Timeline

Apr 19, 2021:

11:33 PM Changeset in webkit [276287] by Antti Koivisto

• 16 edits in trunk

Render tree updates for Text node content mutations should happen during rendering update
​https://bugs.webkit.org/show_bug.cgi?id=222406
<rdar://problem/74822830>

Reviewed by Simon Fraser.

Source/WebCore:

Calls to Text.insertData and similar should not mutate render tree synchronously.
Instead render tree should be updated during the next rendering update along with
any style changes.

These updates already go via RenderTreeUpdater. We just need to save the information
about which nodes need updating so the next rendering update can pick them up.

This seems to help with some performance benchmarks.

• dom/CharacterData.cpp:

(WebCore::CharacterData::parserAppendData):
(WebCore::CharacterData::setDataAndUpdate):

Move in-tree check to the updateRendererAfterContentChange and make it use isConnected().

• dom/Document.cpp:

(WebCore::Document::removedLastRef):
(WebCore::Document::resolveStyle):

Include the text update when updating the render tree.

(WebCore::Document::updateTextRenderer):

Create a text update that will get flushed during the next rendering update.

(WebCore::Document::needsStyleRecalc const):

We need to recalc if there are pending text updates.

• dom/Document.h:
• dom/Text.cpp:

(WebCore::Text::splitText):

Use updateRendererAfterContentChange instead of poking render tree directly.

(WebCore::Text::updateRendererAfterContentChange):

• rendering/updating/RenderTreeUpdater.cpp:
• style/StyleTreeResolver.cpp:

(WebCore::Style::TreeResolver::TreeResolver):
(WebCore::Style::TreeResolver::resolve):

• style/StyleTreeResolver.h:

(WebCore::Style::TreeResolver::TreeResolver):

• style/StyleUpdate.cpp:

(WebCore::Style::Update::addText):

Merge text updates.

(WebCore::Style::Update::addPossibleRoot):

• style/StyleUpdate.h:

(WebCore::Style::Update::roots const):
(WebCore::Style::Update:: const): Deleted.

Refcount the nodes since this now has longer lifetime.

• editing/CompositeEditCommand.cpp:

(WebCore::CompositeEditCommand::deleteInsignificantText):

Ensure we leave deleteInsignificantText with updated render tree. Clients expect that.
Do layout (instead of just style update) for consistency, deleteInsignificantText does one anyway in beginning.

• editing/markup.cpp:

(WebCore::replaceChildrenWithFragment):

Pending text update may ref the node so this refcount assert is not correct.

• style/StyleUpdate.cpp:

(WebCore::Style::Update::addText):

LayoutTests:

These are progressions.

• fast/text/splitText-dirty-lines-expected.txt:
• fast/text/text-combine-surroundContents-crash-expected.txt:
• imported/blink/fast/css/first-letter-range-insert-expected.txt:

Here we were actually drawing text that didn't exist in DOM anymore.

10:19 PM Changeset in webkit [276286] by commit-queue@webkit.org

• 2 edits in trunk/Source/ThirdParty/ANGLE

Build ANGLE dylib into WK_OVERRIDE_FRAMEWORKS_DIR in builds that use it
​https://bugs.webkit.org/show_bug.cgi?id=224785
<rdar://76641662>

Patch by Alex Christensen <​achristensen@webkit.org> on 2021-04-19
Reviewed by Alexey Proskuryakov.

• Configurations/ANGLE-dynamic.xcconfig:

10:15 PM Changeset in webkit [276285] by ysuzuki@apple.com

• 8 edits in trunk/JSTests

[JSC] Rebaseline test results for new ICU
​https://bugs.webkit.org/show_bug.cgi?id=224792

Reviewed by Mark Lam.

This patch updates some intl- tests' expectation since it is changed because of ICU CLDR data change.

• stress/intl-datetimeformat-formatrange-relevant-extensions-ja.js:

(shouldBeOneOfThem):
(vm.icuVersion):

• stress/intl-datetimeformat-formatrange-relevant-extensions.js:

(shouldBeOneOfThem):

• stress/intl-datetimeformat-formatrange-should-not-handle-gregorian-change-date.js:

(shouldBe):
(vm.icuHeaderVersion):

• stress/intl-datetimeformat-formatrangetoparts-relevant-extensions-ja.js:

(normalize):
(shouldBe):
(compareParts):
(shouldBeOneOfParts):
(shouldBeParts):
(Intl.DateTimeFormat.prototype.formatRangeToParts.shouldBeOneOfParts.fmt5.formatRangeToParts):
(Intl.DateTimeFormat.prototype.formatRangeToParts.vm.icuVersion):
(Intl.DateTimeFormat.prototype.formatRangeToParts.shouldBeOneOfParts.fmt7.formatRangeToParts):
(Intl.DateTimeFormat.prototype.formatRangeToParts.shouldBeOneOfParts.fmt9.formatRangeToParts):
(Intl.DateTimeFormat.prototype.formatRangeToParts.shouldBeOneOfParts.fmt11.formatRangeToParts):
(Intl.DateTimeFormat.prototype.formatRangeToParts.shouldBeOneOfParts.fmt13.formatRangeToParts):
(Intl.DateTimeFormat.prototype.formatRangeToParts.shouldBeParts.fmt13.formatRangeToParts):
(Intl.DateTimeFormat.prototype.formatRangeToParts.shouldBeOneOfParts.fmt14.formatRangeToParts):
(Intl.DateTimeFormat.prototype.formatRangeToParts.shouldBeOneOfParts.fmt15.formatRangeToParts):
(Intl.DateTimeFormat.prototype.formatRangeToParts.shouldBeOneOfParts.fmt16.formatRangeToParts):
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt1.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt2.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt3.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt4.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt5.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt6.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt7.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt8.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt9.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt10.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt11.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt12.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt13.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt14.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt15.formatRangeToParts): Deleted.
(Intl.DateTimeFormat.prototype.formatRangeToParts.compareParts.fmt16.formatRangeToParts): Deleted.

• stress/intl-datetimeformat-formatrangetoparts-relevant-extensions.js:

(normalize):
(shouldBe):
(compareParts):
(shouldBeOneOfParts):
(shouldBeParts):

• stress/intl-datetimeformat-formatrangetoparts-should-not-handle-gregorian-change-date.js:

(shouldBe):

• stress/intl-datetimeformat.js:

(shouldBeOneOfThem):

9:49 PM Changeset in webkit [276284] by Chris Dumez

• 5 edits in trunk

REGRESSION (r276189): GPUProcess.WebProcessTerminationAfterTooManyGPUProcessCrashes is crashing
​https://bugs.webkit.org/show_bug.cgi?id=224790
<rdar://problem/76869318>

Reviewed by Darin Adler.

Source/WebKit:

GPUProcess.WebProcessTerminationAfterTooManyGPUProcessCrashes is intentionally repeatedly
killing the GPUProcess. As a result, the GPUProcess may get killed very shortly after
a relaunch and the RemoteAudioDestinationManager::StartAudioDestination synchronous IPC
may fail if it is ongoing at the time of the crash. This would cause m_destinationID to
not get initialized and then get sent as IPC parameter, thus crashing.

pre-r276189, we were not crashing because m_destinationID was not reset on crash and we
would thus send IPC for a destination that does not exist but at least the destinationID
would not be 0. This patch makes sure we don't try and send IPC at all when m_destinationID
is 0.

• WebProcess/GPU/media/RemoteAudioDestinationProxy.cpp:

(WebKit::RemoteAudioDestinationProxy::connection):
(WebKit::RemoteAudioDestinationProxy::startRendering):
(WebKit::RemoteAudioDestinationProxy::stopRendering):
(WebKit::RemoteAudioDestinationProxy::storageChanged):

• WebProcess/GPU/media/RemoteAudioDestinationProxy.h:

Tools:

Re-enable API test now that it is no longer crashing.

• TestWebKitAPI/Tests/WebKitCocoa/GPUProcess.mm:

(TEST):

9:45 PM Changeset in webkit [276283] by Said Abou-Hallawa

• 3 edits in trunk/Source/WebCore

cachedCGColor() and nsColor() are not thread-safe
​https://bugs.webkit.org/show_bug.cgi?id=223033

Reviewed by Chris Dumez.

These functions will break if they are used by a worker thread.

• platform/graphics/cg/ColorCG.cpp:

(WebCore::cachedCGColor):

• platform/graphics/mac/ColorMac.mm:

(WebCore::nsColor):

8:26 PM Changeset in webkit [276282] by Diego Pino Garcia

• 2 edits in trunk/LayoutTests

[GLIB] Unreviewed test gardening. Gardened flaky failures after r224791.

• platform/glib/TestExpectations:

7:26 PM Changeset in webkit [276281] by jer.noble@apple.com

• 2 edits in trunk/Source/WebKit

[iOS] Media playback continues after backgrounding hosting application
​https://bugs.webkit.org/show_bug.cgi?id=224776
<rdar://75707807>

Reviewed by Eric Carlson.

Tested by existing API test: WKWebViewPausePlayingAudioTests.OutOfWindow

When the MediaSessionHelperIOS was moved into the GPU process, no object remains listening for the
UIApplication{Will,Did}Enter{Foreground,Background}Notification rebroadcasted by WebPageIOS.

Rather than just rebroadcast the UIKit notification to all listeners within the WebContent process,
which may or may not be listening, just tell the current PlatformMediaSessionManager directly.

• WebProcess/WebPage/ios/WebPageIOS.mm:

(WebKit::WebPage::applicationWillResignActive):
(WebKit::WebPage::applicationDidEnterBackground):
(WebKit::WebPage::applicationWillEnterForeground):
(WebKit::WebPage::applicationDidBecomeActive):

6:02 PM Changeset in webkit [276280] by Chris Dumez

• 2 edits in trunk/Tools

Unreviewed, temporarily disable GPUProcess.DISABLED_WebProcessTerminationAfterTooManyGPUProcessCrashes
​https://bugs.webkit.org/show_bug.cgi?id=224790

It is crashing. I will renable it with the fix at Bug 224790.

• TestWebKitAPI/Tests/WebKitCocoa/GPUProcess.mm:

(TEST):

5:54 PM Changeset in webkit [276279] by Diego Pino Garcia

• 5 edits in trunk/LayoutTests

[GLIB] Unreviewed test gardening. Update baselines and test expectations after r224791.

• platform/glib/TestExpectations:
• platform/glib/http/tests/xmlhttprequest/simple-cross-origin-denied-events-post-expected.txt:
• platform/glib/imported/w3c/web-platform-tests/xhr/send-entity-body-none-expected.txt:
• platform/glib/security/block-test-no-port-expected.txt:

5:18 PM Changeset in webkit [276278] by Patrick Angle

• 3 edits in trunk/Source/WebInspectorUI

REGRESSION (r268691 && r270134): Web Inspector: Clicking on go-to arrow in Computed panel no longer works
​https://bugs.webkit.org/show_bug.cgi?id=224774

Reviewed by Devin Rousso.

The Styles panel will not always be part of the same sidebar as the Computed panel as of r268691, so we should
look for the Styles panel in WI.detailsSidebar, which will report all of the panels it manages across multiple
sidebars. Additionally, as of r270134, there is no longer a private _visible property for
WI.StyleDetailsPanel, so it can not be used by WI.SpreadsheetRulesStyleDetailsPanel. Instead, visibility
should be determined by checking if the panel is attached and not pending layout.

• UserInterface/Views/ComputedStyleDetailsSidebarPanel.js:

(WI.ComputedStyleDetailsSidebarPanel.prototype.computedStyleDetailsPanelShowProperty):

• UserInterface/Views/SpreadsheetRulesStyleDetailsPanel.js:

(WI.SpreadsheetRulesStyleDetailsPanel.prototype.scrollToSectionAndHighlightProperty):

4:37 PM Changeset in webkit [276277] by aakash_jain@apple.com

• 2 edits in trunk/Tools

[ews] Enabled detailed error log when unit-tests fails
​https://bugs.webkit.org/show_bug.cgi?id=224789

Unreviewed minor unit-test fix.

• CISupport/ews-build/factories_unittest.py:

(TestCase): Set maxDiff to None.

4:03 PM Changeset in webkit [276276] by BJ Burg

• 3 edits
  1 add in trunk/Tools

Can't use Web Inspector on web views made by TestWebKitAPI
​https://bugs.webkit.org/show_bug.cgi?id=147073
<rdar://problem/76708379>

Reviewed by Devin Rousso.

It is necessary to spin a nested run loop at the point in the test where you would
like to remote inspect a WebView. Messages from the remote connection are dispatched
through UIProcess, so if lldb has paused UIProcess, WebInspectorUI will not be able to
get any data from the inspected WebView.

• TestWebKitAPI/DebugUtilities.h: Added.

Add macros to wait for a remote inspector to attach or detach, then drop into
the debugger when it has done so.

• TestWebKitAPI/PlatformUtilities.h: Add missing #pragma once.
• TestWebKitAPI/WTFStringUtilities.h: Force the build to fail noisily if we have

attempted to redefine WTF_STRINGTYPEADAPTER_COPIED_WTF_STRING. Force the correct
ordering between "WTFStringUtilities.h" and <wtf/text/StringConcatenate.h>.

3:20 PM Changeset in webkit [276275] by Wenson Hsieh

• 6 edits in trunk

Rename FloatQuad::isEmpty() to boundingBoxIsEmpty() and reimplement isEmpty()
​https://bugs.webkit.org/show_bug.cgi?id=224769

Reviewed by Tim Horton.

Source/WebCore:

The existing isEmpty() method on FloatQuad only checks that the bounding box of the quad is empty (as
indicated by the comment above the method declaration). This means isEmpty() returns false in cases where
the quad consists of one or more non-rectilinear line segments, even though the quad contains no area.

To address this, we rename the existing isEmpty() helper to boundingBoxIsEmpty() and reintroduce an
isEmpty() method that accounts for quads which are equivalent to non-rectilinear line segments.

No change in behavior.

• inspector/InspectorOverlay.cpp:

(WebCore::drawFragmentHighlight):

• platform/graphics/FloatQuad.cpp:
• platform/graphics/FloatQuad.h:

(WebCore::FloatQuad::isEmpty const): Deleted.
(WebCore::FloatQuad::boundingBoxIsEmpty const):

Tools:

Add an API test to exercise the new FloatQuad::isEmpty() method.

• TestWebKitAPI/Tests/WebCore/FloatQuadTests.cpp:

(TestWebKitAPI::checkIsEmpty):
(TestWebKitAPI::TEST):

2:18 PM Changeset in webkit [276274] by Patrick Angle

• 2 edits in trunk/Source/WebCore

Web Inspector: Grid overlay column line label placement copy/paste error
​https://bugs.webkit.org/show_bug.cgi?id=224777

Reviewed by Devin Rousso.

In grid overlays, a typo meant for all column line labels after the initial line, only the start of the line was
used for placement, where it is more accurate to use the start and end points to match the behavior of the first
line. This had no visual impact because subsequent lines do not currently have their labels adjusted further into
the grid, only flipped into the grid but still referencing the same point. This patch corrects the value of
gapLabelLine to always be a line on which the label can be placed, instead of a single point.

• inspector/InspectorOverlay.cpp:

(WebCore::InspectorOverlay::buildGridOverlay):

2:10 PM Changeset in webkit [276273] by youenn@apple.com

• 3 edits in trunk/LayoutTests

[mac-wk2] LayoutTest fast/mediastream/getUserMedia-webaudio.html is a flaky timeout
​https://bugs.webkit.org/show_bug.cgi?id=170960

Reviewed by Eric Carlson.

Update test to check for noise for more iterations.
Remove timeout expectation since it is no longer timing out in bots.

• fast/mediastream/getUserMedia-webaudio.html:
• platform/mac-wk2/TestExpectations:

1:45 PM Changeset in webkit [276272] by Devin Rousso

• 8 edits
  2 adds in trunk

Web Inspector: Graphics: add support for steps()/spring() CSS timing functions
​https://bugs.webkit.org/show_bug.cgi?id=224654

Reviewed by BJ Burg.

Source/WebInspectorUI:

• UserInterface/Models/Geometry.js:

(WI.StepsFunction): Added.
(WI.StepsFunction.fromString): Added.
(WI.StepsFunction.prototype.get type): Added.
(WI.StepsFunction.prototype.get count): Added.
(WI.StepsFunction.prototype.copy): Added.
(WI.StepsFunction.prototype.toString): Added.
Create a model object for steps() CSS timing function.

• UserInterface/Models/Animation.js:

(WI.Animation.prototype._updateEffect):
Also support effect.timingFunction/keyframe.easing being a steps()/spring() CSS timing function.

• UserInterface/Views/AnimationContentView.js:

(WI.AnimationContentView.prototype._refreshPreview):
Create a UI for steps()/spring() CSS timing functions.

LayoutTests:

• inspector/animation/lifecycle-web-animation.html:
• inspector/animation/lifecycle-web-animation-expected.txt:
• inspector/animation/resources/lifecycle-utilities.js:
• inspector/unit-tests/geometry.html: Added.
• inspector/unit-tests/geometry-expected.txt: Added.

1:35 PM Changeset in webkit [276271] by Chris Dumez

• 5 edits in trunk/Source/WebKit

MotionMark's Canvas-Arcs subtest is broken if the GPUProcess is not yet running
​https://bugs.webkit.org/show_bug.cgi?id=224778

Reviewed by Simon Fraser.

I recently made changes so that the GPUProcess is only launched when it is needed. This means
that until MotionMark's Canvas-Arcs subtest, the GPUProcess is usually not yet running. As a
result, RemoteRenderingBackendProxy::createImageBuffer() ends up launching the GPUProcess
and sending the GPUConnectionToWebProcess::CreateRenderingBackend and
RemoteRenderingBackend::CreateImageBuffer IPC right away. We seem to have a synchronization
issue because when this happens, the rendering for the Canvas-Arcs subtest is visibly broken
and its score is very low (< 6, instead of > 600).

Making the GPUConnectionToWebProcess::CreateRenderingBackend synchronous again (I made it
async recently) seems to address the synchronization issue and restores correct behavior on
MotionMark. I am therefore making this IPC synchronous again in this patch to get benchmark
coverage again while I investigate the root cause offline.

• GPUProcess/GPUConnectionToWebProcess.cpp:

(WebKit::GPUConnectionToWebProcess::createRenderingBackend):

• GPUProcess/GPUConnectionToWebProcess.h:
• GPUProcess/GPUConnectionToWebProcess.messages.in:
• WebProcess/GPU/graphics/RemoteRenderingBackendProxy.cpp:

(WebKit::RemoteRenderingBackendProxy::ensureGPUProcessConnection):

1:16 PM Changeset in webkit [276270] by timothy_horton@apple.com

• 2 edits in trunk/Tools

Fix the macCatalyst TestWebKitAPI build
​https://bugs.webkit.org/show_bug.cgi?id=224780

• TestWebKitAPI/Configurations/TestWebKitAPI.xcconfig:

Unreviewed build fix; link PDFKit since we now enable it.

1:09 PM Changeset in webkit [276269] by Alan Bujtas

• 3 edits
  2 adds in trunk

[LFC][IFC] LineCandidate.inlineContent should be ignored when reverting
​https://bugs.webkit.org/show_bug.cgi?id=224771
<rdar://76760857>

Reviewed by Antti Koivisto.

Source/WebCore:

LineCandidate.inlineContent is a set of candidate runs for the line and when the line breaker says "please revert" (move back to an earlier position on the line)
these runs should all be ignored as they did not make it to the line.
(inlineContentIsFullyCommitted flag got confused when the number of candidate runs matched the number of runs we managed to put on the line as part of the revert. It did not take the "revert" bit into account.)

Test: fast/inline/crash-when-revert-has-trailing-line-break.html

• layout/inlineformatting/InlineLineBuilder.cpp:

(WebCore::Layout::LineBuilder::placeInlineContent):

LayoutTests:

• fast/inline/crash-when-revert-has-trailing-line-break-expected.txt: Added.
• fast/inline/crash-when-revert-has-trailing-line-break.html: Added.

12:42 PM Changeset in webkit [276268] by Amir Mark Jr.

• 2 edits in trunk/LayoutTests

[ wk2 ] http/tests/security/contentSecurityPolicy/report-only-connect-src-xmlhttprequest-redirect-to-blocked.py is a constant text failure
​https://bugs.webkit.org/show_bug.cgi?id=224626

Unreviewed test gardening

• platform/wk2/TestExpectations: Updating expectations as a fix is being worked on.

11:55 AM Changeset in webkit [276267] by Chris Dumez

• 4 edits in trunk

SVG Images launch the GPUProcess unnecessarily
​https://bugs.webkit.org/show_bug.cgi?id=224770

Reviewed by Eric Carlson.

Source/WebCore:

Make sure MediaPlayer::supportsType() returns early for MIME types that can definitely
not be media, to avoid launch the GPUProcess unnecessarily.

• platform/graphics/MediaPlayer.cpp:

(WebCore::MediaPlayer::supportsType):

Tools:

Add API test coverage.

• TestWebKitAPI/Tests/WebKitCocoa/GPUProcess.mm:

(TEST):

11:52 AM Changeset in webkit [276266] by msaboff@apple.com

• 13 edits in trunk/Source/bmalloc

[bmalloc] Enable Adaptive Scavenger for Mac
​https://bugs.webkit.org/show_bug.cgi?id=224706

Reviewed by Filip Pizlo.

Enabled the adaptive scavenger code paths for macOS.
The original reason that the partial scavenging paths were kept for macOS was due
to regression on power tests. To alleviate the power regression, this patch splits
out the adaptive scavenger parameters with macOS specific values.

The parameters are:

The multiplier used to compute the next scavenging wait time based on the

time needed for the prior scavenging.

Minimum wait time between scavenging.
Maximum wait time between scavenging.

The values in the current code are:

Wait time Multiplier: 150
Minimum wait time: 100ms
Maximum wait time: 10,000ms (10 seconds)

The proposed values for macOS, determined using empirical testing.

Wait time Multiplier: 300
Minimum wait time: 750ms
Maximum wait time: 20,000ms (20 seconds)

When tested on various mac variants, this change:

• Provides a 3-5% reduction in memory use on RAMification.
• It is neutral on JetStream2.
• It is neutral to a slight regression on Speedometer2, but there is some variability in those results.

Since macOS was the only platform still using the partial scavenging code path,
the partial scavenging code paths were deleted.

• bmalloc/BPlatform.h:
• bmalloc/Heap.cpp:

(bmalloc::Heap::scavenge):
(bmalloc::Heap::allocateSmallChunk):
(bmalloc::Heap::allocateSmallPage):
(bmalloc::Heap::allocateLarge):
(bmalloc::Heap::scavengeToHighWatermark): Deleted.

• bmalloc/Heap.h:
• bmalloc/IsoDirectory.h:
• bmalloc/IsoDirectoryInlines.h:

(bmalloc::passedNumPages>::takeFirstEligible):
(bmalloc::passedNumPages>::scavenge):
(bmalloc::passedNumPages>::scavengeToHighWatermark): Deleted.

• bmalloc/IsoHeapImpl.h:
• bmalloc/IsoHeapImplInlines.h:

(bmalloc::IsoHeapImpl<Config>::scavengeToHighWatermark): Deleted.

• bmalloc/LargeMap.cpp:

(bmalloc::LargeMap::add):

• bmalloc/LargeRange.h:

(bmalloc::LargeRange::LargeRange):
(bmalloc::LargeRange::setUsedSinceLastScavenge):
(bmalloc::merge):
(): Deleted.

• bmalloc/Scavenger.cpp:

(bmalloc::Scavenger::Scavenger):
(bmalloc::Scavenger::scavenge):
(bmalloc::Scavenger::threadRunLoop):
(bmalloc::Scavenger::timeSinceLastPartialScavenge): Deleted.
(bmalloc::Scavenger::partialScavenge): Deleted.

• bmalloc/Scavenger.h:
• bmalloc/SmallPage.h:

(bmalloc::SmallPage::setUsedSinceLastScavenge):

11:21 AM Changeset in webkit [276265] by Devin Rousso

• 3 edits in trunk/Source/WebInspectorUI

Web Inspector: REGRESSION(?): Network: Request (Object Tree) is sometimes empty
​https://bugs.webkit.org/show_bug.cgi?id=224768
<rdar://problem/76783636>

Reviewed by BJ Burg.

• UserInterface/Views/LocalJSONContentView.js:

(WI.LocalJSONContentView.prototype.renderRemoteObject):
If the WI.RemoteObject is simple enough to be rendered inline, WI.ObjectTreeView won't
show a WI.TreeOutline by default. We have to tell it to use a WI.TreeOutline by passing
forceExpanding = true into the constructor.

• UserInterface/Views/ObjectTreeView.js:

(WI.ObjectTreeView.prototype.showOnlyJSON):
Add some console.assert to help prevent this from happening again.

10:59 AM Changeset in webkit [276264] by Manuel Rego Casasnovas

• 27 edits
  1 delete in trunk

[selectors] Script focus and :focus-visible
​https://bugs.webkit.org/show_bug.cgi?id=224598

Reviewed by Darin Adler.

LayoutTests/imported/w3c:

Update expectations as the tests pass now.

• web-platform-tests/css/selectors/focus-visible-008-expected.txt:
• web-platform-tests/css/selectors/focus-visible-009-expected.txt:
• web-platform-tests/css/selectors/focus-visible-010-expected.txt:
• web-platform-tests/css/selectors/focus-visible-014-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-001-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-002.tentative-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-003.tentative-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-004-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-005-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-006.tentative-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-007.tentative-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-012-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-013-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-014-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-015-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-016.tentative-expected.txt:
• web-platform-tests/css/selectors/focus-visible-script-focus-017.tentative-expected.txt:

Source/WebCore:

Implement :focus-visible behavior when a script moves focus.
An element will match :focus-visible when a script move focus if the last focused element was not focused via mouse click.

This patch makes WebKit behaves like Chromium and Firefox and pass all the related tests.

There's an ongoing discussion about 6 tests that are marked as ".tentative" in the following issue:
​https://github.com/web-platform-tests/wpt/issues/28505

Test: web-platform-tests/css/selectors/focus-visible-script-focus-*

• dom/Document.cpp:

(WebCore::Document::adjustFocusedNodeOnNodeRemoval): Just update FocusOptions initializer.
(WebCore::Document::setFocusedElement): Store if the last element has been focused by mouse click or not.

• dom/Document.h: New member m_latestFocusTrigger.

(WebCore::Document::wasLastFocusByClick const): Method to check status of m_latestFocusTrigger.

• dom/Element.cpp:

(WebCore::Element::focus): When there's a script focus, this checks if the last element was focused
by mouse click in order to make it match :focus-visible.

• dom/FocusOptions.h: Add FocusTrigger enum to know if an element has been focused via mouse click or not.

This refers to the "focus trigger" concept on the HTML spec (see ​https://html.spec.whatwg.org/multipage/interaction.html).

• page/EventHandler.cpp:

(WebCore::EventHandler::dispatchMouseEvent): Pass FocusTrigger:Click in setFocusedElement() call.

LayoutTests:

Update expectations for tests, most of them are passing on Mac. On iOS a bunch timeout so they're skipped, as other :focus-visible tests.

• platform/ios/TestExpectations:
• platform/mac/imported/w3c/web-platform-tests/css/selectors/focus-visible-009-expected.txt: Removed as test passes now.
• platform/mac/TestExpectations:

10:53 AM Changeset in webkit [276263] by Chris Dumez

• 3 edits in trunk/Source/WebKit

Fix races in LibWebRTCCodecs introduced in r276214
​https://bugs.webkit.org/show_bug.cgi?id=224758

Reviewed by Youenn Fablet.

After r276214, LibWebRTCCodecs's createDecoder() / createEncoder() may hop to the main
thread to initialize the connection to the GPUProcess. If releaseDecoder() / releaseEncoder()
were to get called very shortly after, they may win the race and have no decoder / encoder
to release, since they are dispatched directly to the background thread. To address the issue,
we now call ensureGPUProcessConnectionAndDispatchToThread() in releaseDecoder(), releaseEncoder()
and initializeEncoder().

• WebProcess/GPU/webrtc/LibWebRTCCodecs.cpp:

(WebKit::LibWebRTCCodecs::releaseDecoder):
(WebKit::LibWebRTCCodecs::releaseEncoder):
(WebKit::LibWebRTCCodecs::initializeEncoder):

10:51 AM Changeset in webkit [276262] by Darin Adler

• 10 edits in trunk

Nullptr crash in CSSCalcValue::category() via HTMLConverterCaches::floatPropertyValueForNode
​https://bugs.webkit.org/show_bug.cgi?id=221392

Reviewed by Simon Fraser.

LayoutTests/imported/w3c:

• web-platform-tests/css/css-values/minmax-length-percent-serialize-expected.txt:

Updated to reflect 8 tests passing that were failing before.

Source/WebCore:

• css/CSSCalculationValue.cpp:

(WebCore::CSSCalcOperationNode::createCalcExpression const): Pass in a destination category
when creating a CalcExpressionOperation.
(WebCore::createCSS): Pass the destination category from the CalcExpressionOperation when
creating a CSSCalcOperationNode.

• css/CSSCalculationValue.h: Moved the CalculationCategory enumeration from here to

CalculationValue.h.

• platform/CalculationValue.cpp:

(WebCore::operator==): Include destination category when comparing.

• platform/CalculationValue.h: Moved CalculationCategory here. Added a destination

category constructor argument, data members, and getter function to the
CalcExpressionOperation class.

LayoutTests:

• fast/css/calc-parsing-expected.txt: Updated for change below.
• fast/css/calc-parsing.html: Added more test cases so this covers the affected cases of computed

style for calc expressions that mix percentages and numbers. Alternatively, we could remove this
entire test case because the web platform tests also cover this pretty well.

10:20 AM WebKitGTK/2.32.x edited by clopez@igalia.com

(diff)

10:17 AM Changeset in webkit [276261] by Chris Gambrell

• 12 edits
  2 adds
  2 deletes in trunk/LayoutTests

[LayoutTests] Convert http/tests/blink convert PHP to Python
​https://bugs.webkit.org/show_bug.cgi?id=224702
<rdar://problem/76783504>

Reviewed by Darin Adler.

• http/tests/blink/sendbeacon/beacon-cookie-expected.txt:
• http/tests/blink/sendbeacon/beacon-cookie.html:
• http/tests/blink/sendbeacon/beacon-cross-origin-expected.txt:
• http/tests/blink/sendbeacon/beacon-cross-origin.html:
• http/tests/blink/sendbeacon/beacon-cross-origin.https-expected.txt:
• http/tests/blink/sendbeacon/beacon-cross-origin.https.html:
• http/tests/blink/sendbeacon/beacon-same-origin-expected.txt:
• http/tests/blink/sendbeacon/beacon-same-origin.html:
• http/tests/blink/sendbeacon/resources/check-beacon.php: Removed.
• http/tests/blink/sendbeacon/resources/check-beacon.py: Added.

(extensive_strip):

• http/tests/blink/sendbeacon/resources/save-beacon.php: Removed.
• http/tests/blink/sendbeacon/resources/save-beacon.py: Added.

(prettify):
(decode_multipart):

• http/tests/resources/portabilityLayer.py:

(get_post_data):
(get_request):

• platform/ios/http/tests/blink/sendbeacon/beacon-same-origin-expected.txt:
• platform/mac/http/tests/blink/sendbeacon/beacon-same-origin-expected.txt:

10:08 AM Changeset in webkit [276260] by Antti Koivisto

• 2 edits in trunk/Source/WebCore

CSSValuePool should be non-copyable
​https://bugs.webkit.org/show_bug.cgi?id=224764

Reviewed by Alan Bujtas.

• css/CSSValuePool.h:

9:59 AM Changeset in webkit [276259] by mark.lam@apple.com

• 2 edits in trunk/Source/JavaScriptCore

Build fix for Debug -O3 after r276162.
​https://bugs.webkit.org/show_bug.cgi?id=224681
rdar://76698113

Not reviewed.

• runtime/JSObject.cpp:

9:53 AM Changeset in webkit [276258] by Philippe Normand

• 2 edits in trunk/LayoutTests

Unreviewed, GLIB gardening

• platform/glib/TestExpectations: Flag a few more media tests to skip or as flaky crash.

9:46 AM Changeset in webkit [276257] by Chris Dumez

• 2 edits in trunk/Source/WebKit

Add more GPUProcess release logging to facilitate debugging
​https://bugs.webkit.org/show_bug.cgi?id=224761

Reviewed by Darin Adler.

• GPUProcess/GPUProcess.cpp:

(WebKit::GPUProcess::GPUProcess):
(WebKit::GPUProcess::createGPUConnectionToWebProcess):
(WebKit::GPUProcess::removeGPUConnectionToWebProcess):
(WebKit::GPUProcess::initializeGPUProcess):

9:45 AM Changeset in webkit [276256] by Philippe Normand

• 2 edits in trunk/LayoutTests

Unreviewed, GLIB gardening

• platform/glib/TestExpectations: Flag imported/w3c/web-platform-tests/media-source/mediasource-changetype-play-implicit.html as flaky crash.

8:41 AM Changeset in webkit [276255] by Alan Bujtas

• 2 edits in trunk/Source/WebCore

[LFC] The layout containment box establishes an absolute positioning containing block and a fixed positioning containing block
​https://bugs.webkit.org/show_bug.cgi?id=224754

Reviewed by Antti Koivisto.

​https://www.w3.org/TR/css-contain-2/#containment-layout.

"The layout containment box establishes an absolute positioning containing block and a fixed positioning containing block."

• layout/layouttree/LayoutBox.h:

(WebCore::Layout::Box::isContainingBlockForFixedPosition const):
(WebCore::Layout::Box::isContainingBlockForOutOfFlowPosition const):

8:21 AM Changeset in webkit [276254] by Alan Bujtas

• 2 edits in trunk/Source/WebCore

[LFC] Layout containment box establishes an independent formatting context
​https://bugs.webkit.org/show_bug.cgi?id=224753

Reviewed by Antti Koivisto.

​https://www.w3.org/TR/css-contain-2/#containment-layout.

"The layout containment box establishes an independent formatting context."

• layout/layouttree/LayoutBox.cpp:

(WebCore::Layout::Box::establishesIndependentFormattingContext const):

8:18 AM Changeset in webkit [276253] by commit-queue@webkit.org

• 4 edits in trunk/Source/WebCore

Silence various -Wreturn-type warnings
​https://bugs.webkit.org/show_bug.cgi?id=224756

Patch by Michael Catanzaro <Michael Catanzaro> on 2021-04-19
Reviewed by Don Olmstead.

• css/parser/CSSPropertyParserHelpers.cpp:

(WebCore::CSSPropertyParserHelpers::mixColorComponents):

• inspector/InspectorOverlay.cpp:

(WebCore::expectedSizeForLayoutLabel):

• rendering/style/RenderStyle.cpp:

(WebCore::RenderStyle::usedClear):
(WebCore::RenderStyle::usedFloat):

7:16 AM Changeset in webkit [276252] by Wenson Hsieh

• 2 edits in trunk/Source/WebCore

Unreviewed, fix the build after r276181
<rdar://problem/76832595>

Add a couple of missing forward declarations for IntRect and IntPoint.

• page/ContextMenuClient.h:

6:54 AM Changeset in webkit [276251] by youenn@apple.com

• 3 edits in trunk/Source/WebCore

Add assertions to check for no memory allocation in AudioMediaStreamTrackRendererUnit rendering thread
​https://bugs.webkit.org/show_bug.cgi?id=224674

Reviewed by Eric Carlson.

Add ForbidMallocUseForCurrentThreadScope in audio rendering code path to prevent memory allocation.
Rename inputProc to renderingCallback.
Remove use of AudioMediaStreamTrackRendererUnit::singleton and instead use the pointer given to renderingCallback.
This is safe as AudioMediaStreamTrackRendererUnit destructor stops the audio unit.

• platform/mediastream/mac/AudioMediaStreamTrackRendererUnit.cpp:

(WebCore::AudioMediaStreamTrackRendererUnit::createAudioUnitIfNeeded):
(WebCore::AudioMediaStreamTrackRendererUnit::render):
(WebCore::AudioMediaStreamTrackRendererUnit::renderingCallback):
(WebCore::AudioMediaStreamTrackRendererUnit::inputProc): Deleted.

• platform/mediastream/mac/AudioMediaStreamTrackRendererUnit.h:

6:43 AM Changeset in webkit [276250] by Philippe Normand

• 2 edits in trunk/Tools

Unreviewed, WPE Minibrowser build warning fix.

• MiniBrowser/wpe/main.cpp: No need to use typedef for this struct declaration.

6:28 AM Changeset in webkit [276249] by Alan Bujtas

• 3 edits in trunk/Source/WebCore

[LFC] Implement Box::isLayoutContainmentBox
​https://bugs.webkit.org/show_bug.cgi?id=224740

Reviewed by Antti Koivisto.

This is part of ​https://www.w3.org/TR/css-contain-2/#containment-layout

• layout/layouttree/LayoutBox.cpp:

(WebCore::Layout::Box::isLayoutContainmentBox const):
(WebCore::Layout::Box::isInternalTableBox const):

• layout/layouttree/LayoutBox.h:

(WebCore::Layout::Box::isInternalRubyBox const):

6:22 AM Changeset in webkit [276248] by Philippe Normand

• 4 edits
  39 moves
  15 adds in trunk/LayoutTests

Unreviewed, WPE gardening

Share more media tests baselines between the GTK and WPE ports.

• platform/glib/TestExpectations:
• platform/glib/imported/w3c/web-platform-tests/media-source/mediasource-changetype-play-implicit-expected.txt: Renamed from LayoutTests/platform/gtk/imported/w3c/web-platform-tests/media-source/mediasource-changetype-play-implicit-expected.txt.
• platform/glib/imported/w3c/web-platform-tests/media-source/mediasource-changetype-play-negative-expected.txt: Renamed from LayoutTests/platform/gtk/imported/w3c/web-platform-tests/media-source/mediasource-changetype-play-negative-expected.txt.
• platform/glib/imported/w3c/web-platform-tests/media-source/mediasource-changetype-play-without-codecs-parameter-expected.txt: Renamed from LayoutTests/platform/gtk/imported/w3c/web-platform-tests/media-source/mediasource-changetype-play-without-codecs-parameter-expected.txt.
• platform/glib/imported/w3c/web-platform-tests/media-source/mediasource-duration-boundaryconditions-expected.txt: Renamed from LayoutTests/platform/gtk/imported/w3c/web-platform-tests/media-source/mediasource-duration-boundaryconditions-expected.txt.
• platform/glib/imported/w3c/web-platform-tests/media-source/mediasource-getvideoplaybackquality-expected.txt: Renamed from LayoutTests/platform/gtk/imported/w3c/web-platform-tests/media-source/mediasource-getvideoplaybackquality-expected.txt.
• platform/glib/imported/w3c/web-platform-tests/media-source/mediasource-is-type-supported-expected.txt: Renamed from LayoutTests/platform/gtk/imported/w3c/web-platform-tests/media-source/mediasource-is-type-supported-expected.txt.
• platform/glib/imported/w3c/web-platform-tests/media-source/mediasource-remove-expected.txt: Renamed from LayoutTests/platform/gtk/imported/w3c/web-platform-tests/media-source/mediasource-remove-expected.txt.
• platform/glib/imported/w3c/web-platform-tests/media-source/mediasource-removesourcebuffer-expected.txt: Renamed from LayoutTests/platform/gtk/imported/w3c/web-platform-tests/media-source/mediasource-removesourcebuffer-expected.txt.
• platform/glib/imported/w3c/web-platform-tests/media-source/mediasource-sourcebuffer-mode-expected.txt: Renamed from LayoutTests/platform/gtk/imported/w3c/web-platform-tests/media-source/mediasource-sourcebuffer-mode-expected.txt.
• platform/glib/media/W3C/video/canPlayType/canPlayType_codecs_order_1-expected.txt: Renamed from LayoutTests/platform/gtk/media/W3C/video/canPlayType/canPlayType_codecs_order_1-expected.txt.
• platform/glib/media/W3C/video/canPlayType/canPlayType_supported_but_no_codecs_parameter_1-expected.txt: Renamed from LayoutTests/platform/gtk/media/W3C/video/canPlayType/canPlayType_supported_but_no_codecs_parameter_1-expected.txt.
• platform/glib/media/W3C/video/canPlayType/canPlayType_two_implies_one_1-expected.txt: Renamed from LayoutTests/platform/gtk/media/W3C/video/canPlayType/canPlayType_two_implies_one_1-expected.txt.
• platform/glib/media/W3C/video/canPlayType/canPlayType_two_implies_one_2-expected.txt: Renamed from LayoutTests/platform/gtk/media/W3C/video/canPlayType/canPlayType_two_implies_one_2-expected.txt.
• platform/glib/media/audio-controls-rendering-expected.png: Renamed from LayoutTests/platform/gtk/media/audio-controls-rendering-expected.png.
• platform/glib/media/audio-repaint-expected.png: Renamed from LayoutTests/platform/gtk/media/audio-repaint-expected.png.
• platform/glib/media/context-menu-actions-expected.txt: Renamed from LayoutTests/platform/gtk/media/context-menu-actions-expected.txt.
• platform/glib/media/controls-after-reload-expected.png: Renamed from LayoutTests/platform/gtk/media/controls-after-reload-expected.png.
• platform/glib/media/controls-layout-direction-expected.png: Renamed from LayoutTests/platform/gtk/media/controls-layout-direction-expected.png.
• platform/glib/media/controls-strict-expected.png: Renamed from LayoutTests/platform/gtk/media/controls-strict-expected.png.
• platform/glib/media/controls-styling-expected.png: Renamed from LayoutTests/platform/gtk/media/controls-styling-expected.png.
• platform/glib/media/controls-styling-strict-expected.png: Renamed from LayoutTests/platform/gtk/media/controls-styling-strict-expected.png.
• platform/glib/media/controls-without-preload-expected.png: Renamed from LayoutTests/platform/gtk/media/controls-without-preload-expected.png.
• platform/glib/media/media-can-play-ogg-expected.txt: Renamed from LayoutTests/platform/gtk/media/media-can-play-ogg-expected.txt.
• platform/glib/media/media-controls-clone-expected.png: Renamed from LayoutTests/platform/gtk/media/media-controls-clone-expected.png.
• platform/glib/media/media-document-audio-repaint-expected.png: Renamed from LayoutTests/platform/gtk/media/media-document-audio-repaint-expected.png.
• platform/glib/media/media-fullscreen-inline-expected.txt: Renamed from LayoutTests/platform/gtk/media/media-fullscreen-inline-expected.txt.
• platform/glib/media/media-fullscreen-not-in-document-expected.txt: Renamed from LayoutTests/platform/gtk/media/media-fullscreen-not-in-document-expected.txt.
• platform/glib/media/media-source/media-source-resize-expected.txt: Renamed from LayoutTests/platform/gtk/media/media-source/media-source-resize-expected.txt.
• platform/glib/media/video-canvas-alpha-expected.png: Renamed from LayoutTests/platform/gtk/media/video-canvas-alpha-expected.png.
• platform/glib/media/video-controls-rendering-expected.png: Renamed from LayoutTests/platform/gtk/media/video-controls-rendering-expected.png.
• platform/glib/media/video-display-toggle-expected.png: Renamed from LayoutTests/platform/gtk/media/video-display-toggle-expected.png.
• platform/glib/media/video-empty-source-expected.png: Renamed from LayoutTests/platform/gtk/media/video-empty-source-expected.png.
• platform/glib/media/video-no-audio-expected.png: Renamed from LayoutTests/platform/gtk/media/video-no-audio-expected.png.
• platform/glib/media/video-playing-and-pause-expected.png: Renamed from LayoutTests/platform/gtk/media/video-playing-and-pause-expected.png.
• platform/glib/media/video-playing-and-pause-expected.txt: Renamed from LayoutTests/platform/gtk/media/video-playing-and-pause-expected.txt.
• platform/glib/media/video-volume-slider-expected.png: Renamed from LayoutTests/platform/gtk/media/video-volume-slider-expected.png.
• platform/glib/media/video-zoom-controls-expected.png: Renamed from LayoutTests/platform/gtk/media/video-zoom-controls-expected.png.
• platform/glib/media/video-zoom-expected.png: Renamed from LayoutTests/platform/gtk/media/video-zoom-expected.png.
• platform/glib/media/video-zoom-expected.txt: Renamed from LayoutTests/platform/gtk/media/video-zoom-expected.txt.
• platform/gtk/TestExpectations:
• platform/wpe/TestExpectations:
• platform/wpe/media/video-aspect-ratio-expected.txt: Added.
• platform/wpe/media/video-canvas-createPattern-expected.png: Added.
• platform/wpe/media/video-colorspace-yuv420-expected.png: Added.
• platform/wpe/media/video-colorspace-yuv420-expected.txt: Added.
• platform/wpe/media/video-colorspace-yuv422-expected.png: Added.
• platform/wpe/media/video-colorspace-yuv422-expected.txt: Added.
• platform/wpe/media/video-frame-accurate-seek-expected.png: Added.
• platform/wpe/media/video-frame-accurate-seek-expected.txt: Added.
• platform/wpe/media/video-layer-crash-expected.txt: Added.
• platform/wpe/media/video-transformed-expected.txt: Added.

6:12 AM Changeset in webkit [276247] by commit-queue@webkit.org

• 24 edits
  3 adds in trunk

Enable -Wthread-safety, add attributes to custom lock classes, and provide macros to declare guards
​https://bugs.webkit.org/show_bug.cgi?id=221614
<rdar://problem/74396781>

Patch by Kimmo Kinnunen <​kkinnunen@apple.com> on 2021-04-19
Reviewed by David Kilzer.

PerformanceTests:

Add -Wthread-safety to compile flags.

• DecoderTest/Configurations/Base.xcconfig:

Source/bmalloc:

Add -Wthread-safety to compile flags.

• Configurations/Base.xcconfig:

Source/JavaScriptCore:

Add -Wthread-safety to compile flags.

• Configurations/Base.xcconfig:

Source/WebCore:

Add -Wthread-safety to compile flags.

• Configurations/Base.xcconfig:

Source/WebCore/PAL:

Add -Wthread-safety to compile flags.

• Configurations/Base.xcconfig:

Source/WebInspectorUI:

Add -Wthread-safety to compile flags.

• Configurations/Base.xcconfig:

Source/WebKit:

Add -Wthread-safety to compile flags.

• Configurations/Base.xcconfig:

Source/WebKitLegacy/mac:

Add -Wthread-safety to compile flags.

• Configurations/Base.xcconfig:

Source/WTF:

Implement rudimentary support for clang thread safety analysis.
The added macros can be used to declare which member variables or
global variables are locked by which mutexes. The compiler will
check statically that the mutexes are held correctly. The checking
is intra procedural, not global.

• Configurations/Base.xcconfig:

Add -Wthread-safety to compile flags.

• wtf/CheckedLock.h: Added.

Add CheckedLock, a Lock variant that is amenable to static
analysis.
Add a Locker specialization for CheckedLock that is amenable to
static analysis.

Locker<CheckedLock> is a std::scoped_lock. The scoped_lock cannot be aliased,
since it appears that (Apple's) libcxx is not compiled with thread safety
analysis support enabled by default.

New types are needed due Locker move constructor and conditional locking.
The Locker has default usage pattern of:

auto locker = holdLock(m_lock);

This forces dynamism that removes the possibility of simple statical
analysis that thread safety analysis capabilities "mutex" and "scoped_lock"
currently implement. Most likely large fraction of call sites is due to historical
lack of CTAD and as such can be converted to less general form.
Once the pattern is not used by default, CheckedLock can be deleted
and the move dynamism bits of Locker can be moved to some more specific type
("UncheckedLocker").

• wtf/ThreadSafetyAnalysis.h: Added.

Add macro wrappers around clang "mutex" and "scoped_lock" capability attributes.

Tools:

• TestWebKitAPI/Configurations/Base.xcconfig:

Add -Wthread-safety to compile flags.

• TestWebKitAPI/CMakeLists.txt:
• TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
• TestWebKitAPI/Tests/WTF/CheckedLock.cpp: Added.

(TestWebKitAPI::TEST):
Implement a test for testing that CheckedLock compiles.

4:55 AM Changeset in webkit [276246] by Carlos Garcia Campos

• 2 edits in trunk

[GTK] Switch to libsoup3 by default
​https://bugs.webkit.org/show_bug.cgi?id=224741

Reviewed by Sergio Villar Senin.

• Source/cmake/OptionsGTK.cmake:

3:28 AM Changeset in webkit [276245] by youenn@apple.com

• 2 edits in trunk/Source/WebCore

Make RealtimeIncomingAudioSourceCocoa preallocate audio buffer
​https://bugs.webkit.org/show_bug.cgi?id=224672

Reviewed by Eric Carlson.

Instead of allocating the buffer on valid data, we wait for data to be of the correct sample rate.
We preallocate the buffer accordingly in constructor to avoid allocation in the webrtc audio thread.
Default is 1 channel since mono is what webrtc encoders mostly do these days.
Covered by existing tests.

• platform/mediastream/mac/RealtimeIncomingAudioSourceCocoa.cpp:

(WebCore::RealtimeIncomingAudioSourceCocoa::RealtimeIncomingAudioSourceCocoa):
(WebCore::RealtimeIncomingAudioSourceCocoa::OnData):

2:56 AM Changeset in webkit [276244] by youenn@apple.com

• 4 edits in trunk/Source/WebCore

Move from RecursiveLock to Lock in RealtimeMediaSource
​https://bugs.webkit.org/show_bug.cgi?id=224671

Reviewed by Eric Carlson.

RecursiveLock is more expensive as it allocates memory and not useful here.
Covered by existing tests.

• platform/mediastream/RealtimeMediaSource.h:
• platform/mediastream/RealtimeOutgoingAudioSource.h:
• platform/mediastream/RealtimeOutgoingVideoSource.h:

2:53 AM Changeset in webkit [276243] by youenn@apple.com

• 9 edits in trunk/Source/WebCore

Reduce crackling at start of playing a live audio track
​https://bugs.webkit.org/show_bug.cgi?id=218898
<rdar://problem/71625010>

Reviewed by Eric Carlson.

We often hear crackles at the beginning of playing a MediaStreamTrack.
This is due to starting at a point where there is not enough data, so we start rendering and quickly run short of data.
Our heuristic to start was not good since we were not accounting to LibWebRTC audio module which sends audio samples by batch of 5 samples.
So we would have needed to at least have 6 samples to be sure to not run out of data.

What this patch does:

• Add an extra parameter to AudioSampleDataSource so that we do not start until we have at least been pushed a given number of sample count.
• Change LibWebRTAudioModule to send audio samples by batch of 3 (hence 30 ms) instead of 5
• Set this new parameter to 2 for local tracks and 4 for remote tracks
• Add a new boolean to AudioSampleDataSource to know whether we start the track for the first time or not. If we start it, use the new parameter to buffer enough data. Otherwise, use current heuristic.

We also reduce the AudioSampleDataSource buffer to 0.5 seconds instead of 2 seconds, since 2 seconds is too much for real time audio.

Manually tested.

• platform/audio/cocoa/AudioSampleDataSource.h:
• platform/audio/cocoa/AudioSampleDataSource.mm:

(WebCore::AudioSampleDataSource::create):
(WebCore::AudioSampleDataSource::AudioSampleDataSource):
(WebCore::AudioSampleDataSource::pullSamplesInternal):

• platform/mediastream/libwebrtc/LibWebRTCAudioModule.cpp:

(WebCore::LibWebRTCAudioModule::pollFromSource):

• platform/mediastream/libwebrtc/LibWebRTCAudioModule.h:
• platform/mediastream/mac/AudioMediaStreamTrackRendererCocoa.cpp:

(WebCore::pollSamplesCount):
(WebCore::AudioMediaStreamTrackRendererCocoa::pushSamples):

• platform/mediastream/mac/MediaStreamTrackAudioSourceProviderCocoa.cpp:

(WebCore::MediaStreamTrackAudioSourceProviderCocoa::MediaStreamTrackAudioSourceProviderCocoa):

• platform/mediastream/mac/WebAudioSourceProviderCocoa.h:
• platform/mediastream/mac/WebAudioSourceProviderCocoa.mm:

(WebCore::WebAudioSourceProviderCocoa::prepare):

2:49 AM Changeset in webkit [276242] by youenn@apple.com

• 5 edits
  13 moves
  17 adds
  2 deletes in trunk/LayoutTests

Migrate some WebRTC encoded transform tests to WPT
​https://bugs.webkit.org/show_bug.cgi?id=224604

Reviewed by Eric Carlson.

LayoutTests/imported/w3c:

• web-platform-tests/webrtc-encoded-transform/routines.js: Added.

(async createConnections):
(waitFor):

• web-platform-tests/webrtc-encoded-transform/script-audio-transform-worker.js: Added.

(MockRTCRtpTransformer):
(MockRTCRtpTransformer.prototype.start):
(MockRTCRtpTransformer.prototype.process):
(onrtctransform):

• web-platform-tests/webrtc-encoded-transform/script-audio-transform.https-expected.txt: Added.
• web-platform-tests/webrtc-encoded-transform/script-audio-transform.https.html: Renamed from LayoutTests/http/wpt/webrtc/audio-script-transform.html.
• web-platform-tests/webrtc-encoded-transform/script-change-transform-worker.js: Renamed from LayoutTests/http/wpt/webrtc/change-encoded-transform.js.
• web-platform-tests/webrtc-encoded-transform/script-change-transform.https-expected.txt: Added.
• web-platform-tests/webrtc-encoded-transform/script-change-transform.https.html: Renamed from LayoutTests/http/wpt/webrtc/change-encoded-transform.html.
• web-platform-tests/webrtc-encoded-transform/script-late-transform.https-expected.txt: Added.
• web-platform-tests/webrtc-encoded-transform/script-late-transform.https.html: Renamed from LayoutTests/http/wpt/webrtc/webrtc-late-transform.html.
• web-platform-tests/webrtc-encoded-transform/script-metadata-transform-worker.js: Renamed from LayoutTests/http/wpt/webrtc/metadata-transform-worker.js.
• web-platform-tests/webrtc-encoded-transform/script-metadata-transform.https-expected.txt: Added.
• web-platform-tests/webrtc-encoded-transform/script-metadata-transform.https.html: Renamed from LayoutTests/http/wpt/webrtc/metadata-transform.html.
• web-platform-tests/webrtc-encoded-transform/script-transform-worker.js: Renamed from LayoutTests/http/wpt/webrtc/script-transform.js.
• web-platform-tests/webrtc-encoded-transform/script-transform.https-expected.txt: Added.
• web-platform-tests/webrtc-encoded-transform/script-transform.https.html: Renamed from LayoutTests/http/wpt/webrtc/webrtc-transform.html.
• web-platform-tests/webrtc-encoded-transform/script-write-twice-transform-worker.js: Renamed from LayoutTests/http/wpt/webrtc/write-twice-transform.js.
• web-platform-tests/webrtc-encoded-transform/script-write-twice-transform.https-expected.txt: Added.
• web-platform-tests/webrtc-encoded-transform/script-write-twice-transform.https.html: Renamed from LayoutTests/http/wpt/webrtc/write-twice-transform.html.
• web-platform-tests/webrtc-encoded-transform/sframe-keys.https-expected.txt: Added.
• web-platform-tests/webrtc-encoded-transform/sframe-keys.https.html: Added.
• web-platform-tests/webrtc-encoded-transform/sframe-transform-buffer-source-expected.txt: Added.
• web-platform-tests/webrtc-encoded-transform/sframe-transform-buffer-source.html: Added.
• web-platform-tests/webrtc-encoded-transform/sframe-transform-expected.txt: Added.
• web-platform-tests/webrtc-encoded-transform/sframe-transform-in-worker.https-expected.txt: Added.
• web-platform-tests/webrtc-encoded-transform/sframe-transform-in-worker.https.html: Renamed from LayoutTests/http/wpt/webrtc/sframe-transform-in-worker.html.
• web-platform-tests/webrtc-encoded-transform/sframe-transform-readable-expected.txt: Added.
• web-platform-tests/webrtc-encoded-transform/sframe-transform-readable.html: Added.
• web-platform-tests/webrtc-encoded-transform/sframe-transform-worker.js: Renamed from LayoutTests/http/wpt/webrtc/sframe-transform.js.
• web-platform-tests/webrtc-encoded-transform/sframe-transform.html: Renamed from LayoutTests/http/wpt/webrtc/sframe-transform.html.

LayoutTests:

• http/wpt/webrtc/sframe-transform-readable-crash.html: Removed.
• platform/glib/TestExpectations:
• platform/ios-wk1/TestExpectations:
• platform/mac-wk1/TestExpectations:
• webrtc/script-transform.js: Removed.

2:03 AM Changeset in webkit [276241] by imanol

• 3 edits in trunk/Source/WebCore

Enable GL_ANGLE_instanced_arrays WebGL extension in WPE
​https://bugs.webkit.org/show_bug.cgi?id=224438

Reviewed by Darin Adler.

Tested by WebGL WPT.

• html/canvas/ANGLEInstancedArrays.cpp:

(WebCore::ANGLEInstancedArrays::supported):

• platform/graphics/opengl/ExtensionsGLOpenGLES.cpp:

(WebCore::ExtensionsGLOpenGLES::supportsExtension):

1:58 AM CSSContainment edited by rwlbuis@webkit.org

(diff)

1:39 AM Changeset in webkit [276240] by commit-queue@webkit.org

• 4 edits in trunk

[css-flexbox] Table layout disregards overriding height
​https://bugs.webkit.org/show_bug.cgi?id=224665

Patch by Felipe Erias <Felipe Erias> on 2021-04-19
Reviewed by Sergio Villar Senin.

Source/WebCore:

Update table layout to take into account the overriding height set by the element's parent.
This was causing several WPT tests to fail.

• rendering/RenderTable.cpp:

(WebCore::RenderTable::layout):
During layout, set the computed height so that it is at least as large as the overriding height
provided by the element's parent (if any) minus the height that will be taken up by captions.
This additional height will be distributed among the table's sections so its total height matches
the overriding value.

LayoutTests:

• TestExpectations: Removed four WPT tests that now pass.

1:23 AM Changeset in webkit [276239] by Martin Robinson

• 4 edits in trunk/Source/WebCore

[css-scroll-snap] Properly support fractional scroll steps in WebCore::ScrollAnimator::scroll
​https://bugs.webkit.org/show_bug.cgi?id=224176

Reviewed by Simon Fraser.

Stop using the ScrollableArea's position to calculate scroll snap positions in ScrollAnimator::snap.
The position stored in ScrollAnimator is a floating point position, while the one stored in ScrollableArea
is an integer position. This currently isn't an issue, because all callers of ScrollAnimator::scroll
use integer scroll offsets, but this will allow this function to be used in the future for precise
scrolling delta.

No new tests. This doesn't change any behavior, since all callers currently use
integer scroll offsets, but a future change will make use of this fix.

• platform/ScrollAnimator.cpp:

(WebCore::ScrollAnimator::scroll): Get the current scroll position from the ScrollAnimator,
which stores it in floating point. Also, only call into the scroll snap code if we actually
have scroll offsets.
(WebCore::ScrollAnimator::offsetFromPosition): Added this helper.
(WebCore::ScrollAnimator::positionFromOffset): Ditto.
(WebCore::ScrollAnimator::deltaFromStep): Return a delta instead of a position so this helper
can be used with offsets or positions.
(WebCore::ScrollAnimator::positionFromStep): Deleted.

• platform/ScrollAnimator.h:
• platform/mac/ScrollAnimatorMac.mm:

(WebCore::ScrollAnimatorMac::scroll): Use the new helper.j

1:16 AM Changeset in webkit [276238] by Philippe Normand

• 11 edits in trunk

[WPE][GTK] Enable AVIF decoder as experimental feature and unskip tests
​https://bugs.webkit.org/show_bug.cgi?id=224663

Reviewed by Xabier Rodriguez-Calvar.

.:

Make the USE_AVIF option public and enable it as experimental feature.

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

Source/WebCore:

Refactor the AVIF build options out of the GTK port so they can be reused by other CMake ports.

• PlatformGTK.cmake:
• platform/ImageDecoders.cmake:
• platform/image-decoders/avif/AVIFImageDecoder.cpp:
• platform/image-decoders/avif/AVIFImageReader.cpp:
• platform/image-decoders/avif/AVIFUniquePtr.h:

LayoutTests:

• platform/gtk/TestExpectations: Unskip avif tests.

12:46 AM BuildingCairoOnWindows edited by Fujii Hironori

(diff)

Apr 18, 2021:

11:57 PM Changeset in webkit [276237] by Manuel Rego Casasnovas

• 5 edits
  1 delete in trunk/LayoutTests

[selectors] Update :focus-visible tests from WPT
​https://bugs.webkit.org/show_bug.cgi?id=224601
<rdar://problem/76827947>

Reviewed by Darin Adler.

LayoutTests/imported/w3c:

Fix expectations that were broken due to a mistake in the previous commit.
The tests had some extra changes due to webkit.org/224658, the tests were modified but not the expectations.

• web-platform-tests/css/selectors/focus-visible-002-expected.txt:
• web-platform-tests/css/selectors/focus-visible-003-expected.txt:
• web-platform-tests/css/selectors/focus-visible-004-expected.txt:

LayoutTests:

Remove glib expectations as they match the main ones.

• platform/glib/imported/w3c/web-platform-tests/css/selectors/focus-visible-002-expected.txt: Removed.
• platform/glib/imported/w3c/web-platform-tests/css/selectors/focus-visible-003-expected.txt: Removed.
• platform/glib/imported/w3c/web-platform-tests/css/selectors/focus-visible-004-expected.txt: Removed.

11:32 PM Changeset in webkit [276236] by Wenson Hsieh

• 4 edits
  2 adds in trunk

Selected image overlay text should never be visible
​https://bugs.webkit.org/show_bug.cgi?id=224734
<rdar://problem/76806399>

Reviewed by Darin Adler.

Source/WebCore:

Make a slight adjustment to the UA stylesheet to ensure that image overlay text is never visible. See below for
more details.

Test: fast/images/image-extraction/image-overlay-with-selection-styles.html

• html/shadow/imageOverlay.css:

(div.image-overlay-text::selection):

Enforce selected text styles for image overlay text.

• rendering/RenderElement.cpp:

(WebCore::RenderElement::selectionPseudoStyle const):

Make an additional adjustment when computing selection pseudo styles for UA shadow root content. Currently, for
elements inside a UA shadow root, we always immediately ascend to the shadow host; this means that ::selection
pseudo selectors currently don't work in UA stylesheets, since they're skipped when resolving styles, upon
painting selected text.

To fix this, we can let the element's own pseudo styles take precedence over the shadow host's renderer in the
case where we have a selection pseudo style. While we're here, also replace a few raw pointers with RefPtr.

LayoutTests:

Add a new layout test.

• fast/images/image-extraction/image-overlay-with-selection-styles-expected.html: Added.
• fast/images/image-extraction/image-overlay-with-selection-styles.html: Added.

11:13 PM Changeset in webkit [276235] by commit-queue@webkit.org

• 17 edits in trunk

Support contain: layout
​https://bugs.webkit.org/show_bug.cgi?id=223569

Patch by Rob Buis <​rbuis@igalia.com> on 2021-04-18
Reviewed by Darin Adler.

Source/WebCore:

This patch implements layout containment as specified[1].
It adds shouldApplyLayoutContainment to check whether the
element applies for layout containment. Is so, then:

• an independent formatting context is established.
• any overflow is treated as ink overflow.
• an absolute positioning and fixed positioning

containing block is established.

• a stacking context is created.
• for baseline handling, the box is treated as having no

baseline. For grid/flexible box containers/items, this needs
a baseline synthesized from the border edges [2], for buttons it
requires a baseline synthesized from the margin edges [2, 3].

[1] ​https://drafts.csswg.org/css-contain-2/#layout-containment
[2] ​https://drafts.csswg.org/css-align-3/#synthesize-baseline
[3] ​https://www.w3.org/TR/CSS2/visudet.html#propdef-vertical-align

• rendering/RenderBlock.cpp:

(WebCore::RenderBlock::firstLineBaseline const):
(WebCore::RenderBlock::inlineBlockBaseline const):

• rendering/RenderBlockFlow.cpp:

(WebCore::RenderBlockFlow::firstLineBaseline const):
(WebCore::RenderBlockFlow::inlineBlockBaseline const):

• rendering/RenderBox.cpp:

(WebCore::RenderBox::createsNewFormattingContext const):
(WebCore::RenderBox::layoutOverflowRectForPropagation const):

• rendering/RenderBox.h:
• rendering/RenderButton.cpp:

(WebCore::RenderButton::baselinePosition const):

• rendering/RenderElement.h:

(WebCore::RenderElement::canContainFixedPositionObjects const):
(WebCore::RenderElement::canContainAbsolutelyPositionedObjects const):

• rendering/RenderFlexibleBox.cpp:

(WebCore::RenderFlexibleBox::firstLineBaseline const):

• rendering/RenderGrid.cpp:

(WebCore::RenderGrid::firstLineBaseline const):

• rendering/RenderListBox.cpp:

(WebCore::RenderListBox::baselinePosition const):

• rendering/RenderObject.cpp:

(WebCore::objectIsRelayoutBoundary):
(WebCore::shouldApplyLayoutContainment):

• rendering/RenderObject.h:

(WebCore::RenderObject::isAtomicInlineLevelBox const):

• rendering/RenderTable.cpp:

(WebCore::RenderTable::firstLineBaseline const):

• rendering/style/RenderStyle.h:

(WebCore::RenderStyle::containsLayout const):

• style/StyleAdjuster.cpp:

(WebCore::Style::Adjuster::adjust const):

LayoutTests:

Unskip tests that pass now.

• TestExpectations:

7:43 PM Changeset in webkit [276234] by Diego Pino Garcia

• 2 edits in trunk/LayoutTests

[GLIB] Unreviewed test gardening. Marked css-transforms/animation/rotate-composition.html as failure after r276231.

• platform/glib/TestExpectations:

6:56 PM Changeset in webkit [276233] by Fujii Hironori

• 2 edits
  3 adds in trunk/LayoutTests

[WinCairo] Unreviewed test gardening

• platform/wincairo-wk1/http/tests/xmlhttprequest/redirect-cross-origin-post-expected.txt: Added.
• platform/wincairo-wk1/http/tests/xmlhttprequest/simple-cross-origin-denied-events-post-expected.txt: Added.
• platform/wincairo/TestExpectations:

2:14 PM Changeset in webkit [276232] by weinig@apple.com

• 5 edits in trunk/Source/WebCore

GraphicsLayer::distributeOpacity, GraphicsLayer::setOpacityInternal and GraphicsLayer::accumulatedOpacity are never used
​https://bugs.webkit.org/show_bug.cgi?id=224732

Reviewed by Darin Adler.

Remove dead code.

• platform/graphics/GraphicsLayer.cpp:

(WebCore::GraphicsLayer::accumulatedOpacity const): Deleted.
(WebCore::GraphicsLayer::distributeOpacity): Deleted.

• platform/graphics/GraphicsLayer.h:

(WebCore::GraphicsLayer::setOpacityInternal): Deleted.

• platform/graphics/ca/GraphicsLayerCA.cpp:
• platform/graphics/ca/GraphicsLayerCA.h:

11:12 AM Changeset in webkit [276231] by graouts@webkit.org

• 9 edits in trunk

Fix interpolation of the rotate CSS property
​https://bugs.webkit.org/show_bug.cgi?id=224730

Reviewed by Dean Jackson.

LayoutTests/imported/w3c:

Mark 123 WPT progressions.

• web-platform-tests/css/css-transforms/animation/rotate-composition-expected.txt:
• web-platform-tests/css/css-transforms/animation/rotate-interpolation-expected.txt:
• web-platform-tests/css/css-transforms/animation/transform-interpolation-001-expected.txt:
• web-platform-tests/web-animations/animation-model/animation-types/accumulation-per-property-002-expected.txt:
• web-platform-tests/web-animations/animation-model/animation-types/addition-per-property-002-expected.txt:
• web-platform-tests/web-animations/animation-model/animation-types/interpolation-per-property-002-expected.txt:

Source/WebCore:

Implement the CSS Transforms spec about animating the rotateZ() function
for the simplified case where only the angle is interpolated.

• platform/graphics/transforms/RotateTransformOperation.cpp:

(WebCore::RotateTransformOperation::blend):

9:42 AM Changeset in webkit [276230] by youenn@apple.com

• 12 edits in trunk

Blob URLs should use for their owner origin for CSP checks
​https://bugs.webkit.org/show_bug.cgi?id=224535
<rdar://76458106>

Reviewed by Alex Christensen.

Source/WebCore:

Before the patch, we were checking blob origin directly with ancestors.
As per ​https://w3c.github.io/webappsec-csp/#match-url-to-source-expression step 4.1,
we need to get the URL origin, which by spec is the origin of the blob creator.
We only do this for navigation loads as script loads should be kept the current way, as a cross-site scripting protection,
and to remain compatible with other browsers.

Make some refactoring to add helper routines to get origin and secure context state of blob URLs in BlobURL.
Make use of it in MixedContentChecker as a refactoring.
Make use of the helper routine in ContentSecurityPolicySource::matches to fix the bug.

Test: http/tests/security/frame-src-and-blob-download.https.html

• fileapi/BlobURL.cpp:

(WebCore::blobOwner):
(WebCore::BlobURL::getOriginURL):
(WebCore::BlobURL::isSecureBlobURL):

• fileapi/BlobURL.h:
• fileapi/ThreadableBlobRegistry.cpp:

(WebCore::isBlobURLContainsNullOrigin):

• loader/MixedContentChecker.cpp:

(WebCore::MixedContentChecker::isMixedContent):

• page/SecurityOrigin.cpp:

(WebCore::SecurityOrigin::isSecure):

• page/csp/ContentSecurityPolicy.cpp:

(WebCore::ContentSecurityPolicy::urlMatchesSelf const):

• page/csp/ContentSecurityPolicy.h:
• page/csp/ContentSecurityPolicySourceList.cpp:

(WebCore::ContentSecurityPolicySourceList::matches const):

LayoutTests:

• http/tests/security/frame-src-and-blob-download.https-expected.txt: Added.
• http/tests/security/frame-src-and-blob-download.https.html:
• http/tests/security/resources/frame-src-and-blob-download-frame.html: Added.
• platform/mac-wk1/TestExpectations:
• platform/win/TestExpectations:

6:40 AM Changeset in webkit [276229] by Alan Bujtas

• 2 edits in trunk/Tools

Update name in contributors.json
​https://bugs.webkit.org/show_bug.cgi?id=224731

Unreviewed.

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

5:45 AM Changeset in webkit [276228] by cathiechen

• 4 edits in trunk

The implicit aspect-ratio from width and height attributes with float value is not accurate enough
​https://bugs.webkit.org/show_bug.cgi?id=224664

Reviewed by Darin Adler.

LayoutTests/imported/w3c:

The aspect ratio test case with width "0.8" and height "0.2" in img-aspect-ratio.html has passed. This patch doesn't change the behavior of
the original aspect ratio test case(assert_ratio(images[5], 133/106)) which is related to bug 206161.

• web-platform-tests/html/rendering/replaced-elements/attributes-for-embedded-content-and-images/img-aspect-ratio-expected.txt:

Source/WebCore:

The width and height attributes should impact the aspect ratio only not the intrinsic size which should be from the content.
Since computeAspectRatioInformationForRenderBox doesn't change the intrinsic size now, so we can remove it from computePreferredLogicalWidths.

• rendering/RenderReplaced.cpp:

(WebCore::RenderReplaced::computeIntrinsicRatioInformation const): Width and height attributes change intrinsicRatio only, not intrinsicSize.
(WebCore::RenderReplaced::computePreferredLogicalWidths): computeAspectRatioInformationForRenderBox doesn't change intrinsicSize now, so we can remove this.

1:35 AM Changeset in webkit [276227] by ysuzuki@apple.com

• 2 edits in trunk/Source/JavaScriptCore

Unreviewed, build fix
​https://bugs.webkit.org/show_bug.cgi?id=224715

• dfg/DFGDesiredWatchpoints.h:

1:06 AM Changeset in webkit [276226] by ysuzuki@apple.com

• 18 edits in trunk/Source/JavaScriptCore

[JSC] Do not use Bag<> for DFG / FTL watchpoints
​https://bugs.webkit.org/show_bug.cgi?id=224715

Reviewed by Darin Adler.

While Bag<> is useful since its allocated memory will not be moved,
this is really memory-inefficient data structure. Each entry gets a
tail pointer (so adding 8 bytes) and we allocate each entry separately.

In DFG and FTL, we are using Bag<> for watchpoints. But this is not necessary actually: thanks to
concurrent compilers, our watchpoint registration is batched at the end of compilation. This means
that we have a way to know how many watchpoints we should register at that point.

In this patch, we introduce WatchpointCollector. In DesiredGlobalProperties, we run reallyAdd twice
with WatchpointCollector. First time, we just count # of watchpoints. Then we allocate FixedVector<XXXWatchpoint>
and install them. Since we do not (cannot) grow this fixed vector, watchpoint's address will not be changed as required.

We also move DesiredGlobalProperties under DesiredWatchpoints since this basically registers watchpoints.

• bytecode/AdaptiveInferredPropertyValueWatchpointBase.cpp:

(JSC::AdaptiveInferredPropertyValueWatchpointBase::AdaptiveInferredPropertyValueWatchpointBase):
(JSC::AdaptiveInferredPropertyValueWatchpointBase::initialize):

• bytecode/AdaptiveInferredPropertyValueWatchpointBase.h:
• bytecode/CodeBlockJettisoningWatchpoint.h:
• dfg/DFGAdaptiveInferredPropertyValueWatchpoint.cpp:

(JSC::DFG::AdaptiveInferredPropertyValueWatchpoint::AdaptiveInferredPropertyValueWatchpoint):
(JSC::DFG::AdaptiveInferredPropertyValueWatchpoint::initialize):

• dfg/DFGAdaptiveInferredPropertyValueWatchpoint.h:
• dfg/DFGAdaptiveStructureWatchpoint.cpp:

(JSC::DFG::AdaptiveStructureWatchpoint::AdaptiveStructureWatchpoint):
(JSC::DFG::AdaptiveStructureWatchpoint::initialize):

• dfg/DFGAdaptiveStructureWatchpoint.h:
• dfg/DFGCommonData.cpp:

(JSC::DFG::CommonData::validateReferences):
(JSC::DFG::CommonData::clearWatchpoints):

• dfg/DFGCommonData.h:
• dfg/DFGDesiredGlobalProperties.cpp:

(JSC::DFG::DesiredGlobalProperties::reallyAdd):

• dfg/DFGDesiredGlobalProperties.h:
• dfg/DFGDesiredWatchpoints.cpp:

(JSC::DFG::ArrayBufferViewWatchpointAdaptor::add):
(JSC::DFG::SymbolTableAdaptor::add):
(JSC::DFG::FunctionExecutableAdaptor::add):
(JSC::DFG::AdaptiveStructureWatchpointAdaptor::add):
(JSC::DFG::DesiredWatchpoints::addLazily):
(JSC::DFG::DesiredWatchpoints::reallyAdd):
(JSC::DFG::DesiredWatchpoints::areStillValidOnMainThread):
(JSC::DFG::WatchpointCollector::finalize):

• dfg/DFGDesiredWatchpoints.h:

(JSC::DFG::SetPointerAdaptor::add):
(JSC::DFG::GenericDesiredWatchpoints::reallyAdd):

• dfg/DFGGraph.cpp:

(JSC::DFG::Graph::watchGlobalProperty):

• dfg/DFGGraph.h:
• dfg/DFGPlan.cpp:

(JSC::DFG::Plan::reallyAdd):
(JSC::DFG::Plan::isStillValidOnMainThread):
(JSC::DFG::Plan::cancel):

• dfg/DFGPlan.h:

(JSC::DFG::Plan::transitions):
(JSC::DFG::Plan::globalProperties): Deleted.

12:43 AM Changeset in webkit [276225] by ysuzuki@apple.com

• 2 edits in trunk/Source/WebCore

Use binary-search in LocaleToScriptMapping
​https://bugs.webkit.org/show_bug.cgi?id=224727

Reviewed by Darin Adler.

This patch removes HashMaps in LocaleToScriptMapping, and binary-search onto the constant data arrays.
These maps are not frequently used. Keys of the maps can be encoded into uint32_t or uint64_t so that
comparison becomes super cheap and we can initialize this array at compile-time.

We introduce ScriptName(uint32_t) and LocaleName(uint64_t) instead of String. And ues it and generate
sorted constant data array for mappings. We use binary-search to look entry up. Since # of entries are
not huge (~200), comparisons are extremely cheap (uint32_t / uint64_t comparison), and this is not a
hot code, we can just use binary-search here and eliminate HashMaps' memory allocation.

• platform/text/LocaleToScriptMapping.cpp:

(WebCore::PackedASCIILowerCodes::PackedASCIILowerCodes):
(WebCore::PackedASCIILowerCodes::parse):
(WebCore::PackedASCIILowerCodes::operator==):
(WebCore::PackedASCIILowerCodes::operator!=):
(WebCore::PackedASCIILowerCodes::operator<):
(WebCore::PackedASCIILowerCodes::operator<=):
(WebCore::PackedASCIILowerCodes::operator>):
(WebCore::PackedASCIILowerCodes::operator>=):
(WebCore::PackedASCIILowerCodes::value const):
(WebCore::scriptNameToCode):
(WebCore::localeToScriptCodeForFontSelection):

12:14 AM Changeset in webkit [276224] by ysuzuki@apple.com

• 18 edits in trunk/Source/JavaScriptCore

[JSC] Make more DFG/FTL data FixedVector/Vector
​https://bugs.webkit.org/show_bug.cgi?id=224713

Reviewed by Darin Adler.

1. DFG::JITCode::m_osrEntry / DFG::JITCode::m_osrExit / DFG::JITCode::m_speculationRecovery are changed to FixedVector. They are added at compiling time, and after that, these vectors are not modified. So when finalizing, we can easily make it FixedVector. We also change OSREntry::{m_reshufflings,m_expectedValues} to FixedVector and FixedOperands.
2. FTL::JITCode::m_osrExit is changed from SegmentedVector to Vector. We are still using Vector since it also involves osrExitDescriptor. But later, we should merge m_osrExit to osrExitDescriptor. Vector is still better than SegmentedVector since it wastes several entries per segment. SegmentedVector was used to use a direct pointer of OSRExit (this is not possible in Vector since this pointer can be invalidated after growing), but usage of that is fairly limited so that we can just replace them with m_index + osrExit vector.

• bytecode/CodeBlock.cpp:

(JSC::CodeBlock::tallyFrequentExitSites):

• bytecode/Operands.h:

(JSC::Operands::Operands):

• dfg/DFGJITCode.cpp:

(JSC::DFG::JITCode::shrinkToFit):
(JSC::DFG::JITCode::liveRegistersToPreserveAtExceptionHandlingCallSite):
(JSC::DFG::JITCode::validateReferences):
(JSC::DFG::JITCode::findPC):
(JSC::DFG::JITCode::finalizeOSREntrypoints):

• dfg/DFGJITCode.h:
• dfg/DFGJITCompiler.cpp:

(JSC::DFG::JITCompiler::linkOSRExits):
(JSC::DFG::JITCompiler::link):
(JSC::DFG::JITCompiler::noticeOSREntry):
(JSC::DFG::JITCompiler::appendExceptionHandlingOSRExit):

• dfg/DFGJITCompiler.h:

(JSC::DFG::JITCompiler::appendOSRExit):
(JSC::DFG::JITCompiler::appendSpeculationRecovery):

• dfg/DFGOSREntry.h:
• dfg/DFGOSRExit.cpp:

(JSC::DFG::JSC_DEFINE_JIT_OPERATION):

• dfg/DFGSpeculativeJIT.cpp:

(JSC::DFG::SpeculativeJIT::speculationCheck):
(JSC::DFG::SpeculativeJIT::emitInvalidationPoint):
(JSC::DFG::SpeculativeJIT::linkOSREntries):

• ftl/FTLJITCode.cpp:

(JSC::FTL::JITCode::shrinkToFit):
(JSC::FTL::JITCode::validateReferences):
(JSC::FTL::JITCode::liveRegistersToPreserveAtExceptionHandlingCallSite):
(JSC::FTL::JITCode::findPC):

• ftl/FTLJITCode.h:
• ftl/FTLOSRExit.cpp:

(JSC::FTL::OSRExitDescriptor::prepareOSRExitHandle):
(JSC::FTL::OSRExit::OSRExit):

• ftl/FTLOSRExit.h:
• ftl/FTLOSRExitCompiler.cpp:

(JSC::FTL::JSC_DEFINE_JIT_OPERATION):

• ftl/FTLOSRExitHandle.cpp:

(JSC::FTL::OSRExitHandle::emitExitThunk):

• ftl/FTLOSRExitHandle.h:

(JSC::FTL::OSRExitHandle::OSRExitHandle):

• ftl/FTLPatchpointExceptionHandle.cpp:

(JSC::FTL::PatchpointExceptionHandle::scheduleExitCreationForUnwind):

12:10 AM Changeset in webkit [276223] by Chris Dumez

• 3 edits in trunk/Source/WebKit

Update LibWebRTCCodecsProxy to use a Lock
​https://bugs.webkit.org/show_bug.cgi?id=224728

Reviewed by Darin Adler.

Update LibWebRTCCodecsProxy to use a Lock, instead of a std::atomic<bool> that
has to be kept up to date. I think this simplifies the code a bit. Adding / Removing
encoder / decoder is not very hot code as far as I know and there will very rarely
be contention since allowsExitUnderMemoryPressure() is only called on memory pressure.

m_encoder / m_decoder are still always modified from the background thread. However, we
now check from the main thread if they are empty by locking.

• GPUProcess/webrtc/LibWebRTCCodecsProxy.h:
• GPUProcess/webrtc/LibWebRTCCodecsProxy.mm:

(WebKit::LibWebRTCCodecsProxy::close):
(WebKit::LibWebRTCCodecsProxy::createH264Decoder):
(WebKit::LibWebRTCCodecsProxy::createH265Decoder):
(WebKit::LibWebRTCCodecsProxy::createVP9Decoder):
(WebKit::LibWebRTCCodecsProxy::releaseDecoder):
(WebKit::LibWebRTCCodecsProxy::decodeFrame):
(WebKit::LibWebRTCCodecsProxy::setFrameSize):
(WebKit::LibWebRTCCodecsProxy::createEncoder):
(WebKit::LibWebRTCCodecsProxy::releaseEncoder):
(WebKit::LibWebRTCCodecsProxy::initializeEncoder):
(WebKit::LibWebRTCCodecsProxy::encodeFrame):
(WebKit::LibWebRTCCodecsProxy::setEncodeRates):
(WebKit::LibWebRTCCodecsProxy::allowsExitUnderMemoryPressure const):
(WebKit::LibWebRTCCodecsProxy::updateHasEncodersOrDecoders): Deleted.