followup to be much simpler.

NetworkProcess/Downloads/cocoa/DownloadCocoa.mm:

(WebKit::Download::resume):

NetworkProcess/NetworkConnectionToWebProcess.cpp:

(WebKit::NetworkConnectionToWebProcess::createSocketChannel):
(WebKit::NetworkConnectionToWebProcess::establishSWContextConnection):

NetworkProcess/NetworkConnectionToWebProcess.h:
NetworkProcess/NetworkConnectionToWebProcess.messages.in:

NetworkProcess/NetworkDataTask.cpp:

(WebKit::NetworkDataTask::create):

NetworkProcess/NetworkSession.cpp:

(WebKit::NetworkSession::createWebSocketTask):

NetworkProcess/NetworkSession.h:

(WebKit::NetworkSession::removeWebSocketTask):
(WebKit::NetworkSession::addWebSocketTask):

NetworkProcess/NetworkSocketChannel.cpp:

(WebKit::NetworkSocketChannel::create):
(WebKit::NetworkSocketChannel::NetworkSocketChannel):
(WebKit::NetworkSocketChannel::~NetworkSocketChannel):

NetworkProcess/NetworkSocketChannel.h:

NetworkProcess/ServiceWorker/WebSWServerToContextConnection.cpp:

(WebKit::WebSWServerToContextConnection::WebSWServerToContextConnection):

NetworkProcess/ServiceWorker/WebSWServerToContextConnection.h:

NetworkProcess/cocoa/NetworkDataTaskCocoa.h:
NetworkProcess/cocoa/NetworkDataTaskCocoa.mm:

(WebKit::NetworkDataTaskCocoa::NetworkDataTaskCocoa):

NetworkProcess/cocoa/NetworkSessionCocoa.h:
NetworkProcess/cocoa/NetworkSessionCocoa.mm:

(WebKit::NetworkSessionCocoa::hstsStorage const):
(WebKit::NetworkSessionCocoa::NetworkSessionCocoa):
(WebKit::NetworkSessionCocoa::initializeStandardSessionsInSet):
(WebKit::NetworkSessionCocoa::sessionSetForPage):
(WebKit::NetworkSessionCocoa::sessionSetForPage const):
(WebKit::NetworkSessionCocoa::initializeEphemeralStatelessSessionIfNeeded):
(WebKit::NetworkSessionCocoa::SessionSet::initializeEphemeralStatelessSessionIfNeeded):
(WebKit::NetworkSessionCocoa::sessionWrapperForTask):
(WebKit::NetworkSessionCocoa::appBoundSession):
(WebKit::NetworkSessionCocoa::hasAppBoundSession const):
(WebKit::NetworkSessionCocoa::clearAppBoundSession):
(WebKit::NetworkSessionCocoa::isolatedSession):
(WebKit::NetworkSessionCocoa::SessionSet::isolatedSession):
(WebKit::NetworkSessionCocoa::hasIsolatedSession const):
(WebKit::NetworkSessionCocoa::clearIsolatedSessions):
(WebKit::NetworkSessionCocoa::invalidateAndCancelSessionSet):
(WebKit::NetworkSessionCocoa::invalidateAndCancel):
(WebKit::NetworkSessionCocoa::createWebSocketTask):
(WebKit::NetworkSessionCocoa::addWebSocketTask):
(WebKit::NetworkSessionCocoa::removeWebSocketTask):
(WebKit::NetworkSessionCocoa::hostNamesWithAlternativeServices const):
(WebKit::NetworkSessionCocoa::deleteAlternativeServicesForHostNames):
(WebKit::NetworkSessionCocoa::clearAlternativeServices):
(WebKit::NetworkSessionCocoa::initializeEphemeralStatelessSession): Deleted.

NetworkProcess/soup/NetworkSessionSoup.cpp:

(WebKit::NetworkSessionSoup::createWebSocketTask):

NetworkProcess/soup/NetworkSessionSoup.h:

Shared/AuxiliaryProcess.cpp:

(WebKit::AuxiliaryProcess::initialize):

WebProcess/Network/WebSocketChannel.cpp:

(WebKit::WebSocketChannel::create):
(WebKit::WebSocketChannel::WebSocketChannel):
(WebKit::WebSocketChannel::connect):

WebProcess/Network/WebSocketChannel.h:

WebProcess/Network/WebSocketProvider.cpp:

(WebKit::WebSocketProvider::createWebSocketChannel):

WebProcess/Network/WebSocketProvider.h:

WebProcess/Storage/WebSWContextManagerConnection.cpp:

(WebKit::WebSWContextManagerConnection::establishConnection):
(WebKit::WebSWContextManagerConnection::installServiceWorker):

WebProcess/WebPage/WebPage.cpp:

(WebKit::m_lastNavigationWasAppBound):

3:00 PM Changeset in webkit [275115] by ysuzuki@apple.com

5 edits
1 add in trunk

[JSC] Use AppleICU SPI for canonicalization
https://bugs.webkit.org/show_bug.cgi?id=223552

Reviewed by Ryosuke Niwa.

JSTests:

stress/intl-canonical-locale-alias-mapping.js: Added.

(shouldBe):
(Intl.getCanonicalLocales):

Source/JavaScriptCore:

uloc_canonicalize does not perform alias mapping. This is different from ECMA402's canonicalization requirement.
ICU C++ icu::Locale can canonicalize locale ID with alias mapping, but this is not exposed to C API.

In this patch, we adopt AppleICU SPI "ualoc_canonicalForm" added in rdar://74314220. This canonicalization can perform
alias mapping too. We do not extend uloc_canonicalize since this API explicitly says "It does NOT map aliased names in any way."[1].
In [2], we are tracking upstreaming of this new SPI. Once it is upstreamed to the mainline ICU, we will switch to that.

[1]: https://unicode-org.github.io/icu-docs/apidoc/dev/icu4c/uloc_8h.html#a69b148194cf57ac40d4bb15c5b905260
[2]: https://unicode-org.atlassian.net/browse/ICU-21506

runtime/IntlLocale.cpp:

(JSC::LocaleIDBuilder::initialize):
(JSC::LocaleIDBuilder::toCanonical):

runtime/IntlObject.cpp:

(JSC::localeIDBufferForLanguageTagWithNullTerminator):
(JSC::canonicalizeLanguageTag):
(JSC::canonicalizeLocaleIDWithoutNullTerminator):
(JSC::localeIDBufferForLanguageTag): Deleted.

runtime/IntlObject.h:

2:28 PM Changeset in webkit [275114] by Adrian Perez de Castro

1 copy in releases/WPE WebKit/webkit-2.32.0

WPE WebKit 2.32.0

2:27 PM Changeset in webkit [275113] by Adrian Perez de Castro

4 edits in releases/WebKitGTK/webkit-2.32

Unreviewed. Update OptionsWPE.cmake and NEWS for the 2.32.0 release

Source/cmake/OptionsWPE.cmake: Bump version numbers.

Source/WebKit:

wpe/NEWS: Add release notes for 2.32.0

2:23 PM Changeset in webkit [275112] by Wenson Hsieh

12 edits
4 adds in trunk

Allow some image overlay content to render in fully transparent image elements
https://bugs.webkit.org/show_bug.cgi?id=223781
<rdar://problem/75886351>

Reviewed by Tim Horton.

Source/WebCore:

Introduce ImageOverlayController, and use it to render selections in image overlay content when the image
overlay's host element is fully transparent.

Test: fast/images/image-extraction/mac/select-word-in-transparent-image-overlay.html

Sources.txt:
WebCore.xcodeproj/project.pbxproj:
dom/Document.cpp:

(WebCore::Document::willBeRemovedFromFrame):

Add plumbing to allow ImageOverlayController to uninstall its PageOverlay when the document is about to be
detached. See ImageOverlayController:: below.

editing/SelectionRectGatherer.cpp:

(WebCore::SelectionRectGatherer::Notifier::~Notifier):

Have SelectionRectGatherer notify the image overlay controller as well, when selection rects change.

page/ChromeClient.h:

(WebCore::ChromeClient::needsImageOverlayControllerForSelectionPainting const):

Add a new client hook that returns whether or not we should use ImageOverlayController to paint selections.
If not (i.e. we're on iOS, where we use UIKit to draw selections in the UI process), then we'll never install
page overlays, since selectionRectsDidChange is effectively a no-op.

page/ImageOverlayController.cpp: Added.

(WebCore::ImageOverlayController::ImageOverlayController):
(WebCore::ImageOverlayController::selectionRectsDidChange):

When selection rects change, if the selection is inside an image overlay whose host is completely transparent
(or very close to being completely transparent), then remember the selection rects along with the renderer's
background color, and use this to render selection highlights separately, in a PageOverlay.

(WebCore::ImageOverlayController::documentDetached):

If the Document containing the current rendered overlay selection is detached, immediately uninstall the
current PageOverlay. This ensures that the overlays don't persist through navigation.

(WebCore::ImageOverlayController::installPageOverlayIfNeeded):
(WebCore::ImageOverlayController::uninstallPageOverlayIfNeeded):

Helper methods to add or remove the PageOverlay from the current Page.

(WebCore::ImageOverlayController::willMoveToPage):

Uninstall the current page overlay if needed.

(WebCore::ImageOverlayController::drawRect):

Use the information stored in ImageOverlayController::selectionRectsDidChange to render custom selection
rects.

page/ImageOverlayController.h: Added.
page/Page.cpp:
page/Page.h:

Add an ImageOverlayController to the Page.

(WebCore::Page::imageOverlayController):

Source/WebKit:

Implement a new chrome client hook. See WebCore ChangeLog for more information.

WebProcess/WebCoreSupport/WebChromeClient.h:

LayoutTests:

Adjust an existing layout test so that it actually checks what it was intended to check; add a new layout test
to cover the changes in this bug.

fast/images/image-extraction/mac/select-word-in-draggable-image-overlay.html:

Drive-by fix: remove some extraneous imported scripts from this layout test, and additionally simplify the test
so that it doesn't rely on event sender to select text.

fast/images/image-extraction/mac/select-word-in-transparent-image-overlay-expected-mismatch.html: Added.
fast/images/image-extraction/mac/select-word-in-transparent-image-overlay.html: Added.

Add a new layout test with a fully transparent image element that covers a div of the same size, with a
background image that is identical to the image element's image. This emulates the behavior of certain websites
that prompted this fallback image overlay rendering codepath in the first place.

1:37 PM Changeset in webkit [275111] by don.olmstead@sony.com

27 edits in trunk

[CMake] Deprecate using DERIVED_SOURCES_DIR/FOWARDING_HEADERS_DIR directly
https://bugs.webkit.org/show_bug.cgi?id=223763

Reviewed by Michael Catanzaro.

Remove any usages of DERIVED_SOURCES_DIR and FOWARDING_HEADERS_DIR. There are still some
uses in GTK/WPE but those should be phased out.

Sets the directory structure to look like an Apple build in terms of where headers and
derived sources are held. Rather than having one root derived sources and one root
header directory those directories are now per project. This helps catch any errors with
header includes.

Deletes some old Mac CMake code that is no longer relevant around creating forwarding
headers.

For AppleWin the old structure is used because of its internal build which assumes the
old structure.

Source/cmake/OptionsAppleWin.cmake:
Source/cmake/OptionsFTW.cmake:
Source/cmake/OptionsGTK.cmake:
Source/cmake/OptionsPlayStation.cmake:
Source/cmake/OptionsWPE.cmake:
Source/cmake/OptionsWinCairo.cmake:
Source/cmake/WebKitCommon.cmake:
Source/cmake/WebKitFS.cmake:
Source/cmake/WebKitMacros.cmake:

Source/JavaScriptCore: